University of Michigan Energy Survey Results from October 2013 Sample John DeCicco, Ting Yan, Florian Keusch, Patrick Shields, Bruno Vanzieleghem, Michael Sadowsky and Diego Horna Muñoz University of Michigan Energy Institute and Institute for Social Research March 2014 www.energy.umich.edu www.home.isr.umich.edu Table of Contents Introduction 1 Degree of Concern about Energy-Related Problems 2 Views on Reliability 5 Views of Energy Affordability 8 Expected change in home energy bill and its affordability 10 Findings on the affordability of gasoline 14 Conserving energy for cost reasons 19 Views of Energy and the Environment 20 Sources of energy believed to most affect the environment 22 Expectations regarding the impact of energy on the environment 24 Conclusion 25 Acknowledgments 27 Appendix 28 A. Questionnaire Development 28 B. Focus Group Findings 29 C. Sample Design and Margins of Error 32 ii University of Michigan Energy Survey Results from October 2013 Sample Introduction Energy is a major issue that affects consumer wellbeing, the economy, the environment and national security. Public perceptions of energy are buffeted by events ranging from power outages to oil spills, from volatile prices and fears of shortages to promises of plenty, and from concerns about risks to visions of a brighter future through new technology. Surveys are often conducted to address various aspects of energy and the particular issues that arise from year to year. To date, however, there has been no long-running, issue-neutral survey designed to track core consumer concerns about energy over time. To fill this gap, the University of Michigan Energy Institute (UMEI) has joined forces with the university's Institute for Social Research (ISR) to launch a new survey that gauges American consumer attitudes about major energyrelated concerns including affordability, reliability and environmental impact. The U-M Energy Survey is comprised of 18 questions appended to ISR's long-running Surveys of Consumers (SCA). That instrument uses a nationally representative sample targeting 500 U.S. households that are interviewed by telephone on a monthly basis. The Energy Survey will run quarterly as a rider on the SCA in January, April, July and October of each year. The initial fielding of the U-M Energy Survey was in October 2013 and the resulting data are the basis of the analyses reported here. The Surveys of Consumers have been underway for over sixty years and are a proven source of data that indicate the future course of the national economy with a striking degree of accuracy. The Index of Consumer Expectations derived from the SCA is included in the Leading Indicator Composite Index published by the U.S. Department of Commerce, Bureau of Economic Analysis. This official use of the SCA's widely-reported index is a major confirmation of the power of the SCA and its value capabilities for understanding the state of the national economy and forecasting its future course. The resulting survey data become a resource for the social science research community and are made available through ISR's public data archive within about six months after sample collection. 1 Building on this pedigree, the U-M Energy Survey will offer unique insights into consumers’ perceptions and beliefs about energy and how it touches their lives. It inherits the rigorous sample design and statistical power of the SCA, and combining the new data from the energy survey with the long-running SCA core data plus data from standard energy, economic and demographic sources will enable a broad range of analyses and offer a deeper understanding of Americans' attitudes regarding energy than ever before possible. This report summarizes findings from the first fielding of the U-M Energy Survey. As a household-level survey, its responses can be analyzed according to demographic, geographic and other characteristics drawing on data gathered as part of the core SCA. Classification variables used here include region, home tenure and property value, income, and self-reported-knowledge of energy-related issues. Region of residence divides respondents into those from the Northeast, South, Midwest, and West. A plurality of respondents was from the South (37%), around a quarter from the Midwest (25%) and West (22%), and a lesser number from the Northeast (16%). The sample was classified into thirds (terciles) by self-reported household income. Respondents were also classified by home status (home tenure and property value), resulting in four categories including renters plus bottom, middle and top terciles based on self-reported home value; about a quarter of respondents fell into each of these four categories. The Energy Survey also includes a question about respondents' self-perceived knowledge of energy, enabling us to classify the population into consumers who said they knew a lot or a fair amount and those who said they knew a little or nothing about energy issues. In the October 2013 sample, roughly 57% of respondents stated that they knew at least a fair amount about energy and around 43% said that they had little or no knowledge. Responses were weighted so that the results are representative of the U.S. household population; all charts and tabulations reported here are based on weighted frequencies of response. Degree of Concern about Energy-Related Problems We asked a set of similarly worded questions regarding how much consumers personally worried about each of the three energy issues -- reliability, affordability and environmental impact -covered in the survey. These questions were posed near the end of the interview so that the respondents had already become familiar with the topics being addressed. Because these 2 questions were worded similarly, the responses provide an overall gauge of consumers' relative degrees of concern about the three issues. To probe the degree of concern about reliability, the respondents were asked: How much do you personally worry about the reliability of energy? Would you say a great deal, a fair amount, only a little, or not at all? According to the results, approximately 31% of the population worries a great deal or a fair amount about the reliability of energy. We found significant differences by income level; the share of the population that worries a great deal or a fair amount about the reliability of energy in the top income tercile is significantly lower than the one for the population in the bottom income tercile (Figure 1). We did not find any significant differences by region, by home status or by self-reported knowledge of energy. Interviewers asked consumers about their degree of concern in regards to the affordability of energy using the following question: How much do you personally worry about the affordability of energy? Would you say a great deal, a fair amount, only a little, or not at all? We found that on average nearly 55% of the population worries a great deal or a fair amount about the affordability of energy. Results varied by income level, with greater shares of the 3 population in the bottom and middle income terciles expressing higher degrees of concern about the affordability of energy than seen in the top tercile (Figure 2). These results did not vary significantly across regions, by home status or by self-reported knowledge of energy. To probe beliefs about the environmental impacts of energy, consumers were asked the following: How much do you personally worry about the environmental impact of energy? Would you say a great deal, a fair amount, only a little, or not at all? Close to 60% of respondents replied that they worry a great deal or a fair amount about the environmental impact of energy. No significant variation in this level of response was found by region, home status, income or self-reported knowledge of energy. It is notable that consumers are just as concerned about the environmental impacts of energy as they are about its affordability. The values of approximately 60% and 55% for the fraction of respondents expressing higher levels of concern about environmental impact and affordability, respectively, are not significantly different given the sampling error. Both are significantly greater than the 31% of respondents who indicated a great deal or fair amount of concern about the reliability of energy. 4 Views on Reliability To examine consumers’ perceptions of reliability, we started with the question: Considering all sources of energy you usually use in everyday life, how reliable would you say they are -- not at all reliable, slightly reliable, moderately reliable, or very reliable? If clarification was needed, respondents were told that reliability refers to members of the household getting the energy they need when they need it. Overall, around 75% of consumers perceive the energy they use as very reliable. Significant differences were found when cross-tabulating the perceptions of reliability according to home tenure and income. Among respondents whose homes were in the top tercile by property value, a significantly larger-than-average fraction (close to 85%) considers their energy to be very reliable (Figure 3). In contrast, a significantly lower fraction (roughly 63%) of respondents in the bottom tercile by property value considers energy to be very reliable. A similar pattern emerges when the responses are examined by income, with a significant difference between the top and bottom income terciles (Figure 4). We did not find any significant differences in perceptions of reliability across geographic regions or according to the respondents' self-reported knowledge of energy. 5 After inquiring about general perceptions of reliability, the interviewers asked: What specific source of energy were you mostly thinking about when you said that the energy you use is [not at all / slightly / moderately / very] reliable? 6 Interviewers did not offer a list of responses or prompt consumers for any particular source of energy when asking this question. Electricity was by far the dominant response, given by nearly 66% of consumers (Figure 5). Natural gas ranked second at approximately 20% while gasoline, oil and petroleum ranked third at roughly 11% of respondents (replies about the different petroleum-based fuels were aggregated when responses were coded). The sources of energy consumers said that they had in mind when considering reliability showed some regional variation (Figure 6). The fraction of consumers who identified natural gas as a reliability concern was smaller in the South than in the Northeast and Midwest. Conversely, a greater share of Southern respondents mentioned electricity when it comes to reliability. These findings are consistent with the fact that natural gas is used by less than half of households in the South but used by roughly three-quarters of households in the other regions.1 No significant differences in the energy source of concern for reliability was found when examining responses by income, home status or self-stated knowledge of energy. 1 U.S. Energy Information Administration. 2009. Residential Energy Consumption Surveys (RECS); Fuel Used and End Uses by Census Regions (HC1.7) www.eia.gov/consumption/residential/data/2009/ 7 Views of Energy Affordability Consumers were asked about home energy bills (other than for motor fuel) as well as the expected cost and affordability of such expenses in the future. A series of questions were used to determine the degree of increase at which energy bills would become unaffordable and, by implication, the share of the population that would consider their current energy bill to be unaffordable. The first question was: Now thinking about the last time you (or someone else in your household) paid a household energy bill of any kind, how much did that bill cost you? Please do not include your water bill. If clarification was needed, respondents were told that expenses include whatever they might pay for electricity, natural gas, propane, heating oil, or other fuels they might use in their home for lights, appliances, heating and air conditioning, household chores, electronics and entertainment. If they could not remember, they were asked to provide their best estimate. This question allowed us to determine the currently perceived level of energy expenses. Then, consumers were asked the following: What sources or types of energy did that bill cover? If respondents mentioned “gas,” they were asked to clarify whether they meant gasoline or natural gas. We then asked: At what dollar amount would that [type of energy stated by respondent] bill become unaffordable to you (and your family)? By unaffordable we mean that you (and your family) would be forced to make significant changes in the way you live your life. Respondents were asked to report this figure in dollars. If they were unable to provide a dollar value, they were asked how much their bill would have to increase in percentage terms to become unaffordable. Respondents who said that their current bill was already unaffordable were asked the dollar amount at which their energy bill became unaffordable. Analyzing the responses to these three questions enables us to determine the percent increase over their current energy bills at which household energy costs are considered unaffordable. On average, we found that consumers believe that their home energy bills would have to increase by roughly 170% (a factor of 2.7) to become unaffordable. (This result should be interpreted as being relative to energy costs as perceived by the respondents in October 2013, the month the data were taken.) 