CEE Energy Star® HOUSEHOLD
SURVEY REPORT (2000)
Part I - Results
Prime Contractor
The Cadmus Group, Inc.
Subcontractor
XENERGY Consulting, Inc.
for
Energy Star®
Climate Protection Partnerships Division
U. S. Environmental Protection Agency
Washington, D.C.
Contract No. 68-W6-0050; Work Assignment No. 0010AA-29
Deliverable: February 9, 2001

-------
TABLE OF CONTENTS
ACKNOWLEDGEMENTS
EXECUTIVE SUMMARY	ES-1
SECTION 1: INTRODUCTION	1-1
1.1	The Energy Star® Program	1-1
1.2	Purpose of this Study	1-1
1.3	Survey Sponsorship	1-1
1.4	Timing	1-2
1.5	Report Organization	1-2
SECTION 2: METHODS	2-1
2.1	General Approach	2-1
2.2	Questionnaire Design	2-1
2.3	Sampling	2-1
2.3.1	National Sample	2-1
2.3.2	Member Samples	2-3
2.3.3	Combined Sample	2-4
2.4	Analysis	3-5
2.5	Response Rates	3-5
2.6	Precision of Estimates	3-6
SECTION 3: FINDINGS	3-1
3.1	Demographics	3-1
3.2	Understanding of the Energy Star Label	3-3
3.2.1	Coding the Levels of Understanding	3-3
3.2.2	Overall Findings	3-5
3.2.3	Understanding by Subgroup	3-5
3.2.4	Comparison to Other Studies	3-7
3.3	Awareness of the Energy Star Label	3-9
3.3.1	Where People Learned about Energy Star	3-11
3.3.2	Comparison to Other Studies	3-13
3.4	Influence of the Energy Star Label	3-14
3.4.1	Effect of Rebates	3-16
3.4.2	Findings by Subgroup	3-18
3.4.3	Comparison to Other Studies	3-19
SECTION 4: CONCLUSIONS	4-1

-------
TABLE OF CONTENTS
APPENDIX A: SURVEY QUESTIONNAIRE
APPENDIX B: FIFTY-SEVEN LARGEST DMAS
APPENDIX C: SAMPLE WEIGHTING AND VARIANCE ESTIMATION
METHODOLOGY

-------
Acknowledgements
The authors of this study, Miriam Goldberg (XENERGY Consulting, Inc), Mitchell
Rosenberg (XENERGY Consulting, Inc), and Timothy Pettit (The Cadmus Group, Inc)
would like to thank the following organizations and individuals for their contributions:
•	Marc Hoffman, Director for the Consortium for Energy Efficiency, for making the
study data available to EPA for the national analysis.
•	Maureen McNamara of the Environmental Protection Agency ENERGY STAR
Program for project management and oversight of this report.
•	Bonneville Power Administration, the Energy Center of Wisconsin, Wisconsin
Department of Administration, National Grid USA, Northeast Utilities, the
Northwest Energy Efficiency Alliance, Pacific Gas & Electric Company, the
Sacramento Municipal Utility District and United Illuminating for co-sponsoring the
national survey.
•	NSTAR, National Grid USA, Northeast Utilities, Pacific Gas & Electric Company,
Southern California Edison, United Illuminating, Unitil, the Vermont Department of
Public Services, Western Massachusetts Electric and the Wisconsin Department of
Administration for making their local samples available for the analysis.
We also acknowledge the following people for their helpful contributions: Yen Chin,
Seattle City Light; Phil Degens, Northwest Energy Efficiency Alliance; Shel Feldman,
for Wisconsin Department of Administration; Jane Gordon, Northwest Energy
Efficiency Alliance; Ed Hamzawi, Sacramento Municipal Utility District; Marc
Hoffman, Consortium for Energy Efficiency; Ken Keating, Bonneville Power
Administration; Valerie Richardson, Pacific Gas & Electric Company; Shahana
Samiullah, Southern California Edison; Bobbi Tannenbaum, Energy Center of
Wisconsin; and Elizabeth Titus, National Grid.
Finally, thanks also go to Valy Goepfrich (XENERGY Consulting, Inc), data analyst,
and Jane Otto (The Cadmus Group, Inc), technical editor.
1

-------
Executive Summary
In the summer/fall of 2000, The Consortium for Energy Efficiency (CEE) sponsored a national
household mail survey to ascertain consumer awareness and understanding of the Energy Star
label as well as its influence on energy-related purchasing decisions.
CEE is a national nonprofit energy efficiency organization whose membership includes regional,
state, and utility administrators of publicly funded energy efficiency programs. As of the summer
of 2000, eighty-six such entities, representing approximately 40 percent of U.S. households, had
partnered with the national Energy Star program to locally promote (predominantly) Energy
Star qualifying lighting and appliances. The majority of these partners are located in states that
have enacted restructuring or alternative legislation explicitly funding energy efficiency
programming.
In order to facilitate comparison in areas with strong local promotions and areas with little or no
local Energy STAR-related promotions, the sample frame was drawn from the largest media
markets, jointly accounting for about 70 percent of US households. Each media market was
assigned one of three publicity levels and then a sample of media markets was randomly selected
from each publicity category. In addition, some CEE members also fielded the survey in their
local territories to facilitate direct comparison between the effects of the program in their areas
and in the country at large. CEE and participating members made the survey data available to
the EPA for the national analysis developed in this report.
The study is organized into two parts. Part I includes background information, methods,
findings, conclusions, and several appendices. Part II includes detailed cross-tabulations of the
survey response data for each question in the questionnaire.
Summary of key Findings
The key findings are as follows:
•	Nationwide, 41 percent of households have seen the Energy Star label.
•	Nationwide, over one-half of all households, including those that had seen the label
previously and those that were reacting to it for the first time, had at least a general
understanding of the label's message, and 37 percent registered a high degree of
understanding.
•	Nationwide, about 50 percent of those who reported that they bought an Energy Star
labeled product also reported that they were influenced by the label to buy that product.
ES-1

-------
EXECUTIVE SUMMARY
•	Although the overall response rate for the survey was low (10.2 percent), comparative
analyses to U.S. census data and other related studies indicate that the survey data in the
present study represent a reasonable characterization of the current state of Energy Star
awareness, understanding, and influence. Patterns of response in the present study are
similar to other mail surveys, in terms of key comparison statistics such as age and income,
and those same key respondent demographic statistics are reasonably similar to the U.S.
population. Therefore, non-response effects related to demographics are not likely to have
substantially biased estimates. However, non-observable customer attributes, such as concern
about the environment, cannot be fully addressed and may have resulted in the over- or
under-representation of such customers among the survey respondents.
•	Comparisons to four other relevant surveys show reasonably consistent statistics in terms of
mail survey response rate, demographic profile, Energy Star awareness level,
understanding of the Energy Star label, and purchasing intent based on—or influence of—
the Energy Star label. These studies are:
1.	The 1997 Residential Energy Consumption Survey (RECS) conducted by the Energy
Information Administration.
2.	The 1998 Home Appliance Buying Trends Survey conducted by D&R International, Ltd.
for the Department of Energy.
3.	The Pre-Post Energy Star Awareness Tracking Study conducted by XENERGY Inc.
for Pacific Gas and Electric Company (PG&E) in 2000.
4.	Surveys conducted in Wisconsin (WI) and the Pacific Northwest (PNW) concurrent with
the CEE survey.
•	The 1997 RECS shows similar results with respect to the influence of the label.
Comparisons with RECS are useful, given the 80 percent response rate.
In addition, there was a statistically valid difference in Energy Star awareness, understanding,
and influence between areas that had active supplemental promotions for two or more years by a
local or regional energy efficiency program administrator. For those demographics we were able
to observe, non-response effects, if any, would be similar in low and high publicity areas, so that
the observed difference is likely to be reflected in the population at large.
•	Label awareness is much higher in the high-publicity areas than in the low-publicity areas —
52 percent versus 37 percent. The difference between high and low publicity areas was
statistically significant at less than a 1 percent significance (better than 99 percent
confidence) level.
ES-2

