Effects of Air Pollution  on
Public Attitudes and Knowledge
           Oregon State University
               Corvallis
              June 1972

-------
                           ACKNOWLEDGMENTS

     This publication is the final report of a three-year study which was
conducted under the terms of contract number CPA 70^-117, Air Pollution
Control Office, Environmental Protection Agency.  The research began
May 23, 1969 and terminated June 30, 1972.

     Many people have contributed to the completion of various phases
of this work.  Pamela Bodenroeder joined the project in September 1969
as assistant leader and has worked tirelessly on all phases of the study.
Assisting her from time to time were Dorothy Bowers, Betty Lou Brooks,
Diana Coulombe, Alice Doyle, Julia Fisher, Ann Ginn, Gay Greger, Alice
Henderson, Linda Henderson, Kathleen Hulburt, Margaret Kent, Cheryl Miller,
Kathleen O'Dwyer, Penny Schmidt, Lora Seleen, Gretchen Wagner, Debora
Wahlquist and Nancy Young.  Marian Richardson and Joyce Driscoll completed
many of the arduous typing and other secretarial tasks.

     The nature of the research problem interested professionals from
different disciplines and the author greatly profited from an association
with many who contributed to his understanding of the complexity and subtlety
of air pollution processes and effects.  A number of individuals at Oregon
State University generously gave of their time to advise on the many
technical aspects of this study.  These were R. M. Alexander, Air and Water
Resources Center; Richard Boubel, Mechanical Engineering; William G. Brown,
Albert N. Halter and Herbert H. Stoevener, Agricultural Economics; Lyle
Calvin and Norbert Hartmann, Statistics; David Chilcote, Farm Crops; and
E. Wendell Hewson and Wesley Tufts, Atmospheric Sciences.  As well,
Richard E. Hatchard, Columbia-Willamette Air Pollution Authority; Michael
D. Roach, Mid-Willamette Air Pollution Authority; and Thomas E. Waddell,
Environmental Protection Agency, provided most useful professional help
and advice.

-------
  EFFECTS OF AIR POLLUTION ON
PUBLIC ATTITUDES AND KNOWLEDGE
           Robert G.  Mason

        Professor of Sociology

        Oregon State University
             June 1972

-------
                                CONTENTS


Acknowledgments	     i

Summary, Conclusions  and Recommendations   	     1

Introduction  	                    	     9

Theoretical Framework for Public Attitudes and Knowledge	    10
     Public Attitudes and Governmental Action
     Environmental Quality as a Public Issue
     Enforcement and Community Conflict
     Attitudes and Definition of the Situation
     Definition of Air Pollution Situations
     An Air Quality Communication Model

Methods--Content  Analysis and Attitude Change	    29

     Selection of Location and Time
     Development of Operational Measures
     Data for Hypothesis Testing
     Analysis  of Data

Results and Discussion--Content Analysis and Attitude Change	    40
     Results of the Content Analysis
     Analysis  of Visibility Observations
     Analysis  of Attitude Change Scores
     Discussion of  Results

Methods--Air Quality Communication Model	    53

     Development and Evaluation of Operational Measures
     Execution of the Sample  Survey
     Analysis  of Data

Results and Discussion--A Test of the Air Quality Communication Model .    67

     Results of Accuracy of Air Pollution Perceptions
     Empirical Test of the Communication Model
     Empirical Test of the Reformulated Communication Model
     Discussion of  Results

Literature Cited	    90
Appendix A:  Review of  Literature	    95
Appendix B:  Interview Schedules A and B	101

Appendix G:  Frequency Curves of Clean  Air Themes	122

Appendix D:  Verbatim Responses of Individuals Holding Accurate
             or Inaccurate Visibility Perceptions.	  136

-------
              SUMMARY, CONCLUSIONS AND RECOMMENDATIONS






     The purpose of this research was twofold:  First,  a test was made of a



hypothesis concerning the relationship between mass media content and attitude



change.  More specifically, the relationship stemmed  from the hypothesis



that perceived changes in air quality in a community are defined or redefined,



in part,  by the local mass media.  As well,  a  definition or redefinition in the



media is sufficient for people to  change their attitudes toward the topic.



Second,  a  test was made of a theoretical  model concerning the communication



of air quality information to the public.  Specifically,  the model stipulated



that three  variables were mutually interdependent and  were considered



endogenous.  These variables were attitude toward  air quality,  mass media



exposure to information about air quality and level of air quality knowledge.



Variables  considered exogenous  were levels of interpersonal discussion,



social influence, levels of formal and science  education and use of the environ-



ment for recreation purposes.






     An empirical test of the mass media content  - attitude change  hypothesis



involved a content analysis  of two daily newspapers, one each from two



communities.  A panel design was employed to measure attitude change of a



random  sample of residents in the two communities.  These communities



were selected on the basis of known changes in air quality levels during the



late summer.  The  communities  were Eugene  and Salem,  Oregon.   Because



of its location at the base of the Willamette Valley,  Eugene has experienced



poor natural ventilation whereas  Salem,  located in the mid-Willamette Valley,



has  had good  natural ventilation.   Therefore, a contrast in visibility would be



expected to occur when there is air pollution in the  Valley.  Personal inter-



views (for measuring attitudes) were timed in  1970  and in early 1971 to take



advantage  of this known change in air quality.   The  source of air pollution



was agricultural field burning and the measure of air quality employed was



level of visibility.  Visibility observations were completed during  interviewing



periods to  determine  if changes in visibility occurred as anticipated.  The

-------
recording unit for the content analysis was the theme and the unit of analysis



was the article.  Fifteen themes were developed and pretested.   A  total of



3, 054 newspaper articles from an 11-year period (1960-1970) was analyzed.



Frequency of themes, type of article and source of pollution mentioned were



the data employed for determining if definition or redefinition had occurred.



Attitudes toward a'.r quality were measured by interviewing a random  sample



of residents  in each community twice, once when visibility was high and once



when it was low.  Adjusted attitude change scores were employed to determine



if a change in air quality was associated with an observed change in attitude.



Results of  the panel, interview indicated that observed changes in attitudes



were produced primarily by errors of measurement associated with attitude



change  scores and not by changes in air quality.   Air quality did, however,



change  in the anticipated directions,  accorded to analysis of visibility



observations.






     Results  of the mass media analysis  suggested that the  air quality  situation



in Eugene was defined in late 1966 or early 1967;  in Salem,  definition likely



occurred in August 1969-  An air pollution episode occurred in Eugene in



August  1969.  In response, the state  governor banned field burning for one



week and placed control of future burning in the hands of a  state environmental



control agency.  This action  may well have contributed to defining the  situation



for Salem and for the entire state as  well.  One can conclude that the findings



are weak,  but positive,  support for the definitional hypothesis concerning the



relationship  between mass  media content and  attitude change.






     An empirical test of the  communication model involved a personal inter-



view survey  of a random sample of individuals in the two communities



employed for testing the mass media - attitude change  hypothesis - Fugene



and Salem,  Oregon.  Operational measures of variables  in the theoretical



model were based on summated response scores of items asked in  a standard-



ized interview schedule. Items employed for each variable were extensively



pretested in a third community and subjected  to an item analysis.  Reliability



estimates were also calculated.  The personal interview survey was completed

-------
in 1970 and in  c:arly J971 within planned time- limits.  Completion rates were
well within acceptable  levels.   Significance of variables in the model was

tested by two-stage least squares.

     Results of the test of the communication model showed an error  of specifi-
cation in the model.  One exogenous variable, interpersonal discussion, should

have been considered endogenous.  A reformulated model was developed to
include this consideiation.  A new exogenous variable, belief that air pollution

was serious in the community,  was postulated exogenous  to level of attitude.—

It was also anticipated that different models may be operating for individuals
who accurately or inaccurately perceived changes  in visibility which  had

actually occurred.  A  test for pooling data of accurate and inaccurate sub-

groups disclosed  that data  could not be pooled for the two subgroups in Salem.

The model was analyzed separately for each subgroup.  For Eugene,  the
community with an air pollution episode in 1969,  a confounding of the per-

ception measure was discovered which  precluded attempts to pool data on
the basis of accuracy of perceptions.  A relatively large number of respond-

ents compared pollution levels  of 1970 with those of 1969,   rather than with
the previous interviewing periods in the panel.  The communication model
was tested in three instances.  One was for Eugene respondents.  Another
was for  respondents who held inaccurate perceptions in Salem.   A third was
for respondents also in Salem who held accurate perceptions.
     One can conclude that different models were functioning in each instance

and  suggest that each model may represent different stages of the same.

underlying communication process.  Initially,  when changes in air quality

are  not likely to impinge on the  individual's consciousness, attitudes are
— Data from the previous model were used  to test the reformulated model.
This test is less meaningful than if new data had been acquired since the
equations were developed after data from the previous model had been
examined.  Such selection tends to capitalize on relationships which might be
high because of sampling fluctuations.   Thus, caution should be employed in
viewing results of tests of the reformulated model.

-------
                                                                           4
formed primarily by interpersonal discussion and on relatively low levels of



knowledge.   Selectivity in mass media exposure and interpersonal discussion



is likely to  be operating.  As well, knowledge levels are formed partly on



this selective mass media exposure.  Later, when changes in air quality do



impinge on  the individual's consciousness, a lack of sufficient information



in the channels of communication is  likely.  Those with high knowledge levels



are likely to seek out information  from both local  mass media and inter-



personal sources.   Selectivity in information exposure has been abandoned



and attitudes are based primarily  on increased levels  of knowledge.  Finally,



after a  continued period of exposure to deteriorated air quality,  an inter-



dependence has been established between mass media  exposure and level of



knowledge.   A sufficiency of information is available  in the local mass media



and knowledgeable individuals seek out and use this information.  Level of



knowledge is an important basis for  determining level of attitudes  which, in



turn, is a predictor for level of interpersonal discussion.  Those who  are



involved in  interpersonal discussion are also greater users of the  mass media.



The effects of changes in air quality appear to be subtle, cumulative but



nonetheless  real,  when studied in  terms of cumulation  of knowledge and



attitude formation.  The channels  of communication play different  roles at



each stage of the communication process.  An understanding of these roles



should enhance the  efforts of those interested in speeding up the formation



of positive attitudes and the level of  knowledge of people toward this topic.






     The effects of exogenous variables -- social influence, science education,



use of the environment for  recreation and belief that air pollution is a  serious



community  problem -- occurred in their predicted direction for all models.



One effect of exogenous variables  noted was to determine the percent change



of each exogenous variable on the  percent change in attitudes.   Results showed



that the social  influence variable would produce the greatest change in air



quality attitudes for Eugene respondents and for Salem respondents who held



accurate air pollution perceptions.   In these instances  a 1% change in  social



influence would change attitudes toward air quality by  10% in Eugene and by 7%



in Salem.  Effects  of ocher exogenous variables were small.  However,  a

-------
change in use of the environment for recreation was  by far the greatest pro-



ducer  of attitude change among Salem  respondents who held inaccurate air



pollution perceptions.  A 1% change in use of the environment would produce



a 16%  change in attitudes.  Efforts to  sensitize people who hold inaccurate



air pollution perceptions should include attempts to persuade them to use the



environment more for recreation.  This practice  is likely to produce more



positive attitudes toward air quality and,  perhaps, a more accurate perception



of visibility restrictions.  It is concluded that it would be difficult to produce



more favorable air quality attitudes among  respondents holding accurate than



among respondents holding inaccurate visibility perceptions.   Social influence



is considered a more difficult attribute to change  than it  is to persuade an



individual  to use the  environment more for  recreation.






     Moreover,  social influence and interpersonal discussion concerning air



pollution play a rcle  in attitude formation, primarily at the initial stage of



the communication process when perceptions are  not likely to be  accurate.



The importance of interpersonal discussion is likely to be reduced once



knowledge levels and attitudes are well established and after perceptions



concerning air qualify become accurate.  As well,  the positive relationship



between science education and knowledge, found at all stages of the communi-



cation process, suggests that a  knowledge gap exists between the highly- and



poorly-educated segments of the community.  Over time, this gap likely will



become larger, not smaller.  Thus, any program to  employ the mass  media



to reach large numbers of people with information concerning air pollution is



likely  to reach the already well-informed.  These people will be  the primary



beneficiaries of such an effort.   Finally,  the belief that air pollution is



serious -- based on the individual's experience living or working in the



community -- appears  to have a bearing on  attitudes  at all stages of the



communication process.  A pre-condition for the  establishment of a successful



air pollution control  program is that the citizens of an airshed recognize



the scope and magnitude of the problem.  This pre-condition is satisfied when



they perceive odors and loss of  visibility, for example, for what they are and



in  terms which they believe  are  important to them.

-------
     It seems reasonable to set down four recommendations concerning tht;



role of public attitudes and knowledge in air pollution control efforts:






     1.  Air quality goals and standards should be established on the basis



of their social feasibility as well as their economic and technical feasibility.






     The  establishment of goals and standards is defined as  the concentration



of pollutants which the public believes it can accept without adversely affecting



its health or welfare.  Few will argue the relevancy of economic or techno-



logical criteria.  But,  it is argued here that public acceptance of air pollution



goals and standards  implies  the existence of a social definition of air quality



which merits consideration as well.  This social definition includes the role



of public attitudes and knowledge.   It is not argued that air quality goals and



standards  should mirror existing public attitudes or knowledge of the time.



Rather, public  concerns should be  considered in good faith prior to the



establishment of goals and standards.  In many instances the public hearing



requirement for implementation of the  Clean  Air Act of 1970 may well be



sufficient, or result in, the "good faith" requirement stipulated above.   In



other instances, it may well be that public concerns would be more adequately



considered by alternative methods.






     2.  Emission  control programs should be based,  in part, on the existence



of an informed  and supportive public.






     The technology associated with emission control programs  is likely to



become more,  rather than less, complex.  The public is likely to be exposed



to a number of  new ideas constantly and,  if accepted, some of these new  ideas



may help while  some may hinder the improvement of control programs.



Professionals who administer emission control programs should be prepared



to state their cases clearly and publicly and,  if possible,  to  secure public



support for the  changes they wish to make and for those they wish to ignore.



An informed public does not necessarily mean a supportive one;  yet,  it is



difficult to imagine a supportive public that is totally uninformed or is  mis-



informed.  It is quite likely that young people entering the  adult population

-------
are far mon- sophisticated in comprehension of the technical aspects of air



pollution,  for example,  than older citizens.  Greater sophistication of the young



likely stems from mere intensive formal science education and from greater



attention to environmental matters  in the  mass media, particularly news-



papers.   This demographic  implication should not be  ignored by professionals



in charge of emission control programs in their planning of information and



education activities.






     3.   Air pollution control authorities should become skilled in the use



of applied behavioral, science research in carrying out their emission control



programs.






     An objective assessment of the  strengths and weaknesses of  emission



control programs  is considered a necessary condition for the  subsequent



improvement or for the survival of the program itself.  Assume  that a



social definition, including public attitudes and knowledge,  of  control efforts



is an important consideration.  It follows that  this definition should be studied,



analyzed and interpreted to  establish the reference points for  coping with the



public issues which are likely to develop over  the implementation of future



control programs.   It is here that well-planned and professionally conducted



objective assessments of public attitudes,  for  example, will have a bearing.



Professionals who administer emission control programs,  among others,



should become familiar with the strengths and weaknesses  of public opinion



surveys, for example.   These professionals should be able to determine



what  such surveys can do  and cannot do and where they most effectively play



a role in evaluating an emission control program.  How effectively they are



used  depends on the  insight,  skill and judgment of individual administrators.



As well,  these administrators should have sufficient "back-up" expertise



available to evaluate the results of public attitude surveys which  may be



profferred by some interest group.   The quality of such surveys  may render



the results specious or genuine, for example.

-------
    4.  Additional research should be conducted  to provide a more complete



understanding of human reactions and behavior toward air pollution situations



and control efforts.






    An example is the gathering of additional data to verify the formulation



of the communication models developed in this study.  Specifically, models



should be viewed as  stages of an underlying communication process which



are embedded in a particular air pollution situation.  The situation itself



should be studied caiefully prior to  theory construction.  The role of economic



benefits and externalities of a polluter as well as feasibility of controls



should be considered not only for the specification of variables in the model



but also for  the selection of communities or situations employed for empirical



tests.   In addition, the ability of pollutants to impinge on the  consciousness



of individuals should be examined prior  to theory construction for  the same



reasons.  Finally, the model, or models,  should be extended to other



pollution situations where the  role of economic benefits,  externalities,



feasibility of controls and ability of pollutants to  impinge on the individual's



consciousness are likely to  differ.  With sufficient research one may



determine the generalizations required to  explain some aspect of human



behavior toward the phenomenon of envi ronmental pollution.

-------
                                INTRODUCTION






     The role of public knowledge and attitudes  in the administration of air



 pollution control policies and programs has received,  at best,  cursory



 attention among professionals in the  air quality field.  One of the few  to



 recognize that these variables may indeed play a role  is Schueneman [41]— ,



 who has pointed out the contribution of  an informed and active public to a



 successful emission control effort.   The process of setting air quality



 standards and enforcing emission restrictions  has  often relegated  the under-



 standing of people to a residual priority.  More often, attention is  focused



 on the preparation of air quality criteria,  goals and standards; on the



 development of emission goals and standards; on the stipulation of  design



 standards; and,  finally,  on  the establishment of measures  of tests  for



 pollutants, equipment and the ambient air.






     Granted that the establishment of standards is  important, is highly



 technical -- requiring  a specialized competence --  it is assumed nevertheless



 that, to be effective, these  standards must be accepted with a modicum of



 public understanding and support.   Establishment of standards implies that



 the decisions and actions taken by government to conserve (or not to conserve)



 air quality are based on valid knowledge about people and involve a degree of



 judgment about human  behavior.






     This report presents the results of testing  a public information model



 in which attitudes toward and knowledge about air quality play a role.  In



 this report,  first,  trie  linkage between  public attitudes and government action



 is discussed.  Seccnd,  environmental conditions which may have a bearing



 on variables in the model are considered.   Finally, a theoretical model con-



 cerning the acquisition and processing  of air quality information, by people,



 is formulated and tested.
— Numbers in brackets correspond to literature citations,  page 90.

-------
                                                                           10
                        THEORETICAL FRAMEWORK FOR
                     PUBLIC ATTITUDES AND KNOWLEDGE


                     Public Attitudes and Governmental Action


     It is perhaps axiomatic,  as Key [23] has noted,  that all governments must

concern themselves with public opinion;  that they do not maintain their author-

ity without some willing acceptance and conformity from most of those whom

they govern.  He considers popular government a special type in which one

base for authority is the seeking of popular opinion.  Here, government gives

weight to the opinicns of its  citizens in its decisions.  Persons  outside the

governmental establishment have a right to be heard.   The legitimation of

the 'right to be heard1 becomes an ethical,  if not a moral assumption under-

lying popular government and is operationalized into consistent patterns of

action by those who hold power.   Thus,  the linkage between public opinion

and governmental action rests not in a mirroring  of opinion and action but in

the expectation that public preferences and mass interests are considered in

good faith.   It  is also assumed that a two-way flow of communication is estab-

lished between the individual and his government and from government to  its

citizenry for either party to obtain consent --by conveying information or by
persuasion  itself.


     Moreover,  as it seeks to form opinions favoring its policies,  American

government  enjoys no monopoly on information or access to persuasion.

Competing policies are  urged on the public by a variety of sponsors --  special

interest groups, opposition political parties and competing centers of political
and economic power, for example.


     American government may be characterized as functioning within some

vaguely defined limits of public opinion.   This opinion is often unevenly dis-

tributed among people and its translation into support is often a matter  of

conjecture at best.  Still,  governmental decisions must be and are made,

and the  substance of action,  in many instances,  is related to  the contours of

opinions that bear on an issue.  Obviously opinion, in the sense  of a mass

-------
                                                                            11






opinion, haw no immediate bearing on many actions of government.  These



actions represent a fait accompli  simply because no general public opinion



about them exists.  Actions on issues in which opinions are widely held are



another matter,  for  it is here that government often hesitates to act until the



weight  of public opinion has been assessed.  Even when it appears that a



majority favors new policies, government is hesitant to adopt a new course



of action on a major opinion-related issue.  Information from reputable



pollsters,  for example, that a majority favors a change in policy is neither



necessary nor sufficient to compel a political leader to change his position



on the issue.   Reactions of people to  new governmental actions are often



unpredictable and politicians understandably want to be sure that if they act



they will avoid widespread public  hostility.  Moreover,  the social and



economic costs to those adversely affected by new actions  may be sufficient



to guarantee inaction or delay.






