Turning Personal Experience into Political Attitudes: The Effect of Local Weather on Americans’ Perceptions about Global Warming Patrick J. Egan Assistant Professor of Politics and Public Policy New York University patrick.egan@nyu.edu Megan Mullin Assistant Professor of Political Science Temple University mmullin@temple.edu this version: September 2010 We are grateful to Kevin Arceneaux, Larry Bartels, Adam Berinsky, Dimitri Landa, Robert Shapiro, Christopher Wlezien, John Zaller, and seminar participants at Harvard University, New York University, the 2010 EITM Summer Institute at UC Berkeley, the 2009 Midwest Political Science Association meetings, and the 2010 Visions in Methodology Conference for comments on earlier versions. ABSTRACT How do people translate their personal experiences into political attitudes? It can be difficult to explore this question using observational data, because individuals are typically exposed to experiences in a selective fashion, and self‐reports of exposure may be biased and unreliable. In this study, we identify one experience to which Americans are exposed nearly at random—their local weather—and show that weather patterns have a significant effect on people’s beliefs about the evidence for global warming. Weather has an effect as large as those associated with covariates that are standard predictors of political attitudes, including age, education, and race. Furthermore, the effect is strongest among those at mid‐levels of political awareness. Our results reinforce the importance of political sophistication in shaping how individuals respond politically to personal experience, even when an experience has only limited political relevance. 2
Daily life includes many personal experiences that are, on their face, directly relevant to the development of political attitudes. On issues such as crime, the economy, education, health care, public infrastructure, and taxation, large shares of the public are exposed to experiences that could logically be linked to attitude formation. But findings from research examining how these experiences affect opinion have been mixed. Although direct experience—whether it be as a victim of crime, a worker who has lost a job or health insurance, or a parent with children in public schools—can influence attitudes, the impact of these experiences tends to be weak or non‐existent after accounting for typical predictors such as party identification and liberal‐conservative ideology (Citrin and Green 1990; Lau and Heldman 2009; Sears and Funk 1991). Much more powerful in shaping attitudes is the flood of information delivered to us through the tools of mass communication, which convey messages from political elites, experts and other sources, with individual levels of political awareness often playing a key role in moderating the effects of these messages on attitudes (Zaller 1992). Does political awareness play a similar role in moderating the impact of personal experience on attitudes? On the one hand, we might expect political awareness to dampen the effects of personal experience, as the politically aware may be swayed more by elite messages than by their day‐to‐day experiences (Mutz 1992; Weatherford 1983). On the other hand, there is reason to think that political awareness can augment the effect of personal experience on attitude formation. Attention to debates about issues may lead us to view individual experience regarding these issues as political, rather than merely personal (Mutz 1994), and the politically sophisticated may be the only citizens with the cognitive ability to link their personal experience to public policy (Gomez and Wilson 2001). 3
It is difficult to test these competing predictions, because research about the effects of personal experience on attitude formation almost always suffers from two threats to inference. First, individuals are usually exposed to experiences—or in fact may select into these experiences—in non‐random ways that are correlated with the attitudes that may be dependent on the experiences. Second, most research relies on self‐reports of experience, which can be unreliable. Self‐reports can also be the consequences of specific attitudes about politics and public affairs, rather than bias‐
free measures of their causes. In this article, we explore how Americans translate their personal experiences into political attitudes with a natural experiment that avoids both threats to inference. The experience we analyze is fluctuation in local temperatures across the United States. The attitude it affects is Americans’ beliefs in the existence of global warming. Our research strategy is straightforward: we merge local weather records with geocoded survey data. Our approach avoids the first threat to inference because everyone, from the least to the most politically aware, personally experiences the weather. We avoid the second threat by using weather data recorded by scientific weather authorities rather than self‐reports of weather incidence. In short, we exploit the fact that changes in local temperature assign individuals nearly at random to a range of experiences that they may choose to interpret as evidence for or against global climate change. In addition to employing a strong research design, our study is also powerful because—unlike in the case of crime, job loss, or public education—scientific and many political elites consistently advise Americans not to rely on their personal experience of local temperature variation to evaluate arguments for or against the existence of global warming. Furthermore, different regions of the nation can experience quite dissimilar weather conditions simultaneously, meaning that any nationwide message will fail to resonate with every citizen’s personal experience. Therefore, the 4
present case is a particularly hard test of the argument that political awareness facilitates the politicization of personal experience. To the extent that such facilitation is shown to be at work here, we should become more confident that it also takes place in contexts where elites explicitly cite personal experiences of ordinary Americans. Our findings offer new insight into the influence of personal experience on political attitudes. Overall, we find a significant relationship between local weather and beliefs about global warming. For each three‐and‐a‐half degrees Fahrenheit that local temperature rises above normal, Americans become one percentage point more likely to agree that there is “solid evidence” that the earth is getting warmer. The size of the effect is substantial, larger for example than the comparable effects associated with race, age, or education. We also show that political awareness plays a role in moderating the effect of temperature on beliefs. Using several different measures of awareness, we identify a curvilinear effect in which those who are in the middle of the awareness spectrum are more likely to be affected by weather than either the very aware or those who are inattentive to politics. Our results suggest that complex relationships among experience, awareness, and attitudes have yet to be uncovered in other contexts where clean identifications of effects are more difficult to achieve. Personal experience and the formation of political attitudes A casual observer of politics could understandably surmise that among the factors that shape individual political attitudes, personal experience would be the most important. After all, such experience is directly accessible and requires little in the way of information search. One might therefore expect that personal experience would have a large impact on public opinion, particularly among those who are less interested, knowledgeable, or sophisticated about politics. In the realm of economic voting, this reasoning was the basis for early scholarly conjectures that the less 5