8 Significant differences in this result were found by home status and by income. Homeowners in the bottom property value tercile believed that doubling their current energy bill cost would make it unaffordable, whereas for homeowners in the top and middle property value terciles and renters, the bill would need to triple before it would be seen as unaffordable (Figure 7). A large difference was also found between respondents in the top and bottom terciles of selfreported income (Figure 8). While a doubling (100% increase) would push home energy bills to a level viewed as unaffordable by respondents in the bottom income tercile, the bills would have to go up by approximately 250% (a factor of 3.5) before they are seen as unaffordable by consumers in the top income tercile. The home energy bill increase seen as unaffordable did not vary significantly across regions or by self-reported knowledge of energy. This question structure enabled us to estimate the fraction of the population that would view their current energy bills as unaffordable, a situation implied for approximately 5% of respondents. This fraction did not vary significantly by region, home status, income or selfreported knowledge of energy. It is important to note that we did not explicitly ask consumers whether they felt that their home energy expenses were unaffordable, but rather inferred those perceptions based on responses to the series of questions described above. Framing this line of 9 inquiry differently could yield different estimates of the current and expected affordability of home energy expenses. Expected change in home energy bill and its affordability Consumers were also asked what they expect their energy bills to be five years into the future. Interviewers restated the specific energy sources a respondent had identified when describing their energy bill; this was done in order to prompt respondents to estimate the future cost of the particular sources of energy they use in their homes. The question used was: About how much do you expect that [type of energy stated by respondent] bill to cost you five years from now? Respondents were asked to report this figure in whole dollars, but if they were unable to provide a dollar value, they were asked to provide the expectation in percentage terms. Combining the responses to this question with the prior responses about current energy bills enables us to compute the increase in home energy expenses expected across the population. On average, consumers expect that their home energy bills will increase by about 30% in the next five years. Significant differences in expected energy bills were found by region, by home status and by income. Respondents in the West expect their energy bills to increase by nearly 40%, whereas those in the Midwest expect theirs to increase by a roughly 25% (Figure 9). 10 Analyzed according to home status (Figure 10), homeowners generally expect their energy bills to increase by around 25% over current levels, with no significant variation across the terciles of self-reported property value. Renters expect a notably greater increase, foreseeing a roughly 50% increase in their energy bills. 11 Even though energy bill expectations did not vary across the home value terciles, they did vary by household income. The percentage that people in the bottom income tercile expect their energy bill to increase five years from now reached 40% whereas this figure averaged around 25% for the top and middle income terciles (Figure 11). No significant difference in the expected energy bill increase was found by self-reported knowledge of energy. Considering responses to this question plus those to the question about the dollar amount at which home energy bills would be seen as unaffordable, we computed the share of the population who expect home energy costs to become unaffordable in five years. That value averages about 16% overall. Cross tabulating by region reveals that only about 7% of consumers in the Midwest expect energy to become unaffordable in five years, in contrast to approximately 20% in the South and West (Figure 12). We also found significant differences by income and by home status. The proportion of bottom income tercile households who believe that their energy bills will become unaffordable in five years is approximately 30%, compared to less than 15% of middle and top income households (Figure 13). Similarly, the share of homeowners in the bottom property value tercile who expect their energy bills to become unaffordable is nearly 30%, compared to about 10% for 12 those in the middle and top terciles by self-reported property value (Figure 14). No significant difference was found for this topic according to self-reported knowledge of energy. 13 Findings on the affordability of gasoline Price expectations and views on the affordability of gasoline were elicited with a method similar to that for home energy bills. The SCA has long included a question about the gasoline price expected in five years. The Energy Survey rider adds the question: At what price per gallon would gasoline get so high that it becomes unaffordable to you (and your family)? If clarifications were needed, consumers were told that by unaffordable we meant a price per gallon that would force the respondent and his/her family to make significant changes in the way they get around. If consumers stated that the current gasoline price was already unaffordable, we asked them to give the price at which they felt it had become unaffordable. Respondents who were unable to provide a dollar figure were asked to provide an estimate in percentage terms. For such cases, we converted their reply to a dollar value relative to the national average gasoline price of $3.42 per gallon in October 2013, the month the survey was taken.2 2 U.S. Energy Information Administration. “Petroleum & Other Liquids – Weekly Retail Gasoline and Diesel Prices,” accessed February 1, 2014. http://www.eia.gov/dnav/pet/pet_pri_gnd_dcus_nus_m.htm; all prices are in current US$ as of October 2013. 14 On average, in October 2013 U.S. consumers believed that gasoline would become unaffordable at a price of $5.90 per gallon. As one might expect, we found differences by income and by home tenure/property value. We did not find significant differences in response across regions or by self-reported knowledge of energy. Respondents in the top tercile of selfreported household income believe that gasoline would be unaffordable at roughly $6.60 per gallon, a dollar higher than the roughly $5.60 per gallon price considered unaffordable by those in the middle and bottom income terciles (Figure 15). A similar difference was found between respondents in the top and bottom terciles by self-reported property value, with estimates of $6.50 per gallon for the former and $5.40 per gallon for the latter (Figure 16). Compared to the actual average retail gasoline price of $3.42 per gallon at the time of the survey, these results imply that an increase of 85% is considered unaffordable on average across our representative sample of U.S. households. It is notable that this degree of increase in gasoline price is half as large as the average 170% increase at which home energy bills are considered unaffordable (compare, for example, Figure 15 with Figure 8 and Figure 16 with Figure 7). 15 These data related to the perceived and expected affordability of gasoline also enable us to estimate the share of the population that implicitly viewed the gasoline price current at the time of the survey as unaffordable, even though again we did not explicitly ask such a question. When asked the price at which they would consider gasoline to be unaffordable, 3% of respondents stated a price at or below the national average at the time of the survey. No significant variation of this result was found across regions; with this very first sample, we did not compare, say, regional responses to regional fuel price data, an exploration that might be more statistically meaningful once a longer time series of survey data become available. Neither did we find any significant variation by income, by home status or by self-reported knowledge of energy. As was the case for the affordability of home energy bills, framing the questions differently could yield different answers about the perceived and expected affordability of gasoline. The SCA has for many years included questions about gasoline price expectations, asking consumers: Do you think that the price of gasoline will go up during the next five years, will gasoline go down, or will they stay about the same as they are now? 16 About how many cents per gallon do you think gasoline prices will (increase/decrease) during the next five years compared to now? On average, respondents in the October 2013 survey expect the price of gasoline to equal approximately $3.70 per gallon in five years, an increase of 15% over the actual average that month. We did not find significant variation, in either absolute or percentage terms, across regions, by income, by home status or by self-reported knowledge of energy. Combining responses to these gasoline price-related questions yields an estimate that on average 12% of consumers that expect gasoline to become unaffordable in five years. This estimate was found to vary significantly across all the classification variables except region. The fraction of respondents in the top income tercile who expect gasoline to be unaffordable in five years is only 5%, while this fraction averaged around 15% for respondents in the middle and bottom income terciles (Figure 17). The fraction of homeowners in the top property value tercile who expect gasoline price to become unaffordable is also approximately 5%, similar to the fraction of the high income tercile. About 20% of homeowners in the bottom property value tercile anticipate that gasoline will be unaffordable in five years (Figure 18). Approximately 15% of the respondents who professed to know a lot or a fair amount about energy expect gasoline price to be unaffordable in five years, 17 while the corresponding share of those professing to know little or nothing about energy was less than 10% (Figure 19). 18 Conserving energy for cost reasons To explore this topic, consumers were asked: How often do you reduce the energy you use for your home or vehicle for cost reasons -always, often, sometimes, or never? We found that about half of the population reduces energy use for cost reasons always or often. Significant variation was found when cross-tabulating these responses by income and selfreported knowledge of energy. The share of consumers in the bottom income tercile who say they always or often reduce their energy use for reasons of cost was 59%, significantly higher than the 38% share of those in the top income tercile (Figure 20). Of the respondents who stated that they knew a lot or a fair amount about energy, about 55% also stated that they reduce energy use for cost reasons always or often. This compares to 42% among those professing to know little or nothing about energy. No significant variation in this pattern was found by region or by home status. Additionally, we cross-tabulated self-reported energy conservation against the degree of home energy bill increase that would be considered unaffordable. Respondents who said that they always or often reduce their energy use for cost reasons believe that their energy bills would have to increase by about 140% before they become unaffordable. This degree of increase is 19 significantly smaller than that found for respondents who said that they never or only sometimes reduce their energy consumption for cost reasons; for this latter group, it would take an average increase of about 200% before home energy bills were seen as unaffordable. Views of Energy and the Environment To probe attitudes about the effect of energy use on the environment, we asked a sequence of questions designed to elicit consumers' beliefs without leading the respondents to focus on particular forms, sources or characterizations of energy (such as names of fuels or adjectives like "clean," "fossil" or "renewable"). The focus groups had shown us that although lay respondents readily understand that energy use impacts the environment, their knowledge of the mechanisms involved is very limited and their beliefs about how various forms of energy affect different aspects of the environment can differ greatly from expert knowledge of the subject. Therefore, we asked an open-ended question rather than reading a list of energy sources or otherwise portraying specific forms of energy when exploring this topic. A sequence of three questions was used to elicit the overall degree of concern about the impact of energy on the environment, the nature of the concern, and the source or form of energy that respondents believe has the greatest impact. The first question was: Thinking about all sources of energy people use in everyday life, to what extent would you say they affect the environment? Would you say a lot, a fair amount, a little, or not at all? Respondents were next asked: Which one of the following is affected the most by the energy people use in everyday life - air, water, global warming, or personal health? Then, given the individual's answer to this second question, they were asked: What particular source of energy would you say affects [the air / water / global warming / personal health / (other)] the most? If in replying to the second question the respondent gave an answer other than air, water, global warming or personal health, then the respondent's answer was used in the third question. The responses to the first of these three questions showed that a clear majority, averaging about three-quarters of the population, believe energy affects the environment by at least a fair 20 amount. No significant differences in this fraction were observed across regions, income bracket, home status and self-reported knowledge of energy. The environmental impact categories -- air, water, global warming and personal health -listed in the second question were a set that had tested well in the focus groups and cognitive interviews. In the focus groups, respondents described energy impacts on personal health but did not necessarily link them to air or water pollution, and so we did not assume such a cognitive linkage in our questions even though expert knowledge would hold that health impacts can be caused by air or water pollution. Approximately 42% of respondents stated that air is the aspect of the environment most affected by energy use (Figure 21). Global warming was the second most common choice, stated by 24% of respondents. Roughly 16% of respondents view water and personal health as being most affected. We found significant differences when examining these responses by income (Figure 22). Air clearly stands out as the impact of concern for the top and middle income terciles while in the bottom income tercile all four types of impact evidence a similar level of concern. We did not find any significant differences when analyzing responses to this question by region, by home status or by self-stated knowledge of energy. 21 Sources of energy believed to most affect the environment When recording answers to the question about which source of energy the respondents believe affects the environment the most, interviewers coded the replies into broad but not necessarily mutually exclusive categories. If a respondent gave more than one answer, the interviewer asked them to pick the one that they thought had the greatest effect on the environmental impact that they had identified when replying to the second question in this series. If a respondent said “gas,” the interviewer clarified whether they meant gasoline or natural gas. The percentages of each type of response over the wide variety of answers received are listed here in Table 1. Petroleum-related forms of energy such as gasoline and oil itself were those most frequently identified as affecting the environment the most, corresponding to the answers given by about 36% of the respondents. Coal was second with approximately 18% of the responses followed by electricity with roughly 13% among the forms of energy frequently identified. "Fossil fuels" (so termed by the respondents) ranked fourth (7%) followed by natural gas (5%). When respondents answered "fossil fuels," interviewers did not try to clarify which particular fossil fuel the respondents had in mind. 22 Table 1. Responses to the question asking respondents which source of energy they believe affects the environment the most Responses percent Responses (cont'd) percent Gasoline, oil, petroleum 36 People, cities 1.5 Coal 18 Automobile exhaust 1.5 Electricity 13 Pollution 1.3 Fossil fuels 7 Fracking 0.7 Natural gas 5 Power plants 0.7 Emissions 5 All 0.5 Water pollution 1.9 Chemical 0.3 Factories 1.8 None 0.3 Biofuels, wood 1.6 Hydroelectric 0.2 Nuclear 1.6 Don’t know / no answer 4 As shown in Figure 23, combining the generic fossil fuel response with the responses that mentioned particular fossil fuels such as petroleum, coal or natural gas, plus the separate 0.7% of responses that mentioned fracking, yields a subtotal amounting to roughly two-thirds of the population. A category combining oil and gas (petroleum-related fuels plus natural gas) would subtotal to roughly 42% of all responses. 23 When analyzing the responses according to self-reported household income, fossil fuels emerge as the form of energy believed to have the greatest environmental impact by a larger number of respondents in the top tercile than in the bottom tercile (Figure 24). Conversely, respondents from the bottom income tercile were somewhat more likely to view electricity as the form of energy that most affects the environment. No significant variation in the responses to this question were found across regions, by home tenure and property value or by self-reported knowledge of energy. Expectations regarding the impact of energy on the environment The next question asked respondents their view on how the environmental impacts of energy use might change in the future: Thinking about the next five years, do you think the energy people use in everyday life will affect the environment more, affect the environment less, or will the environment impact of energy stay about the same? Approximately 86% believe that environmental impacts due to energy will either increase or remain the same. This pattern remained consistent across regions, by status, by income and by self-reported knowledge on energy. 24 A final question in the set addressing energy and the environment explored the extent to which consumers reduce energy consumption for environmental reasons. The question was: How often do you reduce the energy you use for your home or vehicle for environmental reasons -- always, often, sometimes, or never? Roughly 44% of the population claims to always or often reduce their energy use for environmental reasons. The responses to this question did vary according to self-reported knowledge of energy. The share of the energy-knowledgeable respondents (those who claim to know at least a fair amount about energy) that reduces their energy use always or often is significantly larger than the share of the population does so in the less energy-knowledgeable group (Figure 25). No significant variation in responses to this question was found across regions, by home tenure and property value or by income. Conclusion The first data set produced by the U-M Energy Survey offers intriguing findings about American consumers' attitudes on energy. Among the broad range of issues that span public discussions of energy, we identified three topics -- reliability, affordability and impact on the environment -that the lay population grasped well enough to provide meaningful responses. Multiple facets of 25 consumer attitudes on these topics were explored within the confines of the relatively short interview time (which averaged about 7½ minutes during the October 2013 fielding) available for the energy rider as appended to the SCA. A few notable findings stand out for their statistical significance even with this initial 502 household sample. In probing views on the affordability of energy, one set of questions elicited the cost levels that consumers would see as unaffordable in the sense of requiring their household to make significant changes in their lives. On average, respondents expressed a much greater relative tolerance for an increase in home energy bills than for an increase in gasoline prices. Mindful that such results would be sensitive to market conditions as perceived when the survey was performed (October 2013), we found that home energy bills would have to rise by an average of 170% while gasoline prices would have to rise by but 85% before being considered unaffordable. The survey also asked about price expectations and found respondents believing that, over the next five years, home energy bills would rise by 30% on average in contrast with an average rise of 15% expected for gasoline prices. Another notable finding is that American consumers are just as concerned about the environmental impacts of energy as they are about its affordability. Roughly 60% of respondents said that they personally worry about the impact of energy on the environment a great deal or a fair amount, a fraction not statistically different from the 55% who reported similar levels of concern about the affordability of energy. Significantly fewer respondents expressed such levels of concern about reliability. Looking ahead as new U-M Energy Survey data accumulate, future analyses will reveal which of these initial findings stand the test of time. The growing sample size is likely to reveal a greater number of significant findings based on classification variables such as region, income and other household characteristics. As time goes on, the survey will enable us to observe trends in consumer attitudes about the topics covered and how attitudes shift in reaction to changes in energy markets and the events that inevitably impact the energy sphere from time to time. Thus, by providing objective information on public concerns about energy and what individuals think is important (or not so important), the U-M Energy Survey will generate new insights about consumer beliefs and values as they pertain to energy. Such results will serve the broad public interest and provide an valuable resource that will enable private businesses, policy makers and other stakeholders to better address energy-related challenges in the years ahead. 26 ACKNOWLEDGMENTS The authors are grateful to many individuals for their advice, support and feedback throughout the two-year process leading to the launch of the U-M Energy Survey. The project was made possible with financial support and backing from the University of Michigan Energy Institute and its director, Mark Barteau, and the Institute for Social Research and its director, James Jackson. Paul Gargaro, formerly of UMEI, helped conceptualize the project and initiate the planning. David Lampe of the University of Michigan Office of Research was an early supporter and provided valuable advice throughout the process. Jim Lepkowski of ISR's Program in Survey Methodology helped guide the planning and he and his students Yumi Kim and Yuan Zhang carried out the review of the literature and previous energy-related surveys. Gabriel Tabak provided valuable assistance during the questionnaire development and focus group testing. Richard Curtin's input was crucial from the beginning and he, Rebecca McBee and others on the U-M Surveys of Consumers staff were critical contributors in the final development and testing of the survey questionnaire; Rebecca also went beyond the call of duty by performing the initial analyses of the October 2013 data. Jennifer Puckett of ISR provided outstanding administrative support throughout the process. The U-M Energy Survey team is also grateful to the project's faculty advisors and other individuals who provided input, reviewed the development and early results and otherwise assisted at various stages of the effort. 27 Appendix A. QUESTIONNAIRE DEVELOPMENT In spring and summer 2012, existing surveys asking about energy issues were reviewed and information about specific topics, sponsors and mode of data collection were collected. A question bank with more than 250 questions was compiled as a result of this systematic search. It was found that most existing energy-related surveys were developed either to investigate specific aspects of energy, views of particular energy options and strategies, or in response to energy issues of the day. Some did, however, include general attitudinal questions that were useful for informing survey design. Based on the review and input from U-M Energy Institute staff and faculty affiliates, a straw list of survey topics that targeted general consumer attitudes about energy, independently of specific sources of energy or particular energy market developments, was developed. The topics were then tested in three focus groups conducted at ISR in March and April 2013 (see the summary of the focus groups given below). Based on the focus group findings, a first draft questionnaire was developed, consisting of previously administered questions from the question bank and newly developed questions. All questions were designed to be general enough to allow the tracking of trends over time but specific enough for respondents to answer. The draft questionnaire was then pretested and revised in an iterative process. First, two rounds of cognitive interviews were conducted with 24 individuals in May and June 2013 at ISR. Cognitive interviews include the “administration of draft survey questions while collecting additional verbal information about the survey responses, which is used to evaluate the quality of the response or to help determine whether the question is generating the information that its author intends.”3 After each round of cognitive interviews, the questionnaire was revised based on the problems respondents had with comprehending, mentally processing and responding to the proposed survey questions. Next, the questionnaire underwent further pretesting through expert review by SCA staff and experienced Survey Research Center (SRC) interviews. Following the expert review process, 3 Beatty, P.C. & Willis, G.B. 2007. Research synthesis: the practice of cognitive interviewing. Public Opinion Quarterly, 71: 287-311. 28 the questionnaire was pretested over the phone by the ISR Survey Research Center in early September 2013. Pre-testing resulted in a need to limit the number of questions to fit within the desired seven-minute time allotment added to the SCA interview. Based on the pre-test, the energy survey questionnaire was revised into final form for data collection as part of the SCA in October 2013. The analysis reported here is derived from responses to the 18 questions (which are given in the main text) and we expect to use the same questions through at least the first full year of the U-M Energy Survey, administered in January, April, July and October 2014. B. FOCUS GROUP FINDINGS It is hazardous to design an instrument before listening to consumers.4 With this in mind, we conducted several focus group sessions in order to understand how the public thinks about energy issues and learn the language they use to express their thoughts. This part of the survey development process provides the background knowledge necessary to draft effective questions about energy issues that ordinary consumers could readily understand and answer. Focus groups are group interviews with a small number of participants led by a moderator, who conducts semi-structured discussions on a small number of topics of interest for the survey. We convened three groups with 5-9 participants each. In each focus group, we discussed issues related to the cost of energy, energy reliability, energy security, the economic impacts of energy, the environmental impacts of energy and energy conservation and efficiency. We recruited focus group participants through Craigslist ads, flyers posted in Ann Arbor and surrounding communities, and ads placed in local newspapers. We screened the volunteers so that each group had a balanced mix of individuals in terms of gender, age, residency and home rental or ownership, characteristics that are all likely to be related to individuals' knowledge and views of energy. Table A1 shows the dates and demographic breakdown of participants for the three groups, which were held in meeting rooms designed for focus group research at the Institute for Social Research, University of Michigan, Ann Arbor. The three focus group discussions lasted about two hours each and focus group members were paid an incentive of $35 for their participation. Focus group members were clearly knowledgeable and concerned about the cost of energy. The respondents proved able to track their energy costs and bills, report costs in dollar 4 Krueger, R.A., and M.A. Casey. 2000. Focus Groups: A Practical Guide for Applied Research. Thousand Oaks, CA: Sage Publications. 29 Table A1. Demographic Breakdown of Energy Survey Focus Group Participants Date Total Members Gender Ages Residency Home Tenure Focus Group 1 Focus Group 2 Focus Group 3 March 22, 2013 April 19, 2013 April 26, 2013 9 8 5 4 females 5 males 5 females 3 males 2 females 3 males 2 aged 30-45 6 aged 46-60 1 aged 61+ 1 aged 20-29 3 aged 30-45 3 aged 46-60 1 aged 61+ 1 aged 30-45 3 aged 46-60 1 aged 61+ 4 Ann Arbor; 4 other Washtenaw County 1 Metro Detroit 6 Ann Arbor 2 Oakland County 2 Ann Arbor; 1 other Washtenaw County 2 Wayne County 6 renters 3 owners 5 renters 3 owners 2 renters 3 owners amounts, project changes in costs and describe the cost levels that would require their household to make behavioral changes. They were similarly knowledgable about motor fuel costs, for which they could report the price of gasoline in cents, project changes in price and report the fuel price level that they felt would require behavioral changes. However, respondents could not answer questions about the prices of forms of energy used at home, such as electricity. We therefore decided to design questions that asked respondents the amount of their last energy bill rather than price of energy, as well as questions regarding the changes they expected in their home energy bills in five years, the cost at which their energy bills would become unaffordable in the sense of requiring their household to modify its behavior, and how they would change their behavior if energy bills reached such a level. Respondents also had a fairly strong understanding of energy reliability. However, they thought of reliability exclusively in terms of electricity and confused reliability with energy security when they discussed the reliability of gasoline. With this in mind, we included questions asking about reliability in terms of individual energy use and asked respondents for the source of energy that they had in mind when evaluating the reliability of the energy they use. The focus group results on energy security were much less promising. In general, respondents had great difficulty understanding the topic and discussing it consistently. In the first focus group, respondents did not understand what we meant by energy security and often did not 30 distinguish "security" from "reliability." When discussing the topic, it was clear that different respondents understood energy security in different ways and generally did