-------
EXECUTIVE SUMMARY
•	In the high-publicity areas, a greater proportion had a high understanding compared to the
low-publicity areas (41 percent versus 34 percent) and a lower proportion had no
understanding. The difference between high and low publicity areas was statistically
significant at the 1 percent significance (99 percent confidence) level.
•	The proportion of consumers influenced by the label in purchasing qualifying products was
significantly higher in the high-publicity areas compared with the low-publicity areas, 58
percent versus 35 percent. Thus, where publicity is low, the effect of the label on purchasing
decisions is significantly reduced. The difference between high and low publicity areas was
statistically significant at the 2 percent significance (98 percent confidence) level.
•	The study also found some notable differences in label association for products actively
promoted by the Energy Star partner programs. These results suggest that utilities, market
transformation groups, state administrators, and other organizations involved in
administering publicly funded energy-efficiency programs that integrate Energy Star
messaging are effective allies in spreading the word about Energy Star, in general, and in
promoting these product categories.
Conclusions
The findings in this study confirm that substantial portions of U.S. consumers are aware of and
understand the Energy Star label. Moreover, the label influences purchase decisions, and
publicity efforts improve awareness, understanding, and influence of the label.
This questionnaire provides useful data for comparing publicity levels, and information on the
relative importance of different purchasing factors. The results are probably most useful as a
baseline for future tracking efforts. Future survey efforts that use the same questionnaire will
provide a rich base of time-series data, and an important and useful information source for
utilities, market transformation groups, state administrators, and other organizations, including
EPA and DOE.
The results indicate that the EPA and DOE strategy of partnering with third-party organizations
to build momentum for Energy Star is an effective strategy for building awareness,
understanding, and influence for the Energy Star label.
ES-3

-------
Introduction
1.1 The Energy Star® Program
Energy Star® is a dynamic government/industry partnership program that makes it easy for
businesses and consumers to save money and protect the environment. Currently, the label
appears on over 30 product categories, as well as homes, office buildings, and schools. To
qualify, products and buildings must meet Energy Star criteria, which vary by category, but
generally mean that the product or building ranks in the top 20-30% for energy performance.
Regional energy-efficiency program administrators have become increasingly valuable partners
in educating their constituents about the financial and environmental benefits of Energy Star,
particularly with residential lighting products and appliances. As of the summer of 2000, eighty-
six such entities, representing approximately 40 percent of U.S. households, had partnered with
the national Energy Star program to locally promote (predominantly) Energy Star qualifying
lighting and appliances. The majority of these partners are located in states that have enacted
restructuring or alternative legislation explicitly funding energy efficiency programming.
1.2 Purpose of The Survey
The CEE Energy Star Household Survey of 2000 was designed to obtain information at a
national level on consumer awareness and understanding of the Energy Star label and of its
influence on their energy-related purchase decisions. Research questions of interest included:
•	The media and products on which the Energy Star label was seen;
•	the effect of higher publicity levels on Energy Star label awareness, understanding, and
influence; and,
•	the relationship of household demographics and purchases to label awareness,
understanding, and influence.
1.3 Survey Sponsorship
The Consortium for Energy Efficiency (CEE) and participating members sponsored the survey
instrument and data collection. CEE and participating members made the survey data available to
the EPA for the analysis developed in this report. In addition to contributing to the national
sample, some CEE members also fielded the survey in their local territories to facilitate direct
comparison between the effects of the program in their areas and the country at large.
1-1

-------
SECTION 1
INTRODUCTION
1.4	Timing
Survey instrument design began in the spring of 2000. Data were collected during the summer
and fall.
1.5	Report Organization
This report is organized in two parts. Part I has four sections, in addition to three appendices, and
an Executive Summary. The Introduction (Section 1) is proceeded by Methods (Section 2)—
including questionnaire design, sampling, and survey implementation. Findings from the survey
data analysis are presented in Section 3. Conclusions from the study are given in Section 4. A
copy of the survey instrument is provided in Appendix A. Appendix B contains a list of the 57
DMAs used in the national sample and Appendix C contains details about the weighting and
variance procedures used in the data analysis.
Part II provides detailed cross-tabulations of survey responses by level of understanding of the
Energy Star label, by awareness of the label prior to the survey, and by demographics. These
data tables provide the percent of respondents in each category or the average of numeric
responses. In all cases, the results presented are weighted to the national level.
1-2

-------
Methods
2.1 General Approach
The mail questionnaire and survey sample design were developed for a national sample of
households. In addition, CEE members were invited to field the survey locally to facilitate direct
comparison between the effects of the program in their areas with those in the country at large.
In the analysis, the national sample and CEE member samples were pooled to provide a richer
database for examining the national program. The questionnaire design (Section 2.2) and the
sampling and analysis approach (Section 2.3) are described below.
2.2 Questionnaire Design
The survey was designed to provide information on respondents' awareness of the Energy Star
label, their purchases of Energy Star products, and the influence of the label on those purchase
decisions. The questionnaire also collected data on demographics and sources of information
used when considering energy-related purchases for the home.
An important question in assessing awareness and influence of the label was how the message of
the label was understood, both by those who had been aware of it previously and those who had
not. To gauge understanding, without the filter of previous questions, the questionnaire began
with open-ended questions asking the respondent to look at the label and write down what
message(s) they thought it gave. The remaining questions were closed-ended, requiring simple
yes/no answers, multiple-choice responses, or a selection of all applicable items from a list.
A copy of the survey instrument is included in Appendix A.
2.3 Sampling
2.3.1 National Sample
The national sample was a two-stage sample. In the first stage, Nielsen Designated Marketing
Areas (DMAs) were randomly selected. Nielsen developed DMAs for planning and analyzing
the results of publicity and advertising campaigns. A DMA consists of all counties in which the
largest television viewing share is assigned to stations in that same area. Nielsen uses U.S.
Census statistics for household data within each DMA. These non-overlapping DMAs cover the
entire continental United States, Hawaii and parts of Alaska. There are currently 210 DMAs
throughout the U.S. Only the largest 57 DMAs in the country, accounting for approximately 70
2-1

-------
SECTION 2
METHODS
percent of U.S. households, were included in the first-stage sampling frame for this survey. In
the second stage, a random selection was made from household data for each selected DMA.
The national sample was designed to facilitate comparisons between areas with high levels of
publicity and areas with low levels of publicity. For this purpose, the largest 57 DMAs were
classified into one of three publicity levels in the first stage as follows:
•	High message saturation: Areas in which utilities or other third party organizations
(e.g., utility, state, or regional energy-efficiency systems benefit charge administrator)
based a publicity and/or rebate program on the Energy Star label. This third party
publicity had to include at least 2 of the following: bill inserts, paid ads, retailer
promotion/programs, or rebates resulting in over 500 Gross Rating Points1 (GRPs) for
more than 2 years.
•	Low message saturation: Areas that received only the national-level Energy Star
promotions from EPA and/or DOE.
•	Other: Areas in which national-level efforts were supplemented by additional EPA/DOE
targeted market outreach (PSAs and media outreach) that achieved at least 500 GRPs.
Within each of these publicity strata, a simple random sample of four DMAs was initially
selected. A list of the 57 DMAs by publicity level is provided in Appendix B.
The initial mailing of the survey began in June 2000 to 7,500 households. This mailing was
followed by a reminder postcard two weeks later, then by a subsequent mailing of the survey to
the same sample one week after the postcards were mailed.
Due to low response rates, the CEE committee considered its options for increasing sample size.
A decision was made to combine member sample data with the national data set using
appropriate statistical weighting methods. This combined sample approach is described below in
Section 2.3.2.
The CEE committee also made decision to conduct another survey mailing effort. For this
survey effort, a fresh sample of DMAs was selected only from the other- and low-publicity
strata, excluding areas already covered by the national and member samples. Additional DMAs
were not selected in the high-publicity stratum because all DMAs in that stratum were already
included in either the member or national samples with the exception of the City of Los Angeles.
Additional sample was not selected in the member areas because these already had relatively
large sample sizes compared to the national frame in the combined sample.
1 A gross rating point is an industry standard unit of measurement of advertising audience size, and is equal to one
percent of the total potential audience universe. It is used to measure the exposure of one or more programs or
commercials without regard to multiple exposure of the same advertising to individuals.
2-2

-------
SECTION 2
METHODS
Selecting a fresh sample was considered superior to an additional follow-up effort with the
original sample for two reasons. First, response rates would likely remain low even with some
increase in returns from the additional follow-up. A mailing to a new sample of respondents was
expected to yield a greater increase in sample size than another mailing to the original sample.
Second, drawing a fresh sample of households from additional DMAs would provide data from a
wider set of areas of the national sample and reduce inefficiencies that may have resulted from
clustering. The second survey effort did not include a follow up postcard or survey mailing. As
a result, response rates to the second effort were slightly lower than those for the first sample.
2.3.2	Member Samples
Samples sponsored by CEE members in their own regions were fielded for the following areas:
•	California, excluding the City of Los Angeles
•	Connecticut, excluding Hartford
•	Massachusetts
•	Vermont
•	Wisconsin
In each of these five areas, a simple random sample was drawn from the full set of households.
Unlike the national sample, these surveys included a cover letter. They were mailed
simultaneously with the first national sample.
2.3.3	Combined Sample
Ideally, the CEE member frame would have been identified in advance and the national frame
restricted to areas outside the member frame. However, some of the member surveys were not
planned until after the national effort was underway, and the combination of the national and
member samples was not agreed to until after the first wave of data collection was complete.
Thus, the adjustment for overlap between the two frames was made after the fact.
The combined analysis required the merger of the two data sets: the national and the CEE
member samples. In both the national and CEE member samples, a simple random sample was
selected within a geographic area. Geographic areas were selected by one of two means:
•	In the national sample, DMAs were selected within each publicity stratum.
•	In the member sample, each area was the member's own geographic area.
Together, the national and member samples represent all areas that are either in large DMAs or
are part of one of the five member samples. Some DMAs are in both sample frames. This
overlap is handled by dividing the total represented area into two distinct pieces. The first is the
member frame in its entirety. The second is that portion of the national frame not included in the
2-3

-------
SECTION 2
METHODS
member frame—the national frame minus the member frame. We refer to this second portion as
the "restricted national frame." The Venn Diagram below illustrates the combined sample
classification approach.
The restricted national frame excludes any zip code area that was part of the CEE member frame.
That is, instead of an entire DMA, we consider only the "restricted DMA," excluding any zip
codes that were designated for the member sample. If an entire DMA selected for the national
sample was also part of the CEE member frame, the DMA would be considered only as a part of
the member sample and would be excluded from the restricted national sample. The frames and
samples are summarized in Table 2-1 below.
Exhibit 2-1
Venn Diagram of Combined Sample
Combined Sample
National Sample [ Overlap | Partner Samples
I I = Restricted National Frame
2-4

-------
SECTION 2
Table 2-1
Distribution of Frames and Samples
METHODS

National
Alone
Member
Alone
Combined
# of DMAs
57
27
74*
# households
(millions)
69.9
13.5
72.8*
% U.S. households
69.2%
13.4%
72%*
% of combined
frame
96%
18.5%
100%
Sample
# households
780
2,176
3,496
% of sample
22.3%
77.7%
100%
Note: National Alone and Member Alone do not add due to overlap
2.4	Analysis
The primary analysis consists of calculating means and proportions for various subgroups of
interest and determining the standard errors of these estimates. Means and proportions are
calculated using the expansion weights, which reflect the sampling rates for each portion of the
sample. Standard errors reflect the two-stage structure for the national sample and the
stratification of the CEE members' sample into member areas. The standard errors were
calculated using special-purpose software designed for the analysis of complex survey data
called SUDAAN®. The weighting and standard error calculations are described in Appendix C.
2.5	Response Rates
In the combined sample, the final response rate was 10.2 percent. After a first wave of mailings
and follow-up, response rates of approximately 6 percent were obtained for the national sample
and approximately 13.3 percent for the member samples. The response rate in the second
mailing was 4.5 percent.
Survey response rates for the initial and additional samples are summarized in Table 2-2.
2-5

-------
SECTION 2	METHODS
Table 2-2
Survey Response Rates

National Alone
Member
Alone
Combined
Initial
Additional
Sent
7,500
6,400
20,350
34,250
Returned
491
289
2,716
3,496
Response Rate
6.5%
4.5%
13.3%
10.2%
Possible reasons for the low response rates may include:
•	Declining nationwide response rate to surveys in general;
•	A survey sponsor (CEE) unfamiliar to most respondents; and,
•	The open-ended questions in the beginning require some thought, which could discourage
interest in, and the timely completion of, the survey.
•	The purchased lists of households used as the sample frame in each DMA or member-
sponsored area may have included a large proportion of ineligible or invalid listings.
Response rates are calculated as the ratio of surveys returned to survey addresses mailed,
not as the ratio of returns to valid or eligible addresses in the sample.
•	Poll fatigue due to census and presidential election years.
Potential non-response bias issues, typically associated with similar response rates, are discussed
in Section 3.
2.6 Precision of Estimates
As discussed above, standard errors were calculated for all estimates of population variables
developed from the sample survey data. Throughout the text we use these standard errors to
calculate confidence intervals for population means and proportions. The standard errors are
also the basis for calculations of the statistical significance of differences between sample
subgroups, such as customers subject to the different levels of Energy Star publicity. The
statistical significance indicates the likelihood that the observed differences in the sample reflect
actual differences in the population. The statistics characterizing the precision of the estimates
used are as follows.
• Confidence intervals around sample means and proportions. In reporting sample
means and proportions, we generally provide the 90 percent confidence interval,
indicated as symmetric error bounds around the estimate. For example, 41 percent of
households (+/- 2 percent) reported having seen the Energy Star label prior to
answering the survey. This statement means that we are 90 percent confident that, if all
households in the sampling frame had been surveyed, and assuming no systematic biases
2-6

-------
SECTION 2
METHODS
in the sample, the proportion of the full population that would have reported seeing the
label prior to the survey would fall between 39 percent and 43 percent. By "90 percent
confident," we mean that for 90 percent of the random sample points that might have
been drawn, the calculated interval would include the true value. The interval can be
more loosely interpreted to mean that there is a 90 percent chance that the true population
proportion is between 39 and 43 percent.
Many of the most significant research questions addressed by this study concern differences
in the level of Energy Star recognition, understanding, and use among subgroups of the
sample defined by publicity level, household attributes, and prior experience with the label.
For example, 41 percent of customers in the high-publicity areas recorded a high level of
understanding of Energy Star label, and 34 percent in the low publicity areas, for a
difference of 7 percent. The precision of estimating differences between subgroups can be
characterized two different ways, in terms of confidence intervals, and in terms of statistical
significance tests.
•	Confidence intervals for estimated differences. Standard errors calculated from the
sample data can be used to examine confidence intervals around the estimate of the
difference. In the example just given, the 90 percent confidence interval around the 7
percent difference is 5 percent. This means that, with 90 percent confidence, the
population difference between publicity areas in the percentage of households who have a
high level of Energy Star understanding is somewhere between 2 and 12 percent.
•	Statistical significance (p-value) of differences between subgroup sample means and
proportions. A second way to characterize the precision of estimates of differences
between subgroups is to calculate the statistical significance or p-value of the difference.
The p-value is the probability that the random sample would have produced a difference
as extreme as one observed in the population if the subgroups are not different. That is, a
statistical test is conducted of the null hypothesis that the difference in population means
is zero, against the alternative hypothesis that the difference is not zero. Continuing the
example above, the p-value for the 7 percent difference is 0.02. This means that if there
were no difference between high- and low-publicity groups in the population proportions
who had previously seen the label, there would be only a 2 percent probability that the
sample percentages would be as different as they are. That is, it is very unlikely that the
sample would have shown such a large difference if there is not in fact a difference in the
population. The smaller the p-value, the more unlikely it is that the sample differences
could have been so great just by chance, and the stronger the evidence is that there truly
is a difference in the population. As a final note, a difference is considered statistically
significant at significance level a if the p-value is less than or equal to a. A common
standard in survey research is to specify a significance level of a = 0.05. With this
standard, any difference with a p-value less than or equal to 0.05 would be called
statistically significant.
2-7

-------

3
Findings



3.1 Demographics
As described in Section 2, the national sample was designed to be representative of all
households in the United States in the 57 largest DMAs, while the CEE member samples were
designed to be representative of each member's area. The procedures used for combining the
two samples provided unbiased estimates of the total combined sampling frame—that is, all large
DMAs in the country together with any portions of the partner areas that were in smaller DMAs.
The weighting procedures used to combine the samples implicitly assume that the responding
sample is a random subset of the random sample selected to receive the survey. However, in any
survey effort, individuals who choose to respond to the survey may tend to be different from
those who choose not to respond. The effect of these systematic differences is termed "non-
response bias." The potential for non-response bias increases as response rates fall. As
described in Section 2, response rates for this survey were generally low, making the potential
for non-response bias to be a concern.
To assess this potential, the distribution of key demographic characteristics of the responding
sample, weighted in the same manner as for the rest of the analysis, were compared with national
census data. Age and income distributions are shown in Tables 3-1 and 3-2, respectively.
Table 3-1
Age Distribution for Weighted Sample and National Census Data
(Source: Current Population Statistics for 1998, US Census Bureau)
Percent of Households

Householder/
Respondent Age
Population
Weighted Combined
Sample
% Smpl-% Pop
Pop, 15-24; Smpl, 18-24
5.6%
0.7%
-4.9%
25-34
18.1%
10.0%
-8.2%
35-44
23.1%
18.2%
-4.9%
45-54
19.4%
25.5%
6.1%
55-64
13.1%
16.6%
3.5%
65-74
10.9%
16.4%
5.4%
75 or older
9.8%
12.7%
2.9%
Total (%)
100.0%
100.0%

Total (1,000s)
103,875
63,883

3-1

-------
SECTION 3
FINDINGS
Table 3-2
Income Distribution for Weighted Sample and National Census Data
(Source: Current Population Statistics for 1998, US Census Bureau)
Total Household Annual
Income (before taxes)
Percent of Households

Population
Weighted Combined
Sample
% Smpl-% Pop
Less than $5,000
3.2%
1.4%
-1.8%
$5,000-$9,999
7.1%
3.0%
-4.1%
$10,000—$14,999
7.8%
3.8%
-4.0%
$15,000-$24,999
14.0%
13.8%
-0.2%
$25,000-$49,999
29.2%
28.9%
-0.3%
$50,000-$74,999
18.6%
24.3%
5.7%
$75,000 and over
20.1%
24.8%
4.7%
Total (%)
100.0%
100.0%

Total (1,000s)
103,875
56,861

Although the overall response rate for the survey was low (10.2 percent) raising concerns about
non-response bias, a comparative analysis of respondent demographics and other related studies
indicates that the findings provide a reasonable characterization of the current state of Energy
Star awareness, understanding, and influence. For those demographics we were able to
observe, the demographics for the (weighted) responding sample compared with national Census
data show reasonably similar demographic profiles. Moreover, patterns of non-response in the
present study are similar to residential household mail surveys in general. Younger and lower
income households are less likely to respond to the survey than are older and higher income
households. These broad characteristics should be considered when interpreting the survey
results. Additional demographic detail is provided in the tables in Part II (Questions 14 through
22 in each set of cross-tabs).
A general concern regarding low response rates is whether survey respondents are more likely
than non-respondents to be aware of, understand, or be influenced by the Energy Star label.
This possibility cannot be assessed from the demographic information. Individuals with a
greater interest in energy efficiency might have been more inclined to complete the survey for
the following reasons:
•	The survey was sponsored by an entity identified as the "Consortium for Energy
Efficiency."
•	The introduction to the survey identified EPA and DOE as promoting the program
being assessed by the CEE survey.
•	The first question referred to a picture of the Energy Star label on which "EPA"
and "DOE" appear.
For these reasons, the survey results may overstate or understate the levels of awareness,
understanding, and influence nationwide. However, a review of other studies indicates the
3-2

-------
SECTION 3
FINDINGS
findings in this study are a reasonable characterization of the current state of Energy Star label
awareness, understanding, and influence (See Sections 3.2.4, 3.3.2, and 3.4.3). In addition, the
survey provides useful comparisons across publicity levels, giving information on the relative
importance of different factors. The results also are useful as a baseline for future tracking
efforts, which are likely to have similar non-response effects.
3.2 Understanding of the Energy Star Label
The first two survey questions were used to assess whether the respondent understood the
message of the Energy Star label. A respondent might understand the message even if
previously unfamiliar with the label. The questions asked the respondent to describe in his/her
own words the first, and then any other message(s) that came to mind when looking at the label.
3.2.1 Coding the Levels of Understanding
The open-ended responses were coded into 28 detailed response categories. For purposes of this
analysis, these detailed response categories were initially combined into five levels of
understanding, as follows.
High understanding: The respondent mentioned one or more of the following:
•	Savings
•	Energy efficiency
•	Environmental benefit
•	Product standards.
General understanding: The respondent mentioned one or more of the following:
•	Energy
•	Environment
•	Quality
•	Government backing.
However, the respondent did not specifically mention savings, efficiency, benefits, or standards.
No understanding: No answer, or a response not included in the High or General understanding
level.
Mixed General-No Understanding: The detailed response category includes some General
understanding responses and some No understanding responses.
3-3

-------
SECTION 3
FINDINGS
Mixed High-General: The detailed response category includes some High understanding
responses and some General understanding responses.
The classification of the detailed responses into the five levels of understanding is indicated in
Table 3-3.
Table 3-3
Coding of Levels of Understanding
Description
Understanding
Level*
Energy savings/efficiency/conservation
High
Energy-saving product
High
Product testing, standards, compliance that is energy/environment related
High
Environment - links environmental attributes to items
High
Saves money on operation - links savings to items
High
Savings - not linked to items
High
Government - general
Mixed HDG
Confusion with yellow Energy Guide label
General
Rebate/refund - savings on purchase, not on operation
General
General environmental
General
Names product
General
Electricity/energy
General
General quality product standards (not energy or environmental)
General
World/planet/globe
Mixed GDN
Being told to do something specific
No
Cooperation
No
Design - positive
No
Design - negative
No
General - positive
No
Government - negative
No
Advertising/marketing - negative
No
Universe/stars
No
Skepticism
No
Never saw
No
No/nothing/none
No
Other
No
Don't know
No
Refused
No
3-4

-------
SECTION 3
FINDINGS
3.2.2 Overall Findings
The proportion of respondents at each level of understanding is shown in Table 3-4, by publicity
level and overall. The table shows that, over 50 percent of households had at least a general
level of understanding of the Energy Star label. Forty-one percent of consumers in high
publicity areas displayed high levels of Energy Star understanding compared to 34 percent in
low publicity areas and 38 percent in the "Other" areas. The difference in the proportion with
High understanding between respondents in high- and low-publicity categories is statistically
significant (p = 0.02). We also note that the distribution of sample consumers by level of
understanding is heavily weighted to the two ends of the scale. Thirty-seven percent of
respondents have a high understanding of the label's meaning; 42 percent have no understanding.
Table 3-4
Distribution of Understanding Levels by Publicity Level
Q1 and Q2:
Understanding
of Energy
Star Label
Publicity Category
Total
High
Other
Low
High
41%
38%
34%
37%
High - General
2%
4%
2%
3%
General
10%
8%
11%
9%
General - None
8%
10%
6%
8%
None
39%
40%
46%
42%
Total
100%
100%
100%
100%
# of households
(millions)
15.50
27.84
21.90
65.24
3.2.3 Understanding by Subgroup
Significant differences in the level of Energy Star understanding were found among subgroups
of consumers defined by demographic and housing characteristics, as well as by previous
experience related to Energy Star.
• By awareness of the Energy Star label prior to the survey. Sixty-four percent of
customers who reported having heard of Energy Star prior to the survey displayed high
levels of understanding of the label, compared to 19 percent among customers who
reported that they had not seen or heard about the label prior to the survey. This is a
highly significant difference in level of understanding between the two groups (p <
0.0005). This finding suggests that customers who have seen the label in its intended
contexts (on products or in publicity materials) are much more likely to have a good
understanding of what it means than those encountering it for the first time. Also, it
3-5

-------
SECTION 3
FINDINGS
suggests that the label itself can communicate its intended message to a small but
significant portion of the population.
•	By housing type and tenure. Thirty-nine percent of homeowners displayed high levels
of understanding of the Energy Star label compared to 27 percent of renters (p =
0.005). Similarly, 40 percent of consumers in single-family homes showed a high level
of understanding for the Energy Star label versus 30 percent of consumers in
multifamily housing (p = 0.02). However, there was no significant difference in
understanding between consumers who paid their own energy bills and those who did
not. This pattern of findings suggests that understanding of Energy Star is more
strongly related to customer characteristics such as age and income, which are associated
with single-family home ownership, than to the consumers' perceived opportunity for
financial benefit associated with responsibility for energy bills.
•	By age of respondent. The percentage of consumers with a high level of understanding
clusters in the range of 44 to 51 percent for consumers in the age ranges between 18 and
54. The proportion with high understanding drops in every 10-year age group from 51
percent of 35- to 44-year-olds to 27 percent of 65- to 74-year-olds. Among consumers
over 75, the percentage of respondents displaying a high level of understanding drops
again to 14 percent. Seventy-six percent of these consumers have no understanding
whatsoever of the label.
•	By income. The survey findings suggest that understanding of the Energy Star label is
strongly associated with income levels. Table 3-5 shows a consistent relationship
between income and levels of understanding. Fifty-one percent of customers with annual
incomes above $75,000 showed high levels of understanding versus 25 percent for
customers reporting annual incomes below $25,000 (p < 0.0005).
Table 3-5
Percent of Customers with High Levels of Energy Star Understanding
by Income Level
Income Level
% with High
Understanding
90%
Confidence
Bounds (+/-)
P-Value
Total