politically aware should “vote their pocketbooks”—that is, render a retrospective judgment on the incumbent party on the basis of their own financial situations (e.g., Campbell et al. 1960; Fiorina 1981). Scholars have similarly looked for other links between individuals’ political attitudes and their personal experiences, including such as experiences as losing a job (Sears et al. 1980), having close friends or family serving in Vietnam (Lau, Brown and Sears 1978), having children in public schools (Kinder and Sears 1981), or lacking health insurance (Lau and Heldman 2009). But research on the relationship between personal experience and political attitudes has determined that the ceteris paribus effect of experience is either weak or nonexistent (Citrin and Green 1990; Lau and Heldman 2009; Sears and Funk 1991; Sears et al. 1980). For example, Americans seem to cast their votes more on the basis of their assessment of the national economy than on their personal financial circumstances (Kinder and Kiewet 1979), and self‐interest rarely plays a substantial role in attitude formation except when the stakes are large and well‐defined (Sears and Citrin 1985). One noteworthy exception to this pattern is that recent victims of crime are more likely to perceive neighborhood crime as a serious problem (Marschall 2004) and to favor a law‐and‐order approach to crime prevention and punishment (Sears et al. 1980). In attempting to understand what really drives the formation of attitudes on political issues, we get much farther when we examine the transmission of messages about these issues by the nation’s elites. Research has focused on the process by which these messages are received and subsequently either integrated or rejected by citizens constructing their opinions. The prevailing theory—John Zaller’s “Receive‐Accept‐Sample” (or “RAS”) model—is that the messages’ rate of reception rises with political awareness and sophistication, but that the politically aware only accept messages that are congruent with their personal ideological predispositions (Zaller 1992). 6
The role of political awareness. The importance of political awareness to the development of political attitudes has led to exploration about whether awareness might also moderate the effect of personal experience on attitudes.1 If so, this would help explain why a factor that we would expect on its face to be a strong predictor of attitudes often fails to have much of an effect. Some scholars argue that political awareness should dampen the effects of personal experience. By this logic, the politically aware receive more elite messages and thus are more affected by them than by their day‐
to‐day experiences. The less attentive have little to rely upon but the material of their own lives as they fashion their political beliefs (Conover, Feldman and Knight 1986; Mutz 1992; Weatherford 1983). On the other hand, there is reason to think that political awareness augments the effects of personal experience on attitude formation. Exposure to news reports indicating that fellow citizens are having similar experiences (such as job loss or financial difficulties) can lead individuals to “contextualize” their experiences (Lane 1962) and thus view their challenges as political, not personal, problems (Mutz 1994). Less politically sophisticated people may lack the cognitive ability to link their personal experiences with the broader political landscape and therefore fail to bring experience to bear on attitude formation (Gomez and Wilson 2001). Finally, it is possible that either the dampening or facilitation expectations may be realized, depending on the extent to which the content of news coverage echoes individuals’ personal experiences (Mutz 1998). All of these approaches make the implicit assumption that any moderating effect of political sophistication is monotonic. It is our claim that the effects of certain experiences actually peak among those in the middle of the political awareness spectrum. The claim rests on our hypothesis that two conditions are necessary for the translation of personal experience into political attitudes: first, reception of messages from the political sphere that allow one to contextualize the experience; and second, an openness to attitude change found among those with less structured prior beliefs 1 We use the terms “political awareness,” “political sophistication,” and “political knowledge” interchangeably. Although the concepts are not identical, our argument holds for all three variations. 7
about public affairs. For these reasons, those at the low and high ends of the political awareness spectrum are relatively inoculated from the effects of personal experiences. Highly aware individuals are typically exposed to a variety of news media and political messages and thus could conceivably link those messages to their personal experiences, but their attitudes are already structured along standard ideological, partisan and demographic lines. The attitudes of low‐
awareness individuals are much less structured, allowing personal experience to exert a larger impact. However, they are unlikely to receive and understand messages that allow them to link their experience to national problems. It is those in the middle of the political awareness spectrum who are both attentive enough to receive political messages that contextualize their experience and open enough to attitude change that their beliefs that can be moved by the experience. Threats to inference. By necessity, studies of the effect of individual experience on political attitudes have relied on observational data drawn from sample surveys. This typically causes two problems. First, the use of observational, rather than experimental, data means that individuals are exposed to personal experiences in non‐random ways.2 Second, reliance on sample surveys requires use of survey respondents’ often inaccurate self‐reports of the experience. These problems make it difficult to identify cleanly any effect of experience on attitudes. Experiences such as losing a job, being the victim of a crime, or a change in one’s personal financial situation are distributed among the population in ways that are inevitably correlated with personal characteristics that themselves determine attitudes. Including controls for variables in the analysis is an imperfect solution (e.g., Achen 1987), and strong exogenous instruments for the experiences are often unavailable. Estimates of the effects of personal experience can thus be biased, often in an indeterminate direction. 2 Exceptions to this approach are the few studies that take advantage of experiences that are truly assigned at random, such as the Vietnam draft lottery (Bergan 2009; Erikson and Stoker 2010). 8
Respondents’ self‐reports of experience are subject not only to random measurement error, but in many cases to bias as well. Americans’ perceptions about the state of public affairs tend to be slanted in their party’s favor (Bartels 2002; Gerber and Huber 2009; Shapiro and Bloch‐Elkon 2008), particularly among the politically sophisticated (Shani 2009). Reports of politically relevant personal experiences also may be biased. Although some experiences, such as losing a job or being the victim of a crime, would appear to be relatively immune from either measurement error or biased reporting, many of the measures of personal experience upon which previous research has been based are not. Even a phenomenon as minor as survey item order can affect self‐reports of personal experience. For example, priming attention to politics can “politicize” personal experience by prompting respondents to report their experiences in a manner that rationalizes the political positions already stated (Lau, Sears, and Jessor 1990; Sears and Lau 1983). Evidence from survey experiments confirms that question order has a strong effect on retrospective personal financial assessments (Palmer and Duch 2000). Question wording also might influence how respondents characterize their experience. For example, much of the literature examining “pocketbook” voting relies upon the following survey item, which has been included on the American National Election Studies time series study since 1962: “We are interested in how people are getting along financially these days. Would you say that you and your family living here are better off or worse off financially than you were a year ago?" Given that this is precisely the sort of language used by political candidates to defend their positions, perceptions of one’s own economic circumstances can be more strongly influenced by partisanship than the reverse (Evans and Andersen 2006). Thus the literature on the effect of personal experience on political attitudes leaves several questions on the table. What would such an effect look like, if it could be analyzed in a context free of selection effects and without the bias and errors arising from self‐reports of the experience? And who on the spectrum of political awareness would be most moved by the experience? The present 9