not interpret it in the manner that the term is used by energy experts. For example, some viewed energy security as pertaining to power outages or energy reliability in the future, or to nuclear safety. When we provided a longer background discussion of energy security in the second and third focus groups, e.g., specifically asking respondents about the security of supply for specific sources of energy, we found that then they did relate it predominately to gasoline, oil and petroleum. They could rate levels of energy supply security and name causes of supply disruption, but had only vague ideas of the parties responsible for maintaining undisrupted supply. In light of respondents' struggles with the concept of energy security, and the likely need for lengthy explanations when framing interview questions on the topic, we decided to omit it from the survey. Respondents had even more trouble when it came to making the connection between energy issues and the economy. Respondents didn't even know where to start thinking about this topic and so we decided not to pursue it in the survey. Respondents also had some trouble with the relationship between energy and the environment. While they could articulate their concerns about the environment and clearly believed that energy production and use overall has an effect on the environment, respondents had difficulty clearly connecting their environmental concerns with specific energy issues or forms of energy. This lack of understanding regarding about linkages is consistent with their general lack of knowledge about the production and sources of the energy carriers (such as electricity, other home fuels and gasoline) they they directly use in their everyday lives. Although a few focus group members seemed to have specific knowledge or specific ideas about the issue, in general they demonstrated little knowledge of how energy is produced and delivered, or of the particular environmental impacts of energy production and delivery. Although all believed that energy had environmental impacts, focus group members had diverse views of what the impacts were. Some linked energy to air pollution and water pollution, but others linked energy use directly to personal health problems without mentioning the environmental medium (air or water) that might be involved. Focus group members were clearly aware of the concepts of global warming and climate change but did not have a consistent understanding of what those terms meant. Some saw global warming as a rise in temperatures whereas some saw climate change as meaning a temperatures change in either direction, for 31 example. Some saw the two terms as interchangeable and others believed that global warming causes climate change. Due to the general lack of specific understanding of energy and its environmental impacts, we developed questions that asked respondents only for general impressions of the impact of energy on the environment rather than developing questions that might attempt to explore views on the specific impacts of particular forms of energy. Respondents also took issue with energy conservation and efficiency. While they could provide correct examples for energy conservation and efficiency, they were confused when asked to compare the two terms and did not make clear distinctions between the two. Taking these findings into account, we decided to eschew the use of these terms and instead simply ask respondents how often they reduce their energy usage. Overall, respondents demonstrated only limited knowledge of energy. They did not know where it comes from, how it is delivered, or exactly how it effects the environment. They did not know a priori what was meant by energy security and could not relate what they knew of energy to economic issues. A good number held incorrect beliefs, such as that oil is a common source of the energy they use at home (such as electricity). Focus group members had little consistent understanding of what is meant by "traditional," "renewable" or "sustainable" forms of energy. Respondents were quite able to talk about their direct experiences with energy, such as plugging into a power outlet or their use of gasoline and heating fuels. We decided to tap into these direct experiences in the design of the survey. C. SAMPLE DESIGN AND MARGINS OF ERROR As a rider on the University of Michigan Surveys of Consumers, the U-M Energy Survey inherits that established survey's sample design and statistical characteristics. The Surveys of Consumers (SCA as it is often abbreviated, based on an earlier name Survey of Consumer Attitudes) is an ongoing survey conducted monthly through telephone interviews of approximately 500 adult men and women living in households in the coterminous United States (48 states plus the District of Columbia). The sample is nationally representative and gives each household in the coterminous U.S. an equal probability of being selected.5 The SCA is designed to maximize the detection of changes in attitudes by using a rotating panel sample design within the ongoing monthly survey program. For each monthly sample, an independent cross-sectional sample of 5 Curtin, R.T. 2002. Surveys of Consumers Sample Design. Ann Arbor: University of Michigan Survey Research Center. http://www.sca.isr.umich.edu/fetchdoc.php?docid=24773 32 households is drawn and a subset of the respondents is reinterviewed six months later. As a quarterly rider on the SCA the U-M Energy Survey shares the rotating panel feature. In general, sampling errors for estimates based on SCA questions depend on the number of sample cases on which a particular estimate is based and a variety of other sample design effects. Therefore, no single margin of error or differences in percentages that represent statistical significance can be given that is applicable to the results of all questions and analyses. For the single month (October 2013) sample of 502 households as reported here, the margin of error is ±5 percentage points for estimates based on the full sample and response frequency percentages near 50%; the margin of error can be smaller for frequencies much greater or much less than 50%. Margin of error is defined here as representing a 95% confidence interval. The error bars shown on the charts in this report represent 95% confidence intervals for the specific estimates shown and all differences reported as significant were evaluated using tests with a similar level of confidence. 33