38%
3%

Low
< $25,000
25%
6%

Medium Low
$25,000 - $49,000
36%
6%

Medium High
$50,000 - $75,000
41%
6%

High
> $75,000
51%
6%

Difference High - Low

26%
8%
<.0005
3-6

-------
SECTION 3
FINDINGS
3.2.4 Comparison With Other Studies
Results from the CEE survey were compared with findings from four other studies. These
studies are:
•	RECS: The national Residential Energy Consumption Survey (RECS) conducted by the
Energy Information Administration. Most recent data are for 1997. EIA has not
published any results from the Energy Star questions on the 1997 RECS, but has made
the data available on a public use tape. Results shown here are from analysis of these
data conducted by Roper-Starch.
•	D&R: The 1998 Home Appliance Buying Trends Survey conducted by D&R
International, Ltd. for the Department of Energy.
•	PG&E: The Pre-Post ENERGY STAR Awareness Tracking Study conducted by
XENERGY Inc. for Pacific Gas and Electric Company (PG&E) in 2000. Results shown
here are from XENERGY's report to PG&E. The results described here are for the
residential survey, and unless otherwise noted are for the survey conducted just after the
awareness campaign.
•	WI/PNW: Surveys conducted in Wisconsin (WI) and the Pacific Northwest (PNW)
approximately concurrently with the CEE survey. In Wisconsin, three survey modes
were used: a mail survey using the CEE instrument, a phone survey, and web TV. In the
Pacific Northwest, a web TV survey was used. Results shown here are from a
presentation summary prepared by the Energy Center of Wisconsin with assistance from
Shel Feldman Management Consulting and the Northwest Energy Efficiency Alliance.
Key features of these four studies are summarized in Table 3-6.
3-7

-------
SECTION 3
FINDINGS
Table 3-6
Summary of Studies Compared

RECS
D&R
PG&E
Wl
Wl
Wl
PNW
CEE
Survey year
1997
1998
2000
2000
2001
2002
2000
2000
Mode
In-person
Mail
Phone
Mail
Phone
Web TV
Web TV
Mail
Area
Covered
National
AZ, CA, FL,
MA, TX
PG&E
service
territory
Wisconsin
Wisconsin
Wisconsin
PNW
National
Sampling
Frame
All residential
households
Recent
appliance
purchasers from
major chains
PG&E
residential
customers
Households
with listed
telephones
Households
with listed
telephones
Web TV
panel
Web TV
panel
Households
with listed
telephones in
top 57 DMAs
and member
areas
Response
Rate
81%
11%
-
18%
33%
65%
62%
10%
Comparisons to the present study are described below:
•	The RECS survey did not include questions probing respondent understanding of the
Energy Star message. The survey is included in the summary above, because it is
useful for comparisons of other information collected on the survey, presented below. A
strength of the RECS as a benchmark for the present study is that the RECS is national in
scope and achieves over 80 percent response. However, these data are three to four years
old.
•	The D&R study reported that, of those who recognize the label, 48 percent correctly
interpreted it. Of those who had not recognized the label, 19 percent gave a correct
interpretation. Combined, these findings indicate an overall 32 percent giving a correct
interpretation, across those who were and were not aware. The study did not describe
what was considered a "correct" interpretation. The study had a similar response rate of
11%.
•	In the PG&E study, only respondents who were previously aware of Energy Star were
asked what it meant. Of these open-ended responses in the post-campaign survey, 56
percent were coded as "saves energy, uses less electricity." The total number
corresponding to the present study's definition of "High" understanding is not reported,
but based on the (overlapping) categories reported, at least 63 percent of aware
respondents would be coded as having high understanding in the context of this study. In
the pre-campaign survey, a smaller fraction was aware of the label, but of these at least
3-8

-------
SECTION 3
FINDINGS
72 percent would be coded as having high understanding. In the present study, 57
percent of those who were previously aware of the Energy Star label had high
understanding.
• The WI/PNW study used definitions of "knowledge" that match this study's definition of
high understanding. In that study, the percent coded as knowledgeable were 35 percent
for the WI mail survey, 57 percent for the WI web TV survey after the label was shown,
and 53 percent for the PNW web TV after the label was shown.
Table 3-7 summarizes the percent with high understanding from the studies compared. Overall,
similar results were found in the Wisconsin mail survey and the D&R study to those in the
present study. Why higher proportions of web TV respondents had a high level of understanding
is not clear. The PG&E results for those who were aware of the label are also fairly consistent
with results for aware respondents in the present study.
Table 3-7
Proportions with High Understanding for Studies Compared

D&R
PG&E
WI
WI
WI
PNW
CEE
Survey year
1998
2000
2000
2001
2002
2000
2000
Mode
Mail
Phone
Mail
Phone
WebTV
WebTV
Mail
With label not seen or
described, all
respondents
-
-
-
11%
-
-
-
After label seen or
described, aware
respondents only
-
63%
-
-
-
-
57%
After label seen, all
respondents
32%
-
35%
-
57%
53%
32%
3.3 Awareness of the Energy Star Label
Questions 3 through 5 asked whether the respondent had seen the label before, and if so, where
and on what products. Respondents were considered aware of the label if they reported having
seen it before.
Table 3-8 shows the proportions of respondents who were aware of the label, by publicity level.
The table shows an overall awareness level of 41 percent. Label awareness is much higher in the
high-publicity areas than in the low-publicity areas—52 percent versus 37 percent, a statistically
highly significant difference.
3-9

-------
SECTION 3
FINDINGS
Table 3-8
Awareness of the Energy Star Label by Publicity Level
(Base = All Respondents)
Q3: Have heard or seen Energy Star label
Publicity
category
Estimate
90% Confidence
Bounds (+/-)
Sample
size
P-value
Overall
41%
2.5%
3,394

High
52%
3.5%
1,450

Other
38%
3.8%
1,507

Low
37%
4.8%
437

High-Low
15%
5.9%

< 0.0005
Table 3-9 shows the average number of different products on which customers reported having
seen the Energy Star label. Among those who were aware of the label, the number of products
on which it was seen did not vary significantly by publicity level.
Table 3-9
Number of Products On Which the Energy Star Label Was Seen by Publicity Level
(Base = Aware Respondents)
Q5: Average number of products on which Energy Star label has been seen
(Q3=Yes)
Publicity
category
Estimate
90% Confidence
Bounds (+/-)
Sample size
P-value
Overall
3.97
0.35
1,502

High
3.86
0.38
728

Other
4.28
0.64
599

Low
3.75
0.46
175

High-Low
0.11
0.60

0.76
3-10

-------
SECTION 3
FINDINGS
3.3.1 Where People Learned about Energy Star
Table 3-10 shows on which products respondents recalled seeing the Energy Star label.
Computers and monitors were the most selected product type (52 percent), followed by
refrigerators (43 percent), on which aware respondents recall seeing the label. Other common
products, mentioned by at least 20 percent or respondents were air conditioners, dishwashers,
and washing machines.
Another category mentioned by more than 20 percent of the respondents was the microwave.
The questionnaire included this "ringer" on purpose. This result reflects inaccuracies in
respondents' recognition of labeled products, as there is no Energy Star label for microwaves.
However, the result also indicates a general perception that Energy Star kitchen appliances are
available.
3-11

-------
SECTION 3
FINDINGS
Table 3-10
Proportion Seeing the Energy Star Label on Energy STAR-Labeled Products
(Base = Aware Respondents)
Q5: Products where
Energy Star label
was seen (Q3=Yes)
Publicity Category
Total
High
Other
Low
Central air conditioner
17%
27%
19%
21%
Furnace or boiler
10%
19%
13%
14%
Heat pump
3%
2%
6%
4%
Thermostat
6%
7%
5%
6%
Room air conditioner
20%
28%
15%
22%
Computer or monitor
51%
51%
55%
52%
Computer printer
14%
22%
21%
19%
Copying machine
8%
10%
11%
10%
Fax machine
6%
8%
9%
8%
Scanner
6%
9%
10%
8%
Dishwasher
31%
28%
17%
25%
Refrigerator
55%
40%
34%
43%
Lighting fixture
16%
14%
9%
13%
Washing machine
34%
22%
21%
25%
CFL
18%
8%
13%
13%
Television
18%
29%
25%
24%
VCR
11%
19%
17%
16%
Audio product
5%
7%
4%
6%
Window
16%
13%
19%
16%
Door
5%
8%
6%
6%
Skylight
1%
5%
3%
4%
Insulation
5%
10%
9%
8%
Roofing material
1%
9%
6%
6%
Newly built home
8%
9%
11%
9%
None of these
products
9%
12%
11%
11%
Average
15%
17%
15%
16%
# of households
7,411,478
9,757,661
7,678,135
24,847,275
% of households
30%
39%
31%
100%
Table 3-11 shows the sources where respondents recalled seeing the label. Store displays and
direct mail were most common, at 58 and 41 percent, respectively. Print media and television
each scored approximately 33 percent.
3-12

-------
SECTION 3
FINDINGS
Table 3-11
Proportion Seeing the Energy Star Label in Particular Sources
(Base = Aware Respondents)
Q4: Places seen/heard
about Energy Star
label (Q3=Yes)
Publicity Category
Total
High
Other
Low
Newspaper/magazine
34%
36%
32%
34%
Television
30%
37%
28%
32%
Mail - direct, utility
44%
40%
36%
40%
Store display
62%
56%
57%
58%
Internet
11%
20%
18%
17%
Salesperson/contractor
6%
7%
6%
6%
Friend/neighbor/etc.
4%
9%
7%
7%
Other
24%
19%
25%
22%
None of these sources
6%
3%
3%
4%
Average
24%
25%
24%
24%
# of households
7,493,123
9,699,910
7,479,054
24,672,088
% of households
30%
39%
30%
100%
3.3.2 Comparison With Other Studies
In the 1998 D&R study, 44 percent of respondents recognized the Energy Star label. This rate
is much higher than the national level of 27 percent found a year earlier in the 1997 RECS. For
the South and West Census regions (which account for 89 percent of the D&R sample universe),
the RECS awareness levels were 29 and 30 percent, respectively. One reason for the higher
national awareness in the D&R study may be that the respondents were all recent appliance
purchasers.
The 2000 PG&E study after the awareness campaign found an overall awareness of 63 percent
after their awareness campaign. This overall total includes those (36 percent) who reported in
the phone interview that they had heard of Energy Star before being read a description of the
label, as well as those (27 percent) who initially said they had not heard of the label, but changed
their response after hearing the description. The study did not probe those who initially said yes
to determine if what they had heard of matched a description of the label. The overall awareness
level before the campaign was 37 percent, more consistent with the 41 percent found in the
present study.
3-13

-------
SECTION 3
FINDINGS
The web TV results for WI and PNW after the label was displayed to the respondent were
similar to one another, at 32 percent and 31 percent respectively. On the WI mail survey,
awareness was 26 percent, lower than the same survey instrument yielded nationally on the CEE
study even for the low publicity level stratum. The WI phone survey had awareness of only 17
percent, lower than the PG&E phone survey before the description was read. However, the
PG&E phone survey prior to the awareness campaign had a similar awareness level, 16 percent,
without the description read. A probe with a description of the label was not included in the WI
phone survey.
Table 3-12 summarizes the awareness levels in the different studies.
Table 3-12
Proportions Aware of Energy Star for Studies Compared

RECS
D&R
PG&E
PG&E
WI
WI
WI
PNW
CEE
Survey year
1997
1998
2000-pre
2000-post
2000
2001
2002
2000
2000
Mode
In-person
Mail
Phone
Phone
Mail
Phone
WebTV
WebTV
Mail
With label not
seen or
described
(unprobed)
-
-
16%
36%
-
17%
-
-
-
After label seen
or described
(probed)
27%
32%
37%
63%
26%
-
32%
31%
41%
3.4 Influence of the Energy Star Label
Respondents were considered to have purchased an Energy Star product if they reported
having seen the Energy Star label on a product they purchased within the past 12 months
(Question 7). Table 3-13 shows that about 74 percent of those who were aware of the label and
made a purchase within the past 12 months reported buying an Energy Star product.
No significant difference existed in the proportion of aware respondents making Energy Star
purchases across publicity levels. This lack of difference reflects the fact that awareness is, in
part, the result of making an Energy Star purchase. The total proportion of Energy Star
purchases may be higher in high-publicity areas, but this survey was designed to ask about these
purchases only for customers who were already aware of the label.
3-14

-------
SECTION 3
FINDINGS
Table 3-13
Proportion Who Bought an Energy Star Product
(Base =Product Purchasers Aware of Energy Star)
Q7: Saw the Energy Star label on a product recently purchased
Publicity
category
Estimate
90% Confidence
Bounds (+/-)
Sample
size
P-value
Overall
74%
6%
815

High
77%
6%
379

Other
70%
12%
338

Low
78%
8%
98

High-Low
-0.4%
10%

0.94
No significant difference in Energy Star purchases existed between those with high levels of
understanding of the label and those with low levels of understanding (Table Set 2, Q7). One
reason for this apparent lack of effect is that a respondent can be classified as having a high level
of understanding without having been aware of the label previously. Respondents could be
identified as having made an Energy Star purchase only if they were first classified as
previously aware of the label.
The reported influence of the label on the decision to purchase a product is indicated in Table 3-
14. The table shows that, nationwide, about 50 percent of those who bought an Energy Star
product were influenced by the label to buy that product. The proportion influenced was
significantly higher in high-publicity areas compared with low-publicity areas, 58 percent versus
35 percent (p = 0.02). Thus, where publicity is low, the effect of the label on purchase decisions
is significantly reduced, although the purchase of an Energy Star product may itself lead to
greater label awareness.
3-15

-------
SECTION 3
FINDINGS
Table 3-14
Proportion Influenced by the Energy Star Label To Buy the Product Purchased
(Base = Energy Star Product Purchasers)
Q8: Presence or absence of Energy Star label influenced purchasing decision
Publicity
category
Estimate
90% Confidence
Bounds (+/-)
Sample
size
P-value
Overall
50%
8%
602

High
58%
14%
287

Other
58%
14%
243

Low
35%
8%
72

High-Low
23%
16%

0.02
3.4.1 Effect of Rebates
Many Energy Star products are offered with price incentives from manufacturers, from local
Energy Star partners, or from unrelated utility efficiency programs. Overall, about one-quarter
of those who bought an Energy Star product received a rebate or discount, as shown in Table
3-15. The proportion receiving discounts was twice as high in high-publicity areas compared
with low-publicity areas, possibly reflecting the greater level of energy-efficiency promotion in
those areas. The difference between high and low publicity areas was statistically significant (p
= 0.05).
3-16

-------
SECTION 3
FINDINGS
Table 3-15
Proportion Receiving Incentives for Energy Star Purchases
(Base = Energy Star Product Purchasers)
Q9: Received Discount on Purchase of Energy Star Product
Publicity
category
Estimate
90% Confidence
Bounds (+/-)
Sample
size
P-value
Overall
23%
6%
601

High
30%
7%
285

Other
25%
9%
245

Low
15%
10%
71

High-Low
15%
12%

0.05
Of those who did receive an incentive for Energy Star purchases, 36 percent would have been
"very likely" to buy the product without the incentive, as shown in Table 3-16. Another 32
percent would have been "somewhat likely."
The sample proportion "very likely" to buy without the incentive was higher in low-publicity
areas than in high-publicity areas. However, the difference was not at all significant statistically
(p = 0.33). More analysis of the data will be required to assess the relative importance of rebates
and the label by itself.
3-17

-------
SECTION 3
FINDINGS
Table 3-16
Likelihood of Buying the Energy Star Product without the Incentive
(Base = Energy Star Product Purchaser Who Received Incentives)
Q10: No product
discount, likely purchase
Energy Star product
Publicity Category
Total
High
Other
Low
Very Likely
24%
43%
41%
36%
Somewhat Likely
29%
33%
36%
32%
Somewhat Unlikely
30%
3%
11%
14%
Very Unlikely
17%
21%
11%
17%
Total
100%
100%
100%
100%
# households (millions)
0.83
1.01
0.64
2.48
% households
33%
41%
26%
100%
3.4.2	Findings by Subgroup
Consumers with a high level of understanding were much more likely than those with no
understanding to report that the label had influenced their decision to buy the Energy Star-
labeled product, but the difference was not strongly significant (30 percent versus 12 percent, p =
0.14; see Table Set 2 in Part II). No clear pattern by age (Table Set 5 in Part II) or by gender
(Table Set 6 in Part II) emerged.
3.4.3	Comparison With Other Studies
The D&R study did not ask explicitly about the influence of the Energy Star label on purchase
decisions. The study did find that 78 percent reported that energy efficiency was extremely
important or important to appliance purchase decisions.
The PG&E post-campaign survey found somewhat lower levels, in more focused questions. For
purchasers of refrigerators and clothes washers, respectively, 66 percent and 60 percent were
very or extremely concerned with energy use and operating costs. The importance may have
ranked lower on the PG&E survey because the respondents were first asked an open-ended
question about what attributes they look for when making a purchase. After mentioning
attributes such as size and price as (the most common first mentions) as their first considerations,
3-18

-------
SECTION 3
FINDINGS
respondents would be less likely to identify themselves as very or extremely concerned with
another attribute.
The 1997 RECS did ask explicitly about the influence of the Energy Star label on purchases,
using a question very similar to that on the CEE questionnaire. Of all those who were aware of
the label, 32 percent nationally reported that the label had influenced a purchase decision. In the
present study, 37 percent of those who not only were aware of the label but also purchased an
energy-related product in the past two years reported being influenced by the label. Therefore, in
RECS, the percent influenced would be somewhat lower because the subset of respondents who
were asked the question about influence is larger than the subset of respondents in the present
study (i.e., as a fraction of all those aware of the label, regardless of whether they had made a
recent purchase). Thus, the influence results in the present study are fairly consistent with those
from the 1997 RECS.
Influence results are not available for the WI/PNR studies. The Wisconsin mail survey included
the same questions as the CEE national survey, but the current report does not discuss the results.
Table 3-17 summarizes the influence findings for the RECS and present study.
Table 3-17
Comparisons of Proportions Influenced by Energy Star

RECS (1997)
CEE (2000)
Mode of survey
In-person
Mail
Base = Aware Energy
Star Purchasers
-
50%
Base = Aware Purchasers
-
37%
Base = Aware
Respondents
32 %
-
Base = All Respondents
8%
-
3.5 Information Sources
Table 3-18 shows the sources where respondents reported getting information on heating and
cooling products. The most common sources were Consumer Reports and friends and neighbors,
each at approximately 60 percent. Several sources, including newspaper, television, utility
programs, retailers, and Internet, each were used by about 30 percent of respondents in obtaining
information about heating and cooling products. Little difference was seen across publicity
levels.
3-19

-------
SECTION 3
FINDINGS
Table 3-18
Heating and Cooling Product Information Sources
(Base = All Respondents)
Q12: Heating/cooling
information sources
Publicity Category
Total
High
Other
Low
e.g. Consumer Reports
61%
62%
52%
59%
Other types of magazines
15%
19%
14%
16%
Newspaper
31%
32%
37%
33%
Radio
11%
17%
15%
15%
Television
28%
30%
31%
30%
Utility program
32%
26%
30%
28%
Retailer
32%
30%
26%
29%
Contractor
27%
30%
27%
28%
Friend/neighbor/etc.
57%
62%
57%
59%
Internet
38%
35%
30%
34%
Other source
5%
5%
7%
6%
Average
31%
32%
30%
31%
# households (millions)
14.01
25.37
19.30
58.68
% households
24%
43%
33%
100%
Table 3-19 shows the sources where respondents reported getting information on home
appliances, lighting, and home electronics. The most commonly reported sources were very
similar to those reported for heating and cooling equipment. Utility programs were mentioned
somewhat less frequently, with approximately 20 percent of respondents, and retailers somewhat
more frequently.
3-20

-------
SECTION 3
FINDINGS
Table 3-19
Home Appliances, Lighting, and Electronics Product Information Sources
(Base = All Respondents)
Q12: Non-
heating/cooling
information sources
Publicity Category
Total
High
Other
Low
e.g. Consumer
Reports
63%
62%
54%
60%
Other types of
magazines
21%
26%
21%
23%
Newspaper
29%
35%
38%
34%
Radio
13%
17%
15%
15%
Television
27%
31%
32%
31%
Utility program
25%
18%
22%
21%
Retailer
37%
37%
33%
36%
Contractor
20%
20%
19%
19%
Friend/neighbor/etc.
57%
64%
58%
61%
Internet
36%
37%
32%
35%
Other source
6%
6%
5%
6%
Average
30%
32%
30%
31%
# households (millions)
13.89
24.50
18.90
57.30
% households
24%
43%
33%
100%
3-21

-------
Conclusions
Energy Star qualified products are being promoted by a total of 86 utilities, market
transformation groups, and state administrators representing one-half of U.S. households. This
first national study of the Energy Star label for household products produced several positive
findings:
•	Nationwide, 41 percent of households are aware of the Energy Star label.
•	Over one-half of all households, including those that had seen the label previously and
those that were reacting to it for the first time, had at least a general understanding of the
label's message.
•	Label awareness and understanding are greater in areas with higher promotional activity.
•	The influence of the label on purchase decisions is greater in areas with higher
promotional activity.
•	About one-half of those who purchased Energy Star products were influenced by the
label to buy the product.
•	Roughly 70 percent of those who received an incentive for the purchase of Energy Star
products reported that they would have been somewhat or very likely to have bought the
product even without the rebate.
These findings confirm that a substantial portion of U.S. consumers are aware of and understand
the Energy Star label and are similar to other regional and/or comparably designed studies.
Moreover, the label influences purchase decisions, and local and regional publicity efforts
improve both awareness and understanding of the label.
4
4-1

-------
APPENDIX A
Survey Questionnaire

-------
		^?HwSelwWSurvey
Energy Star "°
promoting energy
«.. -r ;5ix" r:sS4T^^;„*;s"e
are assessing a	pepartment of Ene^' _ompanies. Your
protection Agency,	electric and gas util y ^r... out this
governments, and nun*take a few	it
I 'eeponee is very	to the	°^nJ wil, be
questionnaire, fold t	9 Mge necessary). M <~P
rssss^ 	
1 The Consortium for Energy Efficiency	^
Fold on Dotted Line
Staple or tape here

-------
Household Survey	I
Instructions	i
r)	I
WhojhQulicgnTpIe^^	person in your I
and home electronics purchases.	1
1	Write your brief	|
response neatly or mart^ yQUr
I situation with an X 'n t	d t0 the next
\ refold the questionnaire
1 No postage is necessary.			
1 at 1-800-966-1254.	[/

¦?j\
D rj5
Please look at the Energy Star Label above.
Write the first message that comes to mind when
you see the Energy Star Label.
Please write any other messages that come to mind when you see the Energy Star Label.
Prior to this survey, had you ever heard of or seen this Label? ~ Yes (Proceed to Question ~ on page 2)
~ No (Skip to Questioneei onpaae 3)
Page 1

-------
Please review the following list and mark with an X all places in which you have seen or heard about the
Energy Star Label.
~	In newspapers or magazines	~
~	On television	~
~	On utility inserts or by direct mail	~
~	On displays in stores
On the Internet
From a sales person or contractor
From a friend, neighbor, relative, or
coworker
~	Other
~	None of these sources
~	Don't know
Please review the following list and mark with an X all the products or product literature on which you have seen the
Energy Star Label.
Heating and Cooling Products
Home Appliances/Lighting
Building Materials
~ Central air conditioner
~ Dishwasher
~
Window
~ Furnace or boiler
~ Refrigerator
~
Door
~ Heat pump
~ Lighting fixture
~
Skylight
~ Thermostat
~ Washing machine
~
Insulation
~ Room air conditioner
~ Compact fluorescent light bulb
~
Roofing material

~ Microwave

Home Office Equipment
~ Computer or monitor
Home Electronics
~
Newly Built Home


~ Computer printer
~ Television
~
None of These Products
~ Copying machine
~ VCR


~ Fax machine
~ Audio product


~ Scanner



~
~
Have you purchased any of the products listed in
Question in the last 12 months?
Yes (Proceed to Question ~>
No (Skip to Question E kA on page 3)
For any of the products you purchased, did you see the
Energy Star Label (on the product itself, on the packaging,
or on the instructions)?
~ Yes. On which products did you see the Energy Star label?
(please list all products)
(Proceed to Question ED
~	No (Skip to Question 61 on page 3)
~	Don't Know (Skip to Question HI on page 3)
8 For any Energy Star product(s) you purchased, did the
presence or absence of the Energy Star Label influence
your purchasing decision?
~	Yes (Proceed to Question 0)
~	No (Proceed to Question EJ>
~	Don't Know (Proceed to Question H>
9 if you purchased an Energy Star product, did you receive
rebates or reduced-rate financing?
~	Yes (Proceed to Question EE)
D No (Skip to Question HI on page 3)
~	Don't Know (Skip to Question I on page 3)
10
If rebates or reduced-rate financing had not been available, how likely is it that you would have purchased
the Energy Star product?
~ Very Likely
~ Somewhat Likely ~ Somewhat Unlikely
~ Very Unlikely ~ Don't Know
Page 2
Please continue on the next page

-------
Which of the following products have you purchased in the last 12 months? Please mark with an X all that apply.
Heating and Cooling Products Home Appliances/Lighting
~	Central air conditioner
~	Furnace or boiler
~	Heat pump
~	Thermostat
~	Room air conditioner
Home Office Equipment
~	Computer or monitor
~	Computer printer
~	Copying machine
~	Fax machine
~	Scanner
~	Dishwasher
~	Refrigerator
~	Lighting fixture
~	Washing machine
~	Compact fluorescent light bulb
~	Microwave
Home Electronics
~	Television
~	VCR
~	Audio product
Building Materials
~	Window
~	Door
~	Skylight
~	Insulation
~	Roofing material
~	Newly Built Home
~ None of These Products
Please look at the product types listed below. Please mark with an X the source(s) of information you are most likely to use to
obtain information about that product type. Mark all that apply.
Heating and Cooling Products	Home Appliances/Lighting/Home Electronics
12
~	Consumer Reports and other product-oriented magazines	~
~	Other magazines	~
~	Newspapers	~
~	Radio	~
~	Television	~
~	Electric or gas utility program	~
~	Advice from retailers	~
~	Advice from contractors	~
~	Advice from a friend, neighbor, relative, or coworker	~
~	Internet	~
~	Other	~
~	Don't know	~
Consumer Reports and other product-oriented magazines
Other magazines
Newspapers
Radio
Television
Electric or gas utility program
Advice from retailers
Advice from contractors
Advice from a friend, neighbor, relative, or coworker
Internet
Other
Don't know
Pleas© notes We emphasize that this
survey is strictly confidential. Your re-
sponses will be included with the responses of
other survey participants, and your name will
not be associated with your responses or be
provided to the government or any other party.
13
How many personal computers
are in use in your home?
Number of computers:	
(If vour answer is 0. Dlease skio to Question EE
14
Adding together the use of all
computers in your home, what is
the average number of hours per day that
computers are turned on?
Average number of hours:	
15
How many people live in your
household, including yourself?
(Please count children as well as adults.
Include all members of your household
whether or not they are related to you.)
Number of people in household:	
16
What is your age?
~
~
~
~
18-24
~
55-64
25-34
~
65-74
35-44
~
75 +
45-54


17
What is your gender?
~	Male
~	Female
Are you the person responsible
	 for paying the energy bill(s)
your household?
18
in
~
~
Yes
No
19
Which of the following best
describes your home?
~	Single-family home not attached to others
~	Townhouse or row house
~	Duplex or triplex
~	Apartment (in building with 4+ units)
~	Mobile home
~	Other
How many bedrooms do you
have in your home?
Number of bedrooms:
20
21
Do you or members of your
household own or rent your
present home?
~	Own
~	Rent
~	Occupy but do not pay rent
Please mark the box indicating
the total combined income in the
last 12 months of all family members living
in your household. (Include income before
taxes and deductions from all sources.)
~	Less than $5,000
~	$5,000 - $9,999
~	$10,000 - $14,999
~	$15,000 - $19,999
~	$20,000 - $24,999
~	$25,000 - $49,999
~	$50,000 - $74,999
~	$75,000 and over
Thank you very much for
your assistance
Page 3

-------
APPENDIX B
Fifty-seven Largest DMAs

-------
2
5
6
12
20
26
27
7
8
11
14
15
22
24
28
29
30
32
34
35
36
37
39
40
41
43
44
45
47
48
49
51
52
54
57
	1_
3
4
9
10
13
16
17
18
19
21
23
25
31
33
38
42
46
50
53
55
56
Table B-l
The Fifty-seven DMAs in the National Sample
DMA
Market Name
GRPS through 12/99
LOS ANGELES
375
SAN FRANCISCO-OAK-SAN JOSE
1333
BOSTON
625
SEATTLE-TACOMA
1606
SACRAMNTO-STKTON-MODESTO
126
SAN DIEGO
2515
HARTFORD & NEW HAVEN
816
DALLAS-FT. WORTH
54
WASHINGTON, DC
HOUSTON
TAMPA-ST. PETE (SARASOTA)
465
MINNEAPOLIS-ST. PAUL
101
ORLANDO-DAYTONA BCH-MELBRN
408
BALTIMORE
463
CHARLOTTE
RALEIGH-DURHAM
94
NASHVILLE
CINCINNATI
COLUMBUS, OH
GREENVLL-SPART-ASHEVLLE
SALT LAKE CITY
32
SAN ANTONIO
BIRMINGHAM (ANN, TUSC)
NORFOLK-PORTSMTH-NEWPT NWS
198
NEW ORLEANS
MEMPHIS
WEST PALM BEACH-FT. PIERCE
OKLAHOMA CITY
GREENSBORO-H.POINT-W.SALEM
LOUISVILLE
ALBUQUERQUE-SANTA FE
WILKES BARRE-SCRANTON
295
JACKSONVILLE
DAYTON
LITTLE ROCK-PINE BLUFF
NEW YORK
674
CHICAGO
1010
PHILADELPHIA
605
DETROIT
1517
ATLANTA
914
CLEVELAND
968
MIAMI-FT. LAUDERDALE
1317
PHOENIX
214
DENVER
2228
PITTSBURGH
2016
ST. LOUIS
1421
PORTLAND, OR
1098
INDIANAPOLIS
355
MILWAUKEE
1259
KANSAS CITY
GRAND RAPIDS-KALMZOO-B.CRK
BUFFALO
HARRISBURG-LNCSTR-LEB-YORK
1690
PROVIDENCE-NEW BEDFORD
ALBANY-SCHENECTADY-TROY
FRESNO-VISALIA
LAS VEGAS
375
B-l

-------
APPENDIX C
Sample Weighting and Variance Estimation
Methodology

-------
Sample Weighting and Variance Estimation Methodology
A sampling weight is assigned to each survey respondent. The weight is equal to the
number of households represented by the responding household. This number is the
inverse of the sample inclusion probability.
For households in the CEE member sample, the inclusion probability for each customer
in a member area is the ratio of the number of respondents in the area to the total number
of households in the area. The weight is thus the ratio of the total households in the area
to the total households in the sample. This weight is the same for all households in a
given member area. That is, the weight for member area a is given by
wa = l/pa = (Na/ na),
where
Pa = rial Na
na = number of responding households in member area a
Na = total number households in member area a.
For households in the restricted national sample, the inclusion probability is the product
of two probabilities. First is the probability that the DMA was selected within the
publicity stratum. This probability is the ratio of selected DMAs (or restricted DMAs) to
the number of DMAs in the stratum (in the restricted national frame). The second
probability is the probability that a household was included given that the DMA was
selected. This probability is the ratio of the number of responding households to the
number of households in the restricted DMA.
The corresponding expansion weight is simply the inverse of its overall sampling
probability. Thus, for all sampled homes in (restricted) DMA a from the restricted
national sample, the expansion weight is
Wa = l/pa = (Vpia)(Vp2a)
where
P\a n\ml N\m
n\m = total number of DMAs selected from stratum m for the restricted national
frame
N\m = total number of DMAs in the restricted national frame in stratum m.
P2a = nlaINla
ri2a = number of responding households in the restricted DMA a
Nia = total number of households in the restricted DMA a.
The mean of any variable x of interest over a group g is then calculated from the
observations Xj over all responding households j as
C-l

-------
X	£_/ eg X7 W/ / Zy eg W/.
Variance Estimation
The component estimates and variances were estimated using SUDAAN, which is a
software package developed by Research Triangle Institute for analysis of complex
survey data. The package requires only the specification of the sampling structure,
including stratification, nesting, and first-stage selection probabilities.
Following is the overall sample structure for the combined sample:
1.	The total population is stratified into the member and restricted national
frames.
2.	The member frame is further stratified into the five member areas.
3.	A simple random sample is selected within each of the member strata.
4.	The national restricted frame is stratified by publicity level.
5.	First-stage sampling is of (restricted) DMAs within each publicity stratum.
6.	A simple random sample is selected within each selected DMA.
Because the member and national frames have a different structure, with the two-stage
sampling for the national frame only, estimates and standard errors are calculated
separately for these two portions of the combined frame. The estimates and standard
errors are then combined outside SUDAAN.
The samples in the member and restricted national frames are independent. Thus, the
variance of these two components can be separately estimated. The variance of the
overall estimate is calculated from these two components.
To estimate a total, the overall estimate can be expressed as the sum of totals for the two
components:
Xf = Xtp + Xtr ,
where subscripts P and R, respectively, denote the member and restricted national
samples. The variance of the overall total is then given by
V(XT) = V(XTP) + V(XTR).
To estimate a mean, the overall estimate is a weighted average of the component means
— _ NtpXtp + NtrXtr
7* 	
N +N
TP PR
C-2

-------
where Ntp and Ntr, respectively, denote the number of homes in the member and
restricted national frames. The variance of the overall mean is then
V(XT) = (Ntp /Nt)2V(.XTP) + (Nm /Nt f V(xm),
where Nt = Ntp + Ntr is the total number of homes in the combined frame. These
formulas apply also to the variance of an estimated proportion, viewing a proportion as
the mean of a 0/1 variable.
Subgroup Estimates
Totals, means, or proportions and corresponding variances for any subgroup of interest
can be estimated by using essentially the same formulas given above for overall totals or
means. The difference is that for subsets the population counts Ntp and Ntr used to
combine variances for the two sample components are estimated from the data, rather
than being taken directly from known totals.
C-3

-------