     To be sure,  the appearance of a  crisis may,  at times,  sweep political



leaders into some form of action.   But the action is often  inaction or delay



until the nature of the crisis is fully assessed.  At other times, political



leaders act successfully on controversial matters with what appears to be



marginal public support.  The rule, however,  seems to be that major



governmental actions await development of predominantly favorable public



opinion.






     It is in this vein that one might profitably examine environmental quality



as a public issue and attempt to extract some major variables that would be



of interest for study.






                   Environmental Quality as a Public Issue






    Public  concern about the environment is not new, as any veteran con-



servationist will tell you.   Environmental quality as  a public issue has waxed



and waned in  America  for more than 100 years.   The issue has its historical



roots as a challenge  to the widespread exploitation of the wilderness for



man's use.  Nash [32] and Huth [20]  have chronicled the  adoption of a

-------
                                                                           12


different relationship between man and nature by these environmentalists.

The dominant view expressed at the  time (and still current) was that it is

man's right and duty tx> overcome  nature for his own use; hence, emphasis
was placed upon  the creation and use of technology.   Early environmentalists,

on the other  hand,  organized a new ideology around the goal of preserving
tracts of land in  their natural state for the esthetic,  scientific or spiritual
values inherent in  a wilderness preserve.   The ideology implied restrictions

and cessations of commercial expansion.   Modern environmentalists have

gone beyond  the wilderness-saving goals of their forebears.   A view has
been established that, as a consequence  of the power created by science and
used by technology, the future of mankind in the modern world is in danger.

The value  of scientific progress, technological change and economic growth,

for example,  has been seriously questioned by Boulding [3]  and Commoner [9],
to name a  few.  It  is the codification and documentation of these two lines of

thought that have contributed to the current ideology of present-day ecologists.
                                                                   4
     Concurrent to the development of this new  ideology has been the apparent

swift rise  of environmental quality as a public issue.  One heard almost no
mention of it five or  six years ago in public forums.  Today it ranks among
the top two or  three issues facing  the nation, according to public opinion
pollsters.—    This swift rise in public concern may  stem from several sources.

First, mass media attention  concerning  the environment may well have played
a role in focusing the public's concern on this issue.   One has only to count

the number of  environmentally-related titles in the Readers' Guide to Periodical
Literature  to  note the striking increase in  the  number of these titles.  More-

over,  the number has increased positively  each  year, from 79 in  1950 to 228

in 1969.  Admittedly,  increased media attention does not guarantee increased
— See The Roper Report [ 48] for an up-to-date summary of a  national sample.
Problems of environmental quality ranked second in a survey of  issues con-
fronting a. random sample of Oregon residents.  The  survey was conducted in
April 1972 by this author.   Only problems concerning taxes and school financing
ranked  higher.  Fifty-eight percent of the respondents surveyed mentioned
environmental quality as one of the  two or three most important problems
facing the state today.  Sixty-one percent mentioned taxes and school financing.

-------
                                                                            13
public concern.  Yet, one sees convincing  examples of mass media effects.



Tichenor _et ail. [ 52]  have noted the impact of the publication of Rachel



Carson's Silent Spring [6.] on the subsequent pesticide controversy in the



early 1960's.   Second, as Tichenor et aL  [52],  have noted,  environmental



quality "reflects ideals which have been taught directly,  specifically and



energetically in American schools  in recent years".  They assert that



messages concerning the environment have become important non-controversial



issues to young people.  Compared to family life education, for example,



environmental quality has become a modern "motherhood" issue and  far safer



for teachers to manage.   Third has been the widespread activity of political



leaders  on environmental matters.   While one may argue that political action



is a reflection of public concern, the fact that political leaders have acted



with relative swiftness in attempting  to cope with environmental problems has,



in itself, contributed to socially defining and legitimizing quality of the environ-



ment as an issue.  The relative ease of passage of the Clean Air Acts in 1963,



1967 and 1970, for example,  testifies to the support environmental quality



has among both conservative and liberal political viewpoints.  Increased mass



media attention concerning political action about  environmental matters tends



to generate still further  support.  One finds politicians  of both political parties



vying for honors as guardians of environmental quality.






    While the  evidence seems  clear that there is a public consensus  con-



cerning  environmental quality at the present time, it is  not at all clear that



the public agrees what should be-done.  What apparently has  been rejected is



the recommendation  of some neo-environmentalists who call for drastic



changes  in life styles,  commensurate with a re-ordering of national priorities



concerning the role of science,  technology and economic growth in American



society.   The dominant belief,  as reported by Tichenor  -j^aL  [52]  for example,



is that "technology got us into the environmental crisis  and technology will



get us out".  The adoption of this belief in the technological ethic appears  to



be widespread  and,  to date, few seem to be willing to give up their automobiles,



restrict  their use of  electricity or accept control measures which threaten



their economic self-interest.

-------
                                                                            14
                       Enforcement and Community Conflict






     In the short run there appears to be more support for maintaining



economic security of a community -- even among those most knowledgeable



concerning pollution matters -- than for the enforcement of pollution control



measures which strike at the heart of economic self-interest.  Support for



pollution control measures which  imply unemployment or other forms of



economic hardship  is the least within those communities directly threatened,



according  to Tichenor  e
-------
                                                                            15






also receives most or all of his externalities.   The other is where the com-



munity which receives none or few of the economic benefits of a polluter also



receives most or all of his externalities.  Assume, in the first instance, that



pollution conditions are sufficiently severe to impinge on the public's con-



sciousness.   One would expect conflict about enforcement to occur within



the community if >.t appeared that abatement would jeopardize the economic



health of the community.  The parties to the conflict,  initially, likely would



be among the community's elite.  Higher knowledge and favorable attitudes



toward environmental quality  are associated with high educational levels,



professional or semi-professional occupations and participation in community



organizations.  These people  are more likely,  compared to those  low in these



attributes,  to initially perceive  and assess the implications of a polluter in



terms of the community's long-term welfare.  Those who hold pro-environmental



attitudes are  also likely  to hold pro-community attitudes and may  see nothing



inconsistent with holding both sets.  It  isn't until a  situation ari ses in which



these two attitudes interact or are brought into  relation with one another that



the inconsistency becomes apparent and a  re-evaluation is set in motion.



In the  Tichenor study cited earlier, those interviewed responded to a set of



proposed environmental measures that had a bearing on the economic welfare



of a community.   Support for  the environment was most prevalent  among the



well-educated in all communities.  Furthermore,  it is among the  elite that



one would expect to find  individuals who possess the skill to develop and



propose workable alternatives to continued pollution, to obtain access to



governmental enforcement machinery and,  in general,  to engage in a con-



frontation with a polluter and  other community elites.   Moreover,   the chances



are also good that in many non-metropolitan communities intense  conflict



over pollution abatement is not likely to break out or become public.  For,



as Coleman [8] has'noted, participation in community organizations  tends



to cement the community by providing channels  of communication,  both



among citizens and between citizens and the elite.   These channels also



serve as sources of influence  when conflict threatens to break out.  Individuals



in these organizations hear both sides.of the controversy in face-to-face contact

-------
                                                                            16
and are  subject to persuasive cross-pressures.  Conflict is more likely to
break out in communities where the communication and influence networks
are not available  or are not working.  The evidence also suggests that  the
local  mass media is likely to "manage" conflict by not reporting it.   Olien,
^t &L ,  [35],  in a study of 88 Minnesota communities,  found that the  com-
munity press tends  to support community institutions by ignoring the dis-
ruptive aspects of public life.

     One would not expect the same behavior of people  in a community which
received none or  few  of the  economic benefits  of a. polluter but received all
or most of his externalities.  Again,  assume  that pollution conditions  are
severe enough to  impinge on the public's consciousness.   One would expect
conflict  about enforcement to occur  between communities if it appears  that
abatement would jeopardize the economic health of the polluting community.
The conflict is carried to some higher political institution for resolution,
such as  a regional or state environmental control agency.  One would also
expect to find a high level of consensus within  each community concerning
the issue.   As well,  the local mass  media in each community would be
expected to manage the conflict by reporting the  issue  fully,  emphasizing
each community's side of the controversy respectively.
                               ^
     In both of these instances it is argued that social conflict may serve
the interest of environmental quality.  Conflict is viewed as a method by
which values  concerning  environmental quality are affirmed.  In the long
run society itself is likely to benefit from this  form of conflict.  As Coser
[10]  has noted, conflict is a process through which adjustment to new con-
ditions can be brought about.  It is through  conflict behavior that norms
are modified  or created and the continuation of society is assured under
changed  conditions.   A rigid social system  --  one not permitting conflict --
will block adjustments and will tend to maximize the danger of catastrophic
social breakdowns.

-------
                                                                            17
     The externality concept, as used here,  cannot be applied to predict forms
of conflict for all air pollution situations.   One can conceive of communities
which produce pollutants that mix with emissions from  other communities
in the same airshed.  Pollutants may or may not fall on communities in
proportion to each one's contribution.  An assessment of social effects of
externalities  is difficult at best.   In this instance the community conflict
model may not be appropriate.  One would expect the advancement of an
environmental ideology to become one basis of conflict  rather than community
self-interest per se.

     Moreover, while the effects of externalities may be important in sug-
gesting forms of community conflict, they are not likely to be as useful as an
alternative and related  concept for predicting forms of  conflict resolution.
This alternative concept is  the use of economically and technically feasible
controls for air pollution emissions.

     At times there is considerable interest shown by a polluter or an enforcer
(i. e. , governmental pollution enforcement agency) in reducing  emission levels.
But the technology is not always available to guarantee  lower rates --at least
sufficiently low to reach desired air quality goals.  The availability of econo-
mically and technically feasible emission control devices may well determine
the level of emissions permitted.  The level of control  is considered a function
of the technological effectiveness  of control measures,  of control cost and of
damages averted by lowering pollution  levels.   Applying these  elements,  the
decision rule advanced  by some economists—  is to expand air  pollution control
so that  the additional cost of one more  increment of control is  just equal to
the value  of the additional damage averted.  The point where the total net
benefits (totaj. benefits minus total costs) is a maximumi  defines the optimal
allocation of resources  to air pollution control.  It is recognized that it is
difficult to quantify all the benefits or to quantify some  of them so they are
commensurate with dollar values  employed  to measure costs.   One can, however,
conceive of at least three instances when the presence or absence of feasible
— See, for example,  the recent article by Downing  [13].

-------
                                                                           18






emission control technology would have a bearing on the form of conflict



resolution.






     The first is when both the polluter and enforcer agree that economically



feasible emission technology is available.  Once a polluter  is aware of



available feasible technology he agrees to comply with control standards.



It is a matter of establishing compliance schedules and testing and main-



taining control equipment.  The second instance  is when the enforcer



believes, but the polluter disagrees,  that feasible emission  control devices



are available.  The disagreement arises because some of the pollution control



benefits are "external" to the decision process of the polluter.  In this instance



one alternative for resolution commonly employed is to use  some aspect of



the adversary process.   Procedures of this process include informal and



formal agreements,  binding judicial decisions without use of rules of evidence



or court tests where rules of evidence are employed.  Decisions established



through some adversary  procedure do not necessarily rule  out subsequent



social conflicts,  however.  The ability of an enforcer to impose sanctions



upon a polluter by successful court (or administrative) decision, for example,



has the potential for  considerable social upheaval as enforcers move  to



impose emission restrictions which threaten either  the short- or long-run



economic security of a community.   Whether or  not conflict breaks out may



be more a function of the perceived fairness of enforcement and the effect of



enforcement on the community's future than on mere economic well-being.



The  third instance  is when both the polluter and enforcer agree that feasible



controls are not available. In this instance one is likely  to find, other than



a crash program of zesearch, attempts to modify the behavior of the  public



or the polluter to adjust to severe pollution  conditions.  One sees,  for



example, "smog alerts"  employed in  the Los Angeles and the New  York -



New Jersey metropolitan areas and attempts to  "manage" field burning in



Oregon's Willamette Valley.  Such modifications of  behavior are often viewed



as a temporary solution,  based on the assumption that technology will develop



sufficient hardware to adequately control emissions.   As with many temporary

-------
                                                                            19
         ,  modified behavior patterns tend  to become ch-fined as Ihc Hiatus quo

until another environmental episode underscores their limitations, often

triggering further social conflict.

    The notion of level of technology focuses on the role of some enforcer

as an agent for conflict resolution on the assumption that it is through him

that the laws or authority of government and use of available technology are

operationalized.  Inadequacies of either the availability or application of law

or technology are necessary (but not sufficient) to generate a pollution  episode.

An episode is defined as  the occurrence of  an event which threatens to  produce

an unusually large loss such as damage to human health,  property or status,

and in which the bureaucratic structure is  unable  to  cope with or to controL

with existing laws and technology.  A pollution episode stems from the con-

tamination of some aspect of the environment so it is unfit for intended use.

One effect of a pollution episode is to compel a political leader  to affirm one

set of rival interests  and values of competing groups.  This may be operation-

alized by temporarily suspending or severely regulating the activities of a

polluter,  the public,  or both.
     The social effects of externalities and availability of economically and
technically feasible controls may be summarized by examining their joint
distribution in the following six-fold table:
Availability of economic-
ally and technically feasible
emission controls
Pollution externalities
occur within a polluting
community	
Pollution externalities
 occur outside a
 polluting community
1) Enforcer, polluter
   agree, feasible
   emission controls
   available.
2) Enforcer agrees^ pol-
   luter disagrees,
   feasible emission con-
   trols available.
3) Enforcer, polluter  agree,
   feasible emission con-
   trols not available.
Minimal social con-
flict;  compliance.
"Managed" social
conflict within a
community; adversary
procedure employed.

"Managed" social con-
flict within a community;
behavior management
employed.
 Minimal social
 conflict; compliance.
 Social conflict be-
 twee,n communities
 public; adversary
 procedure employed.

 Social conflict between
 communities public;
 behavior management
 employed.

-------
                                                                          20





     One may well describe the type of social conflict and action concerning air



pollution effects by knowing the nature of the pollution externalities and the



availability of feasible emission controls.   Suppose, for example,  one wished



to study the effects of social conflict concerning air pollution abatement.



He would  likely find these expressed in a situation where feasible controls are



not available; and where pollution externalities occur outside the polluting



community.  Suppose,  on the other hand, one wished  to study the effect of



type of community power structure in  resolving social conflicts brought about



by air pollution abatement.   He would likely examine  communities whose



pollution externalities  occur within the community,  and where feasible



emission  controls are  either not available or where the enforcer and



polluter disagree over the feasibility of available controls.






     Social conflicts over air pollution  effects and controls can,  in themselves,



crystallize  the attitudes of individuals  and groups in affected communities.



Positive or negative attitudes are formed (or changed) from the interaction



between two or  more groups over time. Norms,  or direct guides to posi-



tive or negative action, are established as  the end product of this interaction.



Relations among individuals belonging  to these groups must become standard-



ized as norms toward other groups and toward situations before they are  re-



flected as inter-group  attitudes.   The process of attitude formation or change



concerning  pollution effects thus appears relevant to any study involving an



explanation of human behavior over this issue.






                  Attitudes  and Definition of the Situation
     It has frequently been asserted that a person's observed verbal or non-



verbal behavior is determined by his psychological structuring of events.—



As well,  this structuring  is jointly determined by  the interaction of both



external and internal attributes.  Examples of external attributes are aspects



of objects or events  in the environment and social influences such as inter-



personal  suggestions and  group participation  and pressures.  Internal attri-



butes include personal knowledge,  interests,  motives,  set and socially-



derived factors such as attitudes, beliefs, internalized norms and values.
— See, for example Sherif and Sherif [42].

-------
                                                                          21





Presumably,  if one had sufficient theory to tell how to combine the significant



attributes in their proper  relationship with one another and, knowing the



scores of individuals on each attribute, one could then predict with a reason-



ably high level of certainty each individual's psychological structuring and



his observable behavior toward some event.






     Moreover,  if one had sufficient theory to specify the effects of a change



in external  or internal attributes in terms of changes in psychological struc-



turing,  then one could again predict with reasonable  certainty changes in



each individual's observable behavior.






     Individuals, however, may or may not be clearly conscious of their



psychological structure.   They are often unable to explain their behavior



to anyone.   To be sure, social scientists  may be able, after reviewing inter-



view protocols,  for example,  to reconstruct or infer models which explain



an individual's behavior.  Psychological structuring  does not necessarily



play a role  as a conscious antecedent to the individual's  behavior.   It is a



theoretical  construct which mediates between external and internal attributes



and observable  behavior.






     Fxternal and internal attributes do,  however,  often impinge on the con-



sciousness  of many and the resultant structuring is termed the individual's



"definition of the situation".    This term is derived from W. I. Thomas' [49]



aphorism "if men define situations as  real, they are  real  in their conse-



quences".   This has led workers to investigate how individuals develop common



perspectives and communicate these to each other.   The term implies no



one-to-one  correspondence between an objective event  (or situation) and



people's perceptions of that event—that something intervenes  between



events and people which makes possible a variety of  interpretations  to reality.



The  same event, for example, can have different meanings for different



people and the degree of difference can be expressed as observable behavioral



differences.   As well,  objective events contain common elements through



time for different people (e.-_g- >  many aspects of driving a car along one



smoggy freeway are the same as driving a car along  another smoggy freeway).

-------
                                                                           22





People who enter objective events carry similar backgrounds and attitudes



which make it possible for them to define events in similar ways.  Uniqueness



of events,  while an aspect of any situation,  cannot be ruled as a barrier for



scientific study of them.   What  is required is specification of "something"



which intervenes between events,  or situations, and people.






     Stebbins  [45 ] considers the definition of the situation a "more or less



conscious  synthesis and personal interpretation of the interrelation of the



set of predispositions and the elements  of the subjective situation".   Three



concepts play roles in Stebbins1 formulation:  objective situation,  subjective



situation and  action orientation.  He defines an  objective situation as the



immediate social and physical surroundings and the current physiological



state of the actor (L e^. ,  individual).   The subjective situation is those



components of the objective situation which are seen by the actor to affect



any  one of his action  orientations.   The term "action orientation" refers



to any purpose,  trivial or important, which the actor has as he enters the



objective  situation, such as driving down a smoggy freeway.  The point is



that action orientations serve as guides to behavior as the individual  enters



the objective  situation.  Stebbins asserts that action orientations guide the



selection  of elements from the objective situation for the development of  the



subjective situation.  Subjective situations are postulated to have time limits;



action orientations play a crucial role in time calculations,  since situations



exist as long  as  it takes the individual to manipulate them.   Subjective situa-



tions,  once identified, must be  interpreted and given meaning  In  terms of the



individual's background.   This interpretation,  as noted above,  is the  definition



of the situation and its final formulation is affected by both predisposltional



and  situational factors.   Predispositlonal factors stem  from the actor's past



definitions of situations,  from former and future plans  of action,  from past



and present action orientations  .and from the actor's values and  attitudes,



among others.  Chronologically, the definition of the situation immediately



precedes  the  individual's behavior which has responded,  in turn,  to the sub-



jective situation.   A definition Is completed only after a period of conscious



reflection.  Goal-directed behavior is delayed until the Individual has defined

-------
                                                                           23





the situation.  Action unfolds commensurate with the definition.   Stebbins allows



for trial-and-error behavior during reflection.  Goal-directed behavior,  how-



ever, becomes possible only after meaning is established for  the situation in



terms of the action orientation.   As well,  he notes that definitions vary in



amounts of reflection required;  habitual and routine situations call for less



reflection  than the unusual,  the  relatively new or  crisis situations.   In any



case, definition is a necessary  condition to goal-directed behavior.






     Empirical work supporting  definitional processes has focused on the



effect of defined situations as modifier  of  attitudes and  the procedures



actors employ in  the definitional process.  Gordon [15]   considers the person's



definition of the situation a modifier  of  attitudes toward certain objects.



McHugh £?] , using a laboratory setting, was  able  to demonstrate that differ-



ent actors appear to follow similar procedures for constructing substantially



different definitions.






     It is recognized that explication of the definitional process has not led to



widespread theoretical development and empirical  testing.   One is not likely



at this stage  to find sociologists making precise predictions which are capable



of empirical verification or refutation.  But the definitional formulation pro-



vides a view for examining phenomena capable of affecting   large  numbers of



people simultaneously.  As well,  variables may be selected on the basis  of



definitional literature that are worthy of further study.  It is in this vein  that



the definition of air pollution situations will be examined.






                   Definition  of  Air Pollution Situations
     It can hardly be refuted that some aspect of air pollution -- loss of visi-



bility, odors, perception of particulate matter,  to name a few -- has not im-



pinged on the consciousness of most Americans.  Even before it became



fashionable to express environmental  concerns,  air pollution was (and still



is) defined by some as evidence of employment and economic activity.  To



others, air pollution represents an environmental insult of the worst sort



threatening the  survival  of man and all other creatures living on this planet.

-------
                                                                            24
That multiple definitions are possible from the same  (or nearly the same) ob-
jective pollution situation is supported by studies relating public concern,
attitudes, beliefs and knowledge about air pollution levels, effects and con-
trols.  The work of Crowe [H],  among others,  is  typical.  Moreover,
multiple definitions at one point in time may represent a redefinition for
some --a change in the components of meaning.  One effect of a redefinition
may well be a change in attitude, defined as a change in the positive or
negative evaluative predisposition of an individual toward some aspect of air
pollution.  One  way to examine the postulated association between definition of
the situation and change (or no change) in attitude is to study inputs from the
objective situation that may have a bearing  on definition or redefinition.
One such input  is to study the mass media themes concerning air pollution
which have been published in a community over a relatively long period of
time.  Themes,  or assertions about some aspect of air quality, function in
defining  situations by providing descriptions of the  present.  Concerning themes,
McHugh  [27] asserts that "actors assume before the  fact that a pattern of
meaning will be discovered  in the events they observe.  They are future oriented,
in that they take it for granted they will be able to make something of what is
yet to occur.  They are also past oriented,  in that they take it for granted that
what  has  already occurred will inform the future.   Each of these orientations
folds back as an actor's description of the present. "  One ingredient in  defin-
ing the present  is assumed to be the input of mass media  content, as described
by frequency and direction of themes from the objective situation.   A  rigorous
content analysis should provide a basis for  estimating definitional (or re-
definitional) characteristics of air pollution  situations for individuals  in a
community.  Then if it appears  that an air  pollution situation has been defined
or redefined on the basis  of a mass media content analysis, one can test if
attitudes  toward air quality remain the same or are changed according to
definitional  predictions.  If, for example,  it appears that an air pollution
situation  is not  redefined within a specified time range, then attitudes toward
this phenomena should not change within the same time range.
     The inferential leap from content analysis of mass media themes to
attitude change  of individuals is great, however.  What should be  specified

-------
                                                                         25
as well is how Inputs from the mass media are related to attitude level or
change.  If it is demonstrated empirically that a reciprocal causal link or
links can be  established between attitude level or change, mass media use
and other endogenous variables,  then one is on firmer ground for relating
results of a mass media content analysis to an individual1 s change or lack
of change in  attitude.  The formulation of such a communication model must
be drawn from work outside the definitional literature, however, since the
necessary statements capable of empirical tests are not available.  Research
concerning the communication of scientific information to the public  is likely
to prove more fruitful  and some of this work will be examined next.

                  An Air Quality Communication Model

     In a secondary analysis of science knowledge and  attitude  data from 27
sources,  Tichenor [ 50J hypothesized that the communication process  in
which attitudes and mass media exposure play a role can be  schematised as
follows:
Variables predisposing Exposure Effect
toward exposure variables variable

More
education
-»

More general science
knowledge and more
positive attitude
toward science
-
More exposure to
mass media and
interpersonal
sources
->

More science
knowledge
  Figure 1.  A communication model for accumulation of science knowledge.
     He stipulates that his predisposition variable toward exposure (those
which occur prior to exposure) are represented by two variables.  Level of
education is designated as an antecedent variable, primarily because it occurs
first in the  time sequence.   Formal education establishes an individual's level
of general science knowledge as well as his positive or negative evaluation, or
attitude, toward science.  Exposure variables -- exposure to mass media  and
interpersonal sources -- represent complementary channels through which
additional science  knowledge flows.  Acquisition of such additional knowledge
represents  the effects variable.

-------
                                                                          26
    The sample size of his data base was large enough for him to partial out
or control for the effects of a number of demographic,  attitudinal and exposure
variables in order to determine if postulated effects, such as a level of edu-
cation,  serve as an independent predictor of science knowledge.   His  central
findings suggest that both the predisposition and exposure variables add to
the predictive power of accounting for variation in science knowledge.   His
analysis, however,  did not permit him to state precisely the direction of all
the arrows (J-. £. ,  the effects)  in the above model.  It is still not known if
attitudes toward science  lead a person to acquire more science knowledge
through greater mass media exposure,  or  if these three variables represent
an interdependent system.  It is conceivable,  as Hovland [19]  has pointed
out,  that attitudes may predispose a person to acquire certain types of
information  but such information may also trigger  a change in attitude
which stimulates seeking of still more knowledge.  In addition,  it is also
possible to reconceptualize the scheme  so  that the exposure and  effects
variables represent specific science items, such as mass media exposure
to information and more  knowledge about air quality.  In fact,  this  is the
nature of much of the data with which Tichenor was  dealing.  Predisposition
variables included measures of level of education and other demographic
characteristics,  measures of general science knowledge and attitudes  toward
science.  Measures of  exposure and effects variables,  however, often dealt
with specific science fields, such as mental health,  medicine,  psychology,
fluorine and tooth decay, smoking-cancer link,  etc.  Moreover,  the list of
variables in each class is not exhaustive and  thus can  be  expanded to  include
others which conceivably underlie increased mass media exposure and science
knowledge.  With these considerations,  the Tichenor model might be profitably
reformulated  to describe more precisely the relationship between an individual's
attitudes toward air quality, his level of mass media exposure to air quality
information and his level of knowledge about air quality.

-------
                                                                          27

More
; interpersonal .
discussion (X )

More
with a
(X2)

^
More positive
attitude toward
air quality (Yj
t
experience
ir pollution


*x

^
	 r

Gre
infl


More knowledge
about air ^
quality (Y3)
t
More science
education (X )

ater social More
uence (X ) education (X4)
^ I
More exposure to
	 ^. mass media about
air qualify (Y_)
s
/ Greater use of
environment (X^)



Figure 2.  A reformulated model for accumulation of air quality
           information by the public.


     The model shown in Figure 2 stipulates that three variables are mutually

interdependent and are denoted as "Y" variables in the figure.   These variables

are: attitude toward air quality, mass media exposure to information about air

quality and the level of knowledge about air quality.  The interest  here is not

with a single independent-dependent relationship but rather with a system of

relationships in which the "Y" variables are imbedded and which leads to a

description of the interdependent relationships among these variables.  With

such a model,  then,  the existence of mutual interdependence among attitude,

exposure and knowledge variables can be tested.  Is,  for example, knowledge

dependent on attitude and exposure?  Does  such knowledge, in  turn, affect

attitude and exposure?  The use of a set of simultaneous equations should

provide an answer to these  types of questions.


     The other  variables, denoted as "X's" in the model, are "outside" the

system.  They serve as important predictors and,  as will be noted later

in the section on methods, represent important operational considerations

for analyzing the system empirically.   Direction of arrows does not repre-

sent a causal relationship but only predictions that relationships exist.

-------
                                                                           28
    The model can be described mathematically as a system of simultaneous

equations:
            Y3 = P31Y1 + P32Y2 + ^31X2  +  ^32X5 + ^3X6 + 63     (3)

where

            Y.  is  the observed value for the appropriate endogenous
               variable in the system^

            p..  and v.. are constants, or regression coefficients,
             J       J
            X.  is  an independent predictor or exogenous variable
               outside the system,  and

            e.  is  a stochastic disturbance  term assumed to be inde-
               pendently and normally distributed, with mean equal to
               zero and a constant, finite variance o .  .

     This concludes the discussion of  the theoretical model.  A review of

literature is presented as Appendix A to  support the postulated relationships
within the model.


     Methods for testing the mass media  content - attitude change hypothesis
are presented next, followed by the results and discussion.  Then, method-

ological procedures for testing the communication model are  presented,
followed by the results and discussion.

-------
                                                                            29

          METHODS -- CONTENT ANALYSIS AND ATTITUDE CHANGE

    Methodological considerations necessary to test the mass  media content-
attitude change hypothesis are divided into four phases:

     1.  The selection of locations and times of known air pollution situations
         appropriate for hypothesis testing.
    2.  The development of operational measures of mass media content
         and attitude change.
    3.  The gathering of sufficient data to provide a basis for an empirical
         test of the hypothesis.
    4.  The specification of analytical procedures employed to test the
         hypothesis.
Each phase will be described in its respective order.

                         Selection of Location and Time
    Several conditions  are necessary to test the mass media-attitude change
hypothesis.  One is the  selection of a community or communities where air
pollution effects are severe enough to impinge on the public's consciousness.
Another is a pollution situation which occurs with sufficient regularity that
one can design a panel study and plan personal interviews to capitalize on
this regularity.  A third is a relatively high level of attention to air pollution
sources and effects by the local mass media,  particularly newspapers.

    It seems most likely that one will find these conditions  met in a situation
where pollution externalities occur outside  the polluting community and when
an enforcer and polluter agree that feasible emission  controls are not available.
Social conflicts concerning emission restrictions are  likely to occur,  according
to the discussion in the  introduction (page 19).  This conflict is likely to be
reflected in relatively high levels of mass media attention and public
awareness.

-------
                                                                         30
     Several communities were considered for study shortly after the project
began in early 1969.   The cities of Eugene and Salem,  Oregon, were selected.
Because of its location at the base of the Willamette Valley, Eugene has ex-
perienced poor natural ventilation, whereas  Salem, located in the mid-
Willamette Valley, has had good natural ventilation.  Therefore, a contrast
in visibility would be expected to occur when there is air pollution in the Valley.
Eugene's population was estimated at 109,614; Salem's, 95,748,  according
to the 1970 Census of Population [7].

     One predictable  source of air pollution was agricultural field burning.
Farmers burn approximately 250,000 acres  of post-harvest residue from
grass seed crops in late summer or early fall in the Willamette Valley.  Heat
produced through straw and stubble burning provides at least partial control
of a number of grass seed  diseases and disorders.  As well, other benefits
recognized include effective and inexpensive residue removal, destruction of
weeds and weed seeds and  physiological stimulation of seed yields.  No
economically feasible alternative to field burning has yet been developed.

     The largest  and most concentrated area of grass seed production lies  in
the area between Albany and Eugene (see Figure 3).  Emissions from burning
consist of smoke (carbon particles of various sizes),  ash  and certain gases.
Most individual carbon particles are minute  and cannot be  seen without visual
aid.  Larger particles can be seen easily by persons with normal vision.
Large numbers of minute carbon particles are called "suspended particulates "
and are a major  source of  reduced visibility.  Larger particles, which settle
more readily,  are a  source of soiling.  Relatively large volumes of smoke
emissions from agricultural field  burning were considered an important
stimulus for public concern, particularly in  Eugene.  For this study, a
measure of visibility was employed as the measure of air quality.  For a full
presentation of the scientific basis for visibility, or visual range as it is
sometimes called, see Middleton [31],  Robinson [37], for example, notes
that the deterioration of visibility  is probably the first indication of air

-------
                                                           25
                                                                      31
Figure 3.   Willamette Valley Basin and vicinity.

-------
                                                                            32






pollution about which a citizen becomes aware.   Visibility was selected



because it is most likely to have impinged on the consciousness of individuals



in the study areas, compared to other forms  of air pollution.






     As well, any test of the mass media  content-attitude change hypothesis



must take  into  account any past  definitional situation.  Ideally, one would



like to measure attitudes toward air quality prior to any definition (or



redefinition); wait, hopefully, for events  to be defined or redefined, then re-



measure attitudes of the same individuals.  This ideal situation is not likely



to occur in practice, for the  researcher has no control over mass media



attention to air quality.   An alternative procedure is that, given the inordinate



time involved in  content analysis, one artibrarily selects a time difference



for measuring and remeasuring individuals when attitudes conceivably will



change on  the basis of known changes in air quality.   After all,  Zajonc [54]



asserts that "mere repeated  exposure" of the individual to a phenomenon is



sufficient to enhance his attitude towards it.   Attitude change  scores are



then related to changes  in level  of air quality.  If the mass media content



analysis shows no definition or redefinition within the time period of



measures  for attitude change, then the definitional hypothesis is not sup-



ported  if an attitude change does,  in fact,  occur.  If the content analysis



specifies definition prior to measures of  attitude change, and no measurable



change occurs,  then one would consider this data support for  the definitional



hypothesis.  But,  this support is not as strong as it  would be  if attitude



measures  were obtained prior to and after definition.






     Such a design calls for measures of the same individuals during periods



of high and periods of low visibility.  Two interviewing waves were planned.



The first,  using  Schedule A,  contained measures of  both endogenous and



exogenous variables  of  the model.  The second,  using Schedule B, contained



parallel forms of measures of attitude and knowledge variables.   It was



important  to guarantee  that use  of Schedules A or B  was not correlated with



time of interview or  visibility level.  Otherwise, a confounding of effects



might occur.  Thus,  the sample was drawn so that a random  half of

-------
                                                                           33
respondents was interviewed from each community in the following manner:


                            May 1970       Aug. -Sept. 1970    January 1971
                          % respondents     % respondents    % respondents

Schedule A	         50                 50

Schedule B	                           50               50
     It was anticipated that visibility levels would be high in May 1970 and

January 1971 but low  in August-September 1970.—  It was  also anticipated

that the pattern of high and low visibility would be more pronounced in Eugene

than in Salem, 1. e. ,  Eugene would have lower visibility levels than Salem in

August-September.   Timing interviews in  this fashion would balance any

effects between type  of schedule (A or B) and time of interviewing.   Those

who were interviewed with Schedule A in May 1970 were reinterviewed with

Schedule B in August-September  1970; those who were  interviewed with

Schedule A in August-September  1970 were reinterviewed with Schedule B

in January 1971.  Observers in both communities, trained  by  a representative

of the Oregon State University Department of Atmospheric  Sciences,  completed

visibility observations every two hours on the hour,  beginning at 8 a. m.  and

continuing until dusk.  This meant that seven observations  per day were made

in May and August-September, five  per day in January.  Observations were

completed each day interviewers were in the field.  In  this fashion,  profes-

sionals in  the air quality field could compare levels of  visibility between
interviewing periods and between communities.
— Five year visibility data from U. S. Weather Service stations at the Eugene
and Salem airports were gathered to estimate more precisely the timing for
interviews.  This analysis disclosed that a four-week period from mid-May
to mid-June and a four-week period  in January were consistent periods of
relatively high visibility.   The analysis did not suggest a consistent
timing for periods of low visibility;  however, estimates by professionals
in the  O. S. U. Department of Atmospheric Sciences suggested that late
August to mid-September was likely to be a period of deteriorated air
quality.

-------
                                                                            34
                      Development of Operational Measures

    Mass media themes were defined in the introduction as "assertions
about some aspect of air quality".  Themes served as recording units for
measuring the content of communication;   in this  instance,  mass media
available in Eugene and Salem, Oregon,  the communities selected for study.
The first step in theme  development in this study was to factor analyze
intercorrelations among attitudinal items that were available from a pre-
test conducted among a  random sample of  individuals in Portland, Oregon.
(Details of the pretest will be presented in the section concerning the
methodology for testing the communication model.  Suffice to say that pre-
test data were the  earliest available attitudinal information  that could be
developed in the study. )  The factor analysis disclosed six orthogonal factors
and the first list of potential themes was developed from these factors.  Each
factor was characterized by a single statement.   These statements, or tenta-
tive themes,  could be written to express both a positive and a negative direction
toward some aspect of the  referent, air pollution. A random sample of 52
weeks of newspaper articles, in which some aspect of the referrent "air
pollution" appeared at least once, was selected from an 11-year (1960-1970)
collection of such articles.  These articles had been reproduced from micro-
film copies of the Eugene Register-Guard and  the (Salem) Oregon Statesman.
Two trained coders, working independently, tested the six tentative themes
on a sub-sample of articles,  with the article serving as the unit of analysis.
Discrepancies in coding were resolved by  clarifying  broad terms employed in
the themes,  by rewriting ambiguous themes and by writing new  themes.
Overlapping themes were omitted.   The final list of  15 themes was then tested
on the remainder of the sample articles.   The final version of 15 themes
was written as two lists  to facilitate coding; one list,  "for"  clean air;  the
other, its negation, "against" clean air.  These themes are:

Theme  "For" Clean Air (Code 1)           Themes "Against" Clean Air (Code  2)
1.   Air pollution is a health hazard.       1.   Air pollution is not a health hazard.

-------
                                                                             35
 Theme "For" Clean Air (Code 1)

 2.  Air pollution is a menace to the
     public welfare.

 3.  Air pollution is becoming a
     serious problem and public
     opinion  must pressure govern-
     ment into doing something  to
     control  it.

 4.  Before  the problem of air pollu-
     tion  becomes  too serious, we can
     find the resources and technology
     to eliminate it by working  together
     now.

 5.  Air pollution is a problem  caused
     by industry,  and it is  industry's
     responsibility  to clean up the  air.


 6.  Industry is willing to cooperate
     in cleaning up its air pollution.

 7.  Polluters are willing to use alter-
     native methods of waste disposal
     even if  they are convinced  it is
     neither  economically feasible  nor
     technologically practical.

 8.  Air pollution control agencies  need
     more power if they are to  control
     the problem of air pollution.

 9.  If air pollution control agencies
     are to be effective, they must  be
     organized on a regional basis.

10.  Air pollution may be a problem,
     and industry and economics are
     not as important as clean air.

11.  Industry is a major contributor
     to the air pollution problem.

12.  (Not useful.  See Code 2)
13.   Air pollution has always existed
     and it may be a. problem, and
     something must be done to
     eliminate it.
 Themes "Against" Clean Air (Code 2)

 2.  Air pollution  is not a menace to
     the public welfare.

 3.  Air pollution is becoming a serious
     problem,  but the use of public
     opinion to pressure government is
     not the way to control it.
 4.   (Not useful.   See Code 1)
 5.   Air pollution is not a problem
     caused by industry,  and it is  not
     industry's responsibility to clean
     up the air.

 6.   Industry is not willing to cooperate
     in cleaning up its air pollution.

 7.   Polluters are willing to use
     alternative methods of waste
     disposal if they are convinced it
     is economically feasible or
     technologically practical.

 8.   Air pollution control agencies
     already have enough power to
     control air pollution.

 9-   Air pollution control agencies
     need not be organized on a regional
     basis to be effective.

10.   Air pollution may be  a problem,
     but  industry and economics are
     more important than  clean air.
11.   Industry is only a small contribu-
     tor  to the air pollution problem.

12.   It is difficult to place the blame for
     air  pollution  since it is not con-
     fined to any physical  boundary.

13.   Air pollution has  always existed
     and it may be a problem,  but there
     is nothing that can be done to
     eliminate it.

-------
                                                                            36
Themes "For" Clean Air (Code 1)         Themes "Against" Clean Air (Code 2)

14.  The available means of control-      14.  The available means of controlling
     ling industrial air pollution is             industrial air pollution is ineffect-
     effective or technologically                ive or  technologically impractical.
     practical.

15.  (Not useful.  See Code 2)             15.  Air pollution is not a problem.


     The negation of three themes was not useful and were omitted.  An

example is  the theme  "Air pollution is not a problem".  Its negation, "air

pollution is a problem" is not useful, since all themes in this direction imply

the existence of a problem.


     The measure for  attitude change will be described in the  section on

analytical procedures, presented below, and in the section concerning attitude

measurement which is presented in  the methodology section for the communi-

cation model.


                          Data for Hypothesis Testing


     Community newspapers provided the most appropriate source for

measuring mass media content by counting frequency of themes.  Microfilm

copies of newspapers  were readily available to reproduce selected articles

for analysis.  All newspaper articles -- news stories, features,  editorials,

letters-to-the-editor,  advertisements -- which contained at least one refer-

ence to some aspect of air quality were  reproduced from an 11-year period

(1960-1970).  The  newspapers were the  Eugene Register-Guard and the

(Salem) Oregon Statesman.    Two coders, working independently,  read each

article, noting the presence or absence  of each theme.  Each article was

also coded for type of article (news, editorial, etc. );  length (in column

inches); story placement (front page or other); source of pollution cited

(field burning, industrial,  automobile, commercial,  residential, meteor-

ological, forestry, natural,  other);  and  scope of problem (local,  regional,

state,  out-of-state, national, world).  As well,  month,  year and newspaper

-------
                                                                           37
in which article appeared were coded.  Discrepancies between coders were



resolved by a referee.  At least one coding discrepancy occurred in approxi-



mately 25 percent of the articles.






    A total of 3, 054 articles was coded -- 1, 963 from the Eugene Register-



Guard and 1, 091 from the Oregon Statesman.  Approximately 16 percent of



the articles from each newspaper contained no themes.  These articles



provided procedural information concerning some aspect of air pollution,



such as notices of hearings before regional authorities.   As well, coders



agreed that 5 percent of the articles from each newspaper contained themes



which could not be coded with the available list of themes.  The remaining



79 percent of articles contained at least one theme and these articles pro-



vided the data base for  analysis.






                             Analysis of Data






    Analysis of mass media content frequencies will be described first;



then measurement of attitude change will be presented.  For mass media



content,  analysis consisted of a visual examination of a frequency distri-



bution of total number of articles,  of article  type, of sources of air pollution



and of themes across time.  Time,  in  months, was the abscissa and fre-



quency the ordinate.  A  three-month moving average was employed to



smooth out the frequency curve.  Nine graphs were planned.  First, the



frequency of total number of articles for the  newspapers from each



community was plotted on one graph.  Second, the frequency of article



type --  news story,  feature, editorial, letter-to-the-editor -- was plotted,



one graph for each community.  Third, the frequency of pollution sources



mentioned -- field burning, field burning and other sources, other sources



only,  and no source -- was plotted, one graph for each community.  Fourth,



the "for"  clean air  themes were plotted,  one graph for  each community.



Finally,  the "against" clean air themes were plotted.






    If air quality (or some aspect of air quality) is defined as  an  issue, one

-------
                                                                           38
would expect to find some sustained attention to the  topic over time.  Prior

to being defined as an issue,  one should find frequencies generally low with

no systematic pattern discernable.  At some point in time,  however,  fre-

quencies should increase markedly and continue at relatively high rates with
easily explanable  patterns occurring.   For example, once field buring is de-

fined as an issue,  one should expect increased attention to occur  during the

burning season, compared to the non-burning  season.   Yet,  frequencies

during post-definitional non-burning months should be higher than during

pre-definitional months.—   Themes associated with definition should aid in

conveying  a description of the definition itself.   By examining the frequency

of themes,  one may well be  able to construct a dialogue for describing the

mass media content at any specific period of time and,  by comparison,

note differences among particular time periods.   For example, one would

expect a pre-definitional description of field burning to differ markedly from

a post-definitional description.


     As well, one  can fix the date after which to expect  definition.  Field

burning began during World  War II and reached a  plateau of approximately

240, 000 acres  by  the late 1950's.  Assuming that  the pesticide controversy

of the early 1960's was the forerunner of  the present environmental move-

ment,  and that the publication of Silent Spring played a major role in defining

the pesticide issue, then one should expect environmental issues  to develop

sometime  after 1962,  the date of publication.  Public concern over  air quality

is expected to be one of these  issues.   The intent  of this analysis  is to
— An estimate of grass seed acres burned  in the  Willamette Valley can be
made from acres harvested,  since the practice of burning had been widely
adopted by I960 for all grass seed crops (See Hardison [17]).  Acres har-
vested showed no great fluctuations between 1959 and 1970, according to
Middlemiss and Coppedge [30] and other sources at Oregon State University.
The yearly total grass seed acres harvested were: 1959 -  241, 000; I960  -
235, 700; 1961  - 216, 500; 1962 - 222, 000;  1963 -  215, 300;  1964 - 227, 500;
1965 - 225, 600; 1966 -  228, 100; 1967 - 216, 000;  1968   213, 000; 1969 -
225, 300; 1970  - 242, 700.  Some grain acreage was burned in the I960 decade
but reliable estimates are not available.

-------
                                                                             39
determine the date of definition of air quality as an issue.


     This concludes the analysis proposed for the mass media content fre-

quencies.  Next,  analysis of attitude change scores will be described.


     A regression equation,  employing dummy variables,  was written in an

effort to adjust attitude change scores  of individuals in  each community for

a number of effects,  including the effect of  a change in  air quality.  The

equation was:
where
                        = f(X0'  *!'  X2'
                     Y.  is final attitude  score minus  initial
                           attitude score of the individual,

                     X   is 1 for mean effect,

                     X   is 1 if high visibility occurred first,
                           0 if low visibility occurred first,
                     X   is  1 if Form A used first,
                           0 if Form B used first, and

                     X   is  initial attitude score of the  individual.
    Significance of coefficients was tested by Student's  t_.   A significant t

value for the X  coefficient would provide evidence  that a change in air

quality was associated with a change in attitude.

-------
                                                                          40
                         RESULTS AND DISCUSSION --
                 CONTENT ANALYSIS AND ATTITUDE CHANGE
     Results  in this section are divided into four phases:


            1.  Results of the content analysis.

            2.  Analysis of visibility observations.

            3.  Analysis of attitude change scores.

            4.  Discussion of results.


Each section will be discussed in its respective  order.


                         Results of  the Content Analysis


     Frequency distributions of types of content are presented in Figures 4-22.

Time, in months, is the abscissa and frequency the ordinate for all Figures.

The frequency curve is a three-month moving average.   Figure 4  is the-fre-

quency curve for the total number of articles for Eugene and Salem, respect-

ively.  Figures  7 and 8 are frequency curves of  the sources of air pollution

mentioned for Eugene  and Salem, respectively.  Figures 9-20 are frequency

curves for the "for clean air"  themes for  Eugene and Salem, respectively.

There are two themes per chart,  each theme number corresponding to those

listed  on Page 34.  Curves representing the 12 themes were too concentrated

to present on one chart.  Two  themes were selected which would appear to

vary sufficiently in each graph so one can easily follow  each curve visually.

Figures 21 and 22 are frequency curves for the "against clean air" themes

for Eugene and Salem,  respectively.   Figures 9-22 are presented as

Appendix  C.


     It had been noted in the section on Methods -Content Analysis  (Page 29)

that analysis would consist of a visual inspection of the  different frequency

distributions to determine the approximate date  when air quality was defined

-------
  60  T
  50  "
   40  ••
o
a


3  30
u*

-------
35
30  -
25  "
20
15  • •
10  • •
 5  •
                                                                                                           mi ijitu n it IHJI
1960       1961      1962       1963       1964       1965       1966       1967       1968




                    Figure 5.   Frequency distribution  for type  of  article, Eugene, 1960-1970.
                                                                                                        1969
1970

-------
35
30 ••
25 . •
20  '
15
10  •
 5 • •
1960       1961      1962       1963       1964       1965       1966       1967      1968        1969




                    Figure 6.  Frequency distribution for type of article, Salem, 1960-1970.
                                                                                                                   1970
                                                                                                                          OJ

-------
35  .
30  .
25  . .
20
                                                                                          Field Burning &
                                                                                          Non Field
                                                                                            Burning
 0  MIII i>miiyiiiTiiiMnpnii n ii ii [i n ii rYMTKji IMKTTI nvpi iiYlvi n i| IXTTIIII n [i n n u n n )i irmm i i[li mi n ni| n n n mi t( i
        I960      1961       1962       1963       1964      1965       1966       1967      1968       1969        1970

                     Figure 7.   Frequency distribution for type of  air pollution cited, Eugene, 1960-1970.                4>.
                                                                                                                         •e-

-------
    40  T
    35  4-
    30  -f
    25   4-
o
2J   20
    15   -I-
   10   -I-
    5  +
                                                                                                          Non Field Burning :
                                                                      Field Burning &
                                                                         Non Field
                                                                           Burning
            1960
1961       1962    '   1963       1964       1965       1966        1967        1968       1969

    Figure 8.  Frequency distribution for type of air  pollution  cited,  Salem,  1960-1970.
                                                                                                                        1970

-------
                                                                           46
as an Issue.  It was asserted that one should look for some sustained attention
to the topic over an extended period of time.   Prior to being defined as an issue,
frequencies would be generally low with no systematic pattern discernible.
At some point in time,  however, frequencies  should increase markedly and
continue at relatively high rates with explainable patterns  occurring.

    Applying this very general guide to the data,  it appears that the time
of definition  for Eugene occurred in 1966 or,   possibly,  early 1967.   Fre-
quencies never did reach the low levels prior to these dates --  even in early
1968 when the news was devoted to the presidential primaries.  The pattern
of "for clean air" themes was mixed with some continuing in use from 1966,
others occurring primarily during the  three-month field burning season of
mid-July to mid-October of  each year.  The  only  exception to the definitional
pattern was  the frequency of "against clean air" themes.   Frequency of these
themes disappeared in  the Eugene  press  in early 1968.   The sharp rise of
total mass media attention during the 1969 burning season underscores the
occurrence of an episode in  Eugene.   On August 12, severe visibility re-
strictions  prompted a delegation of Eugene residents to call on  the state
governor who returned with  them to Eugene.   The governor ordered a seven-
day emergency ban on all field burning and called the Environmental Quality
Commission into emergency session.   Following the ban,  a new quota system
was established with farmers drawing  lots to  obtain permits to  burn a share
of the limited daily acreage  allowed.  The quota system was in  effect for
the 1970 and 1971 burning seasons.

    The time of definition in Salem is  less clear-cut.  It is likely to have
occurred in the late summer of 1969,  perhaps as an effect of the episode in"
Eugene.   Actions of the state governor, who resides in Salem,  the  state
capital, may have defined the issue not only for Salem but for the state as
well.   There was an increase in frequency in  1967 but this increase dropped
to near zero in early 1968.  Sustained  use, however, appears after the 1969
burning season; thus,  the issue was likely defined by late 1969-

-------
                                                                         47
     A study of the frequency of themes,  types of articles and types of pollution
sources cited suggests that the air quality issue in both communities was well
defined prior to our interviewing of respondents in 1970.  On the basis of the
definitional hypothesis,  one would not expect attitudes to change during the
three-month interval between interviews.  A null hypothesis  is postulated.

                      Analysis of Visibility Observations

     Visibility observations were grouped by distance and by  source of re-
striction -- natural conditions or air pollution.  The results are  summarized
in Table 5 (for Eugene)  and Table 6 (for Salem).  These results were analyzed
by a faculty member of  the Department of Atmospheric Science?, and he
concluded —

              a.  For Eugene, visual  restrictions due to air pollution
                 were greatest in August-September and least in
                 May and January.  Visual restrictions due to natural
                 conditions were greatest in January.
              b.  For Salem,  visual restrictions due to air pollution
                 were greatest in August-September and least in
                 May and January.  Visual restrictions from natural
                 sources were greatest in January.
              c.  Visibility restrictions due to air pollution in Eugene
                 were greater than in Salem for any distance in August -
                 September.

     Timing of interviews occurred during anticipated levels  of visibility
for each community as measured by visibility observations.  May and January
interviews were during  periods of relatively high visibility; the August-
September interview was during a period of relatively low visibility.   As well,
visibility in Eugene was less than visibility in Salem for the August-September
interviews.
17
  Personal communication from Dr.  E. W.  Peterson, Assistant Professor
  of Atmospheric Sciences, Oregon State University,  April 5,  1971.

-------
                                                                         48
Table 5.   Summary of Visibility Data at Four Distances for Eugene,  Oregon
*
Miles Visibility
O
<2 P
N
O
<3 P
N
O
<4 P
N
O
<8 P
N
***
(N)
May 1970
(7/13)**
. 99
. 01
0
. 99
. 01
0
. 97
. 03
0
. 77
. 23
0
(91)
Aug. - Sept.
(7/17)
. 95
. 05
0
. 92
. 08
0
. 76
. 24
0
. 34
.66
0
(119)
Jan 1971
(5/10)
.86
.02
. 12
. 64
. 10
.26
.48
. 16
. 36
.42
. 14
. 44
(50)
*
   O - Percent of observation marker(s) visible.
   P - Percent of observation marker(s) not visible due to air pollution.
   N - Percent of observation marker(s) not visible due to natural
      conditions (fog, rain,  snow, etc. ).

sjcafc
   Number of observations per day divided by the number of days.

***
   Number of observation-days.

-------
                                                                           49
Table 6.  Summary of Visibility Data at Five Distances  for Salem,  Oregon
J{
Miles Visibility
0
<2 P
N
O
<3 P
N
O
<4 P
N
O
< 5 P
N
O
<8 P
N
***
(N)
May 1970
(7/14)**
.98
. 01
.01
. 98
. 01
. 01
. 93
. 05
. 02
. 98
. 01
. 01
. 92
.06
.02
(98)
Aug. -Sept.
(7/20)
. 97
. 02
.01
. 96
. 02
. 02
. 90
. 09
. 03
. 94
. 04
. 02
. 66
. 29
. 05
(132)
Jan. 1971
(5/10}
. 82
0
. 18
. 56
0
. 44
. 40
0
. 60
. 42
0
. 58
. 28
0
. 72
(50)
*
   O - Percent of observation marker(s) visible.
   P - Percent of observation marker(s) not visible due to air pollution.
   N - Percent of observation marker(s) not visible due to natural
      conditions (fog, rain, snow,  etc. ).
***
 Number of observations per day divided by the number of days.

!e
 Number of observation-days.

-------
                                                                         50
                      Analysis of Attitude Change Scores

     A regression equation (Equation (4),  Page 39) was written to adjust
attitude change scores for regression effects and to determine if a change in
air quality was associated with a change in attitude.   The coefficients obtained
for Eugene and Salem were:

     Y^       = 15. 8529 + 0. 3670 X, + 0. 3829 X.  - 0.  2655 X0      (17)
      Eugene            (P. 490)    1  (0.507)  2  (-6.100)   3
     Y        =  2. 3071 + 0.0295 X  + 2. 2883 X   - 0.  3771 X       (18)
      Salem              .(0.037)      (2.957)     (-9.107)

     Numbers in parentheses under the coefficients are Student's Rvalues.
For Eugene,  only the coefficient associated with regression on initial attitude
scores,  X   was significant.  For Salem,  the coefficients associated with
form effects, X?, and regression,  X ,  were significant.  In neither  com-
munity was the coefficient associated with level of visibility, X  ,  significant.
Thus,  one can conclude that a change in air quality was not associated with
a change in attitude (Y.) in either community.   A high proportion of  the
variance in attitude-change scores  can be accounted for by regression:
effects; i. e. ,  errors of measurement associated with scores of the  dependent
variable and, for Salem, by differences in form effects as well.

                             Discussion of Results
     It was postulated,  on the basis of a content analysis  of the mass media,
that one would not find a change in attitudes occurring during the 1970 inter-
viewing period for  respondents in both communities.  The data presented in
Equations (17) and  (18) support this hypothesis.  It appears that the mass media
had defined the situation well before respondents were interviewed.  If one
had wished to detect an attitude change toward air quality in Eugene,  he should
have completed his  initial interview prior to 1966;   in Salem, prior to 1969,
according to the content analysis.

-------
                                                                          51
     An examination of the frequency curve in Figure 4 (for Eugene only) and
Figure 7 (for field burning only) reveals an interesting pattern associated with
the field burning months of each year (mid-July to mid-October) beginning with
1965.  Mass media attention appears  to increase sharply during these months
each year.  Speculating a bit on this repeated pattern of attention leads one to
ask  if  the media wereresponding to greater visibility losses each year or if
some other phenomenon was operating.  One wonders  why,  after a relatively
long history of field burning and, ostensibly,  reduced  visibility in the Fugene
area,  the mass media began showing  an interest in 1965 and in  subsequent
years.—   The question may have been partially answered in the discussion
concerning the environmental movement and definitional process presented
in the  Introduction.

     Yet,  one must ask further  if increased attention to field burning in Fugene
in the  1965 to 1969 years altered the field burning habits of farmers sufficiently
to contribute to the episode of 1969.

     To answer this latter question, one must assume  as a premise the stipu-
lation; that grass seed farmers are entrepreneurs; j^ ^. , they are risk-taking
decision makers who are motivated by profits.  As entrepreneurs,  they have
established a set of priorities  for farming  practices, including field burning,
to achieve their objective.  The time  of field burning,  then,  is normally
dictated by profits.

     Now,  suppose farmers perceive hostility toward one of their essential
production practices,  field burning,  which threatens their ability to obtain
profits.  Their possible behavior alternatives may be  ranked on the basis of
guarantees that field burning will be completed each year.  Since each farmer
is an entrepreneur, it is  likely that field burning will tend to be completed as
soon as possible after harvest -- in some cases fields just harvested will be
burned before other fields'are harvested.  Suppose,  further,  that rain or other
— See Footnote 1, Page 38, for an estimate of grass seed acres burned yearly
  between 1959 and 1970.   The data indicate that the greatest acreage  burned
  was  in 1959 and in 1970;  the smallest acreage burned was in 1967 and 1968.

-------
                                                                           52

conditions prt-vrnl harvest or field burning when some farmers would normally
harvest or burn.  Potential acres for burning accumulate and are burned im-
mediately when conditions permit.  Visibility restrictions then are likely to
be more pronounced in Eugene and an increase in attention and concern by
community residents follows.

    It may well be that a sequence of events,  such as those described above,
contributed to the episode of 1969-   If this conjecture hats merit, it means
that the process culminating in the episode was set in motion in 1965 when
attention to field burning effects first appeared in any substance in  Eugene.
As well, the  continued arousal  of attention and concern may have contributed
to a worsening of air quality until some form of workable controls was  es-
tablished.

    The process of visualizing the opinions of another group with some
accuracy is termed 'co-orientation1.  In this situation two or more groups
may independently perceive a phenomenon and simultaneously comprehend
their  joint perceptions of the phenomenon. For a more complete description
of the co-orientation  concept and its  role in consensus, see Scheff [39].  The
concept may well be useful in explaining or accounting for episodes in other
air pollution situations as well.

    One would need to gather at least two diverse types of data to evaluate
the efficacy of the co-orientation notion for explaining this particular episode.
One type of data is an evaluation of visibility levels for Eugene prior to and
including the establishment of field burning as  a practice,  and prior to  and
including the establishment of community attention and  concern over the
effects of field burning.   Moreover,  one would suspect  that co-orientation is
operating if there is a lack of correspondence between levels of visibility
and mass media attention  prior to  1965 and a correspondence between these
variables from  1965 to 1969.   The second type of data would be an in-depth
interview of field-burning farmers to determine if their perceptions con-
cerning reactions of Eugene residents altered their decisions about time of
burning.

-------
                                                                           53
             METHODS -- AIR QUALITY COMMUNICATION MODEL







    A theoretical model has been developed which postulates that a person's



attitude, mass media exposure and knowledge about air quality represents an



interdependent system and that this system can be explained by a set of



predictor variables.






    Methodological procedures required to test the model empirically were



divided into three phases:






    1.  The development of operational measures of the variables in the



theoretical model.  Here,  summated response scores  to items asked of the



general public in a standardized questionnaire were used as measures in



all cases.






    2.  The gathering of sufficient data to provide the  basis for an empirical



test of the model.  A personal interview survey of a random sample of



residents living in Eugene and Salem, Oregon,  was completed.






    3.  The specification and estimation of the parameters of a set of



equations by which the model was tested.  A  set of simultaneous  equations



has been specified and a test of significance of variables by two-stage least



squares was planned.






    Each phase will be described in  its respective order.









               Development and Evaluation of Operational Measures







    Appropriate questions (items) were developed to represent operational



measures of the variables in the model.  Specifically,  these variables were



an individual's:

-------
                                                                           54
     a.   attitude toward air quality (attitude)
     b.   level of exposure to messages in the mass media about
         air quality (mass media exposure)

     c.   level of air quality knowledge (knowledge)

     d.   level of exposure to interpersonal information  sources
         (interpersonal exposure)

     e.   level of experience with air pollution (experience)

     f.   level of social influence (influence)

     g.   level of formal education (demography)

     h.   level of formal science education (demography)

     i.   level of use of the environment (use)

     (Note: the terms  in parentheses will be  used to designate each
           variable respectively)

Measurement of  demographic, exposure and  influence variables was straight-

forward and required only pretesting and evaluation.   A first draft of the

questions was written at the outset and incorporated in the pretest question-

naire.


     Measures of the remaining variables were less straightforward.  They

required additional information before the items  could be specified.  First,

measures of attitude and knowledge were considered;  then measures of ex-

perience and use were considered.


     Attitude  and Knowledge.   A content analysis of the local mass media
and magazines published in recent years was employed.   For the mass media,

microfilm tapes  of the daily newspapers in the two target cities were read

independently by teams of two trained  individuals each.  All news stories,

editorials, letters  to the editor,  etc. ,   concerning air pollution were noted.

Newspapers were the Eugene Register-Guard and the  (Salem) Oregon Statesman.

In addition,  the Reader1 s Guide  was surveyed for titles of magazine articles

concerning air pollution.   Time  interval in all cases was the past six years --

January 1, 1964, through December 31,  1969.  Appropriate material from

these sources was  reproduced by xeroxing.   This resulted in 1, 417 articles

-------
                                                                             55





from the Register-Guard, 687 articles from the Statesman and 163 articles



from popular magazines.






     Trained coders performed a content analysis of the xeroxed articles.



The statement served as the unit of analysis.  A statement is  defined as an



expression which denotes a complete idea and which has a referrent -- some



aspect of air pollution,  in this case.  Statements were recorded on cards



and coders independently classified them into either attitude or knowledge



categories and into qualitative similar sub-classes  within each category.



A total of approximately 3, 200 attitude statements and 5, 350 knowledge



statements was available for consideration as items.






     Empirical support of the model requires knowledge of content or inform-



ation presented in the mass  media.  The importance of this requirement can



be noted quickly.  It  is possible  that one will not find a relationship be-



tween mass media exposure and knowledge  when an empirical test is made



of the model.   One explanation might be that items used to test such knowledge



were not related to information available in  the mass media.   By confining



one's attention to areas in which knowledge test items are constructed from



statements in the mass  media,  one can rule  out this alternative explanation.






     Attitude statements were divided into mutually  exclusive sub-categories.



Items that were judged to be unambiguous and representative of various



aspects of the respective attitude categories and that fit the "agree-disagree"



type of response alternative were  selected.   The final list of potential attitude



items was composed of  50 statements.  These  items represented the measure



of the attitude variable in the pretest.






     Knowledge statements were selected that were  judged to be unambiguous



and representative of various aspects of the  respective knowledge categories.



The final list of potential test items was composed of 223  statements.  An



accuracy rating of these statements was performed by a panel of  experts in



the air quality professions from the National Air Pollution Control Administration,

-------
                                                                           56
Raleigh and Durham, North Carolina; and from professionals in Oregon and



California.   A total of 65 raters was asked to cooperate;  57 completed and



returned their ratings.






    Critical ratios were calculated for each item.  A significant CR (at the



. 05 level) was established .as a necessary condition for acceptance of an item



in the knowledge  test battery.  Representativeness of each knowledge category



and ease or appropriateness in an  interview setting were other  conditions



necessary for acceptance.  The final set of knowledge items pretested was



composed of 40 items --20 true and 20 false.







    Experience and Use.   Intensive interviews of a sample of the public



(40 in Eugene, 10 in Salem) were examined for statements concerning the



public's experience with air pollution and use of the  environment.  Probes



were directed toward these two areas. These interviews yielded several



statements about past living and visiting  experiences in areas where the



respondents  believed air quality problems  existed and about use of the out-



doors for work, leisure and living.  Items were developed from these re-



sponses for pretesting as measures of experience and use.







    Fvaluation of Pretest Measures.   A pretest interview  schedule was



developed from items representing measures of the  several variables.   The



schedule was administered by personal interview to  a random sample of 100



adults  in the Portland (Oregon) Metropolitan Area in February 1970.






    Responses were  coded and scores key-punched for  analysis.  Analysis



consisted,  first,  of selection of items from among the 50 attitude state-



ments.  Then estimates of reliability and the construction of  parallel or

-------
                                                                           57
comparable halves (forms) were completed for  the items measuring attitude

and knowledge.—

    Fifty attitudinal items were subjected to an item analysis.   Twenty-
five respondents with the highest total scores and 25 with the lowest total
scores were selected.  These two groups served as criterion groups for

evaluation of individual items.  The following ratio was calculated for

each item:
                             X,  -  X
                       1 =  —~	                         (5)
                            "h      "l
where
               X     is the mean score on a given statement for
                     the high group,

               X     is the mean score on the same statement for
                     the low group,
                2
               S     is the variance  of distribution of responses
                     of the high group  to the  statement,

               S     is the variance  of distribution of responses
                     of the low group to the statement,  and

                     is the number of responde
                     low groups,  respectively.
n , n    is  the number of respondents in the high and
— Parallel or comparable halves (forms) were used to avoid effects of memory
transfer when individuals were remeasured in the full study.   The procedure
for item selection suggested by Horst [18] was followed.  Form reliabilities
reported in Table 1  may be  inflated since they were computed from the same
data used to construct the forms.

-------
                                                                         58
     The value of t is a measure of the extent to which an item differentiates
between high and low groups.  If a t_ equaled 1. 75 or greater,  it was considered
evidence that the average response of the high and low groups to a statement
differed significantly.  This analysis resulted in a deletion of 10 items with
low t_ values.

     Estimated Spearman-Brown split-half (odd-even) reliability coefficients
for the several measures are presented in Table 1.

Table  1.  Estimated Spearman-Brown Split-Half Reliability Coefficients
          for Pretest Measures of Variables in the Theoretical Model

                                       No.  of                 Reliability
        Variable                       Items                 Coefficient
a.  Attitude toward air quality
      (Form A)                          20                     . 83
b.  Attitude toward air quality

c.
d.

e.

f.

g-

h.
i.
j-
k.
(Form B)
Level of exposure to mass media
Level of air quality knowledge
(Form A)
Level of air quality knowledge
(Form B)
Level of exposure to interpersonal
information sources
Level of experience with air
pollution
Level of social influence
Level of education
Level of formal science education
Level of use of the environment
20
3

20

20

2

3
5
1
19
16
, 78
. 44

. 70

. 80

.76

. 30
. 75
	
. 92
. 77
    Reliability estimates  for measures of mass media exposure and experi-
ence with air pollution are low and suggest a need for item revision.  An
examination  of items measuring both variables  disclosed evidence of ambigu-
ous wording  which would likely produce responses based on different frames

-------
                                                                          59

of reference.   New items were written in an attempt to overcome this difficulty,
resulting  in an increased number of items for measuring both variables.

    An item analysis also was completed in an attempt to increase knowledge
test reliability.  Guliksen [16], for example, has shown that the test mean,
standard deviation and  reliability may be estimated from item parameters of
difficulty and reliability index. An attempt was made to increase test relia-
bility lay/making the average item variance smaller or the average item re-
liability index larger.   This practice did not improve test reliability appreci-
ably.   Details are available in a separate publication (see Mason [36] ).

                       Execution of the Sample Survey

    Pretest questions surviving tests of reliability and item analysis consti-
tute the basis for accepting items for use in the sample survey.  Summed
response scores of these items represented measures of variables in the
model.   The new interview schedule constructed for  the sample  survey was
pretested itself to insure that  its make-up maintained rapport between inter-
viewer and respondent.  The final interview schedule employed in  the sample
survey is presented as Appendix B.

    Parallel or comparable forms  had been developed for attitude and know-
ledge measures.  As well,  response alternatives  were reversed for the third
endogenous variable, mass media exposure and items reordered randomly in
the battery.  This produced a  permutation of forms,  or batteries,  as measures
of the three endogenous variables.  These permutations were employed in
order to randomize effects between forms and time of interview.   The number
                            3
of permutations was eight (2  - 8).  Since both Eugene and Salem were divided
into eight census tracts,  one combination of forms was assigned at random to
each census tract for the first, or initial,  interview.

    Sampling of Respondents.    A  two-stage  area sample was employed for
random selection of respondents from each community.  Sample size was

-------
                                                                         60
planned to achieve results having a specified degree of precision,  viz. ,
                               n =
where
                   n  is the sample size,
                   k  is the constant associated with the
                      probability that the  sample «alue will
                      have a relative error no greater than
                      ±D>
                   V  is the coefficient of variation, and
                   D  is the relative error allowed.

     The value of k was arbitrarily set at  3;  V, calculated from pretest;.
data, averaged 0. 30 or less for variables in the model and D was arbitrarily
established at 0. 06.   Solving for n in Equation (6) resulted in a sample size
of 250 for each community.  A design effect of 1. 20 was  assumed,  raising
the sample  size to 300.  A response rate  of 0. 85 for the first interview and
of 0. 77 to 0. 80 for  the second was also assumed.  Application of these rates
increased the sample  size so it ranged from 444 to 464 for each community.

     The latest metropolitan area maps were purchased from the office of the
local planning commission for  each community.  Each map was  divided into
eight areas, following census  boundary lines.  It had been arbitrarily deter-
mined to  select four households at random from each  block  in an  effort  to
reduce effects of clustering or intraclass  correlation.  Random  selection of
14 blocks from each census area would yield 448 households per city for
interviewing -- within the range calculated in Equation (6).

     Both metropolitan areas received an  on-site inspection.  Non-residential
and new residential blocks were noted as  were blocks that were  above- or
below-average in  number  of dwelling units (DU's).  Non-residential blocks
were eliminated prior to numbering  and  new residential blocks added.
Above-average blocks were divided and below-average blocks were combined,
all in an effort to  adjust block  size to a calculated average number of dwelling

-------
                                                                          61
units.   Blocks were numbered in serpentine fashion,  and 14 blocks were
selected at random from each census area, seven blocks for each pair of
interview "waves".

     Sampled blocks received an on-site visit  to 1) ascertain the number of
valid (L e. , occupied) dwelling units,  2) select at random a point to begin
selection of DU's,  and 3) select,  by use of a table of random permutations,
four dwelling  units.  Dwelling units within apartments or other contrived
blocks were selected randomly as well.  Addresses and distinguishing features
at each DU were noted for preparation of  interviewer listing sheets.  One
adult respondent was selected at random wibhin each household by the inter-
viewer,  using a "face sheet" or  "cover sheet" commonly employed for such
purposes.   Proper use of the "face sheet" guaranteed random selection of a
respondent within  each household.

     Completion of Interviews.    The  first interviewing wave began May  15,
1970,  and terminated June  2,  1970.   This was considered the first "high"
visibility condition.—   A total of 224 household addresses was specified in
each city for  each interview.  Oversampling was planned in order to secure,
if possible, more  than 150  completed interviews.  Panel attrition may have
been high for  the  subsequent re-interviewing with Schedule B.   Interviews
for the "low"  visibility condition began  August 26, 1970, and were completed
September  16, 1,970.   The third  (and final) interview wave began January 5,
1971   and terminated January 16,  1971.  This was considered the  second
"high" visibility condition.

     Response  rates for Schedule A were calculated by dividing the number of
successful  completions by 224.  These are reported in Table 2.  Response
rates for Schedule B were the number of usable schedules divided by the
number of usable  Schedule  A's.
— See Page 32 for the interview design employed to test effects of changes
  in air quality.

-------
                                                                         62

Table 2.  Response Rates for Interviewing
Interview
Schedule A
Schedule B

May
1970
77

Salem
September
1970
87
83

January
1971
%
87

May
1970
87

Eugene
September
1970
78
87

January
1971
%
87
      The percent not-at-home and refusals are presented in Table 3.

Table 3.  Percent Not-at-Home and Refusals

                 	Salem	    	Eugene	
                 May   September   January    May   September  January
  Interview       1970      1970         1971      1970    1970        1971
Schedule A
Schedule B
14(9)* 7(6)
9(8)
6(7) 11(11)
8(5) 7(6)
9(4)
*
 Percentages in parentheses  refer to refusals.
     The not-at-home rate for Salem in May and the refusal rate for Eugene
in September,  both with Schedule A,  appear high.   An examination of block
clusters assigned to interviewers did not reveal a pattern of not-at-home1 s
for S;alem.    They appeared to be distributed uniformly across all blocks.
The distribution of refusals in Eugene, however,  was concentrated in blocks
surrounding the University of Oregon and interviewers noted the cause of re-
fusal in some instances as a saturation of previous interviewing by University
students.

     Any effect of Schedule A  non-response may be partially  reduced by the
definition of the sampling universe.   It was defined as residents who live- in
their respective dwelling units for six consecutive months or more after first

-------
                                                                         63





contact.  A set of  items was developed and used for predicting early in



Schedule A if a respondent was likely to move within the next six months.  If



it appeared that he was likely to move, the interview was terminated.






    Since these "likely-to-move" respondents were not defined in the



sampling universe, other dwelling units in the blocks where these respond-



ents lived were sampled at random in an effort to build up the level of



completed  interviews.  It turned out that "likely-to-move" households were



correlated within  their respective blocks.  It was not always possible to



complete four interviews in these blocks.   No more than four "likely-to-



move" respondents in such blocks were included  among the "successful



completions" for the  calculation of Schedule A response rates.






     Assuming that measures to  determine length of residence were valid,  it



is likely that some not-at-home1 s or refusals would include some people not



defined in  the population.  If  so,  any non-response  bias would likely  be re-



duced commensurately.  But,  without confirming information, it is safer to



state that some non-response bias is likely to exist for the Schedule  A inter-



views.  The direction and extent of this bias are  not known.   The Schedule B



response rate averaged 86  percent and effects of non-response bias from



reinterviewing sources are expected to be  less than that for Schedule A.






     Completion of Coding,  Key-Punching,  and Verifi.cation.  The response



rates represented 673  usable Schedule A's and 579  usable Schedule B's.



Efforts employed to transfer the information in these questionnaires to data



processing cards .were, first, to code the questionnaires according to a pre-



determined code key.   Next,  two coders  independently checked each  question-



naire for accuracy.  Third, data were key-punched and verified.  Fourth,  a



print-out was obtained for all data and was proofread and rechecked  with the



questionnaires.   Fifth, a computer column displacement program was applied



and discrepancies sight-checked.  Finally, a second computer print-out  was



proofread and rechecked with the questionnaires.

-------
                                                                          64
                               Analysis of Data

    Analysis consisted of, first, an estimate of reliability for measures
representing variables of the model,  and, second, a test of significance
for variables in the model by two-stage least squares.

    Estimates of Reliability.     Spearman-Brown split-half analyses were
completed for scores representing variables in the model, except for the
measure of attitude change.  In this case, Formula (7) was applied:

                              r   + r    - 2r
                           -   xx    yy    xy                   m
                       r j j ~     T   ~T                           I''
                        dd       2 -  2r
                                       xy
where
                r    is the estimated reliability of the attitude
                 dd   i
                     change   scores,
                r    is the estimated reliability of the Form A
                     scores,
                r    is the estimated reliability of the Form B
                  yy            j
                     scores, and
                r    is the Form A,  Form B correlation coefficient.
                 xy
Estimates of reliability for variables in the model are given in Table 4.
Table 4.  Estimates of Reliability for Variables in the Model
Variable
Attitude change (20 items vs. 20 items)
Attitude level (40 items)
Mass media exposure (newspaper,
magazine, T"V)
Knowledge (40 items)
Interpersonal exposure (neighbors, friends)
Experience with air pollution (5 items)
Social influence (5 items)
Level of science education (19 items)
Use of the environment (16 items)
Reliability
Eugene
. 45
. 90

. 71
. 81
- 79
. 33
. 54
. 91
. 73
Coefficient
Salem
. 52
. 91

. 65
. 75
. 73
. 19
. 58
. 94
. 71

-------
                                                                          65
    Items employed for measurement of the "experience with air pollution"
variable did not intercorrelate particularly high.   Thus,  the reliability coef-
ficient was low.  It was considered too low for the variable to function in the
analysis of the model.  Therefore,  this variable was dropped from further
consideration.   Reliability estimates of other variables were considered
sufficiently high for their inclusion in the subsequent analysis.

    Two-Stage Least Squares.   Equations (1),  (2), and (3) have been written
(Page 28) to describe the theoretical model.  As a number of scholars have
demonstrated (e-_g. ,  Johnston [22] , Valvanis [ 53] ), least squares applied
to each of the above equations will lead to  estimates of the parameters which
are biased and inconsistent.  In addition, such estimators  are less efficient
than those obtained by alternative procedures.

    For example,  consider two- stage least squares:
where
                TT.. = f(P..; \ ..),  and
                 ij      iJ   ij
                v.  = g(e.;  (3. .; v ..).
                 i       i   ij    ij
    Since the e.'s are normally distributed with mean equal to zero,  finite
               1         2
and constant variance (r .  , and the v.'s are linear forms of e 's, then the
                        iL                        i
v.'s themselves are normally distributed,  with mean equal to zero and finite
 1                      2
and constant variance 
-------
                                                                         66
     Having estimated the functions in Equations (8),  (9) and (10), one
proceeded to obtain estimated values for the endogenous variables,  Y,, Y_,
                                    A    A   A                      l   *
and Y ,  and these will be denoted as Y  ,  Y ,  Y .   The next step was to fit
the function:
                Yl=  PllPl2*HXl + "' + *12X2 + el   (H)
                Y2 = P21Y1  + B22Y
by ordinary regression techniques.

     The resulting estimators, the (3 ' s and the -y 's are biased; but are con-
sistent and more efficient than straightforward least squares estimators.

     Having completed this routine, and having acquired estimators of the
several (3's and v 's in the system, and knowing the values of X  '   X,,  it is
possible to solve the original system  (Equations (1),  (2),  and (3)),  for the
values Y,, Y_,  and Y_.   To  distinguish these estimates from those
         J.    Ct       o
obtained from the reduced form  (Equations (8),  (9), and (10)), they will  be
             .    /\        A
             ft   A        A
indicated as  Y^  Y_, and  Y , and  referred to as structural equations:
     Therefore,  parameters of the structural equations were  estimated by the
method of two-stage least squares.  Significance of constants in these equations
was  tested by Student's ^,   recognizing, however, that some degree of bias
is involved.
     By comparing scores of individuals in the two  communities,  the null
hypothesis concerning the difference of each endogenous (Y.)  variable between
communities was  tested.

-------
                                                                          67

                    RESULTS AND DISCUSSION--A TEST
               OF THE AIR QUALITY COMMUNICATION MODEL

    Results  in this section are divided into four phases:

             1.  Results of accuracy of air pollution perceptions.
             2.  Empirical test of the communication model.
             3.  Empirical test of the revised communication model.
             4.  Discussion of results.

               Results of Accuracy of Air Pollution Perceptions

    A condition stressed throughout this  study has  been  the selection of an
air pollution situation which has "impinged on  the consciousness of people
in the community selected  for study".   An operational corollary to this
condition is the contention  that people  are able to compare visibility levels
from  one time period  to another and that  the substance of this comparison
can be ascertained by survey methods.  While visibility differences may well
exist between specified time periods,  as  measured  by  trained observers,
there is no assurance that  accurate perceptions will be distributed uniformly
among people in the community.  It may well be that the  communication
process concerning air quality for those who hold accurate perceptions will
differ from those who hold inaccurate  perceptions or no perceptions at  all.
Before testing the communication model it may be profitable to examine the
nature of perceptions  concerning visibility to determine if analysis of the
model should be adjusted to include effects of perceptual differences.

    An opportunity existed to  examine  such differences in this study.  Per-
ceptual accuracy was  determined by scoring responses to one question  asked
at the end of the second (Schedule B) interview. In  September, for example,
this question was:

    "Within  the last three weeks, would you say that the air pollution
    in Salem (Eugene) is better or worse than when we talked to you
    earlier this  year  in May?"

Responses to this question were compared to the analysis of visibility obser-
vations.   A "worse" response in the September interview and a "better"

-------
                                                                         68

response in the January interview,  for example,  were considered accurate.

Other response codes employed were "no difference" and "don't know" but
these were not used to determine perceptual accuracy or inaccuracy.  These

responses are summarized in Table 7.


Table 7.  Summary of Perceptions  to Visibility
Perception of
Visibility
*
Accurate
Inaccurate
No difference
Don't know
TOTAL
(N)
May-September 1970
Eugene
52
31
14
_ 3
100
(152)
X2 = 15. 58, 3 df;
Salem
53
15
29
3
100
(139)
p< . 01 X2 =
Sept. 1970 - Jan. 1971
Eugene
72
6
18
__ 4
1001
(142)
13. 76, 3 df; p<
Salem
51
12
31
6
100
(146)
. 01
   "Worse" response in May-September interview,   better  response in
   September-January interview.

   "Better" response in May-September interview, "worse " response in
    September-January interview.


     Accuracy of perceptions concerning visibility varied by community (location

effect) and by time of interview (timing effect).  A statistical analysis of the

data in Table 7 suggests a highly significant location  effect,  if one sums the

Chi squares and  degrees  of freedom presented in the table.  Switching the

independent variables, _i. _e. ,  controlling for location  effects and testing for

timing effects, resulted in a non-significant Chi square for Salem respondents
                                    2
but a highly significant Chi square (X  = 28. 90,  3 df;  p<. 001) for Eugene

respondents.  Testing for a possible location x timing interaction effect in-

volved the routine suggested by Snedecor [43] .  This resulted in a pooled

Chi square of 11. 67,  with 3 degrees of freedom,  significant at the . 01 level.

Thus, much of the variation in perceptual accuracy can be accounted for by

three effects --a location effect, a timing effect (for Eugene only) and an

interaction effect.  Location effects had  been anticipated, given the history

-------
                                                                         69
of air pollution in the Eugene area.   Timing and interaction effects, however,
wero a  surprise.  There seemed to be little theoretical basis for predicting
that Eugene should experience a higher level of inaccurate responses in
September than in January.  Responses to an open-ended  item asking re-
spondents to describe how or in what way they believed air pollution was
better were studied for those interviewed with Schedule B in September.
A sample of verbatim responses associated with accurate and inaccurate
perceptions is presented as Appendix D.

     It became clear that one  reason for a "better" response was an evaluation
of attempts by the Oregon Department of Environmental Quality to manage
smoke emissions during the 1970 field burning season.  Daily acreage burning
quotas were established with  burning permitted when winds would likely blow
smoke  away from the Eugene area.  The Eugene Register-Guard carried
several articles on this management proposal prior  to the first interview in
May. It is  conceivable,  therefore,  that facts concerning  the forthcoming
management of field burning became incorporated  as part of the definition of
the situation for many respondents.  If such were the case,  some respondents
may well have ignored the May-September  time frame posed by our perception
question.   Instead, they imposed their own defined time frame which included
a comparison of 1970 management efforts with the pollution episode of the
previous year.  If this were the case,  a "better" response would lead to
spuriously inaccurate responses in the September  Schedule B interview and
to spuriously accurate responses in the January interview.  In the first instance,
respondents would be wrong for the wrong reasons;  in the latter instance,
right for the wrong reasons.

     All September and January Schedule B open-ended responses were
examined to test the definitional hypothesis.  Responses were studied for
statements concerning the management (or lack of management) of field
burning emissions.  If the definitional  hypothesis is  to be  supported,  one
would expect fewer statements among Salem respondents,  compared to
Eugene  respondents.  Smoke  emissions from field burning were less  an issue
in Salem than in Eugene.   Next, a 'better" response  would be associated with a

-------
                                                                         70


positive evaluative statement, je._g. , that field burning was being effectively

controlled,  compared to the previous year.   As well, a "worse" response

would be associated with a negative evaluative statement, _e. jj- ,  that field

burning was not being controlled.  No other response combination should be

present; _L^. ,  one should  not find a "worse"  response associated with a

positive evaluative statement nor a "better" response associated with a

negative evaluative statement.  The data are presented in Table 8.



Table 8.  Relationship Between Perceived Management of Field  Burning

          Smoke Emissions and Accuracy of Visibility Perceptions



                  August-September 1970        	January 1971	
                          5|C                #5{C              SJCSfS               5JC
 Level of        Accurate       Inaccurate       Accurate      Inaccurate
Management   Eugene  Salem  Eugene  Salem   Eugene  Salem   Eugene  Salem
Field burning
controlled
Field burning
not controlled
Total number
of respondents
0
10

(79)
0
5

(73)
40
0

(47)
24
0

(21)
27
0

(103)
7
0

(75)
0
22

(9)
0
6

(18)
0£
    "Worse" response.
>'f>k
    "Better*1 response.



     The data support the definitional hypothesis.  In all cases more Eugene

respondents cited evaluative statements than Salem  respondents.  Moreover,

it does appear that timing and interaction effects were produced,  in part,

by the application of a pre-defined time span in response to our air quality

perception question.   Thus,  any attempt to measure the accuracy of perceptions

in Eugene is likely to be confounded by past definitions  of the situation.  Such

confounding raises questions concerning the validity of  the method employed for

determining perceptual accuracy in Eugene.  This confounding is not apparent

for  the measurement of perceptual accuracy in Salem.   Effects of perceptual

differences  will be considered prior to testing  the communication model in

Salem.

-------
                                                                          71
     An additional source of potential error also was examined.  Known
visibility levels were employed for scoring accuracy of responses concerning
perceptions of air quality.   It may well be that visibility was not employed by
some respondents in their evaluation of air quality.   Odors and health effects
are  some of the possible alternatives, for example.   Open-ended responses
were coded for source and effect of air pollution mentioned.  Analysis dis-
closed that visibility or visibility-related phenomena were cited by 92 percent
of the Eugene respondents and 91 percent of the Salem respondents who prof-
fered perceptual judgments.  It appears  that visibility was,  indeed, an aspect
of air quality that impinged  on the consciousness of most of these respondents.

                 Empirical Test of the Communication Model

     Preliminary analysis of the model involved the plotting of scatter dia-
grams between each exogenous and endogenous variable prior to analysis by
two-stage least squares.  The purpose of such diagrams was to ascertain
visually the order of polynomial fitting the data.  It was determined from
this inspection that  the relationship between all variables was linear and
linear terms were employed for all reduced-form  equations.

     Measurable differences in perceptual accuracy were found among Salem
respondents.   It may well be that the communication process for those who
hold accurate perceptions will differ  from those who hold inaccurate or who
hold "no difference "perceptions.  A  test of  the appropriate residual sums of
squares was made to determine  if data for the three perception subgroups
in Salem could be pooled  and if the  data for both Eugene and Salem could
be pooled as well.—    The procedure Ostle [36]  describes was followed.
Pooling  of data for perception subgroups in Salem was  tested first.   Then,
pooling of data for Eugene and Salem groups was tested.

     When several sets of sample data are available,  one is interested in
determining if a single regression line can be used for all the data.   The
—'I am indebted to Dr.  Norbert Hartmann, Department of Statistics at Oregon
  State University,  for suggesting this procedure and for  completing  the
  calculations.

-------
                                                                         72
first step is to calculate the total residual sum of squares for each reduced
form equation (attitude,  mass media use,  knowledge) in the model for Salem.
The second step is to calculate the residual sum of squares of each perception
subgroup for each reduced form  equation of the Salem data and to sum these
residual sums of squares.  The third step is to subtract the residual sum of
squares  calculated in the second step from the total residual sums of squares
calculated  in the first step.  One now has obtained two residual  sums of
squares  -- one from Step two and one from Step three.  Each of these residual
sums of  squares is divided by its respective degrees of freedom.  Finally,
an F-ratio is calculated for each equation with n] and n  degrees of freedom.
If this  F-ratio is statistically significant,  the  regression  line for that
equation differs among the different subgroups.  If  the F-ratio is not sig-
nificant,  regression lines for each subgroup are considered the same, and
the data  can be pooled.

    The F-ratios  for the different Salem perception subgroups and for the
Eugene and Salem groups are presented in Table 9-

Table 9.   Analysis of Variance, Degrees of Freedom and Significance Levels
          for Residual Sums of Squares of Equations of the Communication
          Model
Equations
Combining over levels of
perception for Salem:
Attitude (Y.)
Mass media exposure (Y_)
Knowledge (Y,)
Combining over communities:
Attitude (Y.)
Mass media exposure (Y,)
Knowledge (Y )
F-Ratio
1. 98
1. 24
<1.00
<1. 00
1. 34
3. 34
Degrees of
Freedom
14,
14,
14,
7,
7,
7,
264
264
264
565
565
565
Significance
Level
NS
NS
NS
NS
NS
p< .01

-------
                                                                        73
     The data show that one can pool the data for perception subgroups in Salem,
but cannot pool the data for Eugene and Salem.  The F-ratio associated with
the knowledge equation is statistically significant.

     Multiple correlations and zero-order correlation Coefficients between
each exogenous and endogenous variable are presented in Table 10.

Table 10.  Multiple Correlation Coefficients and Zero-Order Correlations
           Between Each Exogenous and Endogenous Variable*

                                               Mass Media
                                   Attitude        Use         Knowledge
         Exogenous Variable         (Y.)          (YJ            (YJ
                                      L              C*               .3
                               Eugene
Interpersonal discussion (X )       . 447          . 520           . 497
Social Influence (X3)                . 382          . 379           . 425
Education (X)                     .351          .190           .436
Science education (X  )              . 344          . 238           . 470
Use of environment (XJ            .272          .262           .362
                      b
Multiple correlation coefficient     , 591          . 534           . 622
                              Salem
Interpersonal discussion (XJ       . 416          . 510           . 400
Social Influence (X )                . 308          . 362           . 342
Education (X4)                     .281          .244           .421
Science education (Xg)              . 253          . 277           . 419
Use of environment (X,)            .274          .210           .373
                      b
Multiple correlation coefficient     . 459          . 524           . 538

*p<.05  = . 113;   p< . 01 = .  148.

     All the correlations in  Table 10 are significant and the difficulty this
poses will become apparent when the model, analyzed by two-stage least
squares, is presented.  The structural form of the model is shown in
Equations (19) to  (24).  Numbers in parentheses  under the coefficients are

-------
                                                                           74
Student's t_ values:  (*) denotes . 05 


-------
                                                                            75
Equation (21) of the Eugene model (set of equations).  An alternative pro-
cedure was employed to reformulate the model.   To begin  with, chances are
good that low observed t_ values were produced,  in part, by the effects of multi-
collinearity, JL. e_. ,  at least one variable in an equation was adding little to
the explained variance that was not already accounted for by other variables.
One can check for this possibility by examining the  ratio of the  main diagonal
element of the inverse matrix corresponding to the  i—variable and the re-
ciprocal of the corrected total  sum of squares of the i — variable.—  The ideal,
but uncommon, situation is for this  ratio to equal unity;  however,   unity can
 occur only if all other explanatory variables in the model were completely
 uncorrelated with the i-
 sociological variables.
uncorrelated with the i— variable.  This rarely occurs  in practice with
     It can be shown that the relationship between the i — ratio and the
 variance of the i — variable accounted for by other explanatory variables in
 the model corresponds to the function:
                           D. =  —    -
                            1    1 - R2
                                     ij
where
              D.   is the ratio of the main diagonal element
                  for  the it*1 variable, and
             R..  is the variance of the i*"  variable accounted
               ^  for  by all other explanatory variables.

     Thus, when D. is large,  the variance of  the i — variable  is already
largely accounted for by other explanatory variables in the model.   Variables
with relatively high D. values will tend to have large variances of their esti-
mated coefficients; thus, very low precision of estimation will result -- even
to the point that  one is not  sure  of the sign of the  coefficients.  The values
                         2
of Student's t, D., and R..  obtained for variables in the model are pre-
            -   i        ij
sented in Table 11.  Variables with high  D. values were usually associated
with low Student's  t_ values.  Moreover,  the occurrence of a large
number of relatively high  ( i. e. , greater than 10) D. values was considered
—'I am indebted to Dr. William G.  Brown,  Professor of Agricultural Economics
  at Oregon State University, for suggesting this procedure.

-------
                                                                          76
Table 11.  The Values of Student's Jt_, D. and  R.. Associated with
           Variables  in Each Model    *        IJ
Equation Variable
(19) Xx
Y^
Y3
(20) X3
X4
A
Yl
(21) X5
X6
A
Yl
Y2
(22) Xj
Y2
Y3
(23.) X3
X4
A
(24) X5
X6
*1
*2
Student's t_
-0. 049
0. 346
3. 350
-0. 384
-3. 886

7. 379
1. 833
1. 105
2. 523
-0. 331
1. 577
-0. 916
2. 149
0. 980
-1. 363
6. 216
1. 296
1. 133
2. 268
0. 034
D.
L
53. 659
78. 538
5. 803
2. 289 "
1. 935

3. 588
3. 153
3. 945
15. 585
85. 523
35. 700
58. 880
12. 370
1. 829
1. 600
2. 360
7. 346
2. 183
13. 928
105. 429
R2
. 981
. 987
. 828
. 563
. 483

. 721
. 683
. 747
. 936
. 988
. 972
. 983
. 919
. 453
. 375
. 576
. 864
. 542
. 928
. 990
unacceptable.  Exogenous variables associated with high D. values were noted,
and intercorrelations among exogenous variables examined.   The X exogenous
variable,  interpersonal discussion, had the highest D. value of all exogenous
variables;  as well,  this variable was the only one correlating significantly
with all other exogenous variables.  And,  as noted  in Table 10, this variable
correlated highest with all endogenous variables, except for knowledge (Y,)
in Salem.  It quickly became apparent that this exogenous variable was

-------
                                                                        77
improperly specified and was likely the source of difficulties encountered in
the test of the model.   The interpersonal discussion variable should have
been endogenous.

    A new model was formulated.   Interpersonal discussion was considered
a new  endogenous  variable.  Social influence was specified as exogenous.
Level  of formal education was deleted (it correlated . 80 or higher with level
of science education) and a new exogenous variable, perceived seriousness
of air  pollution in a respondent's community of residence, was added.
This variable was postulated to be positively related to attitude towards air
quality.  Thus,  the new model contained four exogenous and four endogenous
variables.

    Conclusions based on significance tests of the  reformulated model,
however, must be viewed with caution.   The equations were developed after
data from  the previous model had been examined.  Such selection tends to
capitalize  on relationships which might be high because of sampling fluctu-
ations.  This research must be considered exploratory,  due to the tentative
nature of the theoretical model.  Previous work cited to support the model
usually employed  comparisons between two variables considered one at a
time;   no previous work reported has attempted to  estimate the more general
relationships specified among all the  variables simultaneously.   When con-
fronted with this type  of situation,  it is often necessary to experiment with
different hypothesized systems.  Consequently,  one must be particularly
cautious in interpreting tests of significance for the reason cited earlier.
Data from the previous model will be employed to test the reformulated
model.  This test is less meaningful than if new data had been acquired.

           Empirical Test of the Reformulated Communication Model

    Pooling of data for perception subgroups in Salem and of data for
Eugene and Salem groups was completed in a manner similar to that for  the
original model.  The results are presented in Table 12.

-------
                                                                            78
 Table 12.  Analysis of Variance, Degrees of Freedom and Significance
           Levels for Residual Sums of Squares of Equations for the
           Revised Communication Model
Model
F-Ratio
Degrees of
 Freedom
                                                                 Significance
                                                                    Level
 Combining over levels of
 perception for Salem:
Attitude (Y )
Mass media exposure (Y_)
Li
Knowledge (Y )
Interpersonal discussion (Y )
3.
1.
1.
1.
18
10
43
15
10,
10,
10,
10,
270
270
270
270
p<. 01
NS
NS
NS
     The data show that one cannot pool the data among the Salem subgroups.

 The F-ratio associated with the attitude equation is statistically significant.

 Further tests among the different residual sums of squares indicated that

 one could pool the inaccurate  and "no difference" perception subgroups.

 Note that it would be inappropriate to test for combining over communities,

 since the equation for attitudes (Y.) in Salem was significant.


     Multiple correlations and zero-order correlations  between each exogenous

 and endogenous variable of the reformulated model are presented in  Table 13.


     It was noted  in Table 10 that all correlations were statistically significant;

'however, some correlations in Table 13 are not significant.   As well,  several

 of the intercorrelations among the exogenous variables of the reformulated

 model are not significant.   Thus,  the magnitude of the difficulty encountered

 in testing the original model is not likely to be  repeated  in the test of the re-

 formulated model.

-------
Table 13.  Multiple Correlation Coefficients and Zero-Order Correlations
           Between Each Exogenous and Endogenous Variable
Mass Media

Endogenous Variable

Social influence (X )
Science education (X )
Use of environment (X,)
o
Seriousness (X )
Attitude
'V
Eugene*
. 382
. 344
. 212
. 361
Use


. 379
. 238
. 262
. 124
Knowledge


.425
. 470
. 362
. 207
Peer
Discussion
!no<.differenc.e"perceptions)
Social influence (X3)          . 375         . 432         . 431         . 564

Science education (X)         .272         .295         .464         .250
Use of environment (XJ       .409         .312         .434         .375
                      b
Seriousness (X  )              . 349         . 310         . 209         . 281

Multiple correlation coe
    coefficient                .540         .410         .588         .617
                                                 *#
                     Salem (accurate perceptions)
Social influence (X )
Science education (X )
Use of environment (X,)
o
Seriousness (X )
Multiple correlation
coefficient
p<. . 05 = 113; p< . 01 = . 148
**p < . 05 = . 159; p<. 01 = . 208
258
235
154
309
384


. 292
. 254
. 123
. 221
. 359


. 260
. 373
. 304
. 258
. 450


. 509
. 236
. 279
. 187
. 530



-------
                                                                            80
     The structural form of the revised model is presented as Equations (26) -

(37).   Equations (26)  -  (29) are for Eugene; Equations (30) - (32) for Salem

respondents who held either inaccurate or "no difference" perceptions;
                                     *
Equations (34) - (37)  are for Salem respondents who held accurate .perceptions.

Again,  numbers in parentheses under the coefficients are Student's Rvalues:

(*) denotes . 05 

Interpersonal discussion Use of environment (X6) Seriousness of air pollution Mass media use (Y2) Science education (X5) Figure 23. Diagram of the communication model for Eugene.


-------
                                                                              81
     In Equation (26) Air Duality Knowledge (Y  ) and Perceived Seriousness
                                              5                            A
of Air Pollution Ln Eugene (X_) are significantly related to Attitude  Level (Y )
                                 A
•while  Interpersonal Discussion (Y ) is not.  In  Equation (27)  Air Quality

            A                                 A
Knowledge (Y,)  and Interpersonal Discussion (Y ) are significantly  related
                     A                                        A
to Mass Media Use (Y_).   In Equation (28) Mass Media Use (Y,), Use of
                      £*                                        L*

the Environment (X,) and Science Education (X ) are significantly related to
                        A
Air  Quality Knowledge (Y  ).  In Equation (29) Attitude Level  ($) and


Social Influence  (X0) are significantly related to Interpresonal Discussion

(Y4) while Air Quality Knowledge (Y  ) is not.



     The next set of equations  represents the communication  model for Salem


respondents who held inaccurate or "no difference" air quality perceptions:
Y = 75. 5231 -1- 2. 6610 X + 1. 4089 Y + 2. 3386 Y (30)
(2.006)** (1.888)** (1.400)*
^ A
Y2 = -3. 8072 + 0. 3042 X + 0. 1262 Y (31)
(1. 691)** (3. 573)***
^ A
Y = 5. 0937 + 0. 1344.X + 0. 1475 X, + 0. 9432 Y- (32)
(2.770)*** (2.043)** (2.787)***
Y = -0. 1090 + 0. 8132 X + 0. 2381 Y - 1. 2257 Y (33)
(2.487)*** (1.861)** -1.232)
Social
influence

\

Mass media
use (Y2)
Social
(X3)
/
Knowledge

f3)


Science
(X5)
     Figure 24. Diagram of the communication model for Salem respondents with inaccurate or 'no

             difference "air quality perceptions.

-------
                                                                            82

                                             A
     In  Equation (30) Air Quality Knowledge (Y.J, Interpersonal Discussion
 A                                             *
(Y  ) and Perceived Seriousness of Air  Pollution in Salem (X )  are significantly
                          A                                 '    A
related to Attitude Level (Y ).  In Equation (31) Attitude  Level (Y.) and
                                                                       /\
Social influence (X  )  are significantly  related  to  Mass Media Use  ^Y~).
                                   A
In Equation (32) Mass Media Use (Y_),  Science Education (X_)  and Use of-
                                    &                        5
the Environment (X,) are all significantly related to Air  Quality Knowledge

(Y3).  In Equation (33) Attitude Level (Y  ) and Social Influence (X ) are

significantly related to Interpersonal Discussion (Y ) while Mass  Media

Use (Y_) is not.


     The final  set of equations represents the communication model for

Salem respondents who held accurate air quality perceptions:
w
YI = 90. <
V -
Y3=-5.I
V '• =
Social

i '
Interpersonal
785 + 2. 5098 X + 1. 7735 $
(1. 787)** (2. 606)***
* A
5974 + 0. 2464 Y + 0. 3846 Y
(1. 768)** (1. 447)* 4
787 + 0. 0955 X + 0. 1034 X + 0. 19
(1. 751)** (1. 670)** (2. 51
A
i990 + 0. 5212 X + 0. 1271 Y
(5.004)*** (1.480)*
Attitude ^



.
^\ /
^ Knowledge 	

(34)
(35)
40 Y" (36)
6)***
(37)
Seriousness

-------
                                                                         83



       In Equation (34) Air Quality Knowledge (Y ) and Perceived Seriousness


of Air Pollution in Salem (X ) are significantly related to Attitude Level (Y ).


In Equation (35) Air Quality Knowledge (Y_) and Interpersonal Discussion (Y )
                                         «J                                  *

are significantly related to Mass Media Use (Y2).  In Equation (36) Attitude


Level (Y  ), Science Education (X ) and Use of the Environment (X,) are sig-


nificantly related to Air Quality Knowledge (Y ).   In  Equation  (37) Air


Quality Knowledge (Y_) and Social Influence (X ) are significantly re-


lated to Interpersonal Discussion (Y ).




                            Discussion of Results
      Results concerning statistically significant relationships among


endogenous variables will be discussed for each model.   Then comparisons


will be made among the different models.




      For Eugene respondents, the relationship between mass media use


and air quality knowledge  is interdependent,  L e. , one serves as the basis


for the other  and vice versa.   High knowledge levels are associated with


positive attitudes toward air quality which,  in turn,  are related  positively


to interpersonal discussion.  High levels of interpersonal discussion are


positively associated to high levels of mass media use.




      For Salem respondents who held inaccurate or "no difference" air


quality perceptions, positive attitudes toward air quality were positively


associated with high levels of mass media use.  Mass media use was posi-


tively related to air quality knowledge which,  in turn,  was related positively


to attitudes.  An interdependent relationship was found between attitudes and


interpersonal discussion.




      For Salem respondents with accurate  air quality perceptions, an inter-


dependent relationship was found between attitudes  and air quality knowledge.


High knowledge  scores were positively related to mass media use and to


interpersonal discussion.   Interpersonal discussion was positively related


to mass media use.

-------
                                                                          84





     Moreover,  the postulated relationship between  mass media use and



attitudes, as stipulated by the definitional hypothesis, is not supported



by the data.  It may be more correct to state that mass media use is posi-



tively related to  knowledge and knowledge is positively related to attitudes.



This  type of relationship was found for all models.






     It may  be profitable  to compare the different models, particularly since



the analysis of mass media content disclosed such wide differences in the



time  of definition of air pollution as an issue in both communities.  As well,



the role of mass media use appears to vary  according to community and to



accuracy of perception.   For example,  an interdependent relationship was



found in Eugene between  mass media use  and air quality knowledge.   A



dependent relationship was found between mass media use and knowledge •



among Salem respondents holding inaccurate air pollution perceptions.



And,  a dependent relationship was found between knowledge and mass media



use among Salem respondents with accurate  perceptions.






     What may well be going on -- as one  views the  different models along



a postulated time continuum --is the evolution of a single communication



process concerning this issue.






     Initially,  the issue is poorly perceived or defined with the  communication



process functioning much like that of Figure 24.   The direction of the arrows



for the endogenous variables suggest selectivity in information seeking.



Those with  more positive attitudes toward the topic interact or discuss it



more with peers.  As well,  this interaction forms and reinforces their weak



but positive attitudes toward air quality.  Those who hold positive attitudes



are greater users of the  mass media and  it is argued that this use is likely



to be biased or selective  in favor of these weakly-held attitudes.  People



tend to seek out information which supports  their positive attitudes and to



avoid or misperceive messages which threated or fail to support them.



Greater mass media use,  however,  leads to higher levels of knowledge which



tends to form and reinforce positive attitudes.

-------
                                                                          85
     Next,  after the passing of time, the accumulation of information from the
mass media has led to a higher level of knowledge.—   The communication
process now may resemble that of Figure 25.  Mass media use is no longer
selective;  _i. e_. ,  it is not a function of attitude level.  Rather, attitudes are
well established and are now based primarily on level of accumulated know-
ledge.  High knowledge levels now produce greater information seeking in
both the mass media and wi th other people.  It is argued that selectivity
associated with the process in Figure  24  is likely to have been abandoned.
Now, information seeking is unbiased,  with knowledgeable people having
formed well-differentiated and elaborate* cognitive structures concerning
             2/
air pollution.—    Positive attitudes associated with these elaborate cognitive
structures are relatively impervious to change  -- unless there is  a significant
change in the cognitive  structure itself.   Incompatible information is not
avoided or misperceived, for example, but is  likely to be sought out and
discounted through counter-argument.   The mass media is used by both those
high in knowledge and those who are involved in interpersonal discussion.
It may well be  that there is insufficient information available in the mass media
channels to meet the information needs of more knowledgeable people.
Otherwise,  the direction of the arrow  would likely go from mass  media use
to knowledge.

     Finally, the process may evolve to a system much like that of Figure 23.
The interdependence between mass media use  and  knowledge  suggests a suf-
ficiency of information about air  quality in the mass media channels and that
the more knowledgeable people are actively seeking out this information.
— Student's t  values were calculated for attitude and knowledge means between
  Salem accurate and inaccurate or "no difference" . subgroups.   Means of both
  groups for both variables were statistically significant in the predicted
  direction (accurate  means were higher)  at the . 05 level or less.   More-
  over,  it is not likely that these differences were produced as artifacts of
  differences in exogenous variables.  Student's t_ values were also calculated
  to test the significance of difference between means of each exogenous vari-
  able for both groups.  No difference was statistically significant.
2/
— The term  'cognitive structure' refers to the way an individual organizes his
  physical and social worlds,  including all his facts,  concepts,  beliefs,
  expectations  and the pattern of their interactions about the topic.

-------
                                                                            86
Moreover,  higher  knowledge levels support the more positive attitudes toward
air quality which lead to greater interpersonal discussion.  Those who discuss
air pollution with others are also greater users  of the mass media.  Again,
it  is argued that information seeking at this stage of the process is also un-
biased.  Information inconsistent with or in refutation of existing knowledge,
for example,  is  likely to be verified with some authority.  Inaccurate or in-
consistent public statements,  for example,  are likely to be challenged. And,
one is likely to find that the air pollution knowledge levels of  citizens in a
community which has an attentive local mass media is remarkably sophisticated.

    One can conclude that more  than one communication model is  function-
ing in a community at one point in time, depending on perceptions  people
have  about the environment.  As well, whether or not different models
are functioning may well depend  on how the  local mass  media, particularly
newspapers, define the  situation.  Mass media use appears to play a sig-
nificant role in the  accumulation of knowledge which contributes to the
establishment  of attitudes.  The  decisions of mass media gatekeepers  appear
to be  most relevant in determining the extent of knowledge the citizenry of a
community may  have about the local aspect  of a complex issue.  Word-of-
mouth channels can function up to a point, but their ability to  communicate
valid  information of a technical nature is limited, compared to newspapers,
for example.   The outcome  of  efforts to obtain clean air' for  a community
may well rest,  in part,  with the  decisions of these local mass media gate-
keepers  in the selection of information they employ to define  the issue.

    It is doubtful if one will find this particular communication process at
work  in all other types  of air pollution situations.  Consider,  for example,
the situation where the  community which receives both the economic benifits
and the externalities of a polluter and where a polluter  and enforcer disagree
over  the feasibility of controls.  One is not  likely to find a positive linear
relationship between attitudes  and  knowledge among community residents
over  this issue.   Instead, a curvilinear relationship is more  likely -- those
who hold extreme positive or negative attitudes will have relatively high

-------
                                                                            87






knowledge scores.  Attitudes of those with low knowledge scores will likely



be neutral,  L £. ,  tend to have "no opinions".   In this instance the role of the



local mass  media -- particularly newspapers -- may well have a bearing on



any empirical relationship between attitude and knowledge.  For  if the  local



media "manage" the issue by not reporting  it (or not reporting it fully), then



one will need to consider the effects of possible  alternative mass media



sources,  such as state newspapers, in the model.  Otherwise,  an empirical



test of the model may disclose a null relationship between mass media use



and knowledge.  One decision  required in developing such a model would be



to build in the effects of the "localness" of the media  itself.






     The role of exogenous variables should be considered as well.  The role



of social influence and interpersonal discussion  concerning air pollution can



be more precisely defined.  It was recognized in the review of literature that



social influentials expose  themselves not only to pertinent forms of the mass



media,  compared to those less influential,  but are involved more in inter-



personal discussion as well.   Interpersonal discussion, however,  plays a role



in attitude formation at the initial stage of the communication process.   The



importance of interpersonal discussion may be attentuated once knowledge levels



and attitudes  are established.   To be sure,  social influence  is related to inter-



personal discussion at the latter stages of the communication process.   But



interpersonal discussion appears to lead only to  greater mass media use.



This  use  is related  to higher knowledge levels.   A direct relationship between



interpersonal discussion and either knowledge or attitudes is not supported



by the data.   Thus,  any information effort to  improve attitudes concerning air



pollution  should take into account  the significant role of these influentials at



the initial stage of the effort.






     The positive relationship between science education and knowledge, for



example,  suggests that knowledge levels among  the  better educated are likely



to be higher than among the less educated.  And,  over time,  the knowledge gap



between  the highly-  and poorly-educated segments of a community will  likely



become larger, not smaller.  Such a gap stems not  only from higher initial

-------
                                                                           88



knowledge levels but also from the use of the mass media as a learned skill

for obtaining  information.  Tichenor jet aL ,  [5l]  have pointed out the statistical

regularity associated with an increasing knowledge gap for different educational

levels and the implications this has for the "mass" impact of the media;  viz. ,

that of reinforcing or increasing existing inequities  in public knowledge.

Thus,  any program to employ the  mass media in reaching large numbers of

citizens with  information concerning air pollution, for example,   is likely to

reach the already well-informed.   This does not mean that the well-informed

will not  profit from any new information effort.  It suggests that  these individuals

will be the primary beneficiaries of such an effort.



     Use of the environment for recreation purposes, a significant predictor

of air quality knowledge at all stages of the communication process, also plays

a role a-s a source of attitude change as well.  One question likely to arise

when using two-stage least squares is to compare the relative effects of  all

the exogenous variables on each endogenous variable.—   For  example,  if a 1%

change occurred in each exogenous variable,  how much of a percent change

would occur in each endogenous variable?  Using level of attitudes as the

endogenous variable,  an answer to this question showed that social influence

would produce the  greatest change  for Eugene and for Salem respondents who

held accurate air pollution perceptions.  In these instances a 1%  change  in

social influence would change attitudes toward air quality by 10% for Eugene

respondents and by 7% for Salem "accurate" respondents.  Effects  of other

exogenous variables were relatively small.  However, a  change in use of the

environment for recreation was by far the greatest producer of attitude

change among Salem respondents who  held inaccurate or  "no difference" air

pollution perceptions.  A  1% change in use of the environment would produce

a 16% change  in attitudes.  The implication is clear.   One way to sensitize

respondents who hold inaccurate or "no difference" perceptions  is to persuade
                     •
them to  use the environment more  for recreation.  Individuals who make greater

use of the environment for recreation  appear to hold more positive attitudes
—'I am indebted to Dr.  Albert N. Halter,  Professor of Agricultural Economics
at Oregon State University,  for  suggesting this procecure and for completing
the calculations.

-------
                                                                            89

toward air quality.  In turn, they are able to more accurately compare visi-
bility differences and may see the relatively clean air of their recreation area
as an "escape" from the visibility restrictions they perceive at home.

     There is little one can do to change the social influence of an individual.
Thus, producing changes in attitude by changing  the level of this  exogenous
variable would not be particularly effective for Eugene and Salem "accurate"
respondents.  But, for Salem "inaccurate" respondents,  efforts to change
attitudes toward  air quality are likely to produce positive results.  One has to
find some way to persuade these people to spend more  of their recreation
hours out-of-doors away from their home and, perhaps,  to spend more of their
time on recreation activities.

     Finally, the belief that air pollution is serious  -- based on the individual's
experience living or  working in the community -- is a significant exogenous
effect that appears to have a bearing on attitudes at all stages of  the communi-
cation process.   Beliefs based on  experience may well provide a basis for
the acceptance of one of several definitions of air pollution situations with
which an individual may be confronted.   It may be redundant to assert that
the first step in  improving air quality is for a community to recognize that it
has a serious problem.  It  is,  nonetheless, a truism to state that efforts
to improve air quality are not likely to  succeed without this recognition.
A pre-condition for the  establishment of a successful air pollution control
program is  that the citizens of an  airshed recognize the scope and magnitude
of the problem.   This pre-condition is  satisfied when they perceive odors and
loss of visibility, for example,  for what they are and in terms which they
believe are  important to them.

-------
                                                                         90
                          LITERATURE CITED

1.   Gordon Allport and Leo E.  Postman,  "The Basic Psychology of Rumor, "
         in Readings in Social Psychology,  Eleanor E. Maccoby,  T. M.
         Newcomb, and E.  L.  Hartley,  eds. ,  New York:  Henry Holt,
         pp.  54-65,  1958.

2.   E.  Jackson Baur,  "Opinion Change in a Public Controversy, " Public
         Opinion Quarterly, Vol.  26,  pp.  212-226,  1962.

3.   Kenneth Boulding, "The Economics of the Coming Spaceship Earth"
         in  Henry Jarrett, ed. , Environmental Quality in a Growing
         Economy, Baltimore:  The Johns Hopkins Press,  1966.

4.   James E. Brinton and L. N. McKown, "Effects of Newspaper Reading
         on Knowledge and Attitude, "  Journalism Quarterly, Vol. 38,
         pp.  187-196,  1961.

5.   Charles F.  Cannell and James C.  MacDonald,  "The Impact of Health
         News on Attitudes and Behavior, "  Journalism Quarterly, Vol. 33,
         pp.  315-323,  1956.

6.   Rachel Carson, Silent Spring,  New York: Houghton  Mifflin,  1962.

7.   Census of Population,  Advance Report,  PC(VI) - 39,  Oregon, U. S.
         Department of Commerce,  Bureau of Census, December 1970.

8.   James S. Coleman,  Community Conflict,  Glencoe, 111.  The Free
         Press,  1957.

9.   Barry Commoner,  Science and Survival, New York:   Viking,  1967.

10.  Lewis Coser, The Functions of Social Conflict,  New York:  The
         Free Press, 1956.

11.  M.  Jay Crowe,  "Toward a 'Definitional Model1 of Public Perceptions
         of Air Pollution, "  J. of the Air Pollution Control Association,
         Vol. 18,  pp. 154-157,  1968.

12.  Robe.rt C. Davis,  The Public  Impact of Science in the Mass Media,
         Ann Arbor: Survey Research Center,  University of Michigan,
         1958.

13.  Paul B. Downing,  "Solving the Air Pollution Problem:  A Social
         Scientist's Perspective, " Natural Resources Journal, Vol.  II,
         pp.  693-713,  1971.

-------
                                                                          91

14.  Jacob J.  Feldman,  The Dissemination of Health Information,  Chicago:
         Aldine,  1966.

1 5.  Raymond L. Gordon,  "Interaction Between Attitude and the Definition
         of the Situation in the Expression of Opinion, "  American Sociological
         Review,  Vol.  17,  pp.  50-58,  1952.

16.  Harold Guliksen,  Theory of Mental Tests,  New York:  Wiley, 1950.

17.  J. R.  Hardison, "Status of Field Burning and Alternative Methods for
         Disease Control in Perennial Grasses, "  in R. M.  Alexander,  ed. ,
         Agricultural Field Burning in the Willamette Valley,  Corvallis,
         Oregon:  Air Resources Center,  Oregon State University,  pp. 18-24,
         1969.

18.  Paul Horst,  Psychological Measurement and Prediction, Belmont,
         Calif. :  Wadsworth, 1966.

19.  Carl I. Hovland,  "Reconciling Conflicting Results from Experimental
         and Survey Studies of Attitude Change, "  The American Psychologist,
         Vol.  14,  pp.  3-14, 1959.

20.  Hans Huth, Nature and the American: Three Centuries of  Changing
         Attitudes, Berkeley,  Calif. :  University of California  Press,,
         19,57.

21.  Herbert H. Hyman and Paul  B.  Sheatsley,  "Some Reasons Why Infor-
         mation Campaigns Fail,  "  Public Opinion  Quarterly,  Vol.  11,
         pp. 413-423, 1947.

22.  J. Johnston,  Econometric  Methods,  New York:  McGraw-Hill,  1963.

23.  V.  O. Key Jr. , Public Opinion and American Democracy,  New York:
         A, A. Knopf,  1961.

24.  Joseph T. Klapper, The Effects of Mass Communication,  Glencoe,
         111. :   The Free Press,  1961.

25.  Jerome M. Levine and Gardner Murphy,  "The Learning and Forgetting
         of Controversial Material, " J.  of Abnormal and Social Psychology,
         Vol.  38,  pp.  507-517,  1943.

26.  Robert G. Mason,  Selection  of Items for an Air Pollution Knowledge
         Test, Corvallis,  Oregon:  Special Report  324, Agricultural
         Experiment Station,  Oregon State University,  1971.

27.  Peter McHugh, Defining the  Situation, New York:  Bobbs-Merrill,  1968.

-------
                                                                          92

28.  Jack M. McLeod and James W.  Swinehart,  Satellites,  Science and the
         Public: A Report  of a National Survey on the Public Impact of Early
         Satellite Launchings,  Ann Arbor: Survey Research Center,
         University of Michigan,  1959.

29.  Nahum Z.  Medalia,  Community Perception of Air Quality: An Opinion
         Survey of Clarkston,  Washington, Cincinnati:  Environmental
         Health Service, U.S. Department of Health, Education,  and
         Welfare, 1965.

30.  Willis E. Middlemiss and Robert O.  Coppedge, Oregon's  Grass and
         Legume Seed Industry in  Economic Perspective,  Corvallis, Oregon:
         Special Report 284, Cooperative Extension Service,   Oregon State
         University,  1970.

31.  W. E.  K. Middleton, Vision Through the Atmosphere, Toronto:
         University of Toronto Press,  1952.

32.  Roderick Nash, Wilderness and the American Mind.  New  Haven:
         Yale University Press, 1967.

33.  Jum Nunnally,  Popular Conceptions of Mental Health, New York:
         Henry  Holt,  1961.

34.  	, "Experimental Studies of Communicative  Effectiveness, "
         in  Wilbur Schramm, ed. , Studies of Innovation and Communication
         to  the Public, Stanford:  Institute for Communication  Research,
         Stanford University, pp.  175-202, 1961.

35.  Clarice N.  Ol/sin,  et. aL ,  "The  Community Editor's Power and the
         Reporting of Conflict, "  Journalism Quarterly,  Vol.  45,  pp.  243-
         252,  1968.

36.  Bernard Ostle,  Statistics  in Research, Ames, Iowa:   The Iowa State
         University Press,  1963.

37.  Elmer  Robinson,  "Effects of the Physical Properties  of the Atmosphere, "
         in  A. C. Stern,  ed. ,  Air Pollution,  New York:  Academic Press,
         Vol.  I, pp.  349-400,  1968.

38.  Merrill E.  Samuelson,   et.al., "Education, Available  Time and Mass
         Media Use, " Journalism Quarterly,  Vol. 40,  pp.  491-496, 1963.

39.  Thomas J.  Scheff, "Toward a Sociological Model of Consensus, "
         American Sociological Review,  Vol.  32,  pp. 32-46,  1967.

-------
                                                                            93
40.  Wilbur Schramm, "Science News and the Public Mind, " in Wilbur Schramm,
         ed. ,  Studies of Innovation and of Communication to the Public,
         Stanford:  Institute for Communication Research,  Stanford
         University,  pp.  261-286,  1962.

41.  Jean J.  Schueneman, "Air Pollution Control Administration, " in A. C.
         Stern,  ed. ,  Air Pollution, New York: Academic Press, Vol.  Ill,
         pp.  719-795, 1968.

42.  Muzafer  Sherif and Carolyn W. Sherif,  An Outline of Social Psychology.
         New  York:  Hapers, 1956.

43.  George W.  Snedecor, Statistical Methods,  Ames,  Iowa:  The Iowa
         State College Press,  4th edition,  1946

44.  Shirley A.  Star  and Helen MacGill Hughes,  "Report on an Educational
         Campaign:  The Cincinnati Plan for the United Nations, " American
         J. of Sociology,  Vol. 55, pp. 389-400,  1950.

45.  Robert A. Stebbins,  "A Theory of the Definition of the Situation, " The
         Canadian Review of Sociology and Anthropology, Vol. 4,  pp.  148-
         164,  1967.

46.  Arthur C. Stern, "Air  Pollution Standards, " in A. C. Stern, ed. ,
         Air  Pollution, New York:  Academic Press, Vol. Ill,  pp.  601-720,
         1968.

47.  The Quality of Outdoor Recreation: As Evidenced by User Satisfaction,
         Report to  the Outdoor Recreation Resources Review Commission,
         Study Report 5,  Washington:  U. S. Government Printing Office,  1965.

48.  The Roper  Report,  New York:  The Roper Organization,  Inc. ,  March
         1972.
49.  W. I.  Thomas,  The Child in America, New York:  A. A.  Knopf,  1928.

50.  Phillip Tichenor, Communication and Knowledge of Science in the Adult
         Population of the U. S. ,  unpublished Ph. D.  Dissertation, Stanford
         University,  1965.

51.  P.  J.  Tichenor,  ei_ &L ,  "Mass Media Flow and Differential Growth in
         Knowledge," Public  Opinion Quarterly,   Vol. 34,  pp. 159-170,
         1970.

52.  P.  J.  Tichenor,  e±aL,   "Environment and Public Opinion,"  The J.  of
         Environmental Education,  Vol.  2,  pp.  38-42,  1971.

-------
                                                                          94
53.   Stefan Valvanis, Econometrics, New York: McGraw-Hill,  1959-

54.   R.  B. Zajonc,  "Attitudinal Effects of Mere Exposure, " J.  of Personality
        and Social Psychology Monograph Supplement,  Vol. 9,  Part 2,  1968.

-------
APPENDICES

-------
    APPENDIX A





Review of Literature

-------
                                                                           95
                              APPENDIX A
                           Review of Literature

     Few propositions are as widely accepted as the one -which claims a re-
lationship between the  attitudes people have toward some object and the
knowledge they possess about it.   Such a  relationship is assumed in any
attempt to create,  reinforce  or convert attitudes by the dissemination of
factual information.  It is in  this context  that a review of literature will
provide much in  the way of a foundation to support the  theoretical model.
The review will consider first the postulated relationships of variables in
the interdependent system, then the postulated relationships of the predictor
variables.

Effects of the Interdependent System

     Attitudes.    The role of attitudes and attitude change in predicting
levels or changes in  exposure and knowledge has been  studied in several
areas.  Tichenor [50 ] ,  in his re-analysis of the effect of a "threat of science"
attitude scales on science knowledge,  found  that people with lower science
knowledge scores perceived science as more "threatening".  Nunnally
[33, 34 J ,  utilizing laboratory experiments,  discovered that it is relatively
easy to convey new information about mental illness.   Any seemingly authori-
tative and comprehensible information will be accepted by the public.
Attitudes are more difficult to change,  but they can be changed.  Public
anxiety over mental illness concepts is one of the  facets affecting attitude
change toward this topic.  If  anxiety-producing features of the message are
not diminished, adverse attitude toward the  topic develops  and a premature
closure of information  that may be relevant follows.  Acquiring information,
regardless of its validity, often tends  to reduce fear  that may be built up
about mental illness.  Thus,  the mere act of acquiring information leads to
"constructive" attitude change.  Conversely,  when information is invalidated --
through counterargument by scientific authority, for  example  -- anxiety about
the topic tends to increase.  He concludes that it may be unwise to invalidate

-------
                                                                          96
existing public misinformation unless a valid and understandable body of
knowledge can be given in return.

    Mass Media Exposure.   Schramm [40]  considers use of the mass media
to be  an important predictor of scientific information; after school years,
most  of the increment in science knowledge comes from the mass media.
Amount of exposure to the mass media is often used to predict science
knowledge,  with or without controlling for predisposition effects, such as
level  of education.  Brinton and  McKown [4]  found that newspaper sub-
scribers differed significantly from non-subscribers in knowledge concern-
ing fluoridation of a public water supply, at a time and in an area marked
by a dispute over this issue.  They concluded that a high degree  of personal
involvement in an issue and a perceived need for information  are not necessary
for a  person to  expose himself to relevant messages  in the mass media.

    McLeod and  Swinehart [28] have found that media exposure was re-
lated  to level of science information.  Their 1958 (post-Sputnik)  study
showed that mass media exposure  was closely linked to attribution of a
scientific purpose to satellites,  even with the effect of education controlled.
Since the  two  variables (education  and exposure)  were independent pre-
dictors, this provided evidence that mass media  exposure may be viewed
as a distinct effect and not merely an artifact of education.

    Davis [ 12]  was able to describe the amount  and  kind of media used
after  controlling for level of science knowledge.  He  found that highly
informed people were heavy users of the mass media, used more forms
of the print media (newspapers,  books,  magazines,  etc. ) and used them
more intensely  than those less informed.  Attention to the electronic media
was less clear-cut.   Better informed persons were more likely than the
less informed to be  in the radio  and television audience,  but they spent
less total time with  these mediums.  The most extensive attention to radio
and television was among the less  informed.   In a comprehensive survey
on the dissemination of health information, Feldman  [ 14] has noted that

-------
                                                                         97
better informed in dividual a were also newspaper and magazine readers, and
constituted a read/ audience for the health educator.

    Science knowledge can affect mass media exposure and the question of
whether knowledge precedes or follows exposure --  or both -- is often
unresolved.  Studies of selective processes  have frequently suggested that
persons with higher interest and prior knowledge are more likely to expose
themselves to sources of information to gain even more knowledge.  For
example,  Myman and Sheatsley [21]  report that persons showing high
interest in the atomic bomb were more than twice as likely as uninterested
persons to report having been exposed to information about an early atomic
energy report.   Selective processes  also work to "screen out" threatening
messages from  the individual's perceived environment.  Illustrative of
this process are the findings  of Cannell and  MacDonald  [ 5] .   They dis-
covered that readership of articles on health --  particularly those dealing
with the possible relationship between smoking and cancer -- were con-
sistently read by 60  percent of the non-smoking  males but by only 32 per-
cent of the smoking males to whom the information presented was more of
a threat.  Among women, for whom the threat was considerably less,  they
found that the respective readership  percentages were 69 and 59.

    Knowledge.    A widely recognized tenet of social psychology is  that
people tend to select and retain information  which is consistent with their
prior attitudes.   Messages which conflict with established attitudes are often
misperceived, distorted or ignored in order to "fit"  with the pre-existing
attitudes.   Illustrative of this principle is the Allport and Postman [1]  study
on rumor  transmission.  Persons who were shown a picture  of a white man
and a Negro arguing, with the white man holding a knife, later reported that
the Negro was wielding the knife,   consistent with their prior attitude  toward
Negroes.  Levine and Murphy [ 25] presented pro- and anti-Communist
statements to students.  Afterwards,  students tended to recall only informa-
tion consistent with their prior  attitudes:  Pro-Communists were more likely
to remember pro-Communist information, and vice-versa.  Star and

-------
                                                                           98

 Hughes [44], noting a failure of a massive information campaign in Cincinnati
 to increase level of information concerning the United Nations, concluded that
 any increase in exposure was determined largely by peoples' previous
 orientation.   Interest and prior knowledge determined who would attend the
 information  campaign and those who attended were relatively unaffected by
 the campaign itself.   Thus,  what one knows about a subject often can serve
 as a predictor of his attitude about the subject.   Tichenor [50] has supported
 this in two respects.  He found that general science knowledge was related
 to three attitudinal items that were combined into an index of "threat of
 science".  It was noted earlier that those who had lower science knowledge
 scores perceived science as more  "threatening", i. e. , fear that science may
 mean control by a few over  many,  concern for science-caused rapid changes
 and the belief that science tends to break down moral ideas.   In addition, he
 reported that persons with  lower science knowledge scores  tend to hold
 negative attitudes toward science.  This pattern  held at all levels of education
 and was not  associated with level of exposure to  science information in the mass
 media.

     Proposed changes in the environment, however, have formed attitudes
 among individuals who perceived that they might  affect  the outcome of these
 changes.  In his study of a local controversy concerning alternative methods
 of flood control and-water conservation,  Baur [ 2] found that attitudes were
 formed,  in part, through subtle changes in kinds of knowledge about the
 issue.

 Effect of Predictors

     Level of Exposure to Interpersonal Information Sources.   Exposure to
 interpersonal discussion, mass media use and level of  knowledge may com-
plement each other in forming or reinforcing  attitudes toward  air quality.
Information   reaching individuals via the mass media is considered sus-
ceptible  to the mediation  of other people,  as  Klapper [24] has noted.
Moreover, Baur [ 2 ],  in his study of attitude formation and change in a

-------
                                                                          99
public controversy, concluded that opinions emerged and were changed pri-
marily through interpersonal communication, although changes were initi-
ated by formal organizations and projected by the mass media.  Level of
interpersonal discussion is considered a predictor of attitude.

     Experience with Air Pollution.  Experience  refers to the  level of parti-
cipation or living in an area in which a known deterioration in air quality
has occurred.   It is considered a basis for both attitude formation toward,
and knowledge about, air quality.  Medalia [29] , for example, has reported
that length of residence in Clarkston,  Washington,  (an  area subjected to
continuous pollution from a kraft pulp mill) was  strongly related to concern
with air pollution.  Approximately two-thirds of  the persons interviewed
who had lived in Clarkston in 1950 or before were highly disturbed by air
pollution as a problem,  compared to a little over a third of those who had
moved to  the city in 1951 or later.

      Level of  Social Influence..  People  differ  in  their  level of social
influence,  and  a  well-established proposition concerning the behavior
of influentials  is their level  of exposure to the mass media,   particularly
the printed word.  Those high in influence tend to expose themselves to
pertinent  forms of the mass media compared to those less influential,
according to Klapper [24] .   Level of social influence is considered
a predictor  for level of mass media exposure.

     Level of Education.   Satisfactory completion of successive years  of
education provides a person with greater  knowledge and skills.  The knowledge
acquired,  of whatever quality, can only expand a person's total world of
actuality and reality.  Each  increment of  schooling adds to his competence
through the  acquisition of knowledge and Incidental skills.   Further,  advance-
ment into college or university training Is taken  to reflect the potential  of
better performance,  compared to others who are screened from higher
education.   The reading skills learned have Important consequences  In later
life for Increased exposure to the mass media, particularly the printed word.
This fact has been well documented by many communication studies (e._g. ,

-------
                                                                         100
Samuelson, £t aL [38]),    In turn,  it is not at all surprising to find, as
Schramm [40] has reported and as Tichenor [50 ]  has confirmed,  that an
individual's level of education is the chief predictor of his level of exposure
to science information in the  mass media and his amount of science knowledge.

     Level of Science Education.  Level of formal education may not be the
whole story.  It is conceivable that  high school or college graduates  -- with
no formal science education -- may not acquire the vocabulary and content
necessary to evaluate and comprehend science  to the same extent as those
who have completed some formal science courses.   In his secondary
analysis, Tichenor [50 ]  reported that the amount of science education in
high school or college also may be an important predictor of science know-
ledge.   These attributes were strikingly low for high school and college-
educated individuals with no formal science course work,

     Level of the Use of the Environment.    People differ in the percent of
time they spend outdoors and individuals who are outdoors much of their
daylight hours may have a greater opportunity to differentiate and evaluate
the air  quality of where they live, compared to those who are outdoors little,
if any.   This may be particularly true in the realm of outdoor recreation use.
Air quality plays a role in the preference people have for outdoor recreation,
according to a report of the Outdoor Recreation Resources Review Commission
[47] .   Deterioration in air quality was one of  the reasons cited for recreation-
user dissatisfaction.   Thus, it appears  that the quality of the environment is
an important aspect for sustained outdoor recreation use, and the level of
outdoor  use may be a predictor of knowledge toward air quality.

-------
       APPENDIX B




Interview Schedules A and B

-------
    APPENDIX B.   Interview Schedule A
                            101
Budget Bureau Number  85-S70010
Approval Expires March 31.  1971
Schedule A
                                 OREGON  STATE UNIVERSITY
                           Final
Hello, I'm working on a  survey for Oregon State University and I would like to ask you a
lea interesting questions, If* you  don't mind.
1 -

2 _



3 -

4 -





4
3
2
1


3
2
1

Years
X DK, NA
Own
Rent
Other
DK, NA
Years

Yes
No
DK, NA

First, how many years have you, yourself, lived in the
Sale^n (Eugene) Metropolitan Area?
May I ask whether you own or rent the place in which you are
now living?


Approximately how long have you lived in this particular
dwelling unit?
Some people plan to move within six months to a year, while
others plan to stay put in the same home or apartment. How
about you — do you have any plans to move to another place
within, say, the next six months?
NOTE TO INTERVIEWER:  If respondent  answers  "Yes"  to  Q 4 or_ Rents  and has  lived in current
                     DU less  than one  year,  terminate the  interview.

       (ASK OF EVERYONE)
5 -


6 -


7 -




8 -




2
1

3
2
1
4
3
2
1
0
4
3
2
1
0
Yes - Asked
No or DK

Convince
Active part
Listen or DK
A lot
Quite a bit
Little
None
DK, NA
A lot
Quite a bit
A little
None
DK, NA
Looking back over the past few months, can you remember any
time someone outside your immediate family has asked your
opinion about state or community affairs?
When a discussion is on about community affairs, do you
usually listen, take an active part, or try to convince
others especially?
Some people earn all or part of their living by working out-
doors, while others' jobs keep them inside. How about your
family? About how much time, if any, did the chief bread-
winner spend working outdoors for pay last year — a lot,
quite a bit, a little, or none at all?
Some people spend quite a bit of time out-of-doors while
others have other pastimes. Not counting work for pay, how
much time, if any, did you, yourself, spend out-of-doors last
year — none at all, a little, quite a bit, or a lot?


-------
                                                                                    102
     Now,  here are some specific out-of-door activities.   After I read each one, will you
     please tell me how much time,  if any,  you spent participating in that activity during
     the past twelve months — a lot, quite a bit,  a little,  or none at all?
9 -
10 -
11 -
12 -
13 -
14 -
15 -
16 -
17 -
18 -
19 -
20 -
21 -
22 -
23 -
24 -
25 -
Lot
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
Quite
bit
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Little
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
None
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
DK, NA
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Activity
Bicycling

Attending outdoor cultural events
Gardening
Picnicking
Camping
Pleasure driving, sightseeing
Boating or water skiing
Horseback riding
Hunting







Attending outdoor sporting events
Playing golf
Ocean beach activities
Walking for pleasure, hiking,
study, rockhounding, etc.
Swimming
Fishing


nature


Playing outdoor games and sports
Other
(Specify)


26 -  Suppose one of your friends came to you and asked you how he could find out what is
     known about air pollution.   Where would you advise him to go for that information?
     (PROBE for both someone to  talk to and something to read)
     Anything  else?

-------
                                                                                     103
27 -  Do you feel you personally know more or less about air pollution  than most people?
     (PROBE!)
     Anything else?
28 -  Where do you feel that you, yourself, have gotten the most valuable information
     about air pollution?  (PROBE!)
     Anything else?
29 -
30 -
31 -
32 -
33 -
1
2
3
4
5
1
2
3
4
5








Oregon Statesman (R-G)
Capital Journal
Other
None
DK, NA
Press (Reminder)
Other
Other
None
DK, NA
Number

Number





What daily newspaper, or newspapers, if
into your home yesterday?
What weekly newspaper, or newspapers, if
into your home last week?
Not counting car radios , how many radios
do you have in your home?
any , came
any , came
, if any,
And, how many TV sets, if any, do you have in your
home?
Which magazines or publications, if any,
your home this past month? Any others?
came into
       0   None

-------
                                                                                     104
34 -  Listed on this card are several sources that many people use to obtain information.
     From which one of these sources do you believe you get most of your information
     about air pollution?  (HAND RESPONDENT CARD A)  Which is second most in importance?
     Which is j-east in importance?

          Most                            Next                              Least
1 Newspaper
2 Magazine
3 Radio
4 Television
5 Other
6 DK, NA
1
2
3
4
5
6
Newspaper
Magazine
Radio
Television
Other
DK, NA
1
2
3
4
5
6
Newspaper
Magazine
Radio
Television
Other
DK, NA
    I have a list of statements here concerning different sources of information about
    air pollution.  As I read each one, will you please tell me what you know or how
    you feel about them?

           Quite,  Little,   Nothing,
           )uite   Not too   Never      DK
35 -   4      3        2         1        0   How much have you talked to someone at
                                             Willamette University  (Univ. of Oregon) or
                                             Oregon State about air pollution — not at
                                             all, a little, quite a bit, or a lot?

36 -   4      3        2         1        0   How much have you viewed on TV about air
                                             pollution — a lot, quite a bit, a little,
                                             or nothing?

37 -   4      3        2         1        0   How often have you talked to someone at
                                             the state or local Air Pollution Control
                                             Authority about air pollution — never,
                                             not too often, quite often, or very often?

38 -   4      3        2         1        0   How much have you read in the newspaper
                                             about air pollution — nothing, a little,
                                             quite a bit, or a lot?

39 -   4      3        2         1        0   How often have you discussed air pollution
                                             with your neighbors — very often, quite
                                             often, not too often, or never?

40 -   4      3        2         1        0   How often have you listened to something
                                             about air pollution over the radio —
                                             never, a little, quite a bit, or a lot?

41 -   4      3        2         1        0   How much have you talked to your friends
                                             about air pollution — a lot, quite a bit,
                                             a little, or never?

42 -   4      3        2         1        0   How much have you read in a magazine about
                                             air pollution — nothing, a little, quite
                                             a bit, or a lot?

-------
        (ASK OF  EVERYONE)

43 -    2  Yes (Where?	
       3  Yes (Where?     .	
       4  Yes (Where? 	
       1  No (Skip to Q 46)
       0  DK, NA  (Skip to Q 46)
                                                                                     105
Within the past five years, have you  lived  in
any state or states other than Oregon where
you felt there r-ras an air oollution problem
while you were living there?
      (IF YES IN Q 43, ASK FOR EACH AREA)

44 -  How serious did you consider their air  pollution problem to be — not too serious,
      somewhat serious, or very  serious?

     Area(s)
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
45 -  What type or types  of  air  pollution was  there  in this  (these)  area(s)  where you
     used to live — just describe it  in your own words?   (PROBE!)

     (INT:  Write in area(s) with response opposite)

     Area;                                     Response
     Area:
  Response
     Area:
  Response

-------
                                                                                     106
      (ASK OF EVERYONE)
46 -
2 Yes (Where?
3 Yes (Where?
4 Yes (Where?
1 No (Skip to Q 49)
0 DK, NA (Skip to Q 49)
) Within the past five years, have you visited
) any state or states other than Oregon where
) you felt there was an air pollutiori problem
while you were visitine there?
      (IF YES IN Q 46, ASK FOR EACH AREA)

47 -   How serious did you consider their air pollution problem to be — very serious,
      somewhat serious, or not too serious?

      Area(s)
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
48 -   What type or types of air pollution was there in this  (these) area(s) where you
      visited — just describe it in your own words?  (PROBE!)

      (INT:  Write in area(s) with response opposite)

      Area;                                    Response
      Area:
Response
      Area:
Response

-------
                                                                                     107
      (ASK OF EVERYONE)

49 -   2  Yes (Where? 	
      3  Yes (Where? 	
      4  Yes (Where? 	
      1  No (Skip to Q 52)
      0  DK, NA (Skip to Q 52)
 )     Within the past five years, have you lived in
 )     any community or communities in Oregon other
~)     than the one you live in now where you felt
      there was an air pollution problem while you
      were living there?
      (IF YES IN Q 49, ASK FOR EACH AREA)

50 -   How serious did you consider their air pollution problem to be — very  serious,
      somewhat serious, or not too serious?

      Area(s)
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
51 -   What type or types of air pollution was there in  this  (these) area(s) where you
      lived — just describe it in your own words?  (PROBE!)

      (INT:  Write in area(s) with response opposite)

      Area:                                    Response
      Area:
          Response
      Area:
          Response

-------
                                                                                     108
        (ASK OF EVERYONE)

52 -     2  Yes  (Where? 	
        3  Yes  (Where? 	
        4  Yes  (Where? 	
        1  No  (Skip  to Q  55)
        0  DK, NA  (Skip to  Q  55)
Within the past five years, have you visited
any part or parts of Oregon where you felt
there was an air pollution problem while you
were visiting there?
        (IF YES  IN Q  52, ASK FOR EACH AREA)

53 -     How serious did you  consider their  air  pollution  problem to  be  —  not  too  serious,
        somewhat serious,  or very serious?

        Area(s)
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
4
3
3
1
Very serious
Somewhat
Not too
DK, NA
4
3
2
1
Very serious
Somewhat
Not too
DK, NA
54 -   What  type  or  types  of  air  pollution was  there  in this  (these)  area(s) where you
       visited  — just  describe it  in your own  words?   (PROBE!)

        (INT:  Write  in  area(s) with response opposite)

       Area:                                     Response
        Area:
     Response
       Area:
     Response

-------
                                                                                     109
       (ASK OF EVERYONE)

55 -    2  Yes
       1  No (Skip to Q 58)
       0  DK, NA (Skip to Q 58)
                              Now, thinking of the area within a mile  of
                              where you now live in Oregon, do you  feel
                              that there is an air pollution problem in
                              this area?
       (IF YES IN Q 55)
56 -    4  Very serious
       3  Somewhat
       2  Not too
       1  DK, NA
                              How serious do you consider this air pollu-
                              tion problem within a mile of where you now
                              live — very serious, somewhat serious, or
                              not too serious?
57 -  What type, or types, of air pollution problems do you have  in this  area where you
     now live — just describe them in your own words?   (PROBE!)
       (ASK OF EVERYONE)

       Here are some statements that have been made recently about air pollution.  As  I
       read each one, would you please tell me how strongly you agree or  disagree with
       it?  (HAND RESPONDENT LIST OF Q's 58-77)
58 -
59 -
60 -
61 _
         60
       cu a
       0) O
cu
cu
J-l
bO
                     g
                   o P<
                   53 O
0)
0)
t-l
60
rt
CO
co s*,
CU rH
)-i M
00 C
Cfl O
CO M
•H 4J
n w
                                              Statement


                              We don't have  to worry about  air pollution
                              here.  The problem will  solve itself.

                              People must be ready  to  pay increased
                              taxes, utility rates, and higher prices
                              on consumer goods if  we  are to have
                              clean air again.

                              As a good place to live, the  Willamette
                              Valley is deteriorating  from  air pollution
                              year by year.

                              More public support is needed for air
                              pollution control programs.

-------
                                                                                     110

                           0)   >
                     C     a)