ZPA-230-04-39-049
         RADON RISKS: ATTITUDES, PERCEPTIONS AND ACTIONS
                         Lennart Sjoberg

                     Center  for  Risk  Research
                  Stockholm School of Economics
                           August 1989
                           Prepared  for

                          Dr. Ann  Fisher
                   Economics Analysis Division
            Office of Policy, Planning and Evaluation
              U.S. Environmental  Protection Agency
                      Washington, DC  20460
The information in this document has been funded in part by the
United States Environmental Protection Agency  (EPA) under
Cooperative Agreement CR-811075.  It has been subjected to the
Agency's peer and administrative review, and approved for
publication as an EPA document.  Mention of trade names or
commercial products does not constitute endorsement or
recommendation for use.

-------
                             Preface

Indoor radon constitutes a major health risk. As many as 8 million
homes in the USA may have elevated radon levels with accompanying
lung cancer  risks several orders  of magnitude above  many other
environmental hazards. Still,  less than 5 percent of all homes have
been tested  for  radon and  interest  in  the topic seems  to  be
tapering  off,   after  a  rise  following  the  widely  publicized
discovery of extreme radon levels in some homes in early 1985.

     The  U.S.  Environmental  Protection Agency  has sponsored  a
program of research  on radon  risk  communication aimed  at finding
effective  methods for informing  people  about  radon  risks  and
encouraging them to monitor their homes and mitigate if high radon
levels are found.  State level agencies also have sponsored research
on this topic.  Several studies have  now been  completed.  Others,
while still  not'finished,  have produced some; significant interim
results.  This  report  summarizes  this research,  discusses  it
critically and  suggests conclusions  relevant to  radon  policy  as
well as  topics for  further  research. Some  studies performed  in
Sweden also have been  treated.

     I have  written  this  report mostly from  the perspective of a
psychologist, and I have attempted to relate some of the radon risk
communication research to  general  discussions  of  attitude change
and  attitude   measurement.   Some   practical   implications  are
summarized in the appendix.

     Ann  Fisher and  Reed  Johnson have  given  very  valuable  and
detailed  comments on  the  manuscript.  Amos  Tversky made  several
stimulating remarks.


Stockholm, Sweden


August, 1989


Lennart Sjoberg

-------
                         Table of contents
                                                              Page

 Preface 	    i

 List of Figures	v

 List of Tables	vi

 1.  Executive summary  	    1
      1.1.  Introduction  	    1
      1.2.  The media and radon risk	    1
      1.3.  Properties of radon risk	    l
      1.4.  Radon risk compared to life style risks	    2
      1.5.  Studies of risk attitudes and risk perception ...    2
      1.6.  Mitigation of radon risk  .....  	    5
      1.7.  Message format and contents 	    5
      1.8.  Methodology	    6
      1.9.  Social diffusion  	    8
      1.10.  House values and house sales 	    3
      1.11.  Missing information:  groups not studied  	    9
      1.12.  Conclusions  	    9

 2.  Introduction	11

m.  Information,  the media and radon	15

 4.  Characteristics of radon risks 	  17

 5.  Empirical studies of radon risk perception and attitudes  .  20
      5.1.  Introduction	20
      5.2.  Maine study	20
           5.2.1. Design and selected results  .	20
           5.2.2. Evaluation 	  21
      5.3.  Focus groups	...	.21
           5.3.1. Design and selected results	  21
           5.3.2. Evaluation 	  23
      5.4.  NYSERDA Study	23
           5.4.1. Design and selected results  	  23
           5.4.2. Evaluation 	  27
      5.5.  New Jersey study	28
           5.5.1. Design and selected results  	  28
           5.5.2. Evaluation 	  30
      5.6.  New Jersey experiment  on radon information  ....  31
           5.6.1. Design and selected results  	  31
           5.6.2. Evaluation 	  32
      5.7.  Study of effectiveness of communication formats .  .  32
           5.7.1. Design and selected results  	  32
           5.7.2. Evaluation 	  34
      5.8.  Maryland study	35
           5.8.1. Design and selected results  	  35
           5.8.2. Evaluation 	  36

                                 ii

-------
                                                               1
     5.9. Florida risk attitudes study  	  37
          5.9.1. Design and selected results  	  37
          5.9.2. Evaluation 	  33
     5.10. Florida personality study  	  33
          5.10.1. Design and selected results 	  38
          5.10.2. Evaluation  	  40
     5.11. Onondaga study	40
          5.11.1. Design and selected results 	  40
          5.11.2. Evaluation  	  40

6. Empirical studies of mitigation  	  41
     6.1. Introduction to mitigation research 	  41
     6.2. Maine study	41
     6.3. New Jersey study	41
     6.4. Washington study  	  43
     6.5. Colorado study of  testing  and mitigation at time of
          home sale	45
     6.6. Some Swedish experience: the Sollentuna study ...  46
     6.7. Evaluation and conclusions of mitigation studies  .  49

7. Comments on empirical studies of  radon risk perception and
     mitigation	51
8.  Predicting monitoring for radon
9. Factors affecting risk perception	54
     9.1. Risk denial	54
     9.2. Diffusion processes 	  56
     9.3. Assimilation-contrast model of risk perception  .  .  57
     9.4. Risk posterior to action	57
     9.5. Objective vs subjective risk	58
     9.6. Demographic characteristics .  ;	59
     9.7. Cognitive processing limitations  	  .  59
     9.8. A note on terminology: Hazard and outrage 	  59

10. Risk perception and protective action .....-.:....  60

11. Message format and contents 	  62

12. Personality, risk taking, and attitude change 	  63
     12.1. Personality and behavior: general  	  63
     12.2. Vigilant vs. defensive response to threats ....  64
     12.3. Need for cognition and uncertainty orientation .  .  64
     12.4. Health concern	66
     12.5. Emotions and risk perception . . .	66
13. Methodological issues 	  67
     13.1. The definition of risk 	
     13.2. Response rates 	 	
     13.3. Response scales  .  . .	69

                               iii
o /
n

-------
                                                             Page

     13.4. Time perspective	70
     13.5. Attitude measurement	 .   70
     13.6.   More   detailed   information   about   crucial
          decisions	71

14. Discussion and conclusions  	   71
     14.1. Summary and discussion	71
     14.2. Conclusions	80

15. References	 .   32

APPENDIX	93
                                IV

-------
                         List of Ficrures
                                                             Page

Figure 1. Number of published articles about radon  in the New
     York Times. 1980-87	17

Figure  2.  Frequency  of  mitigation as  a function  of radon
     concentration	49

-------
                          List of Tables

                                                             Page"

Table   I.   Distribution   of   radon  exposure   in  various
     countries.	12

Table'II. Reasons  for  not testing for radon (from Weinstein
     et al)	29

Table  III.  Percentage  of persons  who  had mitigated  as a
     function of their  first  floor radon reading. Data from
     Weinstein, Sandman and Roberts  (1988)	41

Table  IV.  Percentage  of  persons  who  had  mitigated  as a
     function of their radon  reading. Data  from Doyle et  al.
     (1989)	  44

Table V. Percentage of persons who had mitigated as a function
     of  the   initial   radon   reading.   Data  from  Akerman
     (1988)	48

Table VI. Response rates  in cited radon studies, measured as
     the percent of  the  studied sample that (a) accepted to
     participate, and  (b) gave usable responses	
                                vi

-------
1. Executive summary
1.1. Introduction

Indoor radon constitutes a major health risk. As many as 8 million
homes in the USA may have elevated radon levels with accompanying
cancer  risks  several  orders  of  magnitude  above  most  other
environmental hazards.  Nevertheless,  less than 5  percent  of all
homes have been tested  for  radon.  Radon in private homes remains
the responsibility of the home owners and there is no legislation
that compels people to reduce this particular risk.

     The  U.S.  Environmental  Protection Agency  has sponsored  a
program  of  research  aimed  at  finding  effective  methods  for
informing people about radon risks and  encouraging  them to monitor
their homes  and mitigate if  high radon levels are  found.  State
agencies also have sponsored  research  on  this  topic.  This report
summarizes  the  research,  discusses it  critically and  suggests
conclusions as well as topics for further research.

     Some  studies  performed  in Sweden are included because of
interesting  differences  between that  country  and the  USA in
reactions to radon risk.  Also,  radon has  been  an issue longer in
Sweden than  in  the USA,  especially radon  emitted from building
material.
1.2. The media and radon risk

Radon was not widely discussed as a health risk in the USA before
1985. That  year  national attention was given  to  a very dramatic
case in the home  of  a  Pennsylvania resident.  The radon level was
much beyond what had previously been believed to be possible from
geological  radon  in  a home. Subsequent screening  has shown that
there is definitely  a  radon problem in many homes, although very
few have had levels as  high as the initial  Pennsylvania case. Radon
was  highly  salient  in the  media  in  1985 but according  to some
authors interest is now tapering off  (see Figure 1, page 17).

1.3. Properties of radon risk

Radon is a risk with certain unique properties. It shares some of
the  characteristics  of all  risks from ionizing  radiation (slow
effects in  the form  of cancer,  a risk from a physical agent that
cannot be sensed). At  the same  time it is a risk that people are
exposed to  in their homes.  Homes are usually perceived  as safe
places  under one's  own control  and  responsibility.  It  is  not
obvious that some person or organization is to  be blamed for radon
risks, in contrast to  such risks as nuclear power plants.

-------
1.4. Radon risk compared to life style risks

Radon risk has been  compared  to  such life-style risks as smoking
or drinking alcohol.  The similarity is that people take these risks
as private individuals. Society is reluctant to  legislate the risk
level  to  which  they  can  expose  themselves.  They  must  act
voluntarily to mitigate or eliminate these risks.

     There are also differences between radon and life-style risks.
First, radon exposure  does not confer  any benefit as smoking and
alcohol do. Second,  it is  not  clear who/ if anyone, is responsible
for radon risk. Radon risks were largely unknown to the us public
before 1985 and most home owners bought their current houses before
then.

     Reducing a  life-style  risk involves making  a  commitment to
action in  spite  of temptations  to  delay action. In  this sense,
there  is  a similarity between  radon and life-style  risks.  Many
people   in   one   of   the   studies   reviewed   here   became
"procrastinators", i.e. they changed  from  indifference towards the
risk to acknowledging  the necessity  of action "in principle". It
is well known that people find  it very difficult to quit a risky
habit and that rationalized delays are very common. It is too early
to tell if the same phenomenon will occur  for  radon risk but it
possible.  On  the other  hand, testing  for radon  and mitigati
require only two  discrete actions while the ex-smoker has to resi
cigarettes several times a day.  People  do  not  have to  quit a habit
(or develop a new one) in order to test and mitigate  for radon.
I
1.5. Studies of risk attitudes and risk perception

There have been several studies of how people perceive and respond
to  radon  risk.  Initially it was  believed  that information about
radon risk might give rise to panic reactions,  but  experience has
not confirmed this concern. On the contrary, people are often quite
indifferent to the  issue,  at least  in  regard to their own homes.
This indifference can be interpreted as a manifestation of a rather
general tendency to deny health risks.  It has been suggested that
denial of radon risk is moderately strong compared to other kinds
of denial*

     Many  people   are  outright  negative  to  radon   testing.
Pennsylvania authorities at  one point went from door to door and
offered free radon testing. Still, about 50 percent refused to have
their homes monitored.  Some  of them  may have been concerned that
rumors of a radon problem would affect property values in an area.

     It is worth  stressing that people  tend to be indifferent
radon risk in spite of the fact that  the EPA action level is
high compared to other risks. The actual risk  level seems to be a

-------
poor predictor  of public response  to  risk.  The  action  level in
Sweden   is   presently   2.5   times  higher  than   the  US  level
corresponding  to  the  lung  cancer  risk of  smoking  a   pack of
cigarettes per day. Still, Swedes are as  indifferent to radon  risk
as Americans.

     As  might be expected on the basis  of these  facts,  there is
usually  only  a  weak  correlation  between  the  scientifically
estimated risk  level,  as physically measured,  and the  level of
perceived risk.  Some people with low levels of  radon in their homes
are quite  concerned while  others,  with quite  high  levels,  are
indifferent. This  low correlation may  partly  arise from the  fact
that the physical  measurements are  not perfect  indicators of the
actual risk  level.  The risk  also depends on how much time people
spend  in the house,  their   age  and other  factors.  Still  it is
reasonable to conclude  that there is  a  weak  correlation between
perceived  and  actual  radon  risk.  There  is  little  evidence
suggesting that people do not believe that the radon readings are
accurate or that the risk assessments are incorrect. They probably
believe  that the risk is there,  in principle and for  people in
general, but that they  themselves  are  for various  reasons  less
vulnerable,   more lucky  or  they  just  do not  care about getting
involved in protecting themselves against another risk, they  just
take it.

     However, the effects of information about radon risk provide
a partly different picture. Two  studies have shown  that people
revise their risk estimates  in the "right" direction, i.e. in the
direction suggested by the readings of radon levels in  their homes.
Still,  other data  suggest that even for risk  revisions there may
be irrational denial factors  at work. There is  some support for the
thesis that people are more  inclined to accept that other people's
homes in their community may be at risk than that their own homes
may be at risk.

     If actual risk is not strongly correlated  with perceived risk,
then what accounts for perceived  risk?  There are some correlations
with background  data. Older  people  are less worried  about radon,
and parents.of small children more so.  In some studies women have
been shown to be more worried than men, but the gender difference
is much smaller  for radon risk than for many other types of risk.
People with a family member with cancer were more concerned about
radon in one study. Some  personality variables have been suggested
as predictors of perceived  radon risk  but so  far little research
has been carried out on this issue.

     These factors are only weakly related to perceived radon risk.
There  is simply  not enough research  available to  support more
definite statements  about what  factors influence perceived radon
risk.

-------
     I suggest that radon risk perception is a special case of e
related risk,  i.e.  it is a risk that  is  closely related to s
conceptions. This hypothesis is derived from evidence that peop
tend to perceive  their homes as extended  parts  of their selves,
especially if  they  have  lived  in  them for a long time.  And there
is a tendency to deny that something as closely related to oneself
as one's home  (or body) could be threatening.

     Data indicate that people with a  longer history of living in
a house are more likely to deny radon  risk. Other supporting data
show that people react strongly to radon  risk when it is imposed
on their neighborhood  - not on their homes. Part of this reaction
is probably a  result of moral indignation over being exposed to a
risk by someone. People strongly resent a loss of  control, and they
tend to perceive that which  they  can control and are responsible
for as benign.

     Communication  about  radon risk with  the  general  public has
been  studied  in  two major  investigations,  one  of  which  was
especially   concerned  with   mitigation.   These   studies   also
investigated the effects  of a major attempt by a television channel
in the Washington,  D.C.,  area to encourage people  to buy test kits,
available at  reduced price in grocery stores. Over 100,000 kits
were quickly  sold  and demand was by  no means exhausted.  About 6
percent of the home owners in the area bought test kits. Howevej
only about half of those who bought the  kits turned them in
analysis and  few people  who had  elevated radon  levels  in the!
homes  reported any  effective  mitigation  beyond  such behavioral
measures as keeping windows open more  often.

     The EPA-sponsored outreach activities in a Maryland community
were  probably more  effective  than the  simultaneous  television
campaign in stimulating  people to  test for radon. The proportion
that tested their homes rose from about 5 to 15 percent. There is
some evidence that the television  campaign was effective in making
people aware of radon risk, but less effective in stimulating them
to act. A similar  finding was reported in a New Jersey study of
smaller scope where it was found that there is a link between risk
perception and action  but that the link  was quite weak.  The weak
effects of  the television campaign are  in  line  with  extensive
experience  from other studies  of  health-related  public service
message*. The effects are usually quite  marginal.  In  addition,
people tend to distrust television.

     A  few  studies have  investigated knowledge about  radon and
found that people are fairly  well  aware of its general properties,
although they  lack specific knowledge. Some misunderstandings are
probably rather common.  People may also  lack knowledge about how
to test for  radon  and whom to ask about  testing and mitigation.
They are especially uncertain about mitigation and its costs.

-------
     It  is  relatively easy  to  inform people about  radon facts.
However, there  is  little  relation between knowledge and tendency
to test  and mitigate. This  finding  is in line  with a generally
noted  lack  of  relationship  between  attitudes  or beliefs  and
knowledge.


1.6.  Mitigation of radon risk

There is a rather  low correlation  between scientifically estimated
risk and  mitigation,  but a  strong correlation  between perceived
risk and mitigation. This is an example of a quite general truth:
people react to what they perceive reality to be rather than what
it really is.

     It  is  not  yet clear how  much mitigation one  can expect in
different  circumstances.  Some  data  (see  Figure   2,  page  50)
certainly suggest that home-buyers are likely to  be among the most
responsive to the  radon  risk message.  On  the other hand,  not all
data on  mitigation are as  negative  as the  ones  reported in the
study of the Washington,  0.  C.,  television campaign. A follow-up
of highly  motivated New Jersey  homeowners found a  high  rate of
mitigation. Swedish  data were in between the two  American data
sets. (Cultural comparisons are risky,  however, since many crucial
aspects, such as  the  economics of housing and mitigation, differ
dramatically between Sweden and the USA).


1.7.  Message format and contents

A few studies have investigated  the  format and contents of radon
risk messages. A major study compared "command" and "cajole" styles
of  communication.  That  study  also  investigated   and  compared
qualitative vs quantitative types of  information.  The results were
somewhat complex  and differences were not large  but there was a
tendency for a command style of communication to be most effective.
That study also  compared the use of  a  Jorief  fact  sheet  as  a
replacement for more extensive brochures.  It was  found that those
who had been given the fact sheet tended to be more concerned about
their risJc level  than others. This was true in spite of the fact
that the fact sheet was only given to those who had very low levels
of radon in their  homes.

     The investigators concluded  against the use of the  fact sheet.
That may be a reasonable conclusion  for that particular group of
homes and their owners.  It may still  be the case  that a short fact
sheet would be  useful for people who have higher levels of radon
measured in their homes.  It has  been found that people,  if given
the  chance  by  rich  information material,  tend   to construct
defensive  denial   conceptions. In other words,  more  information
makes it easier for them to come  up with rationalizations  for lack

-------
of action. Of course, a short fact sheet need not be very
to the one used in the NYSERDA study.

     An experimental study of risk communication formats found that
a brief  statement  about the action  level  was  best for eliciting
continued concern. On the other  hand,  full information about the
risfc, including numerical,  comparison  and graphical information,
was  best for creating  a balanced view.  A major  difficulty was
communicating absolute  risk levels.  A graphical  display seemed
sufficient to make people disregard a difference in absolute risk
level in the range 1-25 (between radon and asbestos risks).

     Another reason for  favoring very brief messages is  that people
have difficulty in handling information and that they are not very
motivated to obtain  it.  Studies  of public service information in
the form of printed brochures usually have shown that most people
do not bother to read them and that those who do read them quickly
forget most of their contents. Other investigators  have  studied the
effects  of level  of risk communicated  and  found  that a  more
threatening  message  tended  to   invite more  risk  denial.  The
conclusion is that the most effective message format and contents
should be brief and not too threatening.

     The conclusion should be regarded  as preliminary.  It is based
on a rather complex set of findings from -radon risk communicatio
studies  and from  communications  research in general,  as well
basic principles of cognitive psychology.


1.8. Methodology

The empirical studies reviewed in the report vary considerably by
methodology. Virtually  all- were  quantitative  and  employed either
mail or  telephone  survey  methods but they differed in many other
respects.

     There  is  evidence  that  people  are  not  very  eager  to
participate in  surveys of this  kind.  Response rates  around 60
percent or lower are common in spite  of  several reminders. Some of
the problems of  low response rates have been  attributed to data
collection during  holidays,  but  there  is  no  evidence  that these
particular studies yielded especially low response rates. Indeed,
the  US  public  is  saturated  with  opinion  polls  and  survey
investigations  and it  is becoming  more  and  more difficult to
recruit respondents for new waves of data collection.

     Women  seem  to be more  likely  to  accept   invitations  to
participate in radon risk studies than  men, and the result is that
females  are  over-represented  in some  of  the  studies. Also, the
samples tend to consist of highly educated and financially securd
people,  possibly because  the studies have focused on home owner^
rather than renters.

-------
     Attitude change has  been  studied  for decades.  It is usually
quite hard to change people's attitudes,  especially to change them
permanently and to affect  behavior. Changes tend to be short lived.
The empirical work on radon so far has not followed attitudes for
a more than a few months,  with  the exception of the NYSERDA study.
Longer follow-up times are called for in order to get a realistic
idea about how persistent attitude changes are.

     It would probably be fruitful  to  investigate in more detail
how people reason with regard  to radon  and testing.  One piece of
information that  is  lacking in these studies  is  why  people test
(there is data on why people do not test). Testing  can be motivated
in many ways,  for example health,  economic factors  or conformity
to expectations  by friends.  The reasons  people have  for testing
could stimulate the formation of hypotheses about effective methods
of risk communication.

     Qualitative   information   about    the    conceptions   and
misconceptions  that  people  have  about  radon would  also  be  of
interest. Among other things, such conceptions could be of crucial
importance for mitigation. Some people  may believe  that radon is
most likely to be  found in the attic rather than  the basement of
a house (since many gases are lighter than air). Ionizing radiation
is also something  that people  know  little about and have several
misconceptions. A study  of this type  is under way  at Carnegie-
Mellon University.

     There have been few attempts to analyze the contents of radon
risk attitudes. So far, most data suggest that health concerns are
most  important.  Many  studies  could have profited  from  a more
extensive mapping of the  components of  risk attitudes,  using one
of the  well  known models  of attitude measurement,  such  as that
devised by Fishbein (Fishbein & Ajzen, 1975).

     Another   possibility  involves  switching   to   the  common
international measureBq/Hr rather than pCi/1.  Since 1 pCi/1 = 40
Bq/m3  such a  switch would  involve  larger numbers that  could by
itself have an effect (whether positive or not).

     Many of the studies reviewed here used a risk  concept that was
not further explained to respondents, who were asked to judge such
perceptions as  size of radon  risk. However, other  research has
shown that people  interpret  the word risk in various ways. There
may be differences among genders and educational strata in whether
respondents  interpret  risk  as  probability,  as the size  of the
consequences  or   as   a   combination  of  these  two  concepts.
Interpretations  of  results may be  misleading if this  factor is
neglected.

     Another  interesting  methodological  aspect  is the response
scale. It is difficult to  translate  the  judgment of seriousness of

-------
a  risk into  subjective probability.  On the  other hand,  it
notoriously difficult to obtain valid judgments of small percei^
risk. As an alternative,  comparative risks often are  used to assi
people in judging small risks.  The radon risk has been compared to
the  risks  of smoking,  for  example.  This approach  is  well worth
trying but one should bear in mind that comparative  risk judgments
are more complex than absolute ones because they involve two risk
levels rather than one. Risks are probably perceived in a context
of several  threatening  events  and conditions and the judgment of
any given  risk may well be  affected by how certain other salient
risks are perceived.  For example,  people with a very risky job may
perceive  their  radon home  risk  as  minor  just because  it  is
implicitly or explicitly compared to a job risk.


1.9. Social diffusion

In many circumstances people are most strongly affected by social
diffusion  of  information  and attitudes,  i.e. what their friends,
neighbors and locally prominent people tell them. For this process
to spread quickly there must be many willing  "informers". However,
in the case of radon there are still very few  people  who have shown
any active  interest  in the issue of  home testing,  and those who
have informed their friends and neighbors have sometimes met with
hostile reactions.
1.10. House values and house sales

There has been  little work  on  property values and house sales as
related to radon.  A Swedish study recently found that radon did not
seem to affect house prices.  Informal  information  from realtors in
the USA  confirms  this finding. However, a  Colorado study showed
that many people are quite concerned about getting a house tested
for  radon  before  they  buy.  Some  50  percent of  the home-buyers
participating in the study  reported that they had done so.  These
people also  tended to  mitigate much  more  oftenr than  those who
bought  test kits  and  monitored  their own  homes. Indeed,  the
prevalence of testing at or above the action level of 4 pCi/1 was
100-200 times as  great  for  home-buyers as  for homeowners.  People
whose employer  was especially  concerned about the radon risk and
those who were in contact with  a knowledgeable-realtor or building
contractor  were  especially  prone to  test  and  mitigate.  These
results were obtained in  a  state  where radon risks have not been
extensively discussed in the media (Colorado).
     The fact that real estate prices are not affected by radon may
be  related to  the prevalent  information  that  a  radon  problem
usually can be  fixed quickly and  fairly  cheaply.  There  has bee
little awareness  so  far  about  how often continued monitoring f,
radon will be required and additional  mitigation will have to
                                8
^4il
a

-------
undertaken. When this  issue  is  given  wider publicity there could
be an effect on real estate prices.
1.11. Missing information: groups not studied

Virtually all research  so  far has been conducted with homeowners
so the  social strata  that have  been  represented have  not been
representative of the whole population. We do not know how people
who rent their homes respond to radon risk, or how involved their
landlords are in monitoring  and mitigating such houses.  There is
also very little work reported about attitudes to radon in schools,
public buildings and workplaces.


1.12. Conclusions

Research  on radon  risk perception and  risk mitigation  has  not
proceeded  very  far,  simply because  the  problem  is so  recent.
However, it is possible to draw some  conclusions on the basis of
existing research results.

(1) .   The  major  problem  is  indifference to   high-level  risk,
especially among people who  already live in a home that may have
an elevated radon level.

(2).  Home buyers are probably  much easier to alert to the issue,
with ensuing radon testing and mitigation  if called for.

(3).  Studies of variations in risk communication material suggest
that  brief,  to-the-point  recommendations about  what  to  do  are
better  than longer  messages  that allow  the  reader to  form  his
personal  opinion.   It   is  just  too  likely  that  these  personal
opinions will be defensive risk denials.

(4).   Even  the most  effective  mass media  campaigns  and outreach
community   programs  investigated  so   far   have  succeeded   in
stimulating only a small fraction of the population of homeowners
to monitor  and  mitigate.  Whether  these  campaigns  have  still
justified their costs has not been determined.  A  community outreach
program  appeared  to be more  promising  than a  major  television
campaign, although they probably reinforced one  another.

(5) .   Further  research  should  be conducted on  the components of
radon risk attitudes and conceptions, in particular why some people
test  and mitigate.  Interpretation of  findings  would  probably be
helped by more information on how people  interpret such key terms
as "risk".

(6).  There  are important ethical aspects of radon risk that have
been neglected in research. Some people  see nobody as responsible

-------
for a radon problem,  others  may blame themselves,  contractors
the  government.  The  owner  of a  house  may  or   may  not  fee
responsible for the  health of  others  who live in it.  At the tin!
of a house sale there is  a  special  ethical problem if the seller
knows there is  a  radon problem. These  are important  matters for
further research.

(7).  There is a total lack of  data  on how renters, as opposed to
homeowners, perceive  risk and  how their landlords  view the radon
issue. There is very little  data on the perception of radon risk
in public buildings and workplaces.  Perhaps some homeowners can be
alerted about  their  home  risks if they get involved  in school
building risks.

(8).  Although the sparse data that exist on the  issue suggest that
radon risk does not reduce property  values,  the issue certainly
needs much more investigation,  will the lack of effect persist when
more  experience  with  mitigation  and  its  costs  and  effects
accumulate? People may at this  time  be overly  optimistic about how
easy it  is  to "fix"  a radon problem permanently.  Do  people see
monitoring  as  a   potential  economic  threat?  Or   will   lack  of
monitoring and disclosure of results  at the time of a house sale
be the real economic threat in  the sense that homes that have not
been tested will sell  for a lower price?
                                10

-------
     "Don't expect too much. People can understand  risk tradeoffs,
     risk  comparisons,  and  risk  probabilities  when   they  are
     carefully explained. But usually  people don't really want to
     understand....Over the long haul,  risk communication has more
     to  do   with  fear,   anger,   powerlessness,   optimism  and
     overconfidence  than with  finding ways to  simplify complex
     information."  (Sandman, 1986, p.  23).
2. Introduction

Radon is a colorless and odorless gas that is  emitted  from uranium
in rock and soil. It can enter houses through cracks  and openings
in foundation walls and floors, drains or ventilation systems, or
by  being released  from water  from  underground sources.  It is
radioactive2.  In addition to  naturally  occurring radon, there is
some radon emitted from building materials. About 10 percent  of the
indoor  radon  in the  USA is  estimated  to  emanate  from building
materials (Krimsky & Plough,  1988).

     There is consensus among experts that prolonged exposure to
high levels of radon can cause lung cancer. This consensus is based
on  data on the  prevalence of  lung cancer among  uranium  miners
(National Research Council, 1988) and general knowledge about the
health effects of ionizing radiation.

     Radon has been named as the most serious  environmental  health
hazard  threatening   the  -American  people  (Lafore,  1987;  us
Environmental Protection Agency,  1987  b; Kerr, 1988),  and  most
experts agree that radon is the most serious indoor pollutant. The
average  indoor  radon  level   in  the USA  is estimated to  be 1.5
picocuries per liter of air3  (pCi/1), the outdoor level to  be 0.2
pci/l.

     Data on  indoor radon  levels are available also from other
countries in  North America  and Europe. The  following  data  (see
     1.  Sea the special issue of American Water Works Association
Journalf 1987, 80. No. 7.

     2 When radon releases its ionizing radiation it is transformed
to so-called  radon daughters,  the radioactive isotopes  P0213 and
P0214 which constitute a more dangerous threat than radon itself.
When they are inhaled and deposited on the bronchial tree they can
be carcinogenic.

     3.  1 pCi/1 means that about two radon atoms per liter of air
decay per minute.

                                11

-------
Table I) were compiled by Akesson, Bergman and Johnson  (undated^
based on  data  from the Swedish  National  Institute for RadiatiB
Protection  (1987) and Nero  (1989).
Table I. Distribution
countries .


Country Mean pCi/1
Sweden
. Norway
Finland
USA
Canada
West Germany
Great Britain

1.3
1.5
0.7
0.7
0.3
of radon exposure in various
Percent of
2.5 pCi/1

10
12
3
0.7
na
homes that exceed
5 pCi/1 10 pCi/1

3 1
4 1
1 0.2
. -<0.1
0.2 <0.1
     It is worth noting that, according to these figures, the radon
problem  seems to  be  somewhat  worse in  the  USA  than in other
countries where measurements are available.

     EPA has estimated that 12 percent of  the homes  in the  USA may
have radon levels exceeding 4 pCi/1,  which is the lowest level for
EPA's official  action  guidelines . Later  estimates  of 20  percent
have been mentioned  (Krimsky & Plough, 1988)..

     Cohen & Gromicko  (no  date)  compiled  a large number of radon
measurements  (about  35  000 homes)  and  presented  results which
probably give  a good  picture  of the situation in  the USA  as a
whole. They found that the  distribution of radon concentration was
skew% with an excess of houses with very high levels. One house in
     \  EPA's action  guidelines are  not  health-based standards.
They are recommendations  based both on health considerations and
what was judged to be achievable in terms  of reducing  indoor radon
levels at reasonable cost.

     2.  Averages were typically twice the median and only 24% o™
the houses were above average.

                                12

-------
1000 was estimated to have  a  radon  level  100 times or more above
the  mean.  They also  found  that  there  was  a  seasonal  variation
(about 60% higher  values  in the winter),  and a low correlation with
the age of the house.  Basements had  2-3 times higher concentration
than  livingrooms.  Weatherization  activities  since  1974  were
estimated to have increased radon levels by  about 15%.

     Lifetime exposure to EPA's  action  level of 4  pCi/1 has been
estimated to carry a 1-5  percent  increased  risk  of lung cancer. It
is estimated that the lung cancer risk of 20 pci/l corresponds to
smoking more than  a pack  of cigarettes per day. Of all lung cancers
not related to smoking,  10-50 percent are  due to radon exposure.
Smokers  are more  seriously  affected by  radon than  nonsmokers
(National Research Council,  1988).


     Risk estimates are  seldom final, of course. Estimates of the
risks  of  low-dose  ionizing  radiation  are  based  on data  from
survivors  of  the Hiroshima  and  Nagasaki  bombs   and  there  is
considerable  controversy about  those data.  Recent  revisions of
interpretations of Hiroshima and  Nagasaki data imply that the risks
are probably between 5 and 15 times as large as previously believed
(Rotblat,  1988).  The extrapolation  of  radon risks  from  data on
uranium miners  is uncertain because  home  radiation generally is
much lower and because  working  in a mine  is  a  very  different
activity from being at rest at home.  However,  it should be noted
that some  homes do  have  levels  as  high as  those  observed in the
epidemiological studies of miners. Also, not  all the time spent at
home is  spent resting.  Some of  it  can  involve  physical exertion
similar to that in mines. It has also been pointed out that there
is no sizable positive correlation between the prevalence of lung
cancer and radon statistics across  regions.

     Since  controlled experiments  on humans  are  ethically  and
legally unacceptable there  is bound to  be continued controversy.
This can be compared with the  risks  of fluoride  in drinking water,
for, which there is still  considerable uncertainty about the issues
despite 40 years of investigation (Hileman,  1988).

     Until 19851,  there was  little attention in the United States
to  high radon  levels  in residential buildings,  resulting from
geological radon.  The problem had been discussed earlier in other
countries  (Sweden and Canada), and  it had  been  mentioned in a UN
     1.   In  December  of  1984,  Stanley  Watras,   a  resident  of
Boyertown, Pennsylvania, set off radiation detectors as he entered
his  place  of  work,   Philadelphia  Electric  Company's  Limerick
generating station, a nuclear power plant. A subsequent test of his
home showed  that  it measured 2700 pCi/1.  The lung cancer risk of
this level of radon exposure has been estimated to be equal to that
of smoking 280 packs of cigarettes per day.

                                13

-------
report in 1977  (United Nations Scientific Committee on Effects
Atomic Radiation,  1977). This  is  in spite of the fact that the
was considerable concern in the 1970's about radon emanating from
man-made sources, such as uranium mill tailings  and phosphate slag
in building materials. The EPA was  criticized in the early 1980's
for its  concern over radon risks  from mill  tailing piles, which
were seen by some as minor in relation to  the  risks from geological
radon.

     EPA had published a risk estimate (5000 -  20 000 lung cancer
deaths per year in the  USA  due to indoor radon) in 1979  (Guimond
et al., 1979). The current official estimate is about the same.

     Since  December  of 1984,  the  problem  has been  intensely
analyzed and it is now known that many homes  in  the USA are likely
to  have  radon  problems,  perhaps  as  many  as  8  million  of  the
nation's 70 million homes. It is the federal policy that every home
in  the USA  should be  tested  for  radon.  There is,  however,  no
legislation  that  forces  any private  citizen  or any government
agency to do so. It is up to the individual to  monitor his or her
own risk level, and to take appropriate action.
     It is relatively  easy  and inexpensive to test for radon. It
is  also  usually rather  straightforward to mitigate  against any
radon problems  that might be discovered in testing,  by means
improved ventilation,  sealing of holes  and cracks, etc1. Stil
very  few  people have  tested their home  for  radon  (less than
percent of the US homes as of November, 1988).
ny
1
     The  population's indifference  to  home radon  risks starkly
contrasts with  the strong  reactions observed in the  New Jersey
community  of Vernon  to the  proposed  deposit  of  soil  somewhat
contaminated  by industrial  radioactive wastes  (Chess  &  Hance,
1988) .  The  enraged people  of Vernon  who threatened  civil  dis-
obedience (and were successful in avoiding having the contaminated
soil deposited in their community) were some of the same people who
did not bother to test for radon  risks  in their homes - risks that
were, according  to  experts, much larger.  Outrage over risks from
toxic waste  has, of course, been noted many  times,  e.g.,  in the
case of Love Canal  (Mazur, 1984).

     Because the radon problem was discovered only recently it is
only natural that rather  little  research  has been  carried out on
the  perception  of  and attitudes toward radon risks.  Still,  the
acute nature  of the problem has  motivated  some  rather extensive
attempts  at mapping  people's reactions  to  radon  risks  and  to
     1.  The question of how often  such repairs need to checked an
improved in the  future has been rather little discussed. A rad
problem is, of course, seldom  solved  for as long a  time as tt
lifetime of the house.
nd|
i
                                14

-------
investigate the effectiveness of risk communication methods  in this
area. This  report reviews these  studies,  draws  conclusions from
them, and suggests topics for further research. The reader  is also
referred to Fisher and Sjoberg (in press).

     I will concentrate on the reactions of homeowners to radon in
their  own homes.  A  few researchers  have  collected  additional
information  about people's  reactions to  radon  risks  in  public
buildings and to taxation aspects. While these problems certainly
are important, I feel that there is as yet too little information
to warrant detailed  coverage. The importance of  radon  risks for
property values is another aspect that so far has not been treated
thoroughly in empirical research.

     Before  discussing  current   research  I  will   sketch  the
development of public radon awareness and attitudes in the USA.


3. Information, the media and radon

Mazur (1987) has described in some detail how radon came to be an
issue on  the public's risk  agenda.  Radon risks did  not receive
national attention in the USA until 1985. Risk estimates had been
published in  professional  journals and by the EPA several years
earlier but with little impact.

     There was (and perhaps still is) a wide spread belief in the
positive effects of  radon. On November 25, 1984,  just before the
Watras incidents, the New York Times had the following to report:

     "People suffering from arthritis, headaches and other ailments
     come from  all  over the  world  to seek  cure in  former gold,
     silver and uranium mines in  southwestern Montana, whose rock-
     walled tunnels emit radon, a radioactive element produced by
     natural  decay  of uranium; among  physicians  reactions range
     from charges of quackery to tacit blessings for patients who
     have no other relief".  (New York Times.  November 25, 1984).

     The New  York Times  Index offers some further information on
the issue. There was a rather alarmist article- published already
in 1980,  citing among other things some Canadian data, but it seems
not to have had much impact on the public.  Then, the newspaper was
almost silent on radon issues until 1984. The New England Journal
of Medicine published an editorial in the spring of  1984, mentioned
by the New York Times on June 7,  1984. The editorial stated that
there may be  up to  100 000 cancer deaths among non-smokers every
year in the USA. The statements made by physicians can be assumed
to be of special importance for public opinion.

     The Watras  incident (see page  13,  note 1)  was  of critical
importance for alerting the  public  to radon  risk.  In addition, a
few weeks before the Watras incident  (in December, 1984) the home

                                15

-------
                                                               * • *•
                                                               i
of  Joel  Nobel,  which  measured  55   pCi/1,  had  received  sofl
publicity, even on national television.                        ^

     The  Watrajj family  evacuated their  house,  "in  some panic"
(Mazur, 1987, p. 39).  The Pennsylvania DER then conducted  a survey
of  some neighboring  houses,  found  some elevated  radon  levels,
although  nothing close  to the  level in  the Watras  house,  and
informed  a  township  commissioners'  meeting on January  7,  1985.
After that,  local news media were quick to report the Watras event
and it  was given wide spread local attention. However, according
to Mazur it was ignored by national media until May, 1985, when a
first page article appeared in the New York Times1, other  national
media  then   picked  up  the story  and concern  about  radon  rose
significantly,  leading to renewed activities by the EPA.

     Mazur points out that the  current risk estimates agree well
with those published  in the  1970's.  Strictly speaking,  radon is
not a  "new"  risk,  and the fact that  it  was ignored for almost a
decade  needs  an explanation.  The risk itself is larger than many
of the  other  risks  that people  have  cared about and that EPA has
regulated,  so  its size  cannot  explain  the  lack of   concern.
According to Mazur, the limited publicity that had occurred in the
Nobel  case,  just  before the Watras  incident,  was  an important
factor  in sensitizing  the   public   (and  decision  makers).  Th
vigilant response by the Pennsylvania OER did much to add  moment '
to the issue. It can be added that the sheer magnitude of radiati
measured in the Watras house  was quite surprising. It established
a possibility of disastrous radiation  in a home,  which had  not been
expected before. Finally, perhaps the connection to nuclear power
(Watras  worked in  a  nuclear power   plant)  may have  added some
attention value.

     Following  the Watras  incident  there  was  considerable mass
media attention  to  radon in the USA,  especially in the East. The
fact that the Watras family had evacuated their home made  a strong
impression.  It has been suggested that apathy might be lessened if
it were made  known-that a nearby family had evacuated their home
(Chess  &  Hance,  1988).  The  contents  of  local  Pennsylvania
newspapers during the first  nine months  of 1985 were compared to
national media  by  Friedman et al. (cited by Lehrer,  1988). They
found that local newspapers gave much more  attention to the radon
issue  than  national  media  but that  headlines  tended not  to be
alarmist, avoiding such words as cancer and threat. Local editors
said they wanted to avoid panic and adverse effects on the housing
market.
     1.  However,  the New York Times  had carried an article aboijH
elevated radon levels in New Jersey homes already on February \l,
1985.

                                16

-------
     According to  Sandman  (1986),  the strong initial response  in
Pennsylvania and New Jersey soon tapered off. in the New  York Tiises
there was  a tremendous  increase  in the number  of radon related
articles in 1985-86, but in 1987  the number dropped,  see Figure  1.
Sales of test kits went down in the winter  of 1987  compared to one
year earlier (Sandman, Weinstein  & Klotz, 1987) .  Sandman explained
this by the facts that radon risks are chronic, not acute, events,
and that  there is  no obvious "villain" or social and  political
conflict involved.
                      so
                      40
                      30
                   •8   20
                      10
                        8081828394888687
   Figure 1.  Number of published articles about radon in the New
   York Times.  1980-87.
     It is clear  that few people actively seek information about
radon. When it comes to media and other sources of  information they
tend to trust local  scientists, the EPA, and possibly newspapers.
The attitude toward  local government agencies varies.

                              -j
4. Characteristics of  radon risks

It is difficult to predict how people will  respond to new hazards.
When indoor radon in homes  was first widely discussed in the USA
there were some expectations among administrators that people might
panic  (Sandman, Weinstein &  Klotz,  1987).  The opposite occurred.
The reasons for this indifference constitute a crucial problem  in
communicating about  radon risks.

     A number of  factors that have been mentioned as explanations
for the lack of urgent responses to radon  risks.

A.There is no "bad guy" responsible for the risks from geological
radon. The  existence of a villain  to  blame for  a  risk has been
singled out as an  important factor in accounting for  public outrage
(Baum, Fleming  &  Davidson,  1983).  An  often quoted comment  by a
                                17

-------
participant in a focus group  is "What are we going to do, sue God?B
(Desvousges & Kollander, 1986). This factor may be  related to the'
lack  of a  discrete  source   -  be  it  an  object or a person  -
responsible for the risk1.

B.  The hazard  is  a  natural one,  not  man-made. People  tend to
underestimate the risk of natural hazards (e.g. Kunreuther et al.,
1978; Lehman & Taylor, 1987). The reason for this underestimation
may be  that nature  is seen as benevolent,  or it is believed that
ve are  biologically adapted  to  it.  In Sweden,  the  initial strong
reaction against radon was probably due to the belief that it was
caused  solely by building materials. When it was later found that
most  of the problem  was caused  by naturally emitted  soil  gas,
protests dwindled.

C. The  hazard cannot be sensed, so  it tends to be underestimated.
It has  been suggested that if radon was not odorless there would
not be  a radon problem.

D. The  illness,  lung  cancer, does not give early warning signals
and takes a very long time to develop.

E. It  is not possible to identify,  in concrete cases, that radon
has  caused  a  person's  lung  cancer.   Lung  cancer deaths  are^
undramatic and occur singly.

F. Persons typically have a long history of benign experience with
their homes.

G. The  choice  of a home is  under one's  own  control, so the risk
may appear to be "voluntary".  This  point can be disputed. Lehrer
(1988)  compared it to such life style risks as smoking which also
calls for protective action by the individuals exposed to the risk.
He judged radon risk to be clearly  involuntary.

H. The  risk is higtoly variable from house to house  and depends on
a number of factors that are hard to understand completely.

I.  Testing  and  mitigation are  complex  and new  activities  that
people  know little about.

     Against  these  factors   that   are  believed  to  reduce  risk
perceptions there are  a  few that  might work  in  the  opposite
direction:

J. The  risk is relatively unfamiliar and new.
     1.  This idea was suggested by Amos Tversky.

                                18

-------
K. Testing  is  simple  and  cheap,  and mitigation relatively so, in
most cases.  These  facts are likely to  moderate  the tendency for
risk denial.

L. The  risk is, after  all,  a cancer  risk and cancer  is  a much
feared illness.

And,  finally, some factors are ambiguous:

M. Extensive mass media attention, such as that recently given to
radon,  should  increase  risk  awareness  and increase the perceived
level of  risk.  On  the other hand, mass  media  attention is never
constant. In the case of radon it is probably tapering off from an
early peak  in 1985-86.


N. Radon  threatens people  in a very personal  and serious manner.
This makes  it  subject to  the phenomenon of risk denial, which is
very common when it comes to health risks.

0.  There  are  no  benefits   from radon  exposure.  This  factor
distinguishes  it clearly from life  style risks,  where there is
usually  a  temptation  due   to  the  pleasure   that follows  upon
consumption.

     Some comments  are  in order.  It is  illustrative  to compare
radon with a well known highly threatening risk, such as the risk
of nuclear power (cf. Sjoberg & Drottz, 1987).

     The  lack of a "bad guy" is a factor that may be more true of
radon than  of  nuclear  power.  It has  some anecdotal  support as
reducing perceived risk, but no theoretical underpinnings.

     Many risks that people  fear strongly cannot be sensed. Both
radon and nuclear power cannot be sensed but  there is widespread
fear  of  nuclear power.  Thus  this  factor can  hardly  explain
indifference to radon risks.

     Cancer with a long latency between exposure and illness is a
characteristic  of  both  radon  and nuclear power risks.  The long
latency factor cannot explain why people ignore radon risks.

     The  more  unique factors  in  the  case of  radon appear  to be
mainly that is  a natural hazard and that  it  is  related to one's
home and under one's  personal   responsibility  and control,  as
compared to nuclear power risks or smoking. Furthermore,  it is easy
to delay action about radon, because there are no obvious adverse
consequences from postponing  testing (or mitigation) for some time.

     Radon  risk communication  also needs to be  viewed  in  the
perspective of risk communication and health promotion in general.
It has been found repeatedly that it is very hard to induce health

                                19

-------
promoting behavior  (Adler & Pittle,  1934).  Some more optimist
results were reported by Viscusi,  Magat and Huber (1986) and Pus
et  al.  (1979).   The  former  was  concerned with  a  laboratory
simulation,  however,  and the  latter  with a  very extensive mass
media  campaign,  much more  extensive  than would  be  realistic to
consider in the USA.

     Thus, risk  communication  is difficult  and  rather little is
known about how  it  should be carried  out most efficiently in any
given field.  In addition,  rirk advisory programs at EPA have seldom
been evaluated (U.S. Environmental Protection Agency,  1987).

     Current EPA sponsored activities constitute a response to this
need for research. Several attempts (by EPA and others) have been
made, and others are underway,  to  study empirically the properties
of radon risk perceptions and attitudes and how risk communication
can be improved. This work is reviewed  in next section.


5. Empirical studies of radon risk "perception and attitudes

5.1. Introduction

In this section, brief  descriptions will  be  given of each of ti
studies carried out so far. The general strategy in reporting thej
studies has been to begin with  a seminal investigation  of  somewhl
limited scope, to continue with preparatory work for more extensive
studies,  then to  deal  with  the three  major  studies   of  risk
perception and radon monitoring that have been  carried  out,  and
finally  to  treat  a  number  of   smaller and  more  specialized
investigations.  A special section, see page 41,  is devoted to work
on mitigation, which is  more  recent and not yet as fully developed.

5.2. Maine study

5.2.1. Design and selected results

Johnson and Luken (1987)  analyzed data  from 230 Maine households,
who  were  interviewed in  the fall  of 1985.   The  respondents  had
participated in a lung cancer epidemiology study.  They had  received
radon measurements of their  homes and a University of Maine radon
information pamphlet.

     Data were collected on risk perceptions,  mitigating activities
and socioeconomic characteristics. The risk judgments were both of
current risk perception and what  the  respondents remembered that
their  risk  perception  had  been  before they obtained the radon
readings.
                                                               I
     Johnson and Luken found that information about radon chang
risk perceptions, and  that  mitigation activities tended to lowe
their  personal  risk  estimates.  However,  they  also  found that

                                20

-------
mitigation was equally likely for those who had obtained low radon
readings  as  for  those  who  had  obtained  high  readings.  The
respondents  greatly  understated  their  radon   risks,  with  few
exceptions. In spite of this, about half of them reported some kind
of mitigation.

     A subset of these data were also analyzed by  Smith and Johnson
(1988). They could use data  from only  117 of the 230 respondents,
due to missing data. It was  found that people adjusted their risk
estimates according to the information they obtained, in line with
a simple Bayesian model.  They  also confirmed the Johnson & Luken
finding that individuals who took some mitigating actions reported
lower risk perceptions after that action.

     Smith  and Johnson  stated that  these  results  suggest that
people may be more rational  than  previously believed and that they
may be open to  new  information about risks  and  adjust their risk
estimates accordingly.

5.2.2. Evaluation

This study is obviously of limited scope.  The sample was small and
consisted  of  a very  special group  of  patients  and  the data
collected were not very extensive. The study was important mainly
because it initiated an interest  at EPA in conducting further work
on radon risk communication.

     The claim to rationality  in people's risk perception made by
Smith and Johnson is similar to a claim made in the NYSEROA1 study,
next section,  and  it will be  discussed more extensively  in that
context.  However,  it  must   be  noted  here  that  the  use  of
retrospective  risk  estimates  is not  very satisfactory.  Clearly,
there is no  support for the implicit assumption that  people can
remember and  correctly report what risk  perception  they  used to
have. On the contrary, Fischhoff's well known studies of hindsight
bias  (Fischhoff &  Beyth,  1975)  have  demonstrated that  people
consistently distort -their  memories of earlier  judgments,  to be
more in  line with  information current at the time  of  judgment.
Smith and Johnson were well aware of these limitations, of course.

5.3. Focus groups
5.3.1. Peaion and selected results

EPA  sponsored  several  focus  groups  on  radon.  Desvousges  and
Kollander (1986) reported initial trials with 6  focus groups. They
were  conducted  with  homeowners  having  different  degrees  of
awareness  of the  radon  problem. The  purpose  was to  evaluate
     '.  New York State Energy Research and Development Authority.

                                21

-------
information materials to be  used  in  the NYSERDA study and to
some preliminary insights into radon attitudes. In the  second
group study (Desvousges  and Cox, 1986) two groups of homeowners who
had tested their homes for radon participated.  Preliminary versions
of two of the NYSERDA booklets were evaluated. A further activity,
preliminary testing of materials to be used  in the Maryland study,
involved  four  groups - two  in  Pennsylvania  and  two  in Maryland
(U.S. Environmental  Protection  Agency,  1987 a). Both testers and
non-testers of different educational  backgrounds participated. The
purpose  of these four  groups  was to  probe why  some paople had
tested for radon while others  had not. Results  from all  of the
focus groups were used to plan further,  more precise and extensive
work.

     Focus groups are useful for getting qualitative information
about the important  factors in any attitude study. They cannot give
information about how common the  opinions are,  or about strength
of relationships, in a population (Desvouges  &  Smith, 1988).

Some of the findings were:

- Quantitative  information was  often called for, as well as risk
comparisons.

- Nonsmokers were more satisfied with comparisons to smoking ri
than smokers who found such  comparisons "confusing".

- Color coding of risk charts was rated favorably.

- People were unhappy and confused when informed about uncertainty
in risk estimates.

- Some people preferred numerical risk estimates,  others verbal
labels and still others graphical information.

- People wanted answers to concrete questions such as "What is my
risk" and "Where can I get information about mitigation".

- Some found the  risks  charts too scary,  others liked them to be
scary.

- Testers had  (with one exception)  spoken to  friends  or family
members before deciding to test.

- Non-testers  showed the familiar denial dynamics,  i.e. they did
not  believe  that they  themselves were  at  risk, and they found
special reasons for  "proving" that view.  (Radon risk is not well
understood by experts, the risk is only one among many, "everyth-
ing gives you cancer",  "Three Mile Island is much more of a dang^
to me").
                                22

-------
- Concern  about property values could  both  motivate testing and
cause people to abstain from testing.

- Many  non-testers believed that mitigation  would be difficult,
expensive and risky.

- Information about radon was mainly obtained from newspapers and
television. EPA and state environmental agencies were  rated highest
in credibility.

5.3.2. Evaluation

These were relatively minor studies  performed  as part of designing
the  NYSERDA and Maryland  studies  to  be described  in following
sections. Work of this kind cannot provide quantitative estimates
of the frequency of various opinions or the importance of factors
determining attitudes  and  behavior  but it can  give  a rough idea
about what factors are important.

Some interesting information from these groups has  not been tested
in subsequent research. For example, there may be individual styles
of   information  processing   since   some ' people   prefer  verbal
information,  others  graphical.  Perhaps  information  could  be
tailored  to  fit  these   styles.  Another idea is  the  use  of
appropriate  comparative  risk.  There  was some indication  that
smokers did not like smoking as a comparative  risk.  Perhaps smokers
tend  to  deny risks  of smoking  and therefore  are confused when
smoking risks are brought in as a reference norm.  At any rate, if
comparative risks  are  used  it seems important to  get independent
measures of how the reference risks are evaluated.
5.4. NYSERDA atudv

5.4.1. Design and selected results

This study (Smith, Oesvousges, Fisher & Johnson, 1987, 1988), was
an attempt to test different approaches to risk communication. The
participants in the study were homeowners  in  the State of New York
who had agreed to participate in a state-wide sample organized by
New  York  State's  Energy  and  Research  Development  Authority
(NYSERDA). NYSERDA's objective was to test a sample of homes in the
state for radon  in  order to map the prevalence of radon problems
in the state.

     Baseline data concerning risk perception and knowledge about
radon were collected while  the radon  monitors were in place. The
same  data  were  collected for  a comparable  non-testing  control
                                23

-------
group'.   Brochures  were  then  sent  to   the   homeowners   in  the
monitoring  group  together with  information  about  their radg^
levels .

     The NYSERDA sample was divided into  several  groups, with each
group receiving a different information  treatment.  Four brochures
contained  information  about   risk  levels  and  mitigation.  They
differed with  respect to two  dimensions: (a)  Whether information
about the  risk  level was given  in  quantitative  or qualitative
format and  (b) the tone of  the. text which was  either directive
("command") or evaluative  ("cajole").

     The command tone emphasized what an expert  (in this case the
EPA)  recommended  as the appropriate  response to  the risk level
while the cajole tone encouraged the  recipient to  form his or her
own opinion on the basis of the information presented.

     The four brochures contained the  same information except that
only the cajole brochures contained information about how to adjust
the  test  result  to  reflect  different  lengths  of  exposure.
Households with  a  radon level of at  least  1  pCi/1 also received
EPA's mitigation brochure , but mitigation was discouraged on the
basis of a 2-3 month reading alone.
     Other groups  received  EPA's  Citizen's Guide or a radon fact
sheet  developed by the  State of New  York (and  similar  to fac
sheets used  in  other  states).  All treatments involved giving t
homeowners the same information except that the fact sheet had le
information. It was,  however,  sent  only to a subset of those who
had test results of less than  1 pCi/1.
wu
i
     Baseline  data  were obtained  in  the summer of  1986  and the
first  follow-up  interviews  were  carried out  six  months  later
(telephone interviews).  A second set of follow-up data was obtained
in   the  period   September,   1987,   to  January,   1988   (mail
questionnaires), after the annual  radon readings had been sent to
the homeowners.

     2300 homeowners participated  in the NYSERDA study, and there
were  252  nontesting homeowners   in  the  comparison  group.  The
response rates  for  those participating in the NYSERDA study were
high: 97 percent in the baseline interview, 91 percent in the first
       Th«r« is some doubt about  the status of this control group.
It should perhaps more appropriately be called a comparison group.
It was selected later than other groups, it had a low response rate
and there may have a been  a self-selection bias in the monitoring
study. The  comparison  group was  selected using  the same protocol
as the original monitoring sample, though.

     2 Two sets of measurements had been carried out: a short-term
measure (2-3 months) and two annual measures (12 months).

     3 Radon Reduction Methods;  A homeowner's guide.

                                24

-------
follow-up and  74  percent in the second  follow-up.  50  percent of
those approached for the comparison group agreed to participate.

     Three major questions were analyzed:

(1)  Which information treatment led to the most learning?

(2)  Which was  best  in helping people form realistic perceptions
of their radon risk?

(3)  Which was most effective in making them  feel  that they had
enough information to make a decision about whether to take action?

     Knowledge about radon increased in all treatment groups, but
the  increase  was  smaller  for those who  only received  the fact
sheet.  There   was  also  an  increase  in  the  comparison group's
knowledge. The  respondents turned  out  to be fairly knowledgeable
when it came to radon measurement, less so with regard to  risks and
mitigation. The four brochures  seemed  to  work about equally well
in  this  respect,   with  a   possible   advantage   for  the  cajole
qualitative brochure, which facilitated learning about radon more
than other brochures.  The  quantitative  brochures were  best for
communicating risk levels. A  higher  radon  reading also seemed to
go together with a higher  level of knowledge.  Older people  (over
40) knew less.  Education was a positive predictor,  and so  was early
awareness  of  radon   (measured  at  the  time  of  the  baseline
measurement).

     Why  did  the  particular NYSERDA  brief fact  sheet work so
poorly? The major reason is probably the very special circumstances
of testing  it. It was used with  people  who had very  low radon
levels  in their  homes.  Hence, the  normative  response was  to
decrease concern about  the risk. The fact sheet  probably did poorly
because  of this peculiar circumstance.  People who have not yet
tested or who have tested and obtained  a high radon reading should
be encouraged  to test or to  mitigate, and such  groups  were not
given the brief fact sheet in the study.

     Personal  radon  risks were  perceived  as  lower  than general
population  risks  of  radon.   Personal ratings  were  also  more
influenced by  information,  and far fewer  people  answered "Don't
know" to the personal risk question in the first follow-up compared
with the baseline. Memory of readings of lifetime risks  from risk
charts was better for homeowners who had received the experimental
brochures than for those who had received the EPA Citizen's Guide.
Those  who had obtained higher  readings  were   more  likely  to
remember incorrectly.

     It was found that  people updated their risk estimate somewhat
in accordance with the  radon reading that they  had received. Young
and well-educated  people were  most likely  to do  so. The updating


                                25

-------
results  were  different  for  'the  first  and  second   follow-up
however.

     For the first follow-up, the radon readings (living area) were
generally low  (average  =  1.39  pCi/1,  range = 0-39.8 pCi/1).  The
average  risk perceptions dropped  by  some 20  percent   (Johnson &
Fisher,  1989) . Johnson  and  Fisher suggested that there still may
be an  overall tendency to overstate  the  posterior risk,  i.e. to
adjust the risk ratings too little downward1.

     This  finding  is  important,  and goes  against the  general
finding  in  other  studies that there  is only a weak relationship
between perceived radon risk and actual risk. However,  this is the
only study as far as I  know  that has investigated changes in radon
risk ratings before and after a reading of the actual  radon level
has been obtained.

     The NYSERDA brochures led  to a lower rating of  radon risk than
the  factsheet,  especially  the quantitative  versions.  Those  who
received the factsheet tended to remain concerned  (in spite of the
low readings, always <1 pCi/1), and were  interested in  paying for
more information and guidance. The estimated willingness to pay was
the  second  highest for this  group,  and  80  percent (the highest
figure of all groups) said they wanted more radon risk information.
Smith et al.  (1988) concluded that giving very little  information
about a  risk  may  increase worries about it rather than the othcfl
way round2.  Those  who made more effective use of the brochures werv
less likely  to demand more information,  as  were,  in  particular,
older respondents.

     For the  second follow-up, the radon readings were somewhat
higher  than  for  the  first  follow-up. The  average living area
measure was comparable, but  the basement reading was twice as high
(average *  3.37 pCi/1,  range » 0-114.9 pCi/1). The  result was that
respondents adjusted their  risk estimates- upwards, especially if
they had received the  command/qualitative NYSERDA brochure. This
brochure encourages thinking about the radon risk as related to a
     1  This may be  an example of the  often observed tendency of
intuitive  judgments to  be  too  conservative  as compared  to the
normative Bayesian model.

     2  However, the contents of  the  information may  be  just as
important. The  fact sheet contained some basic information about
radon  and its  measurement,  and  about EPA's  recommended action
levels. The  fact  sheet  did  not specifically  state  that  a level
below 1 pCi/1 is nothing to worry about. It did state that  the two-
month  reading  "should  not  be  directly  compared"  to  the  EPj
guidelines, because it was not an annual concentration  and becaus
it was for a specific location only, not for a whole house.

                                26

-------
threshold  (of 4 pCi/1). Those obtaining basement values exceeding
the  threshold  consequently  had  a  rather   strong  increase  in
perceived  risk1.

     The conclusion of this report  is that the cajole/qualitative
brochure was  best at stimulating learning about  radon while the
two quantitative brochures did best at providing a more realistic
risk estimate. All booklets did about  equally  well in reducing the
demand for more risk information.

     The findings  indicate  that  none  of the  four employed varia-
tions is best in  all  respects but the over-all performance seems
to be  best for a command/quantitative approach, which combines
clear directions for action with precise information. The NY fact
sheet is judged to be  too brief and also misleading,  since those
homeowners who had less than 1 pCi/1 still were worried  if they had
received the fact  sheet.

     The latter conclusion may have other implications. If people
on the whole underestimate radon  risk it may be that the  fact sheet
format is useful in promoting a more vigilant attitude.

5.4.2. Evaluation

At the  time of writing,  this project is not yet  finished.  Only
preliminary mitigation data have been collected but  they are not
yet analyzed.  The full set of  mitigation data  were collected early
in the spring of 1989,  but no  analysis is available at the time of
writing this report (August,  1989).
                      •
     The study is,  however,  one of the most important and ambitious
attempts at studying radon risk communication.  The design was very
carefully  planned, response  rates  were  excellent  and  the  data
analysis is quite  exhaustive.

     Some limitations should be noted. There are no data  on testing
decisions  because NYSERDA  had  already contacted homeowners  and
gotten  them  to  agree  to participate.   It   is  not  known  how
representative these homeowenrs  were  or how  many were approached
and declined to participate. Furthermore, there is some information
suggesting the NYSERDA  sample homeowners  were not representative
when it comes to radon awareness.

     The choice of the two independent variables (cajole/command
and  qualitative/quantitative)  for  designing  the four brochures
seems a bit arbitrary,  and the report  does not explain in full how
     1  It would be interesting to analyze the data'separately for
those  who  received  readings  crossing  the  threshold,  or  to
investigate  any lack  of linearity  in the  relationship between
perceived risk and basement radon reading.
                                27

-------
the  group  of  experts  reasoned  when  they  recommended  th
particular variables. The researchers state that these factors we'
considered  to  be the  most  interesting ones  because the expert:
thought they might make  a big difference and because they tended
to disagree about how the factors might differ in their effects.
                                                              4
                                                              rtT
     There is, of course, a vast number of possible variations on
risk information messages and  the  empirical  research is, so far,
not very  extensive.  Maybe it would have been  better to start by
performing rather extensive  pilot  work to be  ible  to zero in on
the most  efficient  designs  for  risk communication.  As  it  now
stands, the study gives merely  a single piece of information about
a few selected designs,  which, as it happened, did not differ much.


     The study also takes an unusual approach in expressing concern
that some people may  exaggerate radon risk,  rather than neglect it,
which is by far the most important problem in the area. The brief
fact sheet could  well have been distributed also  to people with
high radon levels.  It would have  been interesting to  see  if it
works well in those  cases.  The present data,  obtained only from
homeowners with very low radon  levels,  suggest  that the fact sheet
or something similar may defend its place as an efficient means of
risk communication when people tend to neglect a risk.
                                                               I
     The finding that  people  tend to revise their risk estimat
in  the  right  direction  is  not  inconsistent  with  a   lack
correlation between  technical and subjective  risk levels, ofte
noted  by  psychologists.  The  reason  why  there  is  no  positive
correlation between  technical and subjective  risk may simply be
that people quite rarely receive explicit risk estimates.  In this
situation, their risk perception is prey to all  sorts of subjective
bias factors and personal tendencies.


5.5. New Jersey study

5.5.1. Design and selected results

This  is  one of  the  earliest and  most ambitious  studies  of the
public's response to the radon threat (Weinstein,  Sandman & Klotz,
1987). Th« authors investigated two samples:

1. Random  sample: Owners of  single-family  homes in New Jersey in
the Reading Prong area or near it.  Data were  collected in April of
1986. People who said they never heard of radon  were excluded. 61.7
percent of eligible  households participated. They tended to have
higher education and higher income than the  overall population of
this region. Mean age was 47.4 years.

2.  Confirmatory sample:  All households who  had  contacted
Jersey's  Department  of  Environmental  Protection  after

                               28

-------
obtained a reading of at least 4  pCi/1  through June, 1986. The net
completion  rate  for  this  sample  was 43.3  percent.   Data  were
obtained in  May-July,  1986.  This sample had very  high levels of
education and income, and more males than in the random sample.

     The respondents  had high trust in authorities.  Many basic
facts about  radon were  well  known, a  bit  less when  it  came to
health-related effects.  The subjects underestimated  the seriousness
of radon health threats. It  was apparently much  easier  to find
reasons to  expect that  one's own  home  had less  radon  than the
average (247 reasons given)  than  why it could have  more radon than
the average  (45 reasons given). The subjects of the random sample
did not seek information actively.

     Emotional reactions were discussed as related  to decisions to
test for radon, unrealistic optimism, explicit emotional reactions,
community interactions and views on blame and responsibility.

     Few respondents had monitored,  and there were  signs of apathy
rather than panic. There was clear evidence of unrealistic optimism
when it came to personal radon risks. Direct ratings of emotional
responses indicated moderate  to  high concern  and moderate worry/
but levels in other aversive emotions (anger, fear, etc) were low.
The  confirmatory sample did  not  show  any different emotional
reactions than the random sample, in spite of having found elevated
radon levels in their homes. Those who had monitored felt high in
concern and  low in  helplessness,  while those  who had  not even
thought  about  monitoring  felt  low  in  concern  and  high  in
helplessness.

     Blaming nobody for the  radon risk  correlated with monitoring,
as did willingness to pay for remediation.

     Weinstein  et  al. asked  about  reasons  for not  testing, and
obtained the following results (see Table II).

     The data suggest that people have not tested because they have
too little information or that th«y have given  a  test low priority.
This is a "rationalistic" pictur§cof their (lack of) action. It is
to be expected that people wish to describe themselves as rational.
For that reason, it may be that the reasons given  in the table to
a larga extent are due to a strategy to exhibit a  rational way of
behaving.

     Klotz, Weinstein and Sandman (undated)  found  that the actual
radon level had little to do with emotional distress. It correlated
only  weakly  with perceived  risk  and  perceived  difficulty  of
remediation.  Respondents  were  also  concerned about  potential
economical losses.

     Intention  to  remediate  was  not correlated with  income and
education, nor  with the cost, feasibility  and  efficacy of radon

                                29

-------
    Table II.  Reasons   for   not   testing   for  radon  (from
    Weinstein et al).

    Reason for not testing                       Percentage

    Don't know what method is best                    45.3
    Don't think I have a problem in my home           42.4
    Just haven't gotten around to it                  35.9
    Don't know how to get the test carried out        32.2
    Will wait: to see what others in community find    26.3
    Wouldn't know what level  is safe anyway           14.9
    Costs too much                                    3.5
    Don't think tests are reliable                    7.1
    Think results wouldn't be kept confidential       6.4
    Reducing radon levels is too expensive            5.8
    Didn't know it was possible                       5.4
    Neighbor's readings were  low                      3.4
    Not interested                 .                   2.4
    I'd rather not know if there is a problem         2.4
    Too much trouble                                  1.4
    Nothing can be done about radon anyway            1.4

    Note. People were asked this question  if they  planned to
    test. were undecided, or thought a test wag not needed.
reduction.  Radon  levels  correlated  weakly  with intentions
mitigate.

                                                   j/2
5.5.2. Evaluation

This study is especially interesting because of its attempt to map

psychological factors of  importance for understanding how people
react to-radon risks.

     The traditional psychometric  approach employed  here differs
somewhat from the more  econometric approach taken in the NYSERDA
study. However,  the methods of the  NYSERDA  investigators are based
on multivariate linear models, just as the ones used by Weinstein
and collaborators. There  is no reason to believe that the choice
of statistical method is very  important for the conclusions drawn
here,  and,  in  particular,  that  it  can  explain  the  somewhat
different pictures of risk perception that have emerged.

     It  is  interesting  to  compare the NYSERDA  and  New Jersey
studies   with  reference   to   their  general   orientations  to
rationality. The NYSERDA investigators looked for rationality and
found  it in risk  revisions,  while the  New Jersey investigators
looked for  lack  of rationality, and  found  it in  risk denial
lack of relationship between subjective and technical risk.


                                30

-------
     Reasons for neglecting to act given by  the respondents in the
New Jersey study may  not  have been the real ones.  People tend to
rationalize and  to hide reasons which  they believe are socially
unacceptable.  This suspicion is  supported  by a general  lack of
rationality in other aspects of the behavior observed here. There
was  little  correlation between  technical  and  subjective  risk
levels,  and  between  mitigation and  intentions  to  mitigate  and
technical risk. Also,  there  was clear evidence  of  risk denial in
these data.
5.6. New Jersey experiment on radon information

5.6.1. Design and selected results

Weinstein, Sandman and Roberts (undated) mailed radon information
brochures and questionnaires  to  400 homeowners  in New Jersey. A
form for ordering a  radon  test kit  at  a reduced price ($ 20) was
included. The brochures differed as to their descriptions of risk
likelihood  and  severity  and the  efficacy  of  radon mitigation
techniques.

     It was found  that the  19 percent of the sample who ordered the
test kit did  not  differ across  the  various brochure treatments.
Subjective  risk estimates and self-reported  concerns correlated
with ordering test kits and  intentions to test.   (All respondents
were asked about their intentions to test, and all were given the
chance to order a test kit).  Perceived mitigation difficulty was
unrelated to test intentions.

     Weinstein,  Sandman, and Roberts had predicted that providing
homeowners with information about radon risk factors would decrease
their concerns and risk estimates2. This was found to be true only
under the  high threat condition,  for  one of the risk measures
(absolute ratings of own risk). The high threat condition involved
depicting the radon risk as quite likely and severe. The prediction
did not hold  for  comparative  ratings where own risk was compared
to community riste* -

     There was a moderate amount of optimistic bias (i.e., people
underestimated the risk) in the radon risk perceptions, as compared
     1.  The response rate was 68 per cent (271 responded) in this
group. The authors do not state how many  refused to participate of
those who were approached.

     2.   The  rationale  behind  this  prediction  was  that  risk
information  provides  material  for  constructing  explanations why
they are not at risk. People use it in order to defend themselves
against anxiety by means of denial and rationalization.

                                31

-------
to ratings of  other  health  hazards on the same scale  (WeinsteiM
1988). This  may  be a sign of risk  denial.   Implications of ris^P
denial for risk  communication  are discussed in section 9.1, page
54.

     Although the treatments in which respondents were exposed to
high risk messages were partly successful in boosting their beliefs
in a high risk level  (with the  exception noted above), the effects
were not strong enough to stimulate an  increased level of testing
behavior. Tt is interesting to  note that this occurred  in spite of
a good fit of a model assuming  that risk information gives rise to
the perception of increased  risk which in turn causes protective
behavior. The model fitted well, but the effect was weak.

5.6.2. Evaluation

This study is mainly of  theoretical interest. For ethical reasons,
the  risk level  communicated  cannot  vary  greatly in practical
campaigns, even if it can be depicted in different ways.

     As  for  methodology,  the  group  of  subjects  was  small.  No
information  is given on how many people refused to participate in
the study. The reduced price of  $20  for a radon  test kit seems
still  to be high, as  compared  to the  price  charged  in  other
studies, such as the Maryland study (see Section 5.8,  page  35).
lower price may have given a quite different result in terms  of t
number of people who ordered a test kit. It is not stated why ttf
particular  price  level  was  chosen.  It  is   possible,  but  not
documented,  that  $20 was a reasonably low price for a test  kit in
New Jersey at the time of the study.
<=i.
I
     The prediction  of a paradoxical decrease  of perceived risk
with an increase  in  information was only partly supported. As an
after-thought, this may  be  a  type of effect that is secondary to
the main effect of risk  information,  which should be to increase
perceived  risk and  risk awareness,  at  least up  to a  point  of
moderate threat.  It  is conceivable that  a very strong threat may
give rise  to  denial,  but such threats are probably uncommon with
radon risk.
5.7. Study of effectiveness of communication formats

5.7.1. Design and selected results

The systematic  study  of  effects  of varying communication formats
in  communicating radon risk was begun by the NYSERDA study. That
study was concerned  with the tone of the  verbal  contents of the
messages and with qualitative versus quantitative risk information.
                                32

-------
     A study by Weinstein,  Sandman  and Roberts (1989)  deals with
the format issue in a very ambitious manner. They investigated two
risks: radon and  asbestos.  Recommended action levels correspond,
in these  two cases  to quite different  scientifically estimated
risks. The  asbestos risk  from  the  action  level  recommended for
schools1  (3  f/1)  is only 1/25 of the risk of 4 pCi/1. The authors
chose to present  both  risks in  comparable numbers,  i.e. asbestos
risk in f/1 and radon risk  in pCi/1.

     Seven different designs of  information  brochures were tested:

- Probability only  (numercial probability of harm)
- Probability plus comparison to smoking risks
- Graphic probability display
- Information about action guideline level only
- Action level, numerical probability and risk comparison
- Action level and detailed action advice
- Action level, probability, risk comparison and advice

     All  brochures  contained   an  initial  3   pages  of  detailed
descriptions of the  risk  (radon or  asbestos).  These formats were
evaluated in several  measures of communication effectiveness.  An
experimental  design  was  used.   A   total   of   1948  subjects
participated. Of those who were intitally contacted, 34.2% agreed
to participate. Of these  67% did respond  to  the questionnaires,
yielding an over-all response rate of 23 percent.

     The subjects  were  given hypothetical  risk information about
their homes.

     There were several interesting findings. Two will be mentioned
here.  First,  information about  the  action  level  induced more
differentiation  among  risks   as   a   whole,   not   only  in  the
neighborhood of the  action level  (some discontinuity  could  be
discerned  around  that   level).  Information  about  the  action
guideline only was the condition giving rise to the highest level
of perceived risk and most concern. The authors point out that this
condition waa the only one lacking a  risk  ladder  or a graphical
probability display. The subjects had  no way of knowing if a value
below or above  the action guideline was little or much different
from it. A  ladder gives strong  cues as to the difference between
an obtained radon  or asbestos reading and the action guideline.

     Second, the condition giving most information came out as the
most  effective  format.   There  was no  evidence of  information
     \  There is no official  action  level  for asbestos in homes.
Perhaps that is part of the reason why radon  risk seems to be more
socially  acceptable  than asbestos  risk.  Radon risk  for schools
might be  more comparable to  asbestos risk  action  level  as here
defined.

                                33

-------
overload. This  condition  contained numerical risk probabilitie:
risk comparisons  (smoking), action guideline and advice.

     The action level only condition thus  appears to be especially
effective in alerting people to a risk, while the most informative
formats  are good at  eliciting  realistic  risk estimates,  i.e.
alerting some people and reassuring others, depending on their risk
readings.

     Third, radon and  asbestos risks were rated as approximately
equivalent  in all conditions,  despite the true 1-25 range, which
was  of course  communicated both in  risk  probabilities  and  in
comparison  risk estimates. The authors suggest that the reason is
the strong cues  provided by the exposure ladder.  The ladder covered
a whole page for  both  radon and asbestos  risks. A smaller ladder
for asbestos, appropriate  for the lower level of risk, might give
a different result.  On  the  other hand, the  condition that gave only
the action  level  also  did not  differentiate between asbestos and
radon, perhaps because there was  no  information about the varying
risk levels in that case.

     The  problem  seems to one  of  using  graphics  for  conveying
absolute risk levels, not  only relative ones (Cleveland & McGill,
1984; Tufte, 1983).

5.7.2. Evaluation                                              (

The study, together with the NYSERDA study, is a good start towards
systematic  investigation of the effectiveness of risk presentation
formats.  As the authors point  out,  there  is  almost  no previous
research  on the  topic -  in  spite  of  the existence of several
manuals on  risk communication.

     The response rate of  this study was low. This means that the
results  cannot  be  safely generalized. On- the other hand,  the
authors state that  the response rate did  not vary systematically
across conditions so the differences between conditions cannot be
due to differential response rate.

     People who had actually tested  their  homes  for asbestos or
radon were  systematically assigned to the  condition  of  the risk
they had  not tested for.  It is  not  stated  how many these people
were, but presumably they were  few.  otherwise,  the  policy might
have created some systematic biases.

     The  hypothetical   nature  of   the   risks  presented  here
constitutes a problem.  It is possible that the paradoxical lack of
a difference between radon and asbestos risks may be due to a low
level of motivation  for the  subjects who  knew  that they were not
actually exposed to those  risks.
                                34

-------
     Work on risk presentation formats might profit from a contact
with psychological scaling  research  where  it  has long been known
that "absolute" judgments always are  affected by context (Parducci
& Perrett, 1971).

5.3. Maryland study

5.8.1. Design and selected results

The study evaluates an attempt to inform people about radon risks
(Desvousges,  Smith &  Rink,  1988,   see  also  U.S.  Environmental
Protection Agency,  1988 a).  Three  communities  in Maryland were
involved: Frederick, Hagerstown and Randallstown.  These communities
were  chosen  because  they  had  elevated  radon   levels,  a  high
percentage  of owner-occupied,  single-family  homes  and  separate
media markets. Residents of Frederick received the most extensive
treatment: media  (radio  and local newspapers)  information plus a
community program. The media program was organized around the theme
"Test now and be  sure"1. This  theme consisted  of the following
parts:

- Radon is a serious health risk: You may be at risk
- Radon testing is easy/inexpensive
- Radon problems can be  fixed
- The State of Maryland has a toll-free number to provide testing
  and mitigation information

     The  community program  involved presentations  to community
organizations,  the placement of posters and  brochures in public
places and a radon awareness week.

     Residents of Hagerstown received the media program only while
residents of  Randallstown  constituted a comparison group used to
approximate a control group. In addition, residents of Hagerstown
and Frederick  received-a utility bill insert that reinforced the
media theme.

     A local television channel, WJLA, carried out a radon campaign
which coincided in time with the efforts of the study, leading to
a rather difficult methodological problem of separating effects of
the television campaign  from those of the intervention.

     In  all three  communities  there were two  samples  of par-
ticipants:  a  panel sample  and  a random sample  interviewed only
after the intervention. The subjects were approached in telephone
interviews. The first wave of data was collected  in December 1987.
Followup data were collected in April 1988.
     '.  On the basis of  the  NYSERDA study,  messages were phrased
more in a command than in a cajole tone.
                                35

-------
     The findings were that

- There was a high awareness of radon, possibly increased  by ttiP
WJLA campaign.

- Testing for radon increased only under the most heavy interven-
tion (Frederick), roughly from 5 to 15 percent, mainly due to the
intervention.

- Mass media  intervention tended  mainly to make people "procras-
tinators",  i.e.  they  now  acknowledged the  problem but  had not
gotten around to doing something about it.

- Knowledge of radon increased (and was the only variable showing
sensitization effects  in the panel samples),  especially  when it
came to general, non-specific questions.

5.8.2. Evaluation

This  is  a  sequel  to  the  NYSERDA  study  in  the  sense that the
information material was designed  on  the  basis of the results in
the previous project.  It  is a major attempt at risk communication.

     The  economic  constraints precluded  extensive use of tele-
vision, but it  is likely, on the  basis of much previous research
on public  service  broadcasts, that  the results  would  have bee
meager anyway (McGuire, 1985). It is well known that public servi^
campaigns on television,  aiming at inducing healthy habits, usuall^
have only marginal success.

     The researchers concluded that the WJLA campaign had probably
only marginal  effects on testing  for radon.  That  statement was
based  on  data analysis  which attempted  to  single out the WJLA
effects and measure them separately from any effects of their own
interventions. It is a bit strange that the official EPA report of
the same study (United States Environmental Protection Agency, 1988
a) states  that  "WJLA's  campaign  was particularly effective in
increasing knowledge,  awareness and testing" (p. 31). It is unclear
what on what  the grounds for this statement are.  The researchers
reported (Desvousges,  Smith  & Rink,  1989)  that WJLA effects were
most clear  for  awareness, doubtful for knowledge and that "there
was no measured effect of the WJLA campaign on testing decisions"
(p. 8-14). Since the official EPA  document and its assertion about
large effects of the WJLA campaign fails to report any supporting
evidence the researchers  seem to be arguing more strongly for their
standing  on  this   issue.   The  WJLA   effects were  small.  This
conclusion is also  in line with a  large amount of previous work on
TV campaigns, see above.
                                36

-------
5.9. Florida risk attitudes study

5.9.1. Design and selected results

This  is  a study of  people  living in  Florida's  three "hot spot"
counties:  Polk,  Hillsborough  and Alachua1  (Valenti  & Ferguson,
1987) . Phosphate  mining is extensive  in Polk, there  is  some in
Hillsborough and  virtually  none in Alachua.  Phosphate mining is
associated with elevated radon levels.

     A sample  of 837  homeowners were  interviewed  by telephone.
Homeowners with  no telephone  or with unlisted numbers  were not
sampled,  which  excluded  about  30  percent  of  the  residents.
Interviews were completed with about  40 percent of the remaining
sample. About 60 percent of the respondents were female. The data
were collected in the fall of  1987.

     76 percent of  the sample were aware  of radon,  males having
heard of  radon  more often than  females. Even though many people
had heard of radon,  most of  them did not feel  they knew much about
it.  They  said they  wanted  more  information,  but did  little to
obtain such information. What they wanted to  know was how to test,
how to mitigate, what level  is safe and where levels are highest.

     Valenti and  Ferguson  point out that  while  people relied on
mass media for  information, these  media did  not have a very high
credibility.  The media create  awareness,  but not. necessarily
concern, and concern is necessary for testing.

     It  was  found  that higher  concern  leads  to  more  blame
attributed to the government  and puts  less responsibility on the
individual. Higher  awareness,  on the  other  hand,  was correlated
with greater perceived responsibility on the part of the homeowner.
The  question is  whether it  is possible  to  increase awareness
without, at the same time, generating increased concern.

     Polk  homeowners  were  most  likely to  say  there might  be
elevated levels of  radon in their county. At the  same time they
were most  likely  to deny that there might be elevated levels in
their own-homesI Males  and  older respondents were most likely to
deny personal risks.

     About 48 percent said they were  very or somewhat concerned
about the  effects  of radon. Less  educated respondents were more
concerned, females more than males, younger more than older. This
matches the pattern of risk denial rather well. Those who were very
concerned were more likely to  say they would test for radon, even
if  a  test would lead to a  drop in market value  of their house.
     1.  In each of these counties  at  least one home had measured
8 pCi/1 or more.

                                37

-------
However, only  3  percent  had  tested their homes for radon.
percent  of  those who  had tested  said  they had  taken  action
reduce radon levels in their homes.

     The subjects were also asked  about responsibility for radon.
Those living in Polk were, as could have been expected, most likely
to blame the  phosphate industry,  but the  most common answer was
"nature" or "nobody". Those  who  were  less concerned  were  more
likely  to  blame nature  or nobody.  In  Alachua,  the  most common
answer about mitigation  was  that the homeowner should pay, while
it was  more common  in Polk and Hillsborough  to  state  that the
government  should pay.  Few seemed to require  that the phosphate
mining industry should pay.

     The respondents were asked  about their confidence in various
potential sources of  information  about radon.  High ratings  were
given  to the  University  of  Florida  scientists,  the EPA,  the
National Centers for  Disease  Control,  and the  American Cancer
Society. Low ratings  were given  to the Florida Phosphate Council
and  "a  company  selling  radon  test kits".  Compared with  male
respondents, females had more trust  in medical sources and media.
More concerned people had more  trust in .medical  sources, in the
media and in experts.  More concerned people also said more often
they wanted more information about radon.

5.9.2. Evaluation

This study  is  relatively ambitious but has some flaws, especially^
in the  low response  rate. The  variable  of  concern is  given  a
somewhat confusing treatment. It is seen as necessary for testing,
on the one hand, and as detrimental because high concern leads to
blaming others, on the other.

     There  may be a  nonlinear relationship between  concern and
testing, with very high levels of  concern being associated with a
decreasing  tendency to test.  Perhaps a high level of concern is
associated  with  a high probability  of  testing and a tendency to
blame  others   for the radon  risks.   At  any  rate,  the  concern
variable needs to  be further  analyzed,  both  theoretically and
empirically. The statistical  analysis in this study was not carried
very far and  it is  likely that  more credible  results could have
been achieved with more ambitious  analysis.

5.10. Florida personality study

5.10.1. Design and selected results

Ferguson, Valenti and Melwani (1988), building  on  Zuckerman's work
on sensation seeking, constructed  a  personality questionnaire for
measuring three dimensions:  adventurousness,   impulsiveness and
rebelliousness. These dimensions resemble  Zuckerman's (1988)
                                38

-------
factors  of  sensation  seeking:   thrill  and  adventure  seeking,
experience seeking, disinhibition and boredom susceptibility.

     Ferguson, Valenti and Melwani found some correlations between
their  personality  dimensions and media  preferences  for getting
information about health hazards. Reliance on television correlated
positively  with  impulsiveness,   while  rejection  of  newspapers
correlated  with the  other  two  dimensions.  Their  scales  were
correlated with health attitude, a measure of concern  for  one's own
health, and with a low  level of concern over food irradiation and
air pollution risks.

     The  survey  also  measured  the  Health  Locus  of  Control
(Wallston, Wallston, Kaplan & Maides, 1976). Persons scoring high
on this index would be  expected to exhibit more concern and more
responsibility  for their  own health.  There were some rather weak
correlations between this  scale and the Ferguson & Valenti scales.

     Ferguson  and  Valenti  (1988) used the three dimensions  of
adventurousness, impulsiveness and rebelliousness to measure risk
aversion. They hypothesized that more risk averse people would rate
risks as larger and that  they would be  especially likely to show
concern if risks were said to be  threatening their children, and
if they were not presented with specific steps to take in order to
mitigate or avoid the risk, and if the risk is presented together
with a simple fear arousing schema rather than   in the context of
other risks.

     They  recruited persons  who  participated in their telephone
survey (see Section 5.9, page 37)  and 706 out of  a total 837 agreed
to participate  in  a panel.  They  were mailed booklets  including
radon  messages.  These messages   varied   as  to alleged  source
(newspaper  vs   government  brochure),   whether  children  were
specifically singled out as being at risk,  whether the radon risk
was associated with nuclear power (simple,   fear-arousing) or with
smoking  and X-rays"  (more elaborated cognitive structure),  and
finally how specific and available were  the  steps to  test for
radon. About  320 persons"responded (38  percent). The researchers
found high risk ratings  when the more complex comparison schema was
activated  (i.e.  smoking and  X-rays)  in  a  newspaper context. They
found high fear ratings  when children were targeted, with  a complex
comparison schema, regardless of  context.  More  information about
radon wa»  requested when  the target  was  an adult rather  than a
child,  especially  if  the source was depicted as  a  government
brochure.  People who were risk  averse requested more information
when the source was a newspaper message with low  specificity. There
were some further interactions between personality and the message
variables, partly different for men and women, but these findings
only partly confirmed the hypotheses. They were complex and seem
to need replication.
                                39

-------
5.10.2. Evaluation

This work is of interest mostly because it is one of the very few
attempts  to  relate personality  variables to attitudes  to radon
risks. The response rate was rather low and the variables seem of
doubtful value since they are based on very few items. The findings
were  suggestive  at best,  and  need replication. There  were many
complex  results,   some  of which  may  be  due   to  sampling  and
measurement errors.
5.11. Onondaaa study

5.11.1. Design and selected result^

Mazur and  Hall  (no date)  conducted a study  of 204 single family
dwellings  in Onondaga  county,  New York.  Radon data were obtained
during  the 1986-87 heating season.  The homes  had been selected
because there were  suspicions  of high radon levels. These suspi-
cions were confirmed,  median  readings being  5.1 pCi/1  and 4.6
pCi/1,  for basement and living areas, respectively. 52 percent of
the  living area readings  were above the .EPA action  level  of 4
pci/l.

     They  found that specific concern about radon, related to t
respondent himself,  was correlated  with information  search
mitigation,  while  general  concern  (concern  without  speci  __
reference to the respondent himself or his home) did not correlate
with  mitigation.  Specific  concern  was  correlated   with  radon
readings (about 0.5) but not with family or media  influence, which
however correlated with general concern.

5.11.2. Evaluation

This study was of limited scope but it did introduce an interest-
ing  distinction between  personal  and  general  concern.  It  is
possible that personal concern can be expected to be stronger for
radon  risks  than  for  many   other  risks   since,  according  to
weinstein, denial  of  such risks is relatively weak.  That may be
the  reason why Mazur  and Hall  found that  specific  concern was
strongly related to technical  risk.  On the other hand,  the finding
is  puzzling since  other  researchers have  failed  to  find  any
relationship.
     1  Some of these correlations were remarkably high, such a
0.76 correlation  between specific  concern and  the  intention
mitigate.
                                40

-------
6. Empirical studies of mitigation

6.1. Introduction to mitigation research

Even if people test for radon and find that  their house has a high
level  of  the gas,  there  is no  guarantee that they  actually do
something  about that  condition. Therefore,  special studies  of
mitigation are  needed. Very few  such studies have been published
so far.

     The problem  of mitigation and risk  perception  is  a special
case of  the problem of the  relationship between  attitudes  and
behavior. People have  been  found in many studies  to behave quite
differently  from what  they  say they will do  or what they prefer
(Wicker,   1969).  Indeed,  Wicker  reviewed data showing  very  weak
relationships between  general attitudes  and  specific behavioral
measures.  On the  other hand,  Fishbein  and  Ajzen  showed  quite
convincingly that there is a strong  relationship between attitudes
and behavior whenever attitudes are measured at the same level of
specificity  as  the behavior.  In  other words,   if  a  specific
protective behavior  is to be predicted from an attitude measure,
that measure should be equally specific.

6.2. Maine study
The first reports on the relationship between mitigation and radon
readings have been disappointing (Johnson & Luken,  1987; Weinstein,
Sandman & Klotz,  1987). People with a higher objective risk level
were not more likely to mitigate.

     According to Weinstein, Klotz and Sandman (1988) "People are
seldom willing to take  preventive measures unless convinced that
their own risk is significant and that^isks would be serious" (p.
796).         -     ~_,

6.3. New Jersey study
One  of the  first major  attempts  to  study mitigation  of  radon
problems was reported  by  Weinstein,  Sandman,  and Roberts (1988).
They followed 123 New Jersey homeowners who had discovered at least
two years previously  (in  1986)  that  they had  a  radon problem, i.
e., a  level  exceeding 4pCi/l. The percentages  of  people in each
exposure category who  did at least  some mitigation  is  shown in
Table III.
      .  Some physical  changes  were required in  order  to qualify
for  the  mitigation  category.  Behavioral  changes,  such  as  more
frequent opening of windows, did not count.
                                41

-------
    Table III.  Percentage  of persons who had mitigated  as  a
    function  of their first  floor  radon reading.  Data  from
    Weinstein,  Sandman and Roberts  (1988).
    First floor reading  (pCi/1)     Percentage who mitigated.

    > 20                                     93
      3-20                                   71
      4-8                                    62
    < 4                                      45
     Note that nearly half in  the group below the official action
level actually did carry out some mitigation.

     Few people retested after mitigation. (A retest is important
because there is no guarantee that the actions taken actually will
reduce  the radon  level) .  A  third of  those who  had  done  some
modifications did  not retest,  and  none  of  those who  relied on
behavioral  methods  (such  as  opening  windows  frequently)  had
retested.

     Of those who had retested their homes following mitigation 9j
percent said modifications  were successful.  There was an averac
drop from 15.2 to 2.0 pCi/1. (First-floor level).  Basement readingl
dropped from an average of 37.3 to 3.9.

     Weinstein, Sandman,  and Roberts estimated that radon remedia-
tion saved between 8 and 13 lives for every 100 households in the
confirmatory monitoring  program with  ratings of  experience and
frequency) and personal experience.1 About 30 percent of the lives
saved were estimated  in homes with  initial radon levels below 20
pCi/1. These estimates should be taken with a grain  of salt, of
course, since  they are based  on a small  sample  and  since these
people were atypical in being among the first to test and mitigate.

     The  best  predictor of  mitigation  was not  the  actual radon
level  measured but how  serious  people  felt  it  to  be.  Other
significant  predictors of  action  were  perceived  likelihood of
health  problems,  of  effects  of radon  on home  property value,
distress du« to worry  over the  radon risk, and perceived personal
susceptibility. The dominant reason for  not acting  was that the
homeowner did not believe that the risk was very  serious. The most
     1.  Several assumptions were behind these figures, of course.
Perhaps the most crucial assumptions were (a)  that people would be
exposed to the measured radon level for a lifetime, and (b) that
mortality is  a linear function of the radon level, starting witlf
2 lung cancer cases per 100 persons  for a level of 4 pCi/1.
                                42

-------
frequently   mentioned   difficulty  of   mitigation   was  getting
information about how radon levels could be reduced.

     The authors  point  out that the group  may be atypical since
they had shown their concern by confirmatory retesting and also by
agreeing to take part in the follow up study.

6.4. Washington study

The most  extensive study  of  mitigation  so far  was  performed by
Doyle et al. (1989). They mailed a questionnaire  to 920 homeowners
in  the  Washington, D.  C., area who had tested  their  homes for
radon. The  tests  had been performed in  the beginning  of 1938 in
response to an extensive television campaign on the WJLA channel.
Test kits  had been made  available  at a  reduced price1  at local
Safeway stores and  through a  newspaper advertisement.  A total of
116,000 kits were  sold,   by  no means exhausting the  demand.  A
testing firm named  Air Check  had  analyzed  the kits  that were
submitted to  them and had reported back  to the homeowners about
their radon levels .

     The test results from Air Check were sent together with some
additional  information.  Homeowners  were told  that  no  mitigation
should be undertaken unless a second test also showed an elevated
radon level. Those with  radon levels exceeding 4 pCi/1 were advised
to perform a second test.  Those with radon levels between 4  and 20
pCi/1 were  sent  a  somewhat  revised  version of  EPA's  booklet "A
Citizen's Guide to Radon11, while those with levels higher than 50
pCi/1 received EPA's booklet "Radon Reduction Methods'1 and  a free
test kit for a re-test.

     Stratified sampling  was employed,  i.e.  an  equal  number of
homeowners was sampled  for the four radon  levels <4  pCi/1, 4-20
pCi/1, 20-50-pCi/1   and >50  pCi/1. The  .77  percent response rate
was quite good. The data were collected in  December 1988.

     In spite of  the strong  recommendation to  perform  a   second
follow-up test, over 85 percent of the  respondents had not done
so. (Due to the design of the study, about 75 percent of the total
sample had obtained radon  levels exceeding  4 pCi/1).

     The results  on mitigation  are  given  in  Table IV.  In this
table,  mitigation  refers  to  any  kind  of  action  taken  by the
respondents, i.e.  including simple behavior changes such as leaving
windows open more  frequently.  Opening  doors  and windows more
     1.  The price was reduced to about half the regular level.

     2.  About 56,000 of the 116,000 that were sold.

                                43

-------
frequently  was,   in  fact,  by  far  the most  common  mitigat;
activity.
    Table IV.  Percentage of persons  who  had mitigated  as  a
    function  of  their radon reading. Data from  Doyle  et al,
    (1989).
    Radon level  (pci/l)     Percentage tfho mitigated

    > 50                                     49
      20-50                       .           42
      4-20                                   14
    < 4                                       3
     These figures are  considerably lower than those reported by
Weinstein, Sandman and Roberts (1988),  see Table III. However, the
latter data refer mitigation  likelihood to first floor readings,
while  Doyle  et  al.   did  not  analyze  their  data  for  differe
responses contingent  on where  in  the house the test kit had be
placed. The data obtained by  Doyle et al. suggest that most te
results refer to basement readings.  It is natural to expect more
concern about a  first floor reading than about  the same reading
from the basement.
= *
i
     Still,  the differences  between the  two  data  sets  are so
extensive  that it  appears  likely that  other  factors  also have
affected the samples differently.  Most important, the Weinstein et
al. sample, consisted of highly motivated  people who were among the
first to  test their homes,  and they were  followed  for a longer
time1 than the Washington group, which was probably more typical in
their level of involvement in the issue of radon risk.

     Doyle et al.  also reported that the reasons homeowners gave
for mitigating their homes were predominantly health related, very
few  referred  to  economic  value and  homesale  consequences.  Of
course, few may have been considering selling their homes at this
particular  point  in  time  (homes  are   sold after  10  years  of
residence, as a national US average).
     '.  The time factor may have been unimportant.  Less than  1 pei
cent of the respondents in  the  Washington study said they planned
mitigation in the future.
                                44

-------
     Doyle at  al.  estimated the effects of  the  WJLA campaign in
terms of how large  a  share  of the  population of  homes in need of
mitigation was actually  mitigated in  a  "credible"  manner.  The
results were about  0.1  percent for houses  with a level between 4
and 20 pCi/1, 0.5 percent for those above.  About half of the test
kits were returned  for  analysis, and  some  kind of mitigation was
performed in slightly less than 50 percent of the cases where the
level  exceeded  20 pCi/1.  However,  only  about  30 percent  of
mitigation went beyond habit changes such as leaving windows open
more frequently. In addition,  those with radon levels between 4 and
20 pCi/1 were much less likely to mitigate, and that group is, of
course, the  major "problem"  group  since  there are  so  many more
homes in this group, even if those  with higher readings constitute
a larger risk for the individuals concerned.

     Summing up,  the  net result in terms  of credible mitigation
was meager. There were two major reasons for this finding:

(1) Few people bought test kits, only about 6.5 percent.

(2) Only a minority mitigated.  The percentage of homes mitigated
was especially low for radon  levels between 4 and 20 pCi/1.

     Doyle et  al.  conclude that "the  likely credible mitigation
resulting from  the program has been so small  as to suggest that
such programs may  be  a  very expensive way  for society to achieve
mitigation"  (1989, pp. 164-165).


6.5. Colorado study of testing and  mitigation at  time of home sale

This is a study by Doyle et al. (1989) of recent home buyers. The
purpose was  to determine  how common testing  is at the  time of
buying and whether such testing leads to mitigation. The authors
also report some information about  common practices among realtors
in  several  States.  Their  own data  were  obtained  in  Boulder,
Colorado. There had been no prior extensive attempts  to inform the
public about radon in that state.
                        L -.' ;
     Doyle et  al.  reported the results of  a telephone survey of
100 home  buyers in Boulder.  They  had to  approach  210  buyers in
order to get 100 interviews, so the response rate'was  slightly less
than 50 percent. The houses had been bought in the fall of  1988 and
the interviews were made in December of that year.

     The number who had tested for radon  in connection with the
sale was 43, most before the  sale closed. Of those who had tested
for radon, 8 reported some kind of mitigation and 5 of those had
retested to check the outcome. Of the 8 mitigators, 7 had had close
contact either with an employer (IBM)  who strongly supported radon
testing  and  mitigation,  or with a  realtor or  a contractor who
provided expertise  in the area.

                                45

-------
     Doyle et al.  pointed out that the overall mitigation rate
this group  was about  15 percent,  to be  compared with  the o.l
percent obtained in the Washington study. (Data refer in both cases
to  those homes  that  had radon  levels  > 4  pCi/1).  Thus,  the
effectiveness  for  stimulating  radon testing  and  mitigation  of
buying a  house  was about 150 times  as  great as  that of  a major
media campaign.

6.6. Some Swedish experience; the Sollentuna study

Concern about radon started earlier in Sweden than in the USA. Alum
shale was frequently employed as a building material before 1975,
and  the  initial   concern was  with  radon emitted by  building
material. Subsequently, geological radon has come into focus. The
National Institute of Radiation Protection has estimated that radon
causes  about  1100 lung  cancer  cases  in  Sweden  yearly  (in  a
population of 8.5  million).  The  figure  is much higher per capita
than the  corresponding  American  estimate of a maximum of 20 000
cases per year. This is somewhat puzzling,  since data indicate that
the home radon levels are probably somewhat higher in the USA than
in Sweden, see page 12, Table I.
     The action  level  recommended by the Institute is 10  pCi/11,
in other words 2.5 times that of the EPA. It is unclear why Swedis
authorities  are  so much  more prone  to accept radon  risks.  T
Swedish action level corresponds to the lung cancer risk of smoki
a pack of cigarettes per day .
• ,
I
     Local authorities inform home owners that this level involves
an  elevated  risk  of  lung  cancer.  The  local  authorities  are
responsible for dealing with radon risks, but their policies vary
greatly. Swedes are just as indifferent to mitigating their homes
for radon as are Americans (Johnson, 1987).
                                          ^ •
     Akerman (1988) reported a study of mitigation and willingness
to pay for mitigation in Sollentuna, a suburb north of Stockholm.
Radon had received considerable attention in this  town. Home owners
could have the local health department conduct a  test and the cost
was 400 SEK, about $65. If the result was above 10 pCi/1 the health
department offered a free retest and provided mitigation advice.

     Akerman obtained  extensive  data from a sample  of  317 homes
that had been tested with results above 10 pCi/1.  The estimated WTP
      .  In Sweden the scale used is that of  Bq/m .  The conversion
formula is 1 pCi/1 =40 Bq/m3.

     2.  However,  a recent  decision  states that the maximum allowe^
concentration in new construction  is 2 pCi/1.                   ™

                                46

-------
in 1986 prices (yearly cost) was 3120 SEK ($510) ,  or 918 SEX ($150)
per  person  since  there  were  on the  average  3.4 persons  per
household.

     The data in this study were obtained  from archives'. There was
no information about  actual mitigation costs, but  experts estimated
the generic costs on  the basis of  detailed  information about radon
readings and technical characteristics of houses.

     The implied WTP of these data was $0.35 for  reducing exposure
by  1  Bq/m per person,  or $13  for a reduction  of 1  pCi/1.  The
general policy of  the National Institute  of Radiation Protection
is that one should be willing to pay 43.50 SEK per annum for a
reduction of 1 Bq/m  ($88.50 for 1 pCi/1),  roughly 6 times as much.
It  is  possible that  this difference reflects the often noticed
difference   between   individual  and  collective   risk  taking.
Collectively taken risks are accepted only  at a  level an order of
magnitude  less  than   that  accepted for individually taken risks
(e.g.  Sjoberg &  Winroth,  1986).  Akerman,   Bergman  & Johnson (no
date) argue that the  estimates are understatements  of the true WTP
values since (a)  all  costs could not be taken into account, and (b)
it is  unknown what risk perceptions  the homeowners actually had,
but likely that they  perceived smaller risks than the expert risk
judgments  on which the  calculations  were  based.  As the authors
stressed,  further work on WTP should make use of  individuals' risk
judgments rather than expert estimates. The problem that will then
arise  is,  of course,   that  unbiased  risk judgments are  very
difficult to get.

     WTP methods can be discussed. Do people really give realistic
estimates of  their willingness to pay?  Oicke,  Fisher and Gerking
(1987)  compared  actual  purchases (of strawberries) with stated
willingness to buy. They  found rather good  correspondence between
the two sets of data. However, we still—do  not know how realistic
such estimates are in different settings  and with more important
choices,  such as  those  involving health  risks. It may  be  that
people are less^likaly to give such ratings in an unbiased manner.
Strategic considerations may easily enter.
                                                     -re
     Akerman  initially  found: only  a  weak correlation  between
mitigation cost and initial radon reading.  This  result is in line
with  some American data,  perhaps most  clearly  the Maine study.
However, after controlling for several independent  variables, most
notably Jcind  of  ventilation1, she  found  a clear relationship
between WTP  and  probability of  mitigation and   initial  radon
reading, see Table V  for  the latter relationship.
     '.   Three  types  were  distinguished:  passive  ventilation,
forced-air exhaust  ventilation and forced-air intake and exhaust
ventilation.

                                47

-------
    Tibia V.  Percentage of  persons who  had mitigated asa I
    function  of  the  initial  radon reading.  Data from Akerman
    (1988) .
    Radon level  (pCi/1)     Percentage who mitigated

      12.5                                    38
      25.0                                    50
      37.5                                    57
      50.0                                    62
      62.5                                    65
      75.0                                    63
      87.5                                    70
     Bergman, Soderqvist and Akerman (1988) also reported a study
of  sales  prices as  a  function of  radon.  They did not  find any
significant  influence  of  radon on  price level,  confirming the
impressions by American realtors reported by Doyle et al. (1989).
However, Bergman et  al. also  pointed out .that the tax assessment
of  houses  had been lowered whenever there was an elevated radon
reading,  leading  to a somewhat smaller tax.  The  tax "discount"
could have been enough to  compensate for the presence  of radojfl
risk.

     Some of the Swedish  results are different  from American data.
The overall picture in Akerman's study was one  of a certain degree
of  rationality.- The reason could be, of course, that her WTP data
were  based not on homeowners'  actual investment  in protective
behavior, but in experts' estimates of generic costs. Presumably,
experts are more rational than p'eople in general. The reason could
also be that Swedes, especially those  in  places  where radon has
been   intensely  discussed,   such  as  Sollentuna,   are   more
knowledgeable than Americans about radon,  or  that they have access
to better advice. Finally, the  reason could be the smaller income
variation in Sweden,  which decreases the confounding effects of how
much money people  can  afford  to spend on mitigation.  Also, state
subsidized loans are available to homeowners in Sweden for radon
mitigation, and even ordinary  bank loans may be cheaper than in the
U.S. because interest is deductible up to 50 percent.

     It should  be  added  that  rationality  is not overwhelming in
AJcerman's data. 30 percent are estimated to ignore mitigation with
an exposure level of 87.5 pCi/1. At this level,  the estimated added
risk of lung cancer approaches 1/2, corresponding to about 10 packs
of cigarettes per day for every family member,  including children.
Willingness to pay data imply an implicit  life  value of some $6000
per family member  at this  level of exposure, a small fraction of
official figures both in Sweden and the USA.


                                48

-------
     The  estimated  over-all implied  life  value,  according  to
Akennan, Bergman & Johnson (no date), was only about  $100 000. This
compares unfavorably with the range $2  million to $7 million per
statistical life save reported for American data  (Fisher, Chestnut
& Violette, 1988).

     The reason could be  the above mentioned risk denial that was
probably present,  or it  could be that people do not believe that
a reduction  of radon below  the  action  level  (10 pCi/1)  has any
positive health effects.  The Swedish information  material does not
mention that a further reduction is of any value.

     A further possibility, not mentioned by the authors, is that
human life simply is less valued, at least  in monetary terms, in
Sweden than  in the USA.  Damage compensation  awarded  by Swedish
courts for loss of life  or various  types of smaller damages tend
to  be  1-3  orders of   magnitude   smaller  than  the  American
corresponding  figures.  It  seems,   however,  that comparisons  of
countries are  quite complicated because  Swedish authorities are
likely to react to prevent damage,  even if  they do  not feel that
individuals should be compensated financially to any great extent
once damage has occurred.


6.7. Evaluation and conclusions of mitigation studies

A major  factor in mitigation is,  of course,  the measured radon
concentration  level  of a  home.  The three studies  of mitigation
discussed above all produced some quantitative information about
mitigation frequency as a  function of radon concentration. They are
plotted in the same graph1 in Figure 2   .

     It can be seen that  the data from  the  study by Weinstein et
al. deviate from  the two  other  data sets in yielding much higher
levels of mitigation. As pointed out above (see page  44) the values
obtained in the New Jersey study refer to  living  room readings and
may  therefore  be misleadingly  high.  In  addition,  that  study
involved a group  of people who  were probably unusually strongly
motivated.

Research on mitigation has so far not been very extensive and most
of the experience is very recent. However,  some observations can
     1.  *-data from Weinstein  et al., *-data from  Doyle et al.,
••data from Akerman et al.

     2.  In the plot of Figure 2 the minimum and maximum levels of
radon  concentration  for  the  two  sets  of American  data  were
estimated by  adding  or subtracting 20% from the  stated  lower or
upper limit, respectively.

                                49

-------
                      1.0


                      oa


                      00


                      04


                      02


                      00
                               ao
oo
                             RM£J> ocrantralon In OQA
  Figure 2.  Frequency  of mitigation  as  a  function of  radon
  concentration.

be made.

(1). Given  that  people have tested and  found an excess level  of
radon  it is  reasonably likely that  if they they  are  going  to
mitigate,  they  do  it within  one  year.  Ho  data  support the
assumption that they may delay mitigation  longer than that.

(2) The  probability  of mitigation has in almost all  studies been
found to be an increasing function of initial radon  reading.
                                            •
(3)  In  a  few  studies  there has  been  an  obscure   raw data
relationship between initial radon reading and mitigation  behavior.
The  reasons could  be  (a)  technical  characteristics of  houses,
perhaps  in  particular type  of ventilation system, (b) the income
variation in the sample, (c) availability of expert advice  and  (d)
variability of risk preferences.

(4)  People  seemed less likely to  re-test to  find  out  if  their
mitigation had been effective, with the possible exception of those
who had just bought a house,  or were in the process of buying one.

(5) The net effect of  media  campaigns on the  number  of people who
mitigate  is  probably  quite small.  The  effect  of encouraging
realtors or employers  at the time of buying a house  seemed in one
small-scale study to be 100-200 times larger.
                                50

-------
7.  Comments  on  empirical  studies of  radon risk  perception
mitigation

Early concern that people would panic over radon has clearly been
found to  be unsupported.  The opposite  is  true: people  are  too
little concerned about radon risk.

     There have been  three major attempts to  investigate  risk
perception and  testing decisions: the NYSERDA,  Maryland  and  Nev
Jersey studies.  Studies  of mitigation have so  far been few,  and
more such studies are needed; some are under way.

     The most successful variables in predicting testing found so
far are those  that measure (a) personal  risk  perception  and  (b)
social influence. Personality variables have not been found to be
very important  when it comes to the perception of radon  risk or
testing decisions.

     The most  common problem in  these studies has been  that of
representativeness  of the  sample  studied.  Probably  because  the
problem is so new  and research resources have been limited there
have  been  few  attempts  to  test  variations  of  messages  in
communication   studies.  It  seems,   however,   that  simple  and
straightforward  messages are  to be preferred.  Messages that  are
too simple can be detrimental  if the purpose is  to reassure people
that the risk  is negligible.  On the other  hand,  the most common
problem is the  opposite one:  how to motivate  people to test and
mitigate if appropriate. Therefore,  it is possible  that  one should
consider quite brief and simple messages.
                                                               •
     Are people at all rational when it comes to the perception of
radon  risks? Psychologists have tended to  emphasize that,  in
general,   people  perceive   risks   incorrectly.   Depending   on
circumstances,  they may  exaggerate  or understate risks. However,
in the Maine and, in particular, the NYSERDA studies it was found
that people  revised their risk estimates in  accordance with the
information that they received.

     This  finding  is not inconsistent with a general skepticism
about people's  subjective risk perceptions. In the cited studies,
there was still virtually zero correlation between subjective risk
and measured home pCi/1 level. Maybe people have such a distorted
view of risks partly because they  seldom  get clear information and
feedbacfc about  risk levels.  The  finding that  they would revise
their risk estimates in a correct  direction  suggests that they sao
treat risk information in a correct manner.  The  problem may simply
be that they seldom receive such information.
                                51

-------
8. Predicting monitoring for radon

Some  of  the studies  treated  above investigated the  tendency of
people to monitor  for radon.  It is a potentially important issue
to determine whether  it is possible to predict who is most likely
to test. However, the problem is difficult to investigate because
so few people have tested, less than 4 percent in the population.
A statistical analysis is not meaningful if the number of testers
is so few as, e.g., in the Florida study (Section 5.9, page 37).

Weinstein finds the following predictors of the intention to test:

- believing that radon is a personal risk
- believing that others are concerned about radon
- a feeling of distress caused by worry over the radon risk
• having more children
- being better educated
- having lived fewer years in home
• being a nonsmoker

     Weinstein also found that several  factors were not predictive
of testing intentions:

• knowledge about radon
- perceived effect of radon on home values
• perceived cost of remediation
- perceived severity of risk
- perceived efficacy of mitigation

     Note especially  the  last two factors  which should have been
efficient according to models of protective behavior, such as the
Health Belief Model (Becker, 1974).

     Those  who  had never thought about testing knew less about
radon. They were quite different from the ones who  had thought
about the issue and found it unnecessary to test.

     Further analyses of testing behavior were reported in the EPA
Interim  Report on the  Radon Risk  Communication Project  (U.S.
Environmental  Protection Agency, 1987  b).  These results  were
derived  from the  New Jersey data  collected by  Weinstein and co-
workers  and  from  a University of  Pittsburgh study of people who
had purchased test kits (N»70,000) or been provided with such kits
at no cost  (N-3,500).

     Major  findings  from New Jersey data,  in addition  to those
already stated, were that

- Testers were affluent and middle-aged.

- People who planned  to  move  within the next two years were more
likely to test for radon.

                                52

-------
- Households in which someone usually sleeps in the basement were
more likely to test.

- Households with children under  10  were slightly more likely to
test.

     Findings based on the University of Pittsburgh study were that

- States  that contained radon  "hot  spots" had more  testers,  as
compared with states that had no such hot spots.

- People who suspected elevated levels before testing tended to be
right.

     The Maryland study (Section 5.8, page 35) reported only three
significant predictors of spontaneous monitoring:

- Level of knowledge about radon1.

- Whether people  asked their physician  a  lot  of  questions about
their health.

- Educational level.

     In conclusion, the following factors have been documented as
being of some importance for the testing decision:

- Demographic  characteristics:  young or middle-aged  people,  and
parents, tend to be more inclined to test.

- Nonsmokers are slightly more inclined to test than smokers.

- The  perception of  a  risk as personal rather than  general is
associated with testingv

-  The  perception  that  others are  concerned  about  testing  is
associated with testing.

- A  fealing of distress over the radon  risks  is  associated with
testing.

- General health concern
     '.  The causal  direction here is debatable.  Other data show
that people do not become more likely to monitor if they get more
information. Possibly, the process of monitoring by itself involves
learning about radon.

                                53

-------
     These findings suggest that risk communication should
personal  risks and  that  social  diffusion  processes  should
employed, whenever possible.

9. Factors affecting risk perception

Few people seem to be  concerned enough  about radon risks to test
their homes and to mitigate.  One  of the most important variables
responsible for this indifference is the perceived risk, which is
low. Therefore, an  analysis of risk perception  is called for in
order to give a background for improved risk communication.

9.1. Risk denial

Risk perception has often  been studied  within the perspective of
cognitive  psychology   and  heuristics   (Sjoberg,   1979).  This
literature is  probably less relevant for  understanding of radon
risk where other factors seem more important.

     People tend  to be indifferent  toward abstract  threats  to
society  (Snarlin,  1987), but  concerned if  threats  are oriented
towards themselves or significant others. At the same time, there
is often denial of risks to oneself, a phenomenon well documented
in health psychology (Weinstein,  1980, 1984, 1987).

     People deny many health risks, among them the risk from
This is an example of a general trend toward exaggerated optimi
when it  comes  to  one's own person  (Perloff,  1983).  Why is there
such a tendency? Kunda  (1987) suggested that people construct self-
serving theories, or evaluate evidence  in a self-serving manner,
and that this  process  is motivationally guided.  It is relatively
easy for most people to come up with ad hoc causal "explanations"
for anything,  after the fact has occurred  (Anderson & Sechler,
1986). At  the  same  time it is increasingly difficult to predict
anything on the basis  of such construals because people generate
such theories  even  on  the basis of  a single instance  (Anderson,
1983).

     Weinstein (1987)  investigated  several factors that had been
suggested  as  causes  of  optimistic bias  in  risk  estimation:
defensive  denial,  self-esteem (as  reflected in  ratings  of risk
preventability and  embarrassment  if exposed to  risk), cognitive
errors (revealed through the correlations of bias with ratings of
experience  and frequency)  and personal  experience.  The latter
concept was measured by responses to the statement: "If you haven't
had this  problem  by the time you're my age, it's not likely to
happen to you".

     Weinstein obtained data from a random sample of 296 New Jersey
residents  (survey  response  rate  68  percent).   They rated  32
                                54

-------
 different health  hazards, making comparative risk judgments1. They
 also  rated the hazards in several  other respects.

      There was no evidence  of defensive  denial.  The  strongest
 determinant of optimistic bias seemed to be personal  experience.
 Variables .related  to  self-esteem  were important  and  cognitive
 errors  seemed  to  enter as  well.  Optimistic  bias  was  largely
 unrelated to  background factors such  as  age,  gender,  race  and
 education.

      Weinstein stressed that  if people believed that there was  the
 slightest chance  that their characteristics (lack of certain risk
 factors)  might enhance  their chances  of avoiding disaster, they
 took  full advantage  of  the opportunity  to  create  optimistic
 beliefs.  In doing  so, they  utilized  specific  factors that were
 characteristic: of their situation. To counteract optimistic bias
 it will   therefore  be  necessary  to  individualize  the  general
 knowledge that people  have about a, risk,  and at the same time avoid
 the influence  of  individualized wishful thinking.

      Weinstein,  Sandman  &  Klotz  (1988)  argued that  radon risk
 differs from  many other  health  risks  because it  (a)  .does  not
 threaten  self  esteem, and (b) is hard to deny  on the  ground that
 the risk  is low because no symptoms  have yet been observed.  (Lung
 cancer  does not occur until  late in life anyway, so the argument
 would  be  inconsistent with medical  science). However, these
 arguments have not been tested empirically. Even if an  owner  who
 bought  his or her house before 1984 cannot,  logically speaking,  be
 blamed  for having a house with an elevated level of radon, it  is
•still  quite  possible  that  the  owner  blames  him-  or  herself.
 Moreover,  despite medical science, people may  argue that  "I have
 lived in  this house  for 30  years and  I have  not developed lung
 cancer  so I don't believe there is a risk"  as a reason to justify
 lack  of  action.   Weinstein  found  that such   reasoning  was  the
 strongest determinant of optimistic  bias.

      Weinstein also found clear evidence of risk denial.  This  was
 apparent  in several of his variables.  For example,  only  6 percent
 of those  who had an opinion  said their radon level was  likely  to
 be more   or  much  more  than the  average  in  their  community.
 Optimistic bias  was  evident in  the  overall  reasoning  of  the
 subjects,   who  used  their knowledge  about  radon  to  construct
      .  The format of  the rating scale was:  "Compared to  other
 men/women my age,  my  chances of getting the problem in  the  future
 are:  much below average,  below average,  ...etc... to much  above
 average".  Seven rating categories were used.  The  responses were
 scored  on a scale  from -3  to +3,  0 indicating  a  rating  of average
 risk.  Evidence of optimistic bias was available at the level  of
 group data whenever average  ratings were <0. Such was the case  for
 31  of the 32 health hazards.

                                55

-------
optimistic scenarios. Among those who had tested for radon, th<
was only a very slight correlation between the judged seriousnel
of the  risk  and the radon level obtained. A similar finding was
reported by Johnson and Luken (1987).

     Weinstein et al. proposed that the  reason why radon risks are
underestimated,  while many  other  environmental hazards  (waste
disposal, nuclear  power,  etc)  are commonly  overestimated  by the
public, is that nobody can be blamed for radon  risks.  This is an
interesting hypothesis, but  so far little has  been presented to
support  it.  In addition, it is not  clear why nobody  should be
blamed  for  the radon  risk.  One could  argue, for  example,  that
planners and  realtors  should have foreseen  the  risk and avoided
building on  sites where  radon  is  likely to be high, or  that a
homeowner should have taken action to reduce the risk. Maybe that
type of blame is unlikely because the risk is so new.


9.2. Diffusion processes
                                                               >*> 0
                                                               i
Risk perception  has been found  to be quite  sensitive  to social
influence such as messages from friends and neighbors. Part of the
spread of  risk attitudes in a population  is therefore dependent
upon social  diffusion.  For social diffusion  to work,  homeowners
must perceive their neighbors to be concerned about radon and the;
must interact with them about the issue (Unger &  Wandersman, 1985)
In the New Jersey study,  people reported a  lack  of knowledge abo
how  their  neighbors viewed  radon  risk  and  a need  for  such
knowledge. Furthermore,  if  they  reported  an opinion, they tended
to say that  they saw their  neighbors as  little concerned. People
who had tested wanted to tell  others about it,  however, and more
often so if  they had obtained  high radon  level test results. But
they said their neighbors did not want to listen.

     Desvousges,. Smith and Rink  (1989) found that one of the best
predictors of testing for radon was whether a person had talked to
someone  else  about  radon.  This  finding  is   open  to  various
interpretations. It  is plausible that testing itself is the cause
of some  verbal  interaction.  On the other hand, other research
strongly  supports the  interpretation that  personal interaction
among people is conducive to behavior change.

     According to Rogers  (1987) new precautions are by themselves
much less  likely to  spread through  a population than  other new
ideas,  because  they entail  immediate  costs  and  no  immediate
benefits. This is in addition to the  negative  experiences of those
people who actually h£y£ tried to talk to their neighbors about
radon testing; they  feel stigmatized.

     Diffusion seemed to  work  in the HJ  study,  but there was too
little "input" for it to spread widely. There were too few people


                                56

-------
who had tested for radon, and those who had tested were sometimes
discouraged from talking about it with their neighbors.

     Mazur and Hall  {no date)  and Mazur  (1981)  argued that risk
perceptions are especially amenable to social influence when they
concern risks that are abstract and not immediately present, while
imminent hazards are perceived in a fairly realistic and rational
manner. People are influenced, if risks  are distant and diffuse,
by the opinions their friends have,  and by mass media.


9.3. Assimilation-contrast model of risk perception

Sjoberg and Drottz (1988) studied various groups of nuclear power
plant personnel and found an  inverse U-shaped relationship, across
groups, between perceived everyday life risks, e.g. traffic risks,
and job risks!. The maximum rated general (everyday life) risk level
was found  for  those who had rated  their job risk as  being  of a
medium size job risk. For low to medium size risks, the correlation
between perceived everyday life  risks and job risk was positive,
and for medium to high job risks it was negative.

     Sjoberg and Drottz suggested that a threat of low to medium
strength increases  all risk estimates by assimilation (enhanced
similarity) while a medium to very strong threat will so dominate
a person's perception  of  risks that  other risks appear as small,
a contrast effect.

     The assimilation part of the model has been supported in other
work.  Johnson and  Fisher  (1989)  found  that  when people  were
informed about radon risks (which happened to be quite low in their
sample) their perception of these risks dropped, but so did their
ratings of other risks.  (The opposite would occur,  according to the
model,  if  people were to be  re-assured about  a risk they had
greatly exaggerated).

9.4. Risk posterior to action

There is considerable evidence that risk perceptions correlate with
action. It is usually  assumed  that  risk  perceptions are prior to
and partially cause action. Mazur and Hall (no date) suggested the
opposite relationship, i.e.  that risk perception is a function of
action,  or  lack  of  action,  and  that   highly discrepant  risk
perceptions may  serve as justification  for lack  of  appropriate
action.

     Although this interpretation may at times have some validity
it seems unlikely that it would be generally true.
                                57

-------
9.5. Objective vs subjective risk

Most risk researchers seem  to  agree that  people's  responses to
threats  and  hazards  are  more  determined  by their  subjective
perception of risk than by the objective,  or true,  level  of risk1.
Mazur  and  Hall  (no  date)   suggested  that   this  psychological
explanation may sometimes be incorrect and that objective risk may
directly drive behavior when the risk is imminent and the response
specific. With more distant or diffuse hazards and more generalized
responses the psychological model would  still  be expected to hold,
according to these authors.

     These notions  can be debated.  First,  it is  unclear  how an
objective  risk could  have  an  effect  on  behavior  except as  a
subjective experience. If  the objective risk is found to  "drive"
behavior,  this is  consonant  with a  subjective  risk that is an
accurate reflection  of the objective risk.  Second, a high degree
of correspondence between  subjective and  objective risk requires
freedom  from emotional  interference with  risk perception.  Such
interference is especially likely whenever the risk is imminent and
large. Denial  of risk  has  been  found to prevail quite generally,
especially for very threatening risks.

     Hence,  it does not seem likely  that  objective  risks  are of
direct relevance except  possibly when  the  risks  are moderate or
small. Information about objective risks have  rarely been found toj
affect people's behavior in published research on the issue.

     There  is a  possibility that the lack  of  effect of  risk
information is due to  an illusion. First,  researchers may be most
concerned  with problematic  cases where  people  have failed  to
respond. Second, social concerns about  lack of response may arise
because a minority  of  people neglect warnings, while most people
do respond in an adequate manner.  Lack of response to warnings and
even to  legislation in even a small  minority  of  people  may be a
social problem,  partly because they may  constitute a  danger to
others (e.g., drunk drivers). Hence, concern about  lack of response
to risk information may be confused with its prevalence.

     It  should be  added that even if  these  arguments  have some
validity in general they do not  seem  to  pertain to radon. Radon is
certainly a case where most people have, so  far, failed to respond
to warnings.  Objective risk does  not appear to be  important in
affecting behavior in  this area.
     1  I bypass the often tricky problem of how to define true risk
  see Freudenberg's (1988) stimulating discussion.
                                58

-------
9.6. Demographic characteristics

Sex differences  are commonly found in risk  studies.  There is no
clear evidence that men  and women experience different levels of
concern when  it comes to  radon,  but some  findings  suggest that
women are more worried over health risks while men are concerned
over the  economic  aspects,  such  as  potential loss  of property
value.  Women seemed to be  better informed about radon risks than
men (U.S. Environmental  Protection Agency,  1987 a).  Couples with
small children,  or expecting a child, seem  to be more concerned
about radon than others.

     Long torm residents in an  area are  usually found to be more
skeptical about the necessity of radon testing than those who_hav«
lived a shorter  time in  an area.  Of course,  those who have lived
longer in an area would tend to be older. They might also'be more
personally involved with their homes and have  a  longer story of
benign exposure to  their environment.

     There are some weak effects on testing of education and non-
smoking, and from having small children;

9.7. Cognitive processing limitations

It has been shown in many studies that people have limited ability
to  process  information  (e.g.  Anderson,  1985)  and that  their
attitudes tend to be determined by a few  salient attributes rather
than a complex net  of many aspects  (Krosnick, 1988)•

     These findings are  somewhat difficult  to reconcile with the
well-known  list  of  risk  dimensions  of Slovic,  Fischhoff  and
Lichtenstein (1981). These  authors suggested  a fairly large number
of risk dimensions. On the other hand, Sjoberg and Winroth (1986)
investigated the moral dimension of risky actions and found that
it  accounted^  for .a very  sizable  proportion of  judgments  of
acceptability of risk* The  predominance, of morality seems to be in
line with general principles of cognitive psychology in the sense
that one single factor accounts for a large share of the variance
of rated risk acceptability.
9.8. A note on terminology; Hazard and outrage

Sandman,  Weinstein  and Klotz  (1987)  introduced a  distinction
between hazard and outrage. Hazard refers to "risk estimates based
on mortality and  other 'objective1  data"  (p.  106),  while outrage
is  based  on  "subjective"  factors.  Sandman  et al.  mentioned
especially  the factors  "fairness,  naturalness,  familiarity and
controllability"  (p. 106).
                                59

-------
     These terms seem to  denote  much  the same as the more
terms subjective and objective rislc. The authors stated that the
wanted to  replace  the traditional terminology  in  order to avoid
encouraging the experts "to see the public as mistaken, misguided,
wrong in its subjective risk assessments" (p.106).  Although their
objective was  worthwhile,  their choice  of  new terminology seems
less than perfect. According to Webster's Ninth New Collegiate Dic-
tionary (1987) outrage means

(1)   An act of violence or brutality.
(2 a) Injury or insult.
(2 b)  An  act that  violates accepted  standards of  behavior  or
      taste.
(3)   The anger and resentment aroused by injury or insult.

     The intention to convey less irrationality by switching from
the  term subjective risk to outrage  hence  seems not to  be well
served by the word outrage.  In  addition,  outrage connotes a rather
strong, emotionally flavored response. Neither of these components
is necessarily true of subjective risk.


10. Risk perception and protective action

One  persistent problem  is  the  lack  of correspondence  betwe*
knowledge and  attitudes on  the one  hand and action on the othei
People  frequently  agree  that  many types of behavior  should
avoided, yet  find it very  unattractive and difficult  to  do so."
Examples are provided by all sorts of addiction.  People agree that
they should test their homes for  radon or change their food habits
but delay such rational and health-promoting actions indefinitely.
The question is why people act this way.

     Sjoberg ^studied addictive  behaviors (smoking, drug addiction,
alcohol abuse .jand excessive eating),  in particular  relapses  in
addiction [see^Sjoberg (1980,  1985) for  reviews  of this work].  Re
found that people.who try to  live up to a  long term commitment,
such  as sobriety1;-frequently  fail  when  they are  emotionally
agitated. Emotional agitation seemed to  be equally disruptive when
it was negative as when  it was positive. The most-crucial factor
in  the  regulation  of  action was  the  ability to  counter-act
emotionally induced short-sighted behavior.

     This kind of behavior was mediated by cognitive deterioration.
People were very good_ at producing excuses for lapses and virtually
never told themselves in  a cold  and rational manner that they no
longer needed to adhere to their commitments.

     Although  this work  is only  indirectly relevant  to health
protective behavior when it comes to radon  there may  some simila-
rities that make generalizations  plausible,  at least for some type*
of behavior that neglect health.  The difference is, of course, thfl

                                60

-------
existence of strong positive attraction in the case of addiction.
Procrastination seemed  to  be a dominant behavior  pattern in the
Maryland stxidy (see Section 5.8, page 35),  and procrastination is
a typical form of irrationality (Silver & Sabini, 1981).

     Weinst«in  (1988)   suggested  a  model  of  stages of  health-
protective behavior.  He contrasted  this stage-oriented,  dynamic
approach with traditional decision-oriented models. Models such as
the  health belief  model  (Becker,  1974)   and  variants,  such  as
Fishbein's theory  of reasoned  action (Fishbein & Ajzen,  1975),
assume that people integrate beliefs and values according to some
simple  decision  rule*  usually a  multiplicative one.  They  then
choose  the option  with the  largest, criterion  value,  e.g.  the
largest expected utility, and act accordingly.

     These nodeIs seem  plausible;  perhaps  they indeed are little
more than common sense.  They  imply that  once people  are credibly
informed about the importance of a  given type of health protective
action, they should behave  rationally. But  people often do not act
rationally, at least not in this sense of the word. Decision model
approaches  are   virtually  useless  for   understanding   health
protective behavior.

     Weinstein's stage model is an  attempt  to account for decision
making as a process involving several stages. It  is broader in its
conception of: the  important  factors in action, as  compared  to
traditional decision models. It includes the following stages:

(1). "Has heard of hazard"
(2). "Believes in significant likelihood for others"
(3). "Acknowledges personal susceptibility"
(4). "Decides to take precaution"
(5). "Takes precaution"

     Although it. still  largely untested, Weinstein's model seems
rather promising and it is  certainly an interesting alternative to
the  traditional  health  belief model. It  is,  however, mainly a
cognitive model* Weinstein argues  for a cognitive model by stating
that "new preventive actions usually  involve high-level cognitive
functioning and advance planning"  (p. 358). In the case of radon,
it is doubtful if one  can  classify the  needed actions in testing
one's  home   as   "high-level   cognitive  functioning",   although
Weinstein illustrates his model with the radon  case. The fact that
a number of steps are postulated does not mean  that there is high-
level  cognitive  functioning.   In  addition,  even if  high-level
cognitive functioning is called for, one can still question a model
that generally ignores emotional influences. Such influences seem
to be especially likely to occur in steps 3 and 4. There is reason
to believe that emotional influences  are more important when more
sophisticated cognitive functioning is involved.
                                61

-------
     In spite of this criticism, Weinstein's paper is recommend
for several insights and suggestions, including the importance „„
prompts and reminders in  eliciting action and the choice of time
horizon.

     Sjoberg used the concept of mental energy to explain relapse
in addictive behavior. According  to him,  mental energy is needed
for the orderly  regulation of action. Under emotional agitation,
competing action tendencies arise,  often  in the fora of tempting
imagery. These action tendencies require additional mental energy
resources for processing,  and since these resources for processing
are  finite,  the cognitive system  will work  at a  lower level,
accepting low quality "excuses* and rationalizations. Hence, the
crucial problea in- pursuing longtent goals is that supply of mental
energy.

     Similarly,  Weinstein  suggested  a "messy desk"  analogy  to
health  protective  actions. Such  actions  must compete  with many
other duties and demanding activities.  Clearly,  the difficulties
of the messy desk can be explained in terms of limiting cognitive
capacity  or in  terms of  limited,  energy  resources.  The latter
approach has  the advantage of  phenomenological  plausibility.  In
addition, most  health protective behaviors are  not cognitively
demanding. Testing one's home for  radon or quitting smoking is not
cognitively difficult to do,  the  difficulty is doing it. This is
why a cognitive  model is  likely to be insufficient in accounting*
for lack of rationality in health protective.behavior.           "I


11. Message format and contents

Message format has been found to be quite important in determining
how people react to a message (Magat, Payne £ Brucato, 1986).

     A traditional issue in work on attitude change, is that of one-
sided  vs two-iilded messages.  A  one--sided»rmessage gives  only
arguments pro op con an issue, while a two-sided gives both types
of arguments. It has been found that one-sided messages are more
efficient when the issue is unfamiliar  (Chu, 1967).

     It is commonly"believed that  messages  that are "personalized"
are more- effective when it comes  to behavior change. But what is
"personalized"? The concept is rather unclear and quite different
aspect* of communication are believed to lead to "personalization".
Thus, personalization of risks has been suggested to occur when (a)
communication  is face-to-face  rather than  via mass  media, (b)
communication contents  directed to parents refer to children at
risk, and (c) whenever the respondent is  made to believe that he
or  she is  personally  at  risk.  These  three  specifications  of
personalized risk are very different.
                                62

-------
     Rogers  (1983)  and  Leventhal  (1970)  studied the vividness of
the information and the personal relevance of the outcomes. While
expected  effects  were   found  in  terns  of  fear  and  perceived
vulnerability, it  was. not clear  that actual protective behaviors
were more likely under fear arousal  (Leventhal, 1986).

     Petty  and  Cacioppo  (1981,  1986)  suggested  a  distinction
between central  and peripheral routes to persuasion.  The central
route involves thinking and elaboration and  leads, according to
the authors,  to more permanent  attitude change.  The peripheral
route is typically evoked in topics of small concern to a person.
Much less and more  superficial processing is involved in this case.
In the  peripheral route, ..efficient  cues may be  "irrational" or
superficial.

     Communication technology now makes it possible to make complex
information accessible in a dialogue mode. A computer program for
communicating radon risk has been  developed by Florig and Morgan
(no date).  The program  simulates  a  house and  shows  how various
factors affect the radon level. It has not yet been evaluated but
it seems like  a  promising development.  There may  be a risk,
however,  in  excess credibility  in  such  a computerized mode of
communication. Real houses  have  many characteristics that cannot
be simulated and generalization from the computer house to a real
house is therefore unjustified however tempting it may be.

     A  final  matter  of  importance  is  to  what  extent  risk
communication needs to be tailormade  to fit  each risk. Can general
principles be designed or does each communication effort need its
own research agenda?

                                         n.
12. Personality, risk taking, and attitude^change

12.1.  Personality and behavior: general  ..

People  vary  greatly in  how they  respond  to threats,  risks and
attempts  at  communication. The  field of  personality psychology
might be expected to explain this variability, but results so far
have been meager. .This is a reflection of general difficulties in
finding trait measures  with sufficient generalizability. Mischel
(1973)  summarized  the field and  came up with  the disconcerting
conclusion that personality dimensions seldom explain more than 10
percent  of  the behavior  variance. Many have since attempted to
disprove this conclusion but so far with little success.

     Notwithstanding,   personality  measures   might  give  some
additional understanding of risk attitudes, even if it is marginal.
For example, Sjoberg and Orottz (1988)  found in  a study of nuclear
power plant personnel that  generalized anxiety scores correlated
about  0.15  with  job  risk  perception.   (This  correlation  is
statistically significant).

                               63

-------
12. 2. _. vigilant vs. defensive response to threats

Recent work  on attitude change has  found personality correlates
for reactions to communicated messages (Eagly,  1981). Janis (1962,
1982) studied the contrast between those who react with vigilance
and those who ignore risks .

     These notions were applied by Stallen and  Thomas (1988) to
how people respond to industrial hazards. They suggested a typology
of reactions to  risk:  secure,  accepting,  vigilant and defensive.
The first two consist  of people who  really do not see a problem.
The  vigilant and the  defensive acknowledge  the existence  of a
problem, but they differ in hope. The defensive lack hope and see
no way they  can personally control the risk,  while the vigilant
have  hope.  The  study  was  of  a  preliminary  character but the
concepts may be fruitful in further work.

12.3. Need for cognition and uncertainty orientation

     Cacioppo  and Petty (1982) suggested a dimension that they
called need for cognition,  based on previous work by Cohen (Cohen,
Stotland & Wolfe, 1957). Need for cognition is a  need to- structure
and understand the world. People who are high in need for cognition
enjoy complex rather than simple tasks,  and may be more responsive
to focused arguments than to peripheral cues.
     Cacioppo^ and Petty devised  a scale for  measuring need
cognition and successfully predicted that those high 'on" the scale
would prefer a complex to a simple task, while the opposite would
hold for those low on the scale.            •-   ...

     Another  somewhat related dimension of potential interest is     ]
uncertainty; orientation  (Sorrentino, Short  £ Raynor, 1984) .           j
According to Sorreiitino, Short and Raynor, uncertainty orientation     j
determines' whether a problem  situation'  fa situation characterized     ,;
by uncertainty about the" self  or the environment) triggers approach     j
or avoidance motivation. People who are  uncertainty oriented would     j
be more^ attracted to such  a  situation, especially if  it is ego     ;*
related (relevant to self esteem) , and would be more influenced by     1
cognitively demanding arguments.  Certainty oriented people would
be more influenced  by 'cognitively simple messages and they would
rely on heuristics rather than elaborate cognitive processing. On
the other  hand,  if the  situation  does  not  pose an important ego
related problem,  these  trends would be reversed, i.e., certainty
oriented   people  would1 use more   elaborate  processing  than
uncertainty oriented people.
                         3(3
     Sorrentino  and coworkers proposed  to measure  uncertainty
orientation by combining a  sentence completion  test  scored for
uncertainty    imagery    with     a  •   questionnaire    measuring
authoritarianism.

                                64

-------
     The  concept of  uncertainty orientation  is  related  to  the
cognitive psychological notion of levels of  processing (Schneider
& Sniffrin, 1977). Cognitive processing can be carried out either
in  an  elaborated,   conscious  manner  or  in  a  more  automatic,
simplistic and undemanding way. Petty and Cacioppo  (1986) suggested
that high-level  processing  is  more  likely with personally highly
relevant and involving thought contents. This notion enjoys a high
level of credibility in current attitude research, but recent work
casts some doubt on its generality.

     Sorrentino  et al.  (1988)  found convincing evidence for this
prediction in an experimental study. Personal relevance increased
systematic processing of message  contents only  for  uncertainty
oriented  subjects, the opposite  was true  for  certainty oriented
subjects.  For the  latter  type  of  subjects  personal  relevance
decreased  systematic  processing. This  is an interesting finding
because it suggests  that the currently  popular notions  of Petty
and Cacioppo (1986)   only if people are uncertainty oriented. In a
practical application to risk communication, we might expect most
people to be certainty oriented and hence most responsive to little
elaborated messages,  provided that  the risk  is  conceived as  a
personal throat.

     It would  be interesting to  test the hypothesis  that a risk
message is most effective if it  is cognitively undemanding and if
the threat is  personalized.  The  hypothesis also implies that the
message is most effective if the risk is abstract and the message
cognitively elaborated. The hypothesis follows from Sorrentino et
al.,  provided  that  most  people  are   certainty  oriented.  For
uncertainty oriented persons the  trends would be opposite. A three-
way interaction between certainty orientation*  elaboration of risk
message and personalized vs. abstract risk is thus predicted.

     It may- be the case that different cultures- differ in uncer-
tainty, orientation.  Sorrentino et al.  suggested  that their data
indicated a lower level of uncertainty orientation  in Canadian than
in US students. This would explain why Petty and Cacioppo, working
with  US  student groups, obtained  results  showing  that  people
process information  in a more sophisticated manner  if  they  are
personally  involved.  Sorrentino  et  al.  also  suggested  that
certainty orientation may be the dominant trend in the population
and that  the hope that  people  are more rational when  they face a
personally important problem  therefore is futile  (Sorrentino  &
Hancock,  1987).
      .  That is,  that personal relevance of an issue increases the
level of processing and makes people more likely to be influenced
by more elaborate argumentation.

                                65

-------
12.4. Health concern

Smith at al (1987)  used a potentially important variable that might
be termed  "health  concern":  a 4-category rating of the statement
"You always ask your physician a lot of questions or regularly read
articles about your health".  The idea of measuring health related
attitudes in order to account for some  of the  individual variation
in response to risk is promising. It measures  individual behavior
at  a level  close  to risk  perception in  meaning.  Research  on
behavior and attitudes has repeatedly found  that the most behavior
relevant attitudes are  those that are  semantically close to the
behavior under study.

     Miller  (1980)  suggested a  variable for measuring monitoring
with regard to  illness  symptoms.  People high in monitoring would
be more prone to see a doctor whenever  they  would be worried about
their health.  Miller, Brody and Summerton  (1988)  reported some
evidence for  this  hypothesis, and also found that those high in
monitoring were  low in  desire to control.  They left that to the
doctors. The  people who tested very early  for radon, in the New
Jersey  study  confirmatory  sample,   appear  to   resemble  the
"monitoring syndrome" described  by. Miller et al. They were quite
concerned and kept extensive  files about radon,  in stark contrast
with the rest of the  population.  The Miller scale has not as far    ,
as I know been applied to the perception of  hazards and technology    :
risks of  but  it  would  be  interesting to do  so,  in  order  tj|    •
understand a part  of  the variance in risk response that might bm    4
attributable to personality.                                    ^    *

12.5. Emotions and risk perception                          :         }
                                                                     1
Emotional states may also influence risk judgments and risk taking.    J
Johnson & Tversky (1983)  found that an experimentally  induced mood    ]
had a global  effect on  risk judgments:  a depressed mood appeared    {
to create higher^.risk judgments. In contrast, Sjoberg and Winroth    j
(1986) found tKat,t|«pressed subjects tended to rate risks as more    ]
acceptable  than  other,  subjects  did.  These   findings  may  be    !
reconciled if a depressed mood is associated with a more tolerant    i
attitude toward oj^ej;,people's risk taking  at the same time as it    >
is associated with aversion for personal risk. Isen et al. (1982)    >
found that positive emotional  states  were associated  with risk    ;
taking in low risk situations but with risk aversion in high risk    -
situations.

     The perception of  an acute threat and consequent high risk
judgment can  be expected to  lead  to anxiety,  at  least in those
people who do not  cope  with the  situation by  means of denial.
Anxiety, in  turn,  is known to lead to a  deteriorated level  of
cognitive functioning (e. g.  Deffenbacher, 1978), possibly because
of a loss of short-term memory capacity (Darke,  1988). The likely
result is  procrastination since people under such circumstances
easily fall prey to rationalized excuses for  inaction.

                                66

-------
13.  Methodological issues

13.1. The definition of risk

Most research reviewed here has utilized risk ratings as dependent
variables without a clear specification of the meaning of the term.
However, risk  is an  unusually ambiguous word.  It  can  denote the
probability of harm,  the size of a loss or a combination of the two
(e.g.  their product).  Sjoberg  and  Orottz   (1988)  asked  their
subjects whether they judged risk as a probability, the size of a
consequence or a combination- of the two. While most subjects chose
the  first  definition,  many opted  for the other two.  There were
systematic differences  in  risk judgments  depending on  the choice
of risk definition. It may be important to pay attention to how the
subjects understand the term risk in interpreting their judgments.

13.2. Response rates                           .

Table VI  indicates that response  rates have been  fairly  low in
these studies.  The table also gives the total number of respondents
which could  be used for  data analysis and  as can  be seen that
number varies by an order of magnitude across samples.

     The last  column gives the final response  rate of all those
that were eligible for  inclusion in the sample. It is the outcome
of several factors which give rise to drop-out.  For  example, in the
Florida risk attitude study (Section  5.9,  page 37) 30  percent of
eligible homeowners  could  not  be reached  on  the telephone* and of
those that  could be reached,  40 percent  agreed to participate.
Hence, the total response rate is given as 28 percent.

     Sometimes the researchers have given conditional rather than
absolute final  response rates. In the  NYSERDA study,  comparison
sample,  followup;, survey^.the  response  rate  is  given as  72.2
percent. However,  this  is in percent of  those who completed the
baseline survey, not of all those who  were eligible for inclusion.
The  latter percentage  is more informative as  to the information
value of  the data obtained-,  and it  is considerably smaller (36
percent).  (Strangely enough,  the results  for the monitored group
are given according to  the principle suggested here).

     I  have tried  to  distinguish,   whenever   relevant,  between
acceptance to participate  and  actual  response.  In  some cases the
reports do  not give any  information  about how many people were
approached  to  give  a  certain number  of people  who  agreed  to
participate. This is a  piece of information that should have been
included.  Even  if  data  do  not  aspire  to  strict  statistical
representativeness it  is  useful to know  if  10 or  90  percent of
those approached accepted to participate.


     The overall picture  is not too  bad.  Several studies reached
respectable response rates.  Some comments about the lower values
are in order, however.


                                67

-------
 	—  ^c.shw.,u^  rates  in  ci-su radon  studies,  measured  as
percent of the studied sample that  (a) accepted to participate,  and
gave usable responses.
  Type of data   Number
Maine study.
Section 5.2*

NYSERDA study
Section 5.4
Monitored group
  Baseline
  1st followup
  2nd: followup* -
Comparison group-
  Baseline
  1st followup

New Jersey study
Section 5.5
Random sample
Confirmatory sample
New Jersey study
of radon information
Section 5.6
                         of  Prop.    Prop.
                         collection     usable     accept, resp
                         method         respondents
Telephone
                                        221
                          Telephone    2231
                          Telephone    2087
                          Mail      ca.1700
New Jersey study
of communication  format
Section 5.7*

Maryland study
Section 5.8  •.•:.;::...:;... '. . 'M:;. •:
Baseline.;:;::  '••••-' •'•'  ' ''^':,^':
FollOWUp      ..     >W^-;:

Florida study  i xi1^""';i'
Section 5.9;.;.-; .. ;--: ;,••::<;•••>

Florida personality   ^ v
study, Section 5.

Onondaga study • ~
Section 53Hfe&*  *
Washington^ D. C,, study
of mitigation
Section 6.4
 Telephone
 Telephone
                          Mail
                          Mail
                          Mail
                           Mail
                                        252
               657
               141
                                        271
                                      1948
                           Telephone   1547
                           Telephone?   1528
                           Telephone    837
Mail
                          Mail
  Mail
                                        320-
                                        204
             709
                                                  No info,
                                                  No info.
                                                  50
                       78
                       47
                        No  info.
                          34
                                                   28
                                                   84
                         No info
                                                   77
                                     .97:^:1111^
                                      9*C:::«ixI:i
                                    50
                                    36-
                                                              79
                                    75
                                   67
                                      38?
                                    81,
                                                              77
Notes. 1. A few of the minorr  exploratory studies are excluded,
       2. All data have not been collected at the time of
          writing.    ,.                       'lr
                                  68

-------
     Desvousges et  al.  were concerned about  their response rate
(which is by no means dramatically low) and suggested that it may
have been related to the fact that they had to carry out their data
collection  during the  Christmas  holidays. However,  since their
response rate is  no worse  than that  obtained by others,  it seems
that the  explanation may  be  less credible.  Perhaps there  is a
general decline in response rates to telephone interviews because
of their increasing use and new computer technologies.

     What are the possible effects of these  low response rates?
Weinstein et al.  argued that  the  people  who respond probably are
more concerned and know more and that, therefore, risk denial may
be underestimated by these data.  On the other  hand,  the better
studies  have examined  some  characteristics  of  respondents  as
compared with the population being studied. They  have often failed
to detect  important differences,  with the  possible  exception of
gender. Women tend  to be over represented among the respondents.
Since women are  known to  be  more-risk averse than men,  this is
likely to create  some bias.


13.3. Response scales

Weinstein  et al.  included  "don't  know"  among  their  response
alternatives and many people chose that alternative, perhaps partly
because the radon issue was  new  and people  in  fact  knew little
about  it.   These  responses were  then treated as missing data,
creating some- problems with correlation statistics and regression
models.

     Krosnick  and  Schuman (1988) showed that  more  intense  or
extreme attitudes are just as vulnerable to such response effects
as effect  of the order of questions, with one exception:  if a
middle, neutral,  response is provided it is usually  very attractive
to those who have-only weak convictions.

     Smith  et al. (1987) obtained explicit ratings of information
materials and found- that homeowners  discriminated little between
them. This  was so in spite of other findings in their study which
clearly showed that one condition (the  fact  sheet)  was  inferior
because  it led  to  less knowledge  and clearly  exaggerated risk
rating**

     It  is an open question  if  responses  to a  question about
willingness to  pay  for   additional  information  give  a  valid
indication  of true demand.

     Some work has indicted consistency between actual purchase and
CV questions, however. Dickie, Fisher and Gerking (1987) found that
there was no difference between the amount of strawberries actually
bought and  the amount that people just said that they would buy.
This result has  some interest but, as the authors point out, it
cannot readily be generalized to WTP measures used for evaluating
environmental protection measures. The latter usually concern much
                                69

-------
more important issues and people can be suspected of letting t
judgments be  influenced by  strategic considerations.

     Methodology may account for the difference between the NYSERDA
and New Jersey studies.  In the NYSERDA study  it was found that
people  exaggerated  the  radon risks while some aspects of data in
the New Jersey  study suggested the opposite. The NYSERDA results
utilized the  absolute level of the  risk scale,  transformed to a
probability  scale.  However, the  absolute level may  not be very
informative   since  scale  levels  are   notoriously sensitive  to
contextual factors. Another big difference may be that the NYSERDA
subjects all  had  agreed to have  their homes tested.  So they-may
have been more health-conscious, or risk averse,  than  the original
NJ  samples.  In  addition,  it   is  unclear that  risk may  be
straightforwardly  translated to  probability.  The  sheer lack of
concern over  radon,  as  shown  by  the lack  of  radon testing,
indicates that  people at  least in that sense  underestimate the
risk.

     Risk  estimates  can  be  given either  in an absolute  or a
comparative manner. Absolute ratings utilize a scale  from,  say, 0
(no risk at all) to 7 (a very high, or severe, risk).  Comparative
ratings are done with reference to a standard, such as the average
radon risk  in the respondent's community  or compared to smoking
risk.

     These two types  of ratings can be expected to have somewhat
different properties. Sjoberg and Drottz (1988)  obtained compara-^
tive and absolute ratings of job risks from nuclear power plant
employees.  They  found  that  comparative  risk   ratings carried
information about both the  reference level and the compared risk.     ,
The  finding  of  Weinstein,  Sandman and Roberts  (undated)  that     >
absolute risk ratings showed the expected negative relationship to     i
risk information, while comparative risk ratings did not, may have     '
been due to  these  different informational  bases behind the two     •
types of ratings.                        .                .             )
              *.-..-,,-.._.          ,                      • •

13.4. Time perspectj-Vfl'      _          .....  _                           I
              . _:   •••-->• : • ' •

Almost  all  studies were  concerned with  short-term  changes  in     :
knowledge, attitude*,  and behavior. The persistence of attitude     ;
change  is, of course, no  trivial matter (Cook & Flay, 1978). For
voluntary changes in behavior (e. g.  testing and  mitigation) it is
probably necessary  to have  lasting changes in  attitude.  In the
radon case, attitude changes need to  last  long  enough to elicit
testing and mitigation.


13.5. Attitude measurement;

In some studies the coverage of attitude about radon risk has been
sketchy. It would be useful to obtain more detailed  information on
the components of the risk attitude. This could be done with the
help of  a well established attitude model such as the  one devised
by Fishbein (Fishbein & Ajzen, 1975).
                                70

-------
     Concern has  been  found to be an  important  variable in some
radon risk,  studies.  There have been doubts  that people can give
valid estimates  of concern or  importance  of attitude components
(Slovic & Lichtenstein,  1971). in the field of political attitudes,
for example,  it has been  hard  to demonstrate that  aspects of a
candidates' policy judged as more important also in fact are more
decisive when it comes to voters'  choices.  Krosnick (1988), showed
that these  difficulties were  largely methodological,  that people
can make  valid judgments of importance, and that more important
dimensions also have a  larger weight in accounting for choice.

     Attitude measurement tends to utilize  rather simple responses
and  does  not usually   involve  attempts  to get more  detailed
information about the  thought process behind risk, attitudes and
perceptions.  Bostron  (1989)  attempted just that, however.  She
collected  extensive  interview  data  about  radon attitudes  and
perceptions from 24 subjects, who were all reasonably knowledgeable
and involved.  She devised an interesting  scheme for scoring the
accuracy of people's notions about radon and came up with several
(preliminary)  conclusions.  People were  found not  to understand
concepts  of radioactive  decay,  to  have scanty  knowledge about
mitigation,  to  be  unclear about  the  effects  of   radon  (many
mentioned cancer unspecifically,  not  lung cancer),  and  to know
little about  the concept of  a  soil gas.  Some  of these findings
could perhaps  have  been obtained with,  a more  conventional and
simple approach, but the results still indicate that  further work
along this line could be of value.

13.6. More detailed information about crucial decisions
                                        •
The studies reviewed here have  collected an impressive amount of
information about, how people perceive radon risks.  Still, there
are some types o£>sinformat±on that are missing.

     Several projects  collected data on reasons  for not testing
for radon,  such as  those shown in  Table  II above  from  the New
Jersey study. However,  I lack comparable data from testers on why
they have tested. Such data  may be even more  informative than
reasons for not testing, which seem  a bit like excuses rather than
real reasons. There  may be several  routes  to testing that differ
in interesting  and important ways,   e.g. health concerns,  social
pressure and property value concerns.

14. Discussion and conclusions

14.1. Summary and discussion

Radon is a risk with certain unique properties.  It shares some of
the characteristics  of  all risks from ionizing  radiation (slow
effects in  the form  of  cancer,  a  risk from a physical agent that
cannot be sensed). At the same time it  is a risk that threatens
people in their  homes  and homes are  usually perceived  as safe
places under one's own  control and responsibility.

     Radon  risk has been compared  to  such life-style  risks as
smoking or  drinking  alcohol.  The similarity is  that people take


                                71

-------
these  risks  as private  individuals.  Society  is reluctant
legislate the risk level to which they can expose themselves. T
must act voluntarily to mitigate or eliminate these risks.

     There are also differences between radon and life-style risks.
First, radon  exposure does  not  confer any benefit as smoking and
alcohol do. Second, it is not clear who,  if anyone, is responsible
for radon risk. Radon risks were largely unknown to the US public
before 1985 and most home owners bought their current houses before
then.

     Reducing  a life-style risk involves making  a commitment to
action in  spite of temptations  to delay action.  In  this sense,
there is a further similarity between radon and life-style risks.
Many  people   in  one  of  the   studies  reviewed  here  became
"procrastinators", i.e.  they changed  from indifference towards the
risk to acknowledging the  necessity  of action "in principle". It
is well known that people  find  it very difficult to- quit a risky
habit and that rationalized delays are very common. It  is too early
to tell if the same phenomenon will occur for radon risk but it is
possible.  On  the  other hand, testing  for radon  and mitigating
require only two discrete actions while the ex-smoker has to resist
cigarettes several times a day.  People do not have to quit a habit
(or develop a new one)  in order to test and mitigate for radon.
                                                               1
     Most  smokers know that  their habit  is unhealthy  but th
procrastinate in spite of their own knowledge of what is in the
best long-term interest. Something similar seems to be going on
the case of radon. More information will therefore  probably not be
sufficient.  It  has even been  argued that more information will
provide people with material for twisted rationalizations for why
they have  not tested  and can go on postponing  it  indefinitely
(Weinstein, Klotz  & Sandman,  1988).  It is furthermore well known
from research on such hazard warnings as for  floods and hurricanes
that people tend  to ignore  them  if the warnings are not directed
to them personally- in a very  concrete manner (Mileti & Sorensen,
1987) or supported by feedback from  friends (Kunreuther et al.,
1978).        —      •"•••              -v

     In the present report I have reviewed several studies of-how*
people  perceive  antf  respond  to  radon  risk.  Initially it  was
believed that information about radon risk might give  rise to panic
reaction*, but  experience has not confirmed this concern. On the
contrary, people are often quite indifferent to the issue, at least
in regard to their own homes. This indifference  can be interpreted
as a  manifestation of  a  rather  general tendency  to deny health
risks and  it has been suggested that denial  of radon  risk is
moderately strong compared to other kinds of denial.

     Many  people  are  outright  negative   to  radon  testing.
Pennsylvania authorities at one point went from door to door and
offered free radon testing. Still, about 50 percent refused to havej
their homes monitored. Some people are concerned that rumors of af
radon problem will affect property values in an area.
                                72

-------
     It is worth stressing  that  people  tend to be indifferent to
radon risk in spite of the fact that  the EPA action level is quite
high compared to other risks. The actual risk level seems to be a
poor predictor  of  public response to  risk. The  action  level in
Sweden is 2.5 times higher than the US level and it corresponds to
the lung  cancer risk  of smoking a  pack  of cigarettes  per day.
Still, Swedes are just as indifferent to radon  risk as Americans.

     As might be expected on the basis of  these  facts,  there is
usually  only  a  weak  correlation  between  the  scientifically
estimated risk  level,  as physically measured, and the  level of
perceived risk.  Some people  with low levels of radon in their homes
are quite  concerned while  others,  with  quite high levels,  are
indifferent. This low  correlation may partly arise  from the fact
that the physical-measurements are not  perfect indicators of the
actual risk level.  The risk also depends  on how much time people
spend  in  the house,  their  age  and  other  factors.  Still  it is
reasonable to conclude that there is a weak correlation between
perceived and actual radon risk.

     There is little evidence suggesting that people do not believe
that the radon readings are accurate or that the risk assessments
are incorrect.  They probably believe that the risk  is  there, in
principle and for people in general,  but that they themselves are
for various  reasons less  vulnerable or more  lucky. It  is also
possible that or they  just  do  not  care  about getting involved in
protecting themselves  against  another risk,  they simply go ahead
and take it.

     The effects of information about radon risk provide a partly
different picture.  Two studies  have shown that people revise their
risk estimates  in  the "right" direction,  i.e. in  the  direction
suggested by the readings of radon levels, in their homes, still,
other  data  suggest  that even for  risk  revisions  there  may be
irrational denial factors at work. There  is some support for the
thesis th^t people are more inclined  to  accept that fi£h££ people's
homes in theia?- community may be at risk than that their own homes
may be at risk.  _;.  ,; „

     If actual risk is  not strongly correlated with perceived risk,
then what accounts for perceived  risk? There are some correlations
with background data.  Older people are  less worried about radon,
and parents of small children more so.  In some studies women have
been shown to be more worried than men, but the gender difference
is much  smaller for the  perception  of  radon  risk than  for many
other types of risk. People with a family member with cancer were
more concerned about radon in one study.  Some personality variables
have been suggested as predictors  of perceived radon risk but so
far little research has been carried out on this issue.

     Nevertheless,  these  factors are  only  weakly  related  to
perceived radon risk. There  is simply not enough research available
to support more definite statements  about what factors influence
perceived radon risk.
                                73

-------
     People seem inclined to  deny radon risk as they do a m
of other health related risks. This reaction  is  in contrast to
attention given to the risks from industrial pollution. Thus, there
is no tendency  for people to  deny risks in general. Only certain
risks are  denied.  It would  be of  interest,  in future  work,  to
investigate why some risks are denied and others exaggerated.  It
may be that risks that are denied are risks where the threatening
agent is associated with the self.

     I suggest that radon risk perception is a special case of ego-
related risk,  i.e. it  is  a risk that  is  closely related to self
conceptions. This hypothesis is derived from evidence that people
tend to perceive their homes as parts of their selves, especially
if they have lived  in them for a long time. And there is a tendency
to deny that something as closely related  to oneself as one's home
could be threatening.  Reactions to radon  risks may therefore  be
related to perception of one's home.  It is conceivable that risk
communication would be more efficient if the risk would be clearly
separated from the home.

     The hypothesis is supported by data that  indicate that people
with a longer history of living in a house are more  likely to deny
radon risk. Other supporting data show that people  react strongly
to radon risk when it is introduced in their neighborhood - not on
their homes*- Part of this reaction is probably a result of moraJ
indignation  over being exposed  to  a risk  by someone.  Peop]'
strongly resent a loss of control, and they tend to perceive
which they can control and are responsible for as benign.

     As an alternative  explanation of perceived radon risk, risk
reactions  could  be  seen as  dependent   on  a number  of  risk
characteristics such as those  suggested by Fischhoff et al. (1979).
The  problem with  that approach  is that  it assumes  that  risks
reactions are based on cognitive information integration. It misses
the  motivational  and emotional  aspects jpj£ risk-, perception.  Few
attitudes are morwel£kely to be dependent ...upon emotional reaction
than risk attitudeV. In addition* it i*. w»lL known  that-attitude*
in general  cannot" be veil explained  by beliefs (McGuire,  1985).
This is not to say that there  is not some  correlation between risk
attitudes  and  risk  characteristics.  The question is what  the
correlation means.  Slovic et al. assumed that  attitude* are caused
by belief*. It i* more likely that both attitudes and beliefs are
caused by an underlying common image  (Sjoberg & Biel, 1983).

     Concern about radon risk may be problematic by  itself. It has
been reported  in some work that higher concern about radon leads
to  more  blame  attributed  to  the  government  and  puts  less
responsibility on the  individual.  Higher awareness, on the other
hand, was correlated with greater perceived responsibility on the
part of the homeowner.  The question is whether it  is possible to
increase awareness without, at the same time, generating increased
concern. More research on this topic could be fruitful.

     Communication about radon risk with  the general public has
been  studied  in two  major  investigations,  one  of which  was
especially   concerned   with   mitigation.  These   studies   also


                                74

-------
investigated the effects of a major attempt by a television channel
in the Washington,  D.C.,  area to encourage people to buy test kits,
available  at reduced  price  in  grocery  stores.  The  television
campaign created an impressive interest in buying test kits. Over
100  000 kits  were  quickly  sold and  demand  was  by  no  means
exhausted.  About 6 percent of  the  home  owners  in the area bought
test kits.  However, only about half  of  those who bought the kits
turned them in for analysis and few people who had elevated radon
levels in their homes reported any effective mitigation beyond such
behavioral measures as keeping windows open more often.

     The EPA-sponsored outreach activities in a Maryland community
were  probably  more  effective than, the  simultaneous  television
campaign in stimulating  people to  test  for radon. The proportion
that tested their homes rose from about 5 to 15 percent. There is
some evidence that the television campaign was effective in making
people aware of radon risk, but less  effective in stimulating them
to act. A  similar finding was reported in a New Jersey study of
smaller scope where it was found  that there is a link between risk
perception and action  but that the link was  quite weak. The weak
effects of the  television  campaign   are  in  line  with  extensive
experience  from  other studies of health-related  public  service
messages. The effects are usually quite marginal.

     A  few studies have  investigated knowledge  about  radon and
found that people are fairly well aware  of its general properties,
although they lack specific knowledge. They may also lack knowledge
about  how  to test for radon  and  whom  to ask about  testing and
mitigation. They are especially uncertain about mitigation and its
costs.            .   '

     It is relatively easy to inform people, about  radon facts.
However, there is  little relation  between.knowledge and tendency
to test and mitigate. This finding  is  in line  with a  generally
noted  lack  of   relationship  between  attitudes  or  beliefs  and
knowledge.      .                  ......
     Risk communication  is a topic of much  current concern. EPA
has published a brochure containing "Seven cardinal rules of risk
communication11 (O^ S. Environmental Protection Agencyr 1988 b). It
is reasonable to ask what relevance such advice has in a concrete
case such as radon risk.

     The Cardinal Rules make a rather pallid  impression, on closer
scrutiny. To be more specific, the communicator is advised to

- Accept and involve the public as a legitimate partner
- Plan carefully and evaluate your efforts
- Listen to the public's specific concerns
- Be honest, frank and open
- Coordinate and collaborate with other credible sources
- Meet the needs of the media
- Speak clearly and with compassion

     It is hard to avoid the impression that  these rules have been
designed for a case  where  the communicator wants to reassure the


                                75

-------
public  that the  risks are  smaller than  they fear.  Such risfl
exist, but the problem with radon risk is the opposite. Still, the
brochure never states  that its applications are only one type of
risk,  it  seems  on  the contrary to  be  aimed  at  very general
application.

     Since  radon  has been  stated as one of the major threats to
public health it would be valuable to have  rules designed for the
case when people need to be alerted to a danger. The fact  that this
has not  been dona already may be due to the  fact that  radon in
homes constitutes no threat to the industry, while public concern
over mean* of energy production aad disposal of industrial waste
does constitute such a threat, at times.

     There  is-an increasing-interest in research on mitigation of
radon  risk. It baa been  found, that there  is  a  rather weak
correlation between scientifically estimated risk and mitigation,
but a stronger correlation between perceived risk and mitigation.
This is an  example of a quite general truth: people react to what
they perceive reality to be rather-than what it really is.
     It  is  not yet clear  how much mitigation one  can expect in
different circumstances*  Some data certainly  suggest that home-
buyers are likely to be among the most responsive to the radon risk
message. On the other hand,  not  all  data on mitigation are at
negative as the  ones  reported in the study of the Washington,
C., television campaign* A follow-up of highly motivated New Jersej
homeowners found a high rate of mitigation.  Some  Swedish data were
in between the two American data sets. (Cultural comparisons are
risky, however, since  many crucial aspects, such as the economics
of housing and mitigation, differ dramatically between Sweden and
the USA).      -••    -...  .:     -,.. -	  .= .*..•

    -A few  studies haver investigated the  format and contents of
radon risk messages.. Av aw^oc study compared "command*"  and* ""cajole/1'
styles of cpmmualcation*.That study also investigated  and compared
qualitative vs quantitative types of  information. The  results were
somewhat complex-and  differences were-not  large but- there was a
tendency for a command; stylesof communication to be-mo»t effective.
That  study  also  compared the  use of  a  brief  fact  sheet as a
replacement for more extensive brochures* It was found that those
who had be*n given the fact sheet tended to be more concerned about
their risk-level than.others.  This was true in spite of the fact
that the* fact sheet was only given to those who had very low levels
of radon in their  homes.

     The investigators concluded against the use of the fact sheet.
That may be a reasonable  conclusion  for that particular group of
homes and their owners. It may still  be the case  that  a short fact
sheet would be useful  for people who have higher levels of radon
measured in their  homes.  It  has been found that people, if given
a basis in rich information material, tend to construct defensive
denial  conceptions. In other words,  more information makes it
easier  for  them  to come  up with rationalizations  for  lack of
action. Of course,  a short fact sheet need not be very similar to
the one used  in the NYSEROA study.


                                76

-------
     Thera have been few attempts to  analyze the contents of radon
risk attitudes. So far,  most data suggest that health concerns are
most  important.  Many  studies could have profited  from  a  more
extensive mapping of the components  of risk attitudes,  using one
of the  veil known models  of attitude measurement, such  as  that
devised by Fishbein (Fishbein & Ajzen, 1975).

     Many of  the  studies reviewed here used a  risk  concept  that
was not further explained to respondents,  who were asked to judge
such variables as size  of perceived radon risk.  However, other
research has shown that people interpret the word risk in various
ways. There may be differences among genders and educational strata
in whether respondents interpret risk as probability,  as the size
of the  consequences-or-  as  a combination  of these two  concepts.
Interpretations of  results may  be misleading  if this  factor is
neglected.

     Another  interesting methodological aspect  is the response
scale. It is difficult to translate the judgment of seriousness of
a risk into subjective probability, and seriousness could refer to
either  probabilities  or consequences. On  the  other hand, it is
notoriously difficult to obtain valid judgments of small perceived
risk. As an alternative,  comparative  risks often are used to assist
people in judging small risks.  Radon  risk has been compared to the
risks of smoking, for example. This  approach is well worth trying
but one  should bear in mind that comparative  risk judgments are
more complex  than  absolute ones because  they involve two  risk
levels rather than one. Risks are probably perceived in a context
of several  threatening  events  and conditions and the judgment of
any given risk may  well  be affected  by how certain other salient
risks are perceived. For example, people with a very risky job may
perceive  their radbn  home risk as minor  just  because it  is
implicitly or explicitly compared to job risk.

     In many  circumstances people are most  strongly  affected by
social  diffusion  of information and attitudes, i.e. what their
friends and neighbors and locally prominent people tell them. For
this  process  to  spread  quickly there  must  be  many  willing
"informers", However,  in the case of radon there  are  still  very
few people who have shown any active  interest in the issue of home
testing, and those  who  have informed their friends and neighbors
have sometimes met with hostile reactions.  It would be of interest
to investigate such obstacles to social diffusion.

     There has been little work on property values and house sales
as related to radon. There are indications that  this aspect  is very
important  for some homeowners.  An  interim report on  radon  risk
communication  states  that  "Some Regions  indicated that  many of
their radon-related telephone calls  focus primarily on the  real
estate implications of radon rather than on health concerns" (U.S.
Environmental Protection Agency,  1987 a, p. 4).

     A  Swedish study  recently found that  radon did not  seem to
affect house prices. Informal information  from realtors  in the USA
confirms this  finding. However,  a Colorado study showed that many

-------
people are quite concerned about getting a house tested for
before they buy. Some 50 percent of the hone-buyers participa
in the  study reported that  they  had done so.  These  people also
tended to mitigate much more often than those who bought test kits
and monitored their own homes. Indeed, the prevalence of mitigation
at or above the action level of 4  pCi/1 was 100-200 tines as great
for  hone-buyers as for  homeowners.   People  whose employer  was
especially concerned about radon risk and those who were in contact
with a knowledgeable realtor or building contractor were especially
prone to test ard nitigate. These  results were  obtained in a state
where radon risks had not been extensively discussed in the nedia.


     The fact that real estate prices  are not affected by radon
nay be related to  the prevalent information  that a radon problem
usually can  be fixed quickly and  fairly cheaply. There has been
little awareness so far about how often continued monitoring for
radon will be required  and additional mitigation will have to be
undertaken. When this  issue is given wider publicity there could
be an effect  on real estate prices.

     Virtually  all  research  so  far  has  been  conducted  with
homeowners so the  social strata that  have been represented have
not been representative of the whole  population* We  do not know
how people who rent their homes  respond, to radon risk,  or how
involved their landlords  are in  monitoring and  mitigating
houses. There is also very little  work reported about  attitudes
radon in schools, public buildings and workplace*.


J.4.2. Conclusions

Research on   radon  risk perception and risk mitigation  has  not
proceeded  very far,  simply  because  the problem is so recent-
However, it  is possible to draw some conclusions on the basis of
existing research results.
                                       :   .-?••« •'•: • iSrtc <. ••= •- •
(1).  The  major  problem  is  indifference  to high-level  risk,
especially among people  who already live in  a home that may have
an elevated radon level.                 _.,_-..

(2). Hose buyers are probably much easier to alert to the issue,
with ensuing  radon  testing and mitigation.

(3). Studies  of variations in risk communication material suggest
that brief,  to the  point,  recommendations about what  to do are
better  than  longer  messages  that allow  the  reader  to  form his
personal  opinion.   It is  just too likely  that these personal
opinions will be defensive risk denials.

(4). Even  the most effective mass media campaigns  and outreach
community  programs  investigated  so   far  have  succeeded  in,
stimulating only a  small fraction of the population of homeowners
to monitor and mitigate. Whether  these campaigns have justified
their costs has not been determined. A community outreach program
                                80

-------
     An   experimental  study   of  risk   communication  formats
(Weinstein, Sandman & Roberts, 1989) found that a brief statement
about the  action  level  was best for eliciting continued concern.
On  the  other hand,  full  information  about the  risk,  including
numerical, risk comparison data and graphical information, was best
for creating a balanced view.  A  major difficulty in this study was
that of  communicating absolute  risk levels.  A graphical display
seemed sufficient to make  people disregard a difference in absolute
risk level in the range 1-25  (between radon and asbestos).

     The NYSERDA fact sheet was somewhat similar to the condition
of stating the action  level only in the study by Weinstein, Sandman
and Roberts (1989). In particular, the fact sheet did not specify
anything about the risk of radon levels lower than  4 pCi/1. one can
conclude that a brief message is  most likely to elicit continued
concern among horn* owners.

     Another reason for favoring very brief messages is that people
have difficulty in handling information and that they  are not very
motivated  to obtain it.  Studies of.public service information in
the form of printed brochures usually have shown that most people
do not bother to read  them and that those who do read  them quickly
forget most of their contents. Other investigators  have studied the
effects  of level  of   risk communicated  and  found  that a  more
threatening message tended to invite more risk denial.

     The conclusion is that the most effective message format and
contents should be brief  and not too threatening.  Messages should
furthermore emphasize that radon  is  a risk  that must  be  taken
seriously. Comparing  it to some other risks, such as  smoking, can
help people understand its size,  rt  should be clear that  it is
simple and cheap to test for radon and relatively simple and cheap,
in most cases, to mitigate. Quantitative information about the risk
level is to be  preferred to qualitative,  and a11 command" tone is
better than a "cajole" tone for eliciting radon testing.

     Another  possibility _ involves  switching   to   the  common
international measure Bq/m3 rather than pci/l. Sine* 1 pCi/1 -40
Bq/m3 such a  switch would involve larger numbers that  could by
itself have an effect (whether positive or not).

     These conclusions should be regarded as preliminary. They are
based  on  a  rather  complex  set  of  findings from  radon  risk
communication studies and from communications research in general,
as well as basic principles of cognitive psychology.

     It  is important  to  determine what the goals of radon risk
communication should  be.  Should it be aimed at inducing behavior
change,  i.e.  testing  for radon and mitigation if the levels are
high? Or  should it aim at informing the public so that they can
make well-informed decisions? If  the  latter path  is  chosen, some
problems arise.  First, it is very hard to know if a person has made
a   well-informed   choice.  Indeed,  the   concept  itself  needs
considerable  clarification.  Second,   information is by  itself
                                77

-------
insufficient to induce much behavior. The studies have found ve.
low correlations between acquired knowledge and testing for radon

     Research methodology is an important issue in a discussion of
risk cosaunication. The  empirical  studies reviewed  in the report
vary considerably in this respect.  Virtually all were quantitative
and  employed  either nail  or telephone  survey methods  but  they
differed in many other respects.

     There  is  evidence  that  people  are  net  very  eager  to
participate in surveys  of  this kind.  Response rates  around 60
percent or lover are common in spite  of several reminders. Some of
the problems of low response rates  have  been  attributed to data
collection during  holidays,  but there is no evidence that these
particular studies yielded especially low response rates. Indeed,
the US public probably is saturated with opinion polls and survey
investigations  and it  is becoming  more- and  more difficult to
recruit respondents for new waves of data collection.

     It  should also be pointed out that several  of  the  major
studies reviewed here reported quite good response rates and that
there  is  no  evidence   of  any  major  bias  in the  samples  of
respondents.

     It can still  be noted that women seem to be more likely tod
accept invitations to participate in radon risk studies than menfl
and the  result is  that they are   overrepresented in  some of the'
studies. Also, the samples tend to consist of highly educated and
financially  secure  people,  possibly  because  the  studies  have
focused on home owners rather than renters.

     Attitude change has been studied  for decades.  It is usually
quite hard to change people's attitudes,  especially to change them
permanently and. to  affect-behavior* Changes'tend to be-short lived.
The empirical work on radon so far has- not followed attitudes for
a more than a few months, with the  exception of the NYSEROA study.
Longer follow-up tines- are called fo* in order-to get a realistic
idea about ban-persistent attitude change* are.

     It would probably  be  fruitful to  investigate  in more deta-il-
how people reason  with  regard to radon and testing.  One piece of
information that is~ lacking in these studies  is whv  people test
(there is data on why people do not test). Testing can be motivated
in many ways,  for  example  health,  econonic factors  or conformity
to expectations, by friends.  The reasons  people have  for testing
could stimulate the formation of hypotheses about effective methods
of risk communication.

     Qualitative    information   about    the   conceptions   and
misconceptions  that people  have  about  radon would  also be of
interest. Among other things,  such  conceptions  could be of crucial
importance for nitigation. Some people  may believe  that radon is!
most likely to be  found in  the attic rather than the basement of
a house (since many gases are lighter than air). Ionizing radiation
is also something  that  people know little about and they tend to
espouse several misconceptions.


                                78

-------
appeared to  be more promising than  a  major television campaign,
although they probably reinforced one another.

(5) . Further research should  be  conducted on  the components of
radon risk attitudes and conceptions, in particular why some people
test and mitigate.  Interpretation of  findings  would probably be
helped by more information on how people interpret such key terms
as ."risk".

(6). There are  important  ethical  aspects of radon risk that have
been neglected in research. Some people see nobody as responsible
for a radon  problem,  others  may bias* themselves, contractors or
the  government.  The  owner  of  a  house  may  or may not  feel
responsible)  for the health of  others who live in it. At the time
of a house sal*- there is  a  special  ethical problem if the seller
knows thero  is a  radon problem. These  are important matters for
further research.    .

(7). There is  a total lack of  data  on  how renters, as opposed to
homeowners,  perceive  risk and  how their landlords view the radon
issue. There is very little  data  on the perception of radon risk
in public buildings and workplaces.  Perhaps some homeowners can be
alerted about, .thair  home risks if  they get involved in school
building risks.

(8). Although the sparse data that exist on the  issue suggest that
radon risk does not reduce property values,  the issue certainly
needs much more investigation.  Will the lack of effect persist when
more  experience  with  mitigation   and  its  costs   and  effects
accumulate?  People may at this time  be  overly  optimistic about how
easy it  is to  "fix"  a radon problem permanently.  Do people see
monitoring  as  a  potential  economic  threat?  Or  will   las&  of
monitoring and disclosure of results at the time of a house sale
be the real  economic threat in the sense that homes that have not
been tested  will sail  for a lower price?
                                81

-------
                          15.  Reference?

Adler, R., & Pittle, D.  (1984).  Cajolery or command: Are education
     campaigns an adequate substitute for regulation? Yale Journal
     of Regulation. i, 159-194.

Akerman, J. (1988). Economic valuation of risk reduction: The case
     of indoor radon. EFI Research Report 91-7258-266-9. Stockholm,
     Sweden: Stockholm School of Economics.

Akernan, J., Bergman, L.,  & Johnson,  F. R. (No date).  Paying for
     safety: Voluntary reduction of residential radon. Unpublished
     manuscript. Stockholm, Sweden:  Stockholm School of Economics.

Anderson, C. A.,  6 Sechler, E.  S.   (1986). Effects of explanation
     and  counterexplanation  on  the  development  and  use  of
     statistical  theories.  Journal  of  Personality  and  Social
     Psychology. 50. 24-34.

Anderson, C.  H. (1983). Abstract  and concrete data in  the per-
     severance  of  social  theories:  When  weak  data  lead  to
     unshakable beliefs. Journal of Experimental Social Psychology.
     12, 93-108.

Anderson, J. R. (1985).
     2nd Ed. New York: Freeman.

Baum, A., Fleming,   R., & Davidson , L. M. (1983). Natural disaster
     and technological catastrophe. Environment and Behavior. 15.
     333-354.

Becker, M. R.  (Ed).' (1974). The health belief model and personal
     health behavior. Health Education Monographs. i(4).

Bergman,  L.,  Sdderqvist,  T.,   &  Akerman, J.  (1988).  Subjective
     valuation  of  radon  reduction.   EFI Research  paper  6358.
     Stockholm, Sweden: Stockholm School  of Economics.

Bostrom, A.  (1989).  Risk  perceptions: An approach to perceptions
     of risk processes,  a mental models analogy and an application
     to radon. Unpublished  manuscript. Pittsburgh, PA: Center for
     Risk Percept ion ..and, Communication, Carnegie-Mellon University.

Cacioppo, J. T.,  & Petty,  R. E.  (1982).  The need for cognition.
     Journal of Personality and Social Psychology. 42. 116-131.

Chaiken,  S.   (1980).  Heuristic  versus  systematic  information
     processing and the use of  source versus message cues in per-
     suasion.'  Journal  of  Personality  and Social  Psychology. 39.
     752-766.
Chess,  C.,  &  Hance,  B.  J.   (1988).  Alerting the  apathetic a
     reassuring  the alarmed:  Communicating about  radon  risk in
     three communities.  Final Report submitted  to  the Office of
     Policy,   Planning,   and   Evaluation,   EPA-230-08-88-036.


                                82

-------
     Washington,  D.  C.:  United States  Environmental Protection
     Agency.

Chu, G.  C.  (1967). Prior familiarity, perceived bias,  and one-sided
     versus  two-sided  communications.   Journal  of   Experimental
     Social Psychology, i, 243-254.

Cleveland,  W.   S.,  &  McGill,   R.  (1984).  Graphical perception:
     theory, experimentation,  and  application to the development
     of  graphical methods. Journal  of  the  Americal statistical
     Association. 22, 531-554.

Cohen, A. R., Stotland, £., 6  Wolfe,  0. M.  (1957). An experimental
     investigation of need for  cognition. Journal of Abnormal and
     Social Psychology. £1, 291-294.

Cohen, B. L., & Gromicko,  N.  (No date). Surveys of radon levels in
     U.  S.  houses.   Unpublished  manuscript.  Pittsburgh,  PA:
     University of Pittsburgh.

Cook, T.  D.,  & Flay,  B.  R.  (1978). The temporal  persistence of
     experimentally induced attitude change: An evaluative review.
     In  L.  Berkovitz  (Ed.),   Advances   in  experimental  social
     psychology (Vol. 11, pp. 2-57). New York: Academic Press.

Cvetkovich, G., & Earle, T. C.  (1988). Platitudes and comparisons:
     A   critique   of  current   (wrong)   directions   in  risk
     communication. Paper read at the Annual Meeting of the Society
     for Risk Analysis, October 30 - November 2,  1988, Washington,
     D. C.

Darke, S.  (1988). Effects of anxiety on inferential reasoning task
     performance* Journal of Personality  and Social Psychology. 55.
     499-505.

Deffenbachar,. J...L. (1978). worry, emotionality, and task-generated
     interference- in test-anxiety:  An empirical test of attentional
     theory. Journal of Educational  Psychology.  70.  248-254.

Desvousges,  W.  H.,  &  Cox, J.  A.  (1986). Radon focus groups:  A
     suanary.   Report,   Research  Triangle   Institute.   Research
     Triangle Park: NC.

Desvousges, W.  R.,  £ Hollander, M.  (1986). Radon focus groups:  A
     summary.   Report,   Research  Triangle   Institute.   Research
     Triangle Park: NC.

Oesvousges, W.  H.,  6 Smith, V.  K. (1988). Focus groups  and risk
     communication:  The  "science*  of   listening  to data. RiaJS
     Analysis.  £, 479-484.

Oesvousges,  W.  H.,  Smith, V.  K.,  6 Rink,   H.  H.,   III.  (1989).
     Communicating   radon risk effectively: Radon  testing   in
     Maryland.  Final Report  submitted  to the Office of Policy,
     Planning,  and Evaluation, EPA-230-03-89-048. Washington,  D.
     C.: United States  Environmental Protection  Agency.


                                83

-------
Dickie, M., Fisher, A., & Gerking, 5.  (1987). Market transactions
     and hypothetical demand data: A comparative study. Journal of
     the American Statistical Association. 8%.  69-75.

Dillman, D. A.  (1978). Mail and telephone surveys; Tha total design
     method. New York: Wiley.

Douglas, M., ft  Wildavsky,  A.  (1982). Risk and culture. Berkeley,
     CA: University of California Press.

Doyle, j.  K., Elliott, S.  R.,  McClelland,  6. ff., Russell, 6. W.,
     & Schulze,  W.  0.  (1989). An evaluation of policies for solving
     the  radon  problem.  Boulder,  CO:  University of  Colorado,
     Boulder, May.
                     •                «c-
Eagly, A. H. (1981). Recipient characteristics as determinants of
     responses to persuasion.  In  R.  E.  Petty, T. M. Ostrom, & T.
     C. Brock (Eds.)» Cognitive responses in  persuasion (pp. 173-
     195). Hillsdale, NJ: Erlbaum.

Edelstein, M. R., ---ft Boyle, V.  (1986). Media and the perception of
     risk.  In W. Makofske  and  M.  Edelstein (Eds.), Radon and the
     environment   (pp.  233-240).  Mahwah,  H.  J.:  Instute  for
     Environmental Studies, Ramapo College of New Jersey.

Ferguson, M. A., ft  Valenti,  J. M. (1988). Risk-taking tendencilV
     and radon messages:  A field experiment testing an information
     processing  model for  risk communication. Paper presented to
     Mass  Communication and  Society  Division,  Association  for
     Education  In  Journalism  and Mass  Communication,  Portland,
     Oregon, July.

Ferguson, M. A., Valenti, J. M.,  ft Melwani, 6.  (1988). Effects of
     adventurousnessy impulsiveness, ft rebelliousness on health and
     environmental  risk-taking   attitudes'.   Paper  submitted  to
     International Communication Association,  Health Communication
     Division, November 1.

Fischheff,  B.,  ft Beythy R.  (1975). "I  kxfeV it  would  happen1*  -
     reaebered probabilities of once-future things. Organizational
     Behavior and Human Performance. 13. 1-16.

Fischhoff, B., Slovic, P.,  Lichtenstein,  S.,  Read,  S., ft Combs, B.
     (1979).  How safe is  safe enough?  A psychometric  study of
     attitudes  towards  technological risks  and benefits. Policy
     Sciences. 2* 127-152.

Fishbein, M., ft  Ajzen, I. Belief, attitude,  intention and behavior.
     Reading, Mass.: Addison-Wesley.

Fisher, A.,  ft  Sjoberg,  L. Radon  risks:  People's perceptions anV
     reactions. In Majumdar, Schmalz ft Miller (Eds.), Environmental
     radon:  Occurrence,  control  and health  hazards.  Pittsburgh:
     Pennsylvania Academy of Science, in press.
                                84

-------
Fisher, A., Chestnut,  L. ,  &  Violette,  D.  M. (1988). The value of
     reducing risks  of  death:  A note on new evidence. Journal g£
     Policy Analysis and Management. &, 88-100.

Florig, H.  K.,  & Morgan, M.  G.  (No  date).  A simulated house for
     learning   about  radon.   Pittsburgh,   Pa:  Carnegie  Mellon
     University, Department of Engineering  and Public Policy.

Freudenberg,  W. R.  (1988).  Perceived  risk,  real  risk  - social
     science and the art of probable risk assessment, scieneg. 242.
     44-49.

Guimond, R., Elliott, W., Fitzgerald, J., Windham,  S., & Cuay, P.
     (1979). Indoor radiation exposure due to Radiua-226 in Florida
     Phosphate:  lands. Washington, D. C.: Environmental Protection
     Agency, July.

Hileman, B. (1988).  Fluoridation of water.  Chemical and Engineer-
     ing News.  £$(31),  26-42.

Isen, A. M.,  Means, B., Patrick, R.,  & Nowicki,  G. (1982). Some
     factors influencing decision making strategy and risk taking.
     In M.  S.  Clark 6 S. T. Fiske (Eds*)/ Affect and cognition  (pp.
     243-261).  Hillsdale, NJ: Erlbaua.

Janis, I.  L.  (1962). Psychological  effects of warnings.  In G. W.
     Baker  & E. W.  Chapman (Eds.),  Man and society and disaster.
     Hew York:  Basic Books.

Janis, I.  L.  (1982). Stress,  attitudes, and decisions> New York:
     Praeger Books.

Janz, N. K.,  &  Becker, M. H.  (1984).  The  health belief model: A
     decade later. Health Education Quarterly. 11.  1-47.

Johnson, E., 4-Tyersky, A. (1983). Affect,  generalization and the
     perception of  risk.  Journal  of Personality and  Social
     Psychology. 45. 20-31.

Johnson, P. R., &  Fisher,  A.  (1989). Conventional  wisdom on risk
     communication  and  evidence  from  a  field  experiment.  Risk,
     Analvain.  £, 209-213.

Johnson, F. R.,  & Luken, R. A.  (1987).  Radon risk  information and
     voluntary protection: Evidence from a natural experiment. RisJs
     Analvaia.  2, 97-107.

Johnson, F.  R., Fisher,  A.,  Smith, V. K., £  Oesvousges,  w. H.
     (1988).  Informed choice or  regulated risk?  Lessons  from a
     study in radon risk communication. Environment. lfi(4), 12-35.

Johnson, R.  (1987). Radon control  policies in  Sweden:  A status
     report. EPA Document 870722.

Kerr, R. A. (1988).  Indoor radon: The deadliest pollutant, science.
     240.  606-608.


                                85

-------
Klotz,  M.   L.,  Weinstein,  N.  D. ,  &  Sandman,  P.  M.   (Undateo^H
     Promoting  remedial  response  to  the  risk  of  radon:  Are
     information campaigns  enough? Unpublished paper.

Krimsky, S.,  &  Plough,  A.  (1988).  Environmental  hazardsi  Com-
     municatinas risks as a social process.  Dover, Massachusetts:
     Auburn.

Krosnick, J. A.  (1988). The role  of attitude importance in social
     evaluation:  A  study  of policy   preferences,  presidential
     candidate  evaluations,  and voting  behavior.  Journal  of
     Personality and Social Psychology. 5{jf  196-210.

Krosnick,  J.  A.,   &  Schuman,   H.   (1988).   Attitude  intensity,
     importance,  and  certainty   and  susceptibility  to response
     effects. Journal  of Personality  and  Social Psychology. 54.
     940-952.

Kunda, Z. (1987). Motivated inference:  Self-serving generation and
     evaluation  of causal  theories.  Journal of  Personality and
     Social Psychology. £i, 636-647.
Kunreuther, H.,  Ginzberg,  R., Miller, L.,  Sagi,  P., Slovic, P.,
     Borkan, B., & Katz, N. (1978).  Disaster insurance protection;
     Public policy lessons. New York: Wiley.

Lafore, M. (1987). Radon. The invisible threat.  Emmaus,  Pa: RodaJ
     Press.

Lau, R.,  Kane,  R.,  Berry, S., Ware, J.,  &  Roy, D.  (1980). chan-
     neling health: A review of the evaluation of televised health
     campaigns. Health  Education Quarterly.  7_*  56-89.

Lehman, D. R.,  & Taylor, S. E.  (1987).  Date with an earthquake:
     Coping with  a  probable,  unpredictable  disaster. Personality
     and Social Psychology Bulletin. 13.  546-555.

Lehrer, B.  (1989).  The radon response:   A  literature  review and
     discussion. Risk Assessment and Risk Management. 1988.

Leventhal, H.  (1970).   Findings and theory  in  the study of fear
     conunications. In L. Berkowitz (Ed.),  Advances in  social ex-
     perimental psychology. Vol. 5. New York: Academic  Press.

Leventhal, H.  (1986).   Health  psychology:  A social  psychological
     perspective.  In  L.  Berkowitz (Ed.),  A  survey  of social
     psychology (3rd ed.), (pp. 448-476).  New York: Holt,  Rinehart
     & Winston.

Magat,  w.  A.,  Payne, J. W.,  & Brucato,  Jr.,  P.  F. (1986). How
     important is information format? An experimental study of hom«
     energy  audit  programs.
     Management. £, 20-34.
                                86

-------
Maher, E. F. (1988). Characterization and control of Radon-222 and
     its progeny  in buildings. Aviation Space and  Environmental
     Medicine. 12(11), Section II, A93-A99.

Makofske,  W.,  &  Edelstein,  M. (Eds.)  (1987)  Radon  and the en-
     vironment. Mahwah, NJ: Ramapo College.

Mazur,   A.   (1981).  Tha  dvnamiea   of  technical   controversy.
     Washington,  DC: Communications  Press.

Mazur, A.  (1984). The journalists and technology: Reporting  about
     Love Canal and Three Mile Island. Minarva. 22.,  45-66.

Mazur,  A.   (1987).  Putting radon  on the public's  risk agenda.
     Science. Technology. & Human Values. 12. 86-93.

Mazur, A.  (No date).  Does public perception of risk explain the
     social  response  to  potential  hazard?  Quarterly Journal  of
     Ideology.

Mazur, A.,  &  Hall,  G.  (No date).  Effects of social  influence and
     objective risk on response to radon. Unpublished manuscript.

McGuire, W. J. (1985). Attitudes and attitude change. In G. Lindzey
     & E. Aronson (Eds.), The Handbook of Social  Psychology. Vol.
     2.  3rd Ed. (pp. 233-304). New York: Random House.

Mileti,  ,D.,  &  Sorensen,  J.  H.   (1987).  Natural  hazards and
     precautionary behavior. In N. D. Weinstein (Ed.), Taking care;
     Understanding  and  encouraging  self-protective behavior. New
     York:  Cambridge University Press.

Miller,  S.  M. (1980). When  is a little information a  dangerous
     thing?  Coping with stressful life-events  by monitoring vs.
     blunting. In S.._Levine  & H.  Ursin  (Eds.), coning  and health
      (pp. 145-J.6?)'." New York: Plenum Press.

Miller,  S.  M., Brody, D.  S.,  & Summertbn,  J.  (1988).   Styles  of
     coping with  threat:  Implications for health. Journal of Per-
     sonality and Social  Psychology. 54.  147-148%

Misch*!,   W.   (1973).   Toward   a   cognitive   social  learning
     reconceptualization of personality. Psychological Review. &£,
     252-283.

National Research Council, Committee on  the  Biological  Effects  of
     Ionizing Radiation  (1988). Health effects  of radon and  other
     internally deposited aloha-emittera fBEIR IV).  Washington,  D.
     C.: National Academy Press.

Nero, A. (1989).  Earth, air,  radon and home.  Phvaiea Today. ii(4),
     32-39.

Office of Radiation Programs, US Environmental  Protection Agency
      (1986),,  Press  release, August 14.
                                87

-------
Parducci, A.,  & Perrett,  L.  F.  (1971).  Category rating
     Effects of relative spacing and frequency of  stimulus values.
     Journal of Experimental Psychology.  8.9 f  427-452.

Perloff, L. s.  (1983). Perceptions of vulnerability to victimiza-
     tion. Journal of Social Issues. 39.  41-61.

Petty, R. E., & Cacioppo, J. T.  (1981). Attitudes and persuasion:
     Classic and contemporary approaches. Dubuque, Iowa: Brown.

Petty, R. 2., 6 Cacioppo, J.  T.  (1986). The elaboration likelihood
     modal  of  persuasion.  In  L. Berfcowitz   (Ed.),  Advances in
     experimental  social psychology (Vol.  19) (pp) .   Mew York:
     Academic Press.

PusJca,  P.,  Tuomilehto,  J., Salonen, J.,  NeittaanmaJci ,  L. , Maki,
     J., Virtamo, J.,  Nissinen,  A., Koskela, K. , Takalo, T.  (1979).
     Changes in coronary risk factors during a comprehensive five-
     year  community program  to  control  cardiovascular diseases
      (North Karelia Project).  British Medical Journal, i,  No. 6198,
     1177-1178.

Radon   Risk  Communication  Project'.   (1987).  Interim  Report.
     Washington,  0.  C. :  United  States  Environmental Protection
     Agency, Office of Policy,  Planning and Evaluation.

Rogers, R. W. (1983). Cognitive  and psychological processes in  fea?
     appeals and attitude changes: A revised  theory of protection
     motivation. In J. T. Cacioppo and R. E.  Petty  (Eds.), Social
     psvehophvs iologv . Boston:  Reidel.

Rogers, R. W. (1987). Cognitive  and physiological processes in  fear
     appeals and  attitude change: A revised theory of protection
     motivation.  In J.  T. Cacioppo  6 R.  E.  Petty (Eds.),
     PsyehophysiolQcrv  (pp. 153-176) . New York: Guildford Press.

Rotblat, J. (1988). A tale of two cities. New Scientist. 111(1594),
     46-50.

Sandman, P.  (1986).  Explaining environmental risk. Some notes on
     environmental  risk  communication.  Washington, D. C. : United
     States  Environmental  Protection  Agency,  Office  of Toxic
     Substance*.

Sandman, P.  M. ,  We ins te in,  N. 0.,  6 Klotz,  M. L.  (1987). Public
     response  to  the  risk   from  geological  radon.  Journal of
                   .  37.  93—108.
Schneider, W., & Sniff rin, R. M.  (1977). Controlled and  automatic
     information  processing:!.  Detection,  search and attention.
     Psychological Review. 84. 1-66.

Seven Cardinal Rules of Risk Communication.  (1988). Washington,  D.
     C. : United States Environmental  Protection Agency.
                                88

-------
Sharlin, H. I.  (1987).  EDB:  A case study in the communication of
     health risJc. In B.  Johnson & V. Covello (Eds.), The sogj^i fln^
     cultural construction of  risfr.  Boston: Reidel.

Short, Jr» , J.  (1984).  Toward  the  social transformation of risk
     analysis. American Sociological Review. 49. 711-725.

Silver, M., ft Sabini, J.  (1981). Procrastinating.  Journal for tha
     Theory of Social Behaviour. lj. 207-221.

Sjoberg* L. (1979). Strength of belief and risk. Policy Sciences.
     II, 39-57.

Sjoberg,  L.   (1980).  Volitional problems  in carrying  through a
     difficult decision. Aeta  Psychologies . ±S, 123-132.
Sjoberg, L.  (1985).  Motivation and: compliance: Row assessed, how
     important.  In J.  Hirsch ft T.  B.  Van Itallie (Eds.), Recent
     Advances  in Obesity Research  IV  (pp. 275-280) .  London: John
     Libbey.

Sjoberg, L, , ft Biel,  A. (1983).  Mood and belief-value correlation.
     Acta Psychologies. 53. 253-270. y.

Sjoberg, L.. ,  & Orottz, B.-M.  (1987).  Psychological reactions to
     cancer risks after the Chernobyl accident. Medical  Oncology
     and Tumor Pharmacotherapv. £,  259-271.

Sjoberg, L. ,  & Drottz, B.-M.  (1988).  Radiation risks: Knowledge
     and subjective experience among nuclear power plant employees.
     Paper  read  at  the Annual Meeting  of the  Society  for Risk
     Analysis, October 30 - November 2, 1988, Washington, 0.

Sjoberg, L. ,  ft Winroth, E.  (1986).  Risk,  moral value of  actions,
     and mood. Seandj^-navian Journal of Psychology.. 27. 191-208.

Slavic,, P., ft Lichtenstein, s.  (1971). Comparison of Bayesian and
     regression  approaches to the study of information, processing
     in judgment. Organizational  Behavior and Hunan Performance.
     1, 649-744.

Slovicv  P.*  Flschhbff,  B.,  ft  Lichtenstein,* S.  (1981).  Charac-
     terizing  perceived risk.  In R.  W.  Kates  ft  C.  Hohenhemser
     (Ed*.),,  Technological  hazard  management.  Cambridge, Mass.:
     Oelgeschlager,  Gunn ft Hain.

Smith, V.  K., ft Johnson, F.  R.  (1988).  How  do risk perceptions
     respond  to  information?  The case of  radon. The Review of
     Economies and Statistics.  70.  1-8.

Smith,  V.  K. , Desvousges,  W.  H.,  ft  Fisher,  A., ft  Johnson,  R.
     (1987).  Communicating  redon risk  effectively:  A mid-course
     evaluation. EPA-2 3 0-07-87-029.
                                89

-------
Smith, V. K. , D«svousges, w. H. ,  Fisher,  A.,  &  Johnson, R.
     Learning about radon's risJc. Journal of  Risk and Uncertainty
     I, 233-258.

Smith, V.  K. , Desvousges,  W.  H., Johnson,  F. R.,  &  Fisher, A.
     (1988).   Can  public   information  programs   affect   risk
     perception? Paper submitted for publication.

Sorrentino, R. M. , & Hancock, R.  D. (1987). The role of information
     and affective value for social influence: A case for  the study
     of individual differences.  In M. P. Zanna, J. M. Olson,  & C.
     Pi Herman (Eds.),  Social influence;  The  Ontarior  Symposia
     (Vol. 5, pp. 244-268). Hillsdale, NJ: Erlbaum.

Sorrentino, R. M. ,  Bobocel, D. R. ,  Gitta,  M. Z., Olson,  J. M. , &
     Hewitt, E. C. (1988) .  Uncertainty orientation and persuasion:
     Individual differences in the  effects of personal  relevance
     on.  social  judgments.  Journal  of  Personality  and  Social
     Psychology. ££, 357-371.

Sorrentino, R. M., Short, J., &  Raynor, J. O.  (1984). Uncertainty
     orientation: Implications for  affective and cognitive views
     of achievement behavior.  Journal of  Personality  and Social,
     Psychology. ±£, 189-206.
Stallen,  P.  J.  M. ,  &  Thomas,  A.  (1988).  Public  concern abofl
     industrial hazards. Risk Analysis. &, 237-245.

Swedish National Institute for Radiation Protection  (1987) . Radon
     i  bostader.  SSI-rapport  87-17.  Stockholm,  Sweden:   Statens
     stralskyddsinstitut.

Tesser, A. (1978). Self -generated attitude change. In L. Berkowitz
     (Ed.), Advances in experimental social psychology (Vol.11, pp.
     289-338). New York: Academic Press.

Tufte,  E.   R.'  (1983).  The  visual   display  of  quantitative
     information. Cheshire, CT: Graphics Press.

U.S.  Environmental  Protection  Agency  (1986).  Radon  reduction
     methods; A home owner's guide. OPA-86-005. Washington, D. C. :
     U.S. Environmental Protection Agency.

U.S.  Environmental  Protection  Agency  (1987  a). Evaluating and
     improving EPA's risk advisory programs.  Washington,  0. C. :
     U.S.  Environmental  Protection Agency,   Program  Evaluation
     Division.

U.S. Environmental  Protection Agency  (1987 b) . Radon  risk com-
     munication project.  Interim  report.  Washington, 0. C. : U.S.
     Environmental Protection Agency,  Office of Policy, Planninjj
     & Evaluation.                                              m

U.S. Environmental Protection Agency (1987 c) .  unfinished business;
     A   comparative   assessment  of   environmental   problems.
     Washington, 0.  C. : U.S. Environmental Protection Agency.
                                        «

                                90

-------
U.S. Environmental Protection Agency (1988  a). Region 3/OPPE/stati»
     of Maryland  Radon  Risk Communication Prelect; An Evaluation
     of Radon  Risk Communication Approaches.  Washington,  D. C. :
     U.S.  Environmental  Protection  Agency,   Office   of  Policy,
     Planning, and Evaluation, Program Evaluation  Division.
U.S. Environmental Protection Agency (1988 b) . Seven Cardinal
     of Risk Communication. Washington, D.  C. : U.S. Environmental
     Protection Agency.

U.S.  Environmental  Protection Agency and  Centers  for Disease
     Control  (1986) •  A citizen's guide, to  radon;  What it is and
     what  to  do  about it. OPA-86-004.  Washington,  D.  C. :  U.S.
     Environmental Protection Agency.

Unger, D. , 6 Wander sman, A. (1985).  The importance  of  neighboring:
     The social, .cognitive, and affective components  of neighbor-
     ing. American Journal of Community Psychology. 13. 139-169.

United Nations Scientific Committee  on Effects of Atomic Radiation.
     (1977) . Sources  and  effects of ionizing  radiation. New York:
     United Nations.          a

Valenti, J. M. , 6 Ferguson, M. A. (1987).  Radon awareness,  concern,
     responsibility, and source confidence: A survey of home-owners
     in  three  Florida counties— Alachua,  Hillsborough  6  Polk.
     Unpublished manuscript.  Gainsville, FL: University of Florida,
     College of Journalism and Communications.

Valenti, J. M., & Ferguson, N. A. (No date).  Risk-taking tendencies
     and  radon messages:  A   field  experiment.  Paper No. II-P4.
     University of Florida, Communication Research  Center, College
     of Journalism and  Communications^

Viscusi, W. K.,  Magat, w. A.,  &  Huber,* J.T.U986)  . Informational
     regulation of consumer health  risks:. An  empirical .evaluation
     of hazard warnings.  Rand Journal  of Economies. 17. 351-365.

Wallaten,  T.  S. t  6,  Budescu,  D.  V.  (1983).  Encoding subjective
     probabilities:   A  psychological  and  psychometric  review.
             nt Seifme*.  29.  151-173.
Wallston,  B.  S., Wallston,  K. A.,  Kaplan,  6., &  Maides,  S. A.
      (1976) .  Development and  validation  of the Health Locus of
     Control  (HLC)   Scale.   Journal  of  Consulting  and Clinical
     Psychology. 44, 580-585.

Weinstein,  N.  D. (1980). Unrealistic  optimism about future  life
     events. Journal of Personality and Social Psychology. i9_,  806-
     820.

Weinstein, N. D. (1984). Why it won't happen to me:  Perceptions of
     risk  factors and  susceptibility.  Health Psychology. I,  431-
     457.
                                91

-------
Weinstein, N. D. (1987) .  Unrealistic optimism about susceptibi
     to health problems:  Conclusions from a community-wide sam
     Journal of Behavioral Medicine.  3^0.  481-500.

Weinstein, N.  D.  (1988).  The precaution adoption process. Health
     Psychology. 1, 355-386.

Weinstein, N. D., Klotz, M. L., & Sandman,  P. M.  (1988). Optimis-
     tic  biases in public perceptions  of the  risk  from radon.
     American Journal of Pyfrljic Health.  78.  796-800.

Weinstein, N.  D.,  Sandman, P. N.,  4  Klotz, M.  L. (1987). Public
     response to the risk from radon,  1986.  Final Report. Rutgers,
     The State University of New Jersey: Cook College.

Weinstein, N.  D., Sandman,  P. M.,  4 Roberts,  N.  E.  (Undated).
     Determinants of self-protective behavior:  Home radon testing.
     Unpublished.  Rutgers, The State University of  New Jersey:
     Cook College.

Weinstein,  N.   D.,  Sandman,  P.  N.,   &  Roberts,  N.   E.  (1988).
     Homeowner  radon  mitigation.  Report   to   the   Division  of
     Environmental Quality, New Jersey Department of Environmental
     Protection.

Weinstein,  N.   D.,  Sandman,  P.  M.,   &  Roberts,  N.   E.  (1989
     Communicating effectively about changes in risk.  Final Rep
     to  the   Office  -of  Policy  Analysis,  U.S.  Environmen
     Protection Agency.

Wicker, A. W.  (1969).  Attitudes vs. actions; The relationship of
     verbal  and overt behavioral responses to attitude objects.
     Journal of Social Issues. 25.  41-78.

Wiio, J., Puska, P. Koskela, K., 4 McAlister, A.  (1986).  Television
     in national health, promotion in-Finland: The 'Keys to health*
     TV  program in 1982.  In D.  S. Leathar, 6.  B.  Hastings,  K.
     O'Reilly 4 J. K. Davies  (Eds.),

Zuckerman, M.  (1988). Behavior and  biology:  Research  on sensation
     seeking and reactions  to the media.  In L. Donohew,  H.  E.
     Sypher,  4  E.  T.  Higgins  (Eds.),   Communication,  social
     eomition. and affacrfc  (pp. 173-194). Hillsdale,  NJ: Lawrence
     Erlbaum Associates.
                                92

-------
                             APPENDIX

                     COMMUNICATING RADON RISK.

        Practical implications of recent research findings

This appendix provides  a  summary of principles of effective risk
communication applied to radon risk, based on current research.

     The basic principles of good communication are a first step
for  successful  communication  of radon  risk.  Guidance  on these
principles  is  readily  available  elsewhere   (e.g  EPA's  "Seven
Cardinal Rules of Risk Communication").

     Research on persuasion and attitude change has shown that it
is difficult to change attitudes and even more difficult to change
behavior. Knowledge  is  easier to communicate  but knowledge is by
no means sufficient  to cause a change  in  behavior or attitudes.
Fortunately, it has still been found that skillful and enthusiastic
attempts at communication of radon risks have had some success.

     People  often have biased  perceptions  of  risks.   They  may
dismiss or exaggerate them,  depending on many factors. Unlike such
risks as nuclear power  or toxic wastes, the largest problem with
radon is that people tend to dismiss or deny the risk.

     There are several reasons why people tend to deny and dismiss
radon risk:

- the risk is associated with one's home which is usually perceived
as a source of security and as part of one's identity.

-  acknowledging  the risk  carries  with  it  both, economic  and
psychological  costs:  it requires  testing and possibly  costly
mitigation, and also worry about health effect*.
             '•(•'•..    :  •              . - .        T . ; .
- adverse effects will not show up for a long tin*, and: people are
more concerned about the short run.

• the risk is perceived as natural rather than man-made.

     The objective of any radon risk communication program should
be to •rouso -  but not alarm - people  and motivate them to take
appropriate action. The following points can help in designing such
a program:

1. People are more likely to acknowledge and act  on a risk if they
(a) see it as a serious threat, and  (b) perceive that measures to
mitigate the risk exist and  are within their  reach.  Both these
points must be formulated clearly. First;, people need to be aware
of the  risk.  Radon cannot be perceived  with the senses, and its
effects, if any, are delayed. People need to be informed that even
if they have already been exposed to radon for a long time, it is
never too late to mitigate  since the body  can repair some damage
if exposure is  decreased so any  reduction  in  exposure reduces
risks.  Make  then aware  that  test  kits  are  easy  to get  and


                                93

-------
inexpensive.  Try  to  diminish  fears  that  mitigation   is
expensive and complicated.

     This suggests the following strategy:

A.  Raise  awareness: "Say,  have you  heard  about Radon?  It  is a
colorless and odorless radioactive gas that seeps into some homes
and can cause lung cancer".

B.  "It is easy  to find out whether your home has dangerous radon
levels - testing  is inexpensive and you can get a do-it-yourself
test kit at  your  nearby hardware store,  supermarket, or discount
store."
C.  "You  can stop worrying if  your radon level is  low.  If it is
high, mitigation is affordable and effective."

D. "What if you have already been living for^a long time in  a house
with high  radon levels?"  (Provide  material-about lungs repairing
damage).
E. "But I only plan to live here for a  few years, at the most, why
should I mitigate?" (Point out that mitigation apparently protects
resale  value -  however,  there  is little  research yet  on this
problem. Point  out that mitigation is a  way of protecting one's-
investment in a house, and a way of avoiding unpleasant surprisfl
when selling it).
2. People  frequently dismiss potential hazards when  it comes to
their own personal risk, yet accept that others, unrelated to them
but otherwise siail&r, are at risk. This tendency may be enhanced
when _the probability is small  (say, less  than 0.1)  or uncertain.
Sometimes these difficulties may be avoided by accumulating risks
over a longer time period or by  framing them in appropriate terms,
e. g., the number of  people in the community likely to develop lung
cancer from a.given.level of radon, exposure.

3. People  dislike uncertainty  and may use  it  as an  excuse for
disregarding a  risk  message that  indicates  a  lack  of complete
understanding of a  risk  or  conflicts among  experts.  Of course,
honesty is essential. The best strategy is to  formulate the message
that although experts do not have  a complete understanding of all
issues  involved,   they  do  agree  about  important  practical
conclusions.

4. Because it is hard to understand probabilities,  people prefer
safety thresholds. They dislike being faced with  a  complex decision
involving  trade-offs between various  value dimensions  that are
difficult to compare. Many people  like to have someone they trust
make a difficult  decision for them or to  be advised according ta
a simple rule of the  kind "act if value exceeds x". This means thaM
EPA1 s action guidelines will often be treated  as safety standards!

5. Personalized risks, i. e. messages that are concrete and vivid
and that depict people with whom one can identify, are essential.


                                94

-------
6. Direct social influence from friends and neighbors is probably
the most  powerful  way of affecting behavior.  However,  it is not
easy to initiate such a process and, in the case of radon, people
may be unwilling to tell others about their experiences. One tactic
that has  worked is for community leaders  to  support the testing
program by publicizing tests  of  their own homes and to encourage
others  to test.  Try to  encourage  people who  have tested and
mitigated  to  tell  their  friends  and  neighbors  about  their
experience,, It should be seen as  a smart and sensible thing to do.
It may help to associate testing with scientific and technological
advances.

7. There is some evidence that too little information may make some
people unnecessarily anxious about radon risk.  Too much information
may, on th« other  hand, be detrimental because it is too much to
absorb. People  are exposed to an enormous amount  of mass media
information.  The obvious  conclusion  is  that it  is  important to
formulate complete, but concise,  messages.

8. Use as many  media  channels as possible. Design the message so
as to attract attention, make it concrete and repeat it often. A
utility  bill  flyer  had  encouraging  results.  Posters  required
considerable'effort to place. Effectiveness of PSA's on radio has
been difficult to  judge. EPA has produced a slide show,  including
an audio tape and script. It is available through the Radon Office,
phone 202/475-9605.

9. Research has shown that people are especially averse to certain
losses. Here is a  strategy that takes advantage of that tendency:

"Are you worried about what  radon may do to you and your family?
Radon risk is one  worry you  can  get rid  of1  A radon test of your
home will either relieve you or all future worry over the issue -
 a likely outcome  -  or it will be  the first  step in permanently
removing a threat  to you and your family".

This gives  a  positive message rather  than a  negative  one, while
doing nothing leaves one more thing to worry about.
10.  SOB* data suggest  that women are  more likely to  carry out
testing  than men,  and that they are particularly concerned about
their children's health so work especially with women and emphasize
that  tasting  for  radon  is very  easy and requires no  special
technical skill.

11. As pointed out earlier, information campaigns for motivating
voluntary changes  in behavior are likely to have limited success.
Legislation has often turned out to be necessary in order to make
a  difference,  as  in the  case of  seatbelt use  in automobiles.
Effective reduction of radon risks may involve

(a)  requiring  construction techniques  that  reduce likely  in-
filtration of  radon; and
                                95

-------
                             ngl
(b)  requiring testing at the time of a real estate sale, prob
after the buyer takes possession to avoid inaccurate low readin

     Summing up, we conclude that risk communication is difficult
but by no means impossible. Important results have been achieved,
and the principles discussed here have shown promise in practical
applications. Further results are expected  fron on-going research,
especially on mitigation behavior.
96

-------