study allows us to answer these questions in a more satisfactory way than permitted by previous work. Public Opinion on the Existence of Global Warming In the few studies that have examined opinion about the existence of global warming, evidence suggests that people’s values and political predispositions have a bigger impact than factual information on judgments about the nature and extent of the problem.3 Age, liberal ideology, general pro‐environment attitudes, and being nonreligious are associated with existence beliefs, while the effect of scientific knowledge about the causes and consequences of climate change is weak and inconsistent (Bord, O’Connor and Fisher 2000; Kellstedt, Zahran and Vedlitz 2008; but see Curry, Ansolabehere and Herzog 2007).4 People’s perceptions of the threat posed by global warming appear to be guided more by the polarized discourse of political elites than by the scientific consensus affirming that climate change is taking place. We hypothesize that an additional factor—the personal experience of local temperature variation—
has an independent effect on attitudes about the existence of global warming. Only a handful of studies have examined this relationship, and none have done so with either the comprehensive approach or theoretical line of inquiry undertaken here. In lab experiments, the experience of heat, regardless of whether it takes place indoors or outside, has been shown to make people more likely to say that global warming is a proven fact (Risen and Critcher 2010). Using evidence from a national telephone survey, Krosnick et al. (2006) find that respondents who perceive that local 3 Other research on climate change opinion has focused on explaining risk perceptions (Brody et al. 2007; Kellstedt, Zahran and Vedlitz 2008; Krosnick et al. 2006; Leiserowitz 2006) or support for government policies and voluntary action to mitigate the problem (Curry, Ansolabehere and Herzog 2007; Lubell, Zahran and Vedlitz 2007; O’Connor et al. 2002; O’Connor, Bord and Fisher 1999). 4 Inconsistent results about the effects of scientific knowledge may be attributable to research design, as these studies rely on either self‐reports about a respondent’s level of knowledge or performance on factual information items, both of which may be effects rather than causes of interest in and attitudes about the climate change problem. 10
temperature has increased in recent years are more likely to believe that global warming will occur in the future if nothing is done to stop it. Data from their survey of Ohio residents showed that this relationship was particularly strong among respondents with low levels of education and high trust in scientists. But without an objective measure of temperature conditions, the authors could not rule out the possibility that their results were attributable to projection, and that people who already believe that a global warming threat exists may perceive associated changes in weather. Brody et al. (2007) geocoded the location of respondents in a 2004 survey to measure the effect of actual climate change vulnerability on respondents’ perceptions of risk. They found that vulnerability to floods, sea level rise, and other natural hazards had an effect on risk perceptions, but a measure of long‐term rising temperature in the respondent’s region did not. Given the modest long‐term change in temperatures experienced thus far in the United States and the modest cross‐sectional variation in that change, we would not expect to see a measurable impact on Americans’ attitudes. We look instead at the much more dramatic short‐term fluctuations in temperature that individuals experience as part of their daily weather and ask whether that immediate experience affects perceptions of the global problem. Our approach represents a substantial advancement over previous work in that we use surveys of nationally representative samples of Americans, we employ official weather records instead of self‐reports of the experience of weather, and we hypothesize and test the theory that individuals’ reactions to this kind of personal experience will vary with their levels of political awareness. Overall, we expect to see a positive association between a respondent’s recent experience with high temperatures locally and her belief that the global climate is warming. But we expect weather’s effect on beliefs to be heterogeneous, with political awareness moderating the extent to which people rely on their experience with temperature in forming their beliefs about global warming. Those who are the least politically aware hear little about global warming and are therefore 11
unlikely to draw a connection between unusually hot or cold weather and the issue of climate change. The most politically aware are quite likely to make the connection, but they have strong prior beliefs about the issue that curtail the effect of personal experience on their attitudes. It is those in the middle of the political awareness spectrum who are both engaged enough to learn about the phenomenon of global warming and to hold weak enough prior beliefs about the issue that their attitudes can actually be changed. Data Opinion on climate change. Our data on Americans’ attitudes about climate change come from five national surveys sponsored by the Pew Research Center in June, July, and August 2006, January 2007, and April 2008. The surveys were conducted by telephone with a representative national sample of American adults living in the contiguous 48 states. In each survey, respondents were asked the following question: “From what you’ve read and heard, is there solid evidence that the average temperature on earth has been getting warmer over the past few decades, or not?” Americans’ responses to this question were largely in the affirmative: on average over the five surveys, 74 percent of respondents agreed that the earth is getting warmer. Local weather. Our goal was to measure the short‐term weather conditions that respondents might call to mind when asked whether evidence supports the existence of global warming. The variable we used to capture these conditions is the departure from normal daily local temperature, measured in degrees Fahrenheit, averaged over the week prior to the date on which the respondent was interviewed by Pew.5 That is, we calculated 5 We ran the analyses here with alternate versions of this variable, including departures from normal local temperature on the day of interview, the day before the interview, and the two weeks and three weeks prior to interview. All produced similar results to the weekly measure used here. 12
ddt _week i
1 7
∑tempij − normaltempij , 7 j =1
where ddt_weeki is the local temperature variation experience by respondent i, tempij is the respondent i’s local temperature j days before his or her interview, and normaltempij is the normal local temperature for that date, calculated over the period 1971‐2000. Data on local weather conditions (both past and present) come from the National Climatic Data Center (NCDC), part of the National Oceanic and Atmospheric Administration in the U.S. Department of Commerce. The NCDC compiles meteorological data over time from weather stations worldwide in order to produce climatic profiles of locations or regions. Zip code information in the Pew dataset allowed us to attach weather data to individual respondents. Pew asked each respondent to supply his or her five‐digit zip code, and almost all did so (97.5 percent). Of these, we were able to match nearly all (95.6 percent) to a Zip Code Tabulation Area (ZCTA), a geographic entity developed by the U.S. Census Bureau to provide a spatial representation of U.S. Postal Service zip codes.6 By overlaying a map of 448 U.S. weather stations onto a map representing ZCTA boundaries, we identified the weather station closest to each respondent’s zip code and used temperature data from that station. The mean distance from a ZCTA to its associated weather station is 0.43 degrees in latitude/longitude, or approximately 25‐30 miles.7 6 U.S. Postal Service zip codes are a way of classifying linear features such as street segments, address ranges, and delivery points to facilitate mail delivery. The Census Bureau created ZCTAs for the first time in the 2000 Census to provide an areal representation of zip codes and allow calculation of Census data at the zip code level. The distinct origins and data types of zip codes and ZCTAs result in some degree of spatial mismatch. As a robustness check for those respondents whose reported zip codes we were unable to match to the NCDC dataset, we also calculated the mean daily temperature readings for ZIP3 ZCTAs, a larger geographic unit that entirely incorporates all five‐digit ZCTAs beginning with the three‐digit prefix. Most of the respondents we could not match to a five‐digit zip code (89.0 percent) successfully matched to a ZIP3 ZCTA based on the first three digits of the zip codes they supplied to Pew. We repeated the analyses reported here with a dataset that included these respondents. As expected, the results were substantively similar but not as strong due to the less precise nature of the additional weather data. 7 For the April 2008 survey wave, weather data availability required that we merge respondents to a subset of the weather stations (416 out of 448), but the mean distance to the nearest weather station is essentially unchanged (0.44 degrees). 13
The temperature departure measure captures variation in local weather conditions both cross‐
sectionally and over time. Weather patterns exhibited a fair amount of variance while Pew’s surveys were in the field. The median value of the departure from historical temperatures for respondents in our dataset was 2.9 degrees Fahrenheit; the 5th and 95th percentiles of this variable were ‐4.3 and 14.7 degrees, respectively. We considered the possibility that rather than being affected by absolute deviations in local temperature, individuals might instead respond to change relative to the typical amount of volatility in their local weather.8 To account for potential differences in expectation about weather patterns, we also collected data on the standard deviation in average monthly temperature over the period 1971‐2000.9 Dividing the ddt_week measure by the local standard deviation produced an indicator of temperature departure measured in standard deviations. Similar results were obtained when temperature was measured in this fashion. The exogeneity of local weather to political attitudes. We have good theoretical reasons to believe that—conditional on geographic and temporal weather patterns—Pew’s respondents can be treated “as if” they had been randomly assigned to short‐term weather variation (Dunning 2008). The exogeneity of weather variation to social processes has long made it a favored instrument in quasi‐experimental work in the social sciences. Two types of tests provide confidence that our estimates of weather’s effect on global warming beliefs are unbiased by the omission of unobserved confounders. First, we found no discernable relationship between weather and observable covariates that previous research has determined are associated with attitudes on global warming (age, sex, race or ethnicity, attendance of religious services, education, party 8 We thought it possible, for example, that respondents who live in the upper Midwest, where temperature variability is highest, would be less responsive to a given level of departure from normal temperature than respondents who live in climates with more predictable weather patterns. In other words, five‐degree departure from normal on a January day may be more noticeable in Arizona than in Minnesota. 9 Data on standard deviation in temperature are not available from weather stations. Instead, we used data from 344 climate divisions, geographic entities developed by the NCDC that represent areas with fairly uniform climate conditions. We assigned each ZCTA to the climate division in which a majority of its land area lay. 14
identification, and ideology) after accounting for geography and time.10 Second, if the observed relationship between weather and global warming beliefs were spurious and attributable to unobserved covariates, we would expect to see similar relationships hold between weather and other political attitudes. We do not: placebo test estimates of the association between local weather patterns and (1) respondents’ assessments of the decision to invade Iraq and (2) their approval of George W. Bush’s job as president confirmed that the relationships between local weather and these variables were insignificant and operated in ideologically inconsistent directions.11 The Effect of Local Weather on Beliefs about Global Warming The plots in Figure 1 display the simple bivariate relationship between temperature and beliefs about the evidence for global warming. To construct the figure, we divided the cases in our dataset into 100 centiles on the basis of the ddt_week variable and calculated the percentage of respondents in each centile who agreed that there is solid evidence for global warming. In the figure, the mean of ddt_week within each of the centiles is plotted on the x‐axis and the percent of respondents in the centile believing in global warming’s existence is plotted on the y‐axis. We summarize the relationship both parametrically and non‐parametrically. The left‐hand plot includes a line that traces the predicted values of opinion obtained from a bivariate probit estimation on the individual‐
level data. The right‐hand plot traces the aggregate relationship with a nonparametric smoother drawn with the lowess technique (Cleveland 1993). Both plots include 95‐percent confidence bounds constructed using bootstrapped resamples of the data. The figure shows a significant and substantial relationship between the two variables: as local temperatures rise above normal, so 10 Wave‐by‐wave Wald tests comparing a model including only state‐of‐residence fixed effects with a model that also incorporated the individual covariates generally failed to reject the null that the covariates were jointly insignificant predictors of ddt_week. 11 In ordered probit models with state and survey‐wave fixed effects and controls for a set of demographic and political variables, warmer weather is associated with negative attitudes regarding the Iraq War (p = .42) but with positive attitudes about George W. Bush’s performance as president (p = .06). 15
does the percentage of Americans believing that global warming is a reality. The smoother indicates that the relationship between weather and global warming opinion is stronger in extreme weather and holds for both unusually hot and cold temperatures. To rule out the possibility of a spurious relationship, we control for variables that may be correlated with both geography and opinion with a series of ordered probit regressions in which respondents’ assessments of the evidence for global warming are the dependent variable.12 In Table 1, we model the simple bivariate relationship between weather and attitudes, then add fixed effects for wave of survey and state of residence, and finally estimate a fully saturated model that includes a block of demographic and political covariates. The estimated effect of weather on beliefs about climate change is statistically significant across all specifications, and its size remains unchanged with the introduction of control variables. Our estimates from the fully saturated model indicate that a shift in local temperature from the 5th to 95th percentile (that is, from 4.3°F below normal to 14.7°F above normal) increases the probability that Americans believe there is solid evidence for global warming by 5.3 percentage points, an effect that is significant at the 95 percent level of confidence.13 The size of this effect is substantial—both when considered alongside the effects of other covariates and when compared to other effects found in the literature on attitude formation. Figure 2 12 The dependent variable is scored 1 for “no” responses, 2 if the respondent volunteers that the evidence is mixed or they don’t know, and 3 for “yes.” Results are robust to treating the dependent variable as dichotomous. 13 Throughout this paper, we estimate effects of variables with a statistic called the “average predictive comparison” (Gelman and Hill 2007: 101‐103). This statistic is calculated by generating predicted probabilities for each observation in the dataset at two values of the variable of interest (for example in the case of temperature, 4.3°F below normal and 14.7°F above normal) while holding all other variables and their effects constant at their actual values. The difference in these two predicted probabilities is known as the “predictive difference” for any given observation, and the statistic reported is the average predictive difference across all observations. Because they take into account how all variables in the model covary in a dataset, average predictive comparisons are more representative estimates of the size of a variable’s effect than those generated by the typical approach of holding all variables constant at their medians or other values. We use bootstrapping techniques to generate reported confidence bounds. 16
provides a sense of the relative size of the effects of local temperature compared to other variables associated with opinion on climate change. It is a graph depicting the estimated effects associated with individual demographic, political and other variables—that is, the change in probability of believing in global warming accompanying a shift from the minimum to maximum value of that variable. The figure shows that the effect of temperature is substantial compared to other variables typically considered important in shaping Americans’ attitudes on the environment. A shift in local temperature from the 5th to 95th percentile results in opinion change that is substantially larger than the ceteris paribus differences in opinion between whites and blacks, between those belonging to the youngest and oldest age groups, and between the least and most educated. The estimated five‐point shift in opinion associated with weather variation is notable compared to the size of other shifts in attitudes that have been found noteworthy by political scientists, including the change in vote intention associated with a sharp boost in presidential campaign advertising (Shaw 1999); the increase in Americans’ approval of the president’s handling of the economy in the wake of a burst of positive economic news coverage (DeBoef and Kellstedt 2004); and shifts in opinion on public affairs following broadcasts of relevant TV news stories (Page, Shapiro and Dempsey 1987). The Moderating Effects of Political Awareness We now turn to tests of our hypothesis about how the effects of temperature on attitudes about global warming should be moderated by individuals’ levels of political awareness. Unfortunately, Pew did not ask factual information questions of the kind most favored by political scientists as measures of political awareness, nor did Pew ask its interviewers to assess respondents’ political knowledge. Our solution is therefore to employ three other proxies for political awareness used widely in the literature on the formation of political attitudes: (1) self‐reported attention to the news, (2) opinionation (i.e., respondents’ willingness to offer directional responses to survey questions about public affairs), and (3) strength of partisanship. None of these variables directly 17
measures the concept of awareness or sophistication, but evidence in existing literature demonstrates their strong association with those concepts and supports use of these measures as proxy variables. News attention. Self‐reported attention to the news has consistently been shown to be associated with higher levels of political knowledge (Bennett 1989; Delli Carpini and Keeter 1996; Hamill, Lodge, and Blake 1985; Luskin 1990), and it has been used in previous research as a proxy for knowledge (Brady and Sniderman 1985).14 We use a measure of news attention derived from batteries of questions Pew included in three of the five survey waves. In each survey, respondents were provided a list of five or six stories covered by news organizations over the previous month, and asked how closely they had followed each news item. Exploratory factor analyses found that responses to these questions all loaded on a single factor in each wave of the survey, and that scale reliability coefficients were high (Cronbach’s alphas ranged from .70 to .80). We therefore used responses to these questions to create an index, which we then converted into percentiles for comparability across surveys. We pooled our respondents and divided them into terciles of low, medium and high news attention.15 Opinionation. Opinionation refers to a respondent’s willingness to offer directional responses to attitudinal survey questions rather than choose the “don’t know” option. Saying “don’t know” is a form of satisficing, rather than optimizing, in survey response. It is more likely if a respondent has less practice thinking about the question topic, does not have pre‐formulated opinions, or finds the topic unimportant (Krosnick 1999). Previous research has demonstrated that those who have less 14 Although some scholars have expressed concern that self‐reported media attention measures exposure to the news rather than reception of the messages delivered in the news, recent evidence suggests that self‐
reported attention measures both exposure and processing of news content (Slater, Goodall, and Hayes 2009). 15 See the Appendix for details about construction of these scales and their validity. 18
factual knowledge about politics offer a “don’t know” response more frequently (Delli Carpini and Keeter 1996; Faulkenberry and Mason 1978; Krosnick 1999; Krosnick and Milburn 1990; Zaller 1991). Opinionation is also associated with political interest, efficacy, partisan intensity, and exposure to the news media (Atkeson and Rapaport 2003; Francis and Busch 1975; Krosnick 1999). We constructed our opinionation measure by counting the number of times each respondent supplied a “refused” or “don’t know” response to any survey question having to do with public affairs. We excluded survey questions about global warming, and any items that might invoke social desirability effects making respondents reluctant to offer an opinion on sensitive topics such as race, gay rights, or immigration (Berinsky 1999). The number of questions appropriate for inclusion in the opinionation index varied by survey, from a low of 13 questions to a high of 47 questions. We assigned each respondent a score based on the number of questions to which the respondent answered “don’t know.” Scale reliability on this index was very high (Cronbach’s alphas ranged from .70 to .91). We converted these scores into percentiles for comparability across surveys and then divided our respondents into terciles of low, medium and high opinionation.16 Partisanship strength. The final variable we use to proxy for political awareness is a respondent’s self‐reported level of partisanship. Empirical research consistently shows that political engagement and information rises with the strength of partisanship (Bennett and Bennett 1993; Bennett 1994; Campbell et al. 1960; Converse 1964; Miller and Shanks 1996).17 Pew measures party identification with a standard branching format question in which respondents are first asked 16 A weakness of using opinionation to proxy for political awareness is the fact that some respondents are more willing to offer opinions than others, holding constant political knowledge. Women, racial minorities, and those with low education and income are more likely to offer a “don’t know” response (Atkeson and Rapaport 2003; Francis and Busch 1975). We minimize the effects of this bias on our estimation by including standard demographic controls in our models and running separate analyses for each tercile of opinionation. 17 This relationship may exist because those who are more politically aware are more likely to form strong party attachments; alternatively, party may serve as an important reference group that guides and constrains issue attitudes (Jacoby 1988); or it may be that factors in place early in life such as interest, cognitive ability, and parents’ political engagement contribute to both knowledge and strength of partisanship (Highton 2009; Luskin 1990). 19
whether they are a Republican, Democrat or Independent. Those who respond “Independent” or volunteer another party, no preference, or don’t know and those who refuse to answer the question altogether are then asked, “As of today do you lean more to the Republican Party or the Democratic Party?” We group those who respond “Republican” or “Democrat” to the first question into the category “partisans.” Those who respond with a party affiliation to the second question we group into the category “leaning partisans.” Those who answer the second question by continuing to refuse to claim a party identification (a volunteered response, because participants are not explicitly offered this choice) we call “independents.” Education as a proxy for cognitive ability. One variable we do not consider to be an appropriate proxy for political awareness in the present context is educational attainment, even though studies typically find a strong relationship between political sophistication and education (Bennett 1989; Delli Carpini and Keeter 1996; Hamill, Lodge, and Blake 1985; Highton 2009; Jennings 1996). The difficulty here is that educational attainment is also correlated with the cognitive ability to evaluate evidence rather than rely upon heuristics (Chaiken 1980; Petty and Cacioppo 1986)—and short‐
term local weather is a rather poor piece of evidence for the existence of global climate change that we expect to be appealing as a heuristic to those with less cognitive ability.18 Therefore we can’t be sure if any relationship uncovered between education and the effect of weather on attitudes is due to the effect of political awareness or instead should be chalked up to differences in the ability to evaluate the quality of weather as a piece of information. To be sure that this ability to evaluate evidence does not confound our findings, we performed our analyses with an interaction between educational attainment and local weather and obtained very similar results to those shown here. 18 To be sure, short‐term weather is not a completely meaningless signal of global warming: scientists agree that one of the primary manifestations of climate change, both now and in the future, is a rise in global surface temperatures, with an increase in the probability of milder winters and more frequent, intense, and long‐
lasting heat waves (NASA 2009; Meehl et al. 2009; Trenberth et al. 2007). Nevertheless, any short‐term local weather pattern—even a spell of particularly hot or cold weather—is extremely low‐quality information compared to the large body of scientific evidence documenting climatic warming. 20
As shown in the Appendix, the effect of weather tended to be dampened among those with greater educational attainment, but our findings about the moderating effect of political awareness persist. This robustness check shows that the movement in attitudes caused by weather is more than merely a priming effect found only among the less educated. Even after accounting for this effect, the translation of the experience of weather into attitudes is shaped profoundly by the extent to which individuals are aware of and engaged in the world of public affairs. Validity check: 2004 American National Election Studies. As a validity check for our political awareness proxies, we turned to the most recent national survey for which data are available on standard political information items (the 2004 American National Election Studies) to examine the relationships among political knowledge, educational attainment, and our three proxy measures for political awareness. As shown in the Appendix, these measures—news attention, opinionation, and partisanship strength—are strong, statistically significant, and independent predictors of political knowledge, and they remain so after controlling for educational attainment, giving us additional confidence that they are good proxies for political awareness in our study. Results. We estimated separate models for each level of political awareness and replicated the analyses for each of our three proxies. Each model retains all covariates and fixed effects included in the fully saturated specification in Table 1, Model III. By estimating separate models for each level of political awareness, we account for the fact that associations between attitudes on public affairs and many demographic controls tend to vary by political awareness (Althaus 1998; Bartels 1996; Zaller 1991). Results are shown in Table 2, which displays the ordered probit coefficients associated with weather for each level of political awareness replicated over each of our three proxies. The results confirm our predictions: those in the middle of the political awareness spectrum (that is, those in the middle tercile of news attention, the middle tercile of opinionation, 21
and leaning partisans) are the only group whose beliefs about global warming are significantly affected by weather. Further confirmation of the validity of our proxies for political awareness is provided by the fact that attitudes on global warming generally become more structured as the analysis moves from the lowest to highest levels of news attention, opinionation, and strength of partisanship (as measured by the expected proportional reduction in prediction error generated by each model). In order to compare the size of effects across subgroups, we used the models in Table 2 to estimate the average change in global warming beliefs caused by a shift in local temperature from 4.3°F below normal to 14.7°F above normal for each level of political awareness, again replicated over our three measures. These effects and their 95‐percent confidence bounds are displayed in Figure 3, which quite clearly shows how the effect of weather on attitudes is curvilinear in political awareness. A shift from unseasonably cold weather to unseasonably hot weather causes beliefs in global warming to increase by 11 percentage points among those in the middle tercile of news awareness, by 10 points among those in the middle tercile of opinionation, and by 14 points among leaning partisans. Across all three of our proxies, the estimated effects for those in the middle of the political awareness spectrum are higher than those found among those with low or high political awareness, and only these effects are statistically significant. Thus political awareness plays a key role in moderating how Americans translate their personal experience with weather into beliefs about global warming: those in the middle of the political awareness scale are the most moveable. Conclusion This paper presents results that offer both topical relevance and theoretical interest. Global climate change is one of the most important public policy challenges of our time. But it is a complex issue of 22
low salience with which Americans have little direct experience in their day‐to‐day lives. As they try to make sense of this difficult issue, many people use fluctuations in local temperature to reassess their beliefs about the existence of global warming. The effect of weather on beliefs is significant and substantively large. This is an interesting and important finding in itself, and a discovery that should trouble anyone interested in engaging the public in a thoughtful debate about global warming. Our approach also permits the exploration of a more general theoretical question about how individuals use their personal experiences to form political attitudes. Political scientists’ attempts to answer this question have long been handicapped by reliance on self‐reports of experiences that are distributed among the population in ways that are correlated with unmeasured causes of attitude change. Short‐term weather variation offers a rare opportunity to identify cleanly the effects of an exogenous experience on political attitudes among a general population sample. These findings complement what is known about attitude change while deepening our understanding of how it operates. Previous work—most notably the RAS model—has focused on the effects of messages transmitted by elites. These messages function as stimuli whose effects depend upon whether the messages are received and accepted, with political awareness playing a strong role in determining message reception and acceptance. Our focus is not elite messages but instead personal experience. Here, the stimulus is particularly hot or cold local weather—a stimulus that is received in a fashion entirely uncorrelated with political awareness. Yet political awareness still plays an important moderating role: the translation of experience into political attitudes requires a degree of sophistication about politics and public affairs that is strong enough to contextualize experience, but not so strong as to be associated with hardened attitudes. 23
How might our findings generalize to other personal experiences? We expect the curvilinear effect of political awareness to emerge most frequently when a public issue is considered newsworthy enough to merit the media’s attention but nevertheless the nation’s liberal and conservative elites do not invoke personal experience with regard to the issue. This pattern leaves personal experiences open to interpretation among those at mid‐levels of political awareness, who are both cognizant of the issue and open to tapping their own experiences in forming judgments about it. One circumstance where this may take place regularly is when an issue is rising on the public agenda but has yet to be personalized by elites. More generally, our result represents a lower bound on both the extent to which individuals translate events in their lives into political attitudes and the persistence of an event’s effects. After all, the few messages received by Americans linking local weather to climate change are inconsistent and often do not correspond to contemporaneous experience, in contrast to elite messages during, say, a crime wave or an economic recession. Furthermore, the average effect of weather on attitudes is dampened by the fact that people expect the weather to be variable, so that even an extreme weather event may not make a lasting impression. The nature of this evidence is always in flux, making opinion on global warming’s existence among those at mid‐levels of awareness as changeable as the weather. In other issue domains, personal experiences (such as being a victim of crime or losing a job) can be much more vivid and have a greater impact on one’s life. For these experiences, the effects are likely to have more endurance. Finally, we think this work should inspire further research on political attitude change that takes advantage of sources of exogenous variation in personal experiences. Our research suggests that personal experience has substantial effects on political attitudes, and that rich discoveries await those who can explore these questions in ways that permit clean identification of these effects.
24
Table 1. The Structure of Beliefs about Global Warming Ordered probit. DV: Opinion on whether there is “solid evidence” for global warming (scored “no” = 1; “mixed”/”some”/DK/ref = 2; “yes” = 3) Variable Departure from normal local temperature (°F), week prior to survey
Sex: Male Race/Ethnicity: Black Race/Ethnicity: Hispanic Race/Ethnicity: Not White, Black or Hispanic
Age: 18‐24 Age: 25‐34 Age: 35‐44 Age: 55‐64 Age: 65 or over Education: high school diploma or less Education: college grad Education: post grad Party ID: Republican
Party ID: lean Republican Party ID: lean Democrat Party ID: Democrat Ideology: very conservative Ideology: conservative Ideology: liberal Ideology: very liberal Attend services: never Attend services: seldom Attend services: few times per year Attend services: weekly Attend services: more than weekly State and survey wave fixed effects included?
Goodness of fit statistics: expected percent correctly predicted expected proportional reduction in error I II 0.010***
(0.003)
0.010* (0.004) No
Yes 58.2% 0.1% 58.7% 1.4% For details about analysis, see notes on following page.
III 0.010* (0.004) ‐0.220*** (0.035) 0.045 (0.068) 0.185* (0.080) 0.116 (0.084) ‐0.049 (0.077) ‐0.068 (0.064) ‐0.082 (0.056) 0.028 (0.054) ‐0.012 (0.051) 0.079 (0.044) 0.049 (0.048) 0.114* (0.056) ‐0.445*** (0.067) ‐0.218*** (0.041) 0.199** (0.062) 0.160 (0.097) ‐0.319*** (0.060) ‐0.127 (0.069) 0.306*** (0.069) 0.386*** (0.061) 0.121 (0.072) ‐0.103 (0.066) ‐0.043 (0.061) ‐0.065 (0.056) ‐0.163* (0.064) Yes 62.1% 9.6% Notes to Table 1 N = 6,726. Estimated quantities significantly different from zero at *p<.05; **p<.01; ***p<.001 (two‐
tailed tests, robust standard errors). Excluded categories are the medians or modes of each variable: female, white, between 45‐55 years old, some college, Independent, moderate, attends services a few times per month. Included in the analyses, but not shown, are estimates of cutpoints and (in Model III) coefficients estimated for each variable for those who responded “don’t know” or did not provide a response. Goodness of fit statistics (expected percent correctly predicted and expected proportional reduction in error) are calculated as proposed by Herron (1999) and implemented by the epcp routine in Stata (Lawrence 2009). 26
Table 2. How Political Awareness Moderates Weather’s Effect on Beliefs about Global Warming (Replicated Over Three Proxies for Political Awareness) Ordered probit. DV: Opinion on whether there is “solid evidence” for global warming (scored “no” = 1; “mixed”/”some”/DK/ref = 2; “yes” = 3) PROXY 1: NEWS ATTENTION Departure from normal local temperature (°F), week prior to survey LOWEST MIDDLE HIGHEST .0156 (.0104) .0212* (.0084) .0120 (.0086) Expected percent correctly predicted Expected proportional reduction in error N 58.9%
6.6%
1,143
64.9% 13.1% 1,541 66.5%
16.4%
1,556
LOWEST MIDDLE HIGHEST .00521 (.0077) .0196** (.0072) .00915 (.0072) PROXY 2: OPINIONATION Departure from normal local temperature (°F), week prior to survey Expected percent correctly predicted Expected proportional reduction in error N 60.1%
7.9%
1,939
64.0% 10.6% 2,056 67.4%
18.7%
2,440
LEANING PARTISANS PROXY 3: STRENGTH OF PARTISANSHIP Departure from normal local temperature (°F), week prior to survey INDEPENDENTS ‐.00129 (.0112) .0279*** (.0082) .00531 (.0051) Expected percent correctly predicted Expected proportional reduction in error N 58.8%
10.3%
673
65.2% 9.2% 1,784 62.2%
11.9%
4,231
Analyses include all control variables included in Table 1, Model III. Coefficients significantly different from zero at *p<.05; **p<.01; ***p<.001 (two‐tailed tests; robust standard errors). Analyses exclude completely determined cases. 27
Figure 1. Americans’ Beliefs about the Evidence for Global Warming, by Departure of Local Weather from Normal Temperature in Week Prior to Survey source for weather data: National Climate Data Center source for survey data: Pew Research Center Pr(agree that earth is getting warmer)
PARAMETRIC ESTIMATION (PROBIT)
NON‐PARAMETRIC ESTIMATION (LOWESS)
.9
.9
.8
.8
.7
.7
.6
.6
‐10
‐5
0
5
10
15
20
25
‐10
‐5
0
5
10
15
20
25
departure from local normal daily temperature (degrees F)
in week prior to survey
Pr(agree there is "solid evidence"
that earth is getting warmer)
95% confidence bounds
Points indicate mean proportion agreeing that the earth is getting warmer within each of 100 temperature centiles. Each point represents approximately 70 survey respondents. Smoother drawn with .3 bandwidth. 28
Figure 2. Effects of Variables on Americans’ Beliefs about the Evidence for Global Warming source: predictions generated by Model III, Table 1 .215
PARTY
IDENTIFICATION*
.182
.082
.064
.053
0
GENDER
HISPANIC
BLACK
.010
.010
RELIGIOSITY
LOCAL TEMPERATURE
.051
.014
IDEOLOGY*
AGE
effect size
EDUCATION
.05
95% confidence bounds
.1
.15
.2
average effect of variable
on beliefs about global warming
.25
*Indicates size of this variable’s effect is significantly different from that of local temperature (p = .05). Effects are the average predictive differences in the probability of agreeing there is evidence for global warming accompanying shifts in variables as follows: party identification: Republican to Democrat; ideology: very conservative to very liberal; religiosity: attend services more than weekly to never attend services; gender: male to female; local temperature: ‐4.3°F to 14.7°F; Hispanic: compared to white, not Hispanic; black: compared to white, not Hispanic; education: HS or less to post‐graduate; age: 18‐24 to 65 or over. In calculating the effect of a variable, all other variables were held constant at their actual values across all observations; the average effect is reported along with bootstrapped confidence bounds. For additional details, see Table 1. 29
Figure 3. Weather’s Effect on Global Warming Beliefs by Three Measures of Political Awareness source: predictions generated by models in Table 2 effect of temperature shift (from 5th to 95th percentile)
on belief that there is "solid evidence" for global warming
NEWS
ATTENTION
OPINIONATION
.2
.2
.1
.091
.2
.1
.108
0
.046
.030
.142
.1
.102
.060
0
STRENGTH
OF PARTISANSHIP
0
.029
‐.007
‐.1
‐.1
lowest
middle
‐.1
highest
lowest
tercile
middle
tercile
highest
inds.
leaning
party ID
Figure displays the estimated average marginal effect caused by a shift in local temperature from ‐
4.3°F to 14.7°F on the probability of agreeing there is evidence for global warming. In calculating the effect of a variable, all other variables were held constant at their actual values across all observations; the average effect is displayed along with bootstrapped confidence bounds. For additional details, see Table 2. partisans
30
REFERENCES Achen, Christopher H. 1987. The Statistical Analysis of Quasi­Experiments. Berkeley: University of California Press. Althaus, Scott L. 1998. Information Effects in Collective Preferences. American Political Science Review 92:545‐558. Atkeson, Lonna Rae, and Ronald B. Rapoport. 2003. The More Things Change the More They Stay the Same: Examining Gender Differences in Political Attitude Expression, 1952‐2000. Public Opinion Quarterly 67: 495‐521. Bartels, Larry M. 1996. Uninformed Votes: Information Effects in Presidential Elections. American Journal of Political Science 40:194‐230. Bartels, Larry. 2002. Beyond the Running Tally: Partisan Bias in Political Perceptions. Political Behavior 24:117‐50. Bennett, Stephen Earl. 1989. Trends in Americans’ Political Information, 1967‐1987. American Politics Quarterly 17: 422‐435. Bennett, Stephen Earl. 1994. Changing Levels of Political Information in 1988 and 1990. Political Behavior 16: 1‐20. Bennett, Stephen Earl, and Linda L. M. Bennett. 1993. Out of Sight, Out of Mind: Americans' Knowledge of Party Control of the House of Representatives, 1960‐1984. Political Research Quarterly 46: 67‐80. Bergan, Daniel E. 2009. The Draft Lottery and Attitudes Towards the Vietnam War. Public Opinion Quarterly 73: 379‐384. Berinsky, Adam J. 1999. The Two Faces of Public Opinion. American Journal of Political Science 43: 1209‐30. Bord, Richard J., Robert E. O’Connor, and Ann Fisher. 2000. In What Sense Does the Public Need to Understand Global Climate Change? Public Understanding of Science 9 (3):205‐18. Brady, Henry E. and Paul M. Sniderman. 1985. Attitude Attribution: A Group Basis for Political Reasoning. American Political Science Review 79: 1061‐78. Brody, Samuel D., Sammy Zahran, Arnold Vedlitz, and Himanshu Grover. 2007. Examining the Relationship Between Physical Vulnerability and Public Perceptions of Global Climate Change in the United States. Environment and Behavior 40 (1):72‐95. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American Voter. New York: John Wiley and Sons. Chaiken, Shelly. 1980. Heuristic versus Systematic Information Processing and the Use of Source versus Message Cues in Persuasion. Journal of Personality and Social Psychology 39:752‐766. Citrin, Jack and Donald P. Green. 1990. The Self‐Interest Motive in American Public Opinion. In Samuel Long, ed., Research in Micropolitics 3: 1–28. Greenwich, CT: JAI Press. Cleveland, William S. 1993. Visualizing Data. Summit, N.J.: Hobart Press. Conover, Pamela Johnston, Stanley Feldman, and Kathleen Knight. 1986. Judging Inflation and Unemployment: The Origins of Retrospective Evaluations. Journal of Politics 48:565–88. Converse, Philip E. 1964. The Nature of Belief Systems in Mass Publics. In David Apter, ed., Ideology and Discontent. New York: The Free Press. Curry, T.E., S. Ansolabehere, and H. J. Herzog. 2007. A Survey of Public Attitudes Towards Climate Change and Climate Change Mitigation Technologies in the United States: Analyses of 2006 Results. Cambridge, MA: MIT LFEE. 32
De Boef, Suzanna and Paul M. Kellstedt. 2004. “The Political (and Economic) Origins of Consumer Confidence.” American Journal of Political Science 48:633–649, October 2004 Delli Carpini, Michael X., and Scott Keeter. 1996. What Americans Know About Politics and Why It Matters. New Haven, CT: Yale University Press. Dunning, Thad. 2008. Improving Causal Inference: Strengths and Limitations of Natural Experiments. Political Research Quarterly 61:282‐93. Erikson, Robert S. and Laura Stoker. 2010. Caught in the Draft: Vietnam Draft Lottery Status and Political Attitudes. Working paper. Evans, Geoffrey and Robert Andersen. 2006. The Political Conditioning of Economic Perceptions. Journal of Politics 68: 194‐207. Faulkenberry, G. David, and Robert Mason. 1978. Characteristics of Nonopinion and No Opinion Response Groups. Public Opinion Quarterly 4: 533‐43. Fiorina, Morris P. 1981. Retrospective Voting in American Elections. New Haven: Yale University Press. Francis, Joe, and Lawrence Busch. 1975. What We Now Know About “I Don't Knows." Public Opinion Quarterly 39: 207‐218. Gelman, Andrew and Jennifer Hill. 2007. Data Analysis Using Regression and Multilevel/Hierarchical Models. New York: Cambridge University Press. Gerber, Alan S. and Gregory A. Huber. 2009. Partisanship and Economic Behavior: Do Partisan Differences in Economic Forecasts Predict Real Economic Behavior? American Political Science Review 103: 407‐26. Gomez, Brad T. and J. Matthew Wilson. 2001. Political Sophistication and Economic Voting in the American Electorate: A Theory of Heterogeneous Attribution. American Journal of Political Science 45: 899‐915. 33
Hamill, Ruth, Milton Lodge, and Frederick Blake. 1985. The Breadth, Depth, and Utility of Class, Partisan, and Ideological Schemata. American Journal of Political Science 29: 850‐870. Herron, Michael C. 1999. Postestimation Uncertainty in Limited Dependent Variable Models. Political Analysis 8:83‐98. Highton, Benjamin. 2009. Revisiting the Relationship between Educational Attainment and Political Sophistication. Journal of Politics 71: 1564‐1576. Jacoby, William G. 1988. The Impact of Party Identification on Issue Attitudes. American Journal of Political Science 32:643‐661. Jennings, M. Kent. 1996. Political Knowledge over Time and across Generations. Public Opinion Quarterly 60: 228‐52. Kellstedt, Paul M., Sammy Zahran, and Arnold Vedlitz. 2008. Personal Efficacy, the Information Environment, and Attitudes Toward Global Warming and Climate Change in the United States. Risk Analysis 28:113‐26. Kinder, Donald R., and D. R. Kiewiet. 1979. Economic Discontent and Political Behavior: The Role of Personal Grievances and Collective Economic Judgments in Congressional Voting. American Journal of Political Science 23: 495‐527. Krosnick, Jon A. 1999. Survey Research. Annual Review of Psychology 50: 537‐67. Krosnick, Jon A., and Michael A. Milburn. 1990. Psychological Determinants of Political Opinionation. Social Cognition 8: 49‐72. Krosnick, Jon A., Allyson L. Holbrook, Laura Lowe, and Penny S. Visser. 2006. The Origins and Consequences of Democratic Citizens' Policy Agendas: A Study of Popular Concern about Global Warming. Climatic Change 77:7‐43. Lane, Robert E. 1962. Political Ideology: Why the American Common Man Believes What He Does. New York: Free Press of Glencoe. 34
Lau, Richard R., and Caroline Heldman. 2009. Self‐Interest, Symbolic Attitudes, and Support for Public Policy: A Multilevel Analysis. Political Psychology 30: 513‐537. Lau, Richard R., David O. Sears, and Tom Jessor. 1990, Fact or Artifact Revisited: Survey Instrument Effects and Pocketbook Politics. Political Behavior 12: 217–242. Lau, Richard R., Thad A. Brown, and David O. Sears. 1978. Self‐Interest and Civilians’ Attitudes Toward the War in Vietnam. Public Opinion Quarterly 42: 464–83. Lawrence, Christopher N. 2009. epcp: Display Classification Accuracy for Nonmetric Dependent Variable Models. Available at www.cnlawrence.com/data/epcp.zip Leiserowitz, Anthony. 2006. Climate Change Risk Perception and Policy Preferences: The Role of Affect, Imagery, and Values. Climatic Change 77:45‐72. Lubell, Mark, Sammy Zahran, and Arnold Vedlitz. 2007. Collective Action and Citizen Responses to Global Warming. Political Behavior 29:391‐414. Luskin, Robert C. 1990. Explaining Political Sophistication. Political Behavior 12: 331‐361. Marschall, Melissa. 2004. Citizen Participation and the Neighborhood Context: A New Look at the Coproduction of Local Public Goods. Political Research Quarterly 57: 231‐44. Meehl, Gerald A., Claudia Tebaldi, Guy Walton, David Easterling, and Larry McDaniel. Relative Increase of Record High Maximum Temperatures Compared to Record Low Minimum Temperatures in the U.S. Geophysical Research Letters 36. Miller, Warren E., and J. Merrill Shanks. 1996. The New American Voter. Cambridge, MA: Harvard University Press. Mutz, Diana C. 1992. Mass Media and the Depoliticization of Personal Experience. American Journal of Political Science 36: 483‐508. Mutz, Diana C. 1994. Contextualizing Personal Experience: The Role of Mass Media. Journal of Politics 56: 689‐714. 35
Mutz, Diana C. 1998. Impersonal Influence: How Perceptions of Mass Collectives Affect Political Attitudes. Cambridge: Cambridge University Press. National Aeronautics and Space Administration (NASA), Goddard Institute for Space Studies. 2009. GISS Surface Temperature Analysis. Accessed March 2009 at http://data.giss.nasa.gov/gistemp/2008/. O'Connor, Robert E., Richard J. Bord, and Ann Fisher. 1999. Risk Perceptions, General Environmental Beliefs, and Willingness to Address Climate Change. Risk Analysis 19:461‐71. O'Connor, Robert E., Richard J. Bord, Brent Yarnal, and Nancy Wiefek. 2002. Who Wants to Reduce Greenhouse Gas Emissions? Social Science Journal 83:1‐17. Page, Benjamin I., Robert Y. Shapiro, and Glenn R. Dempsey. 1987. What Moves Public Opinion? American Political Science Review 81:23‐44. Palmer, Harvey D., and Raymond M. Duch. 2000. Do Surveys Provide Representative or Whimsical Assessments of the Economy? Political Analysis 9: 58‐77. Petty, Richard E. and John T. Cacioppo. 1986. The Elaboration Likelihood Model of Persuasion. Advances in Experimental Social Psychology 19:123‐205. Risen, Jane L. and Clayton R. Critcher. 2010. The Effect of Ambient Temperature on Belief in Global Warming. Typescript. University of Chicago. Sears, David O., and Jack Citrin. 1985. Tax Revolt: Something for Nothing in California. Cambridge, MA: Harvard University Press. Sears, David O. and Carolyn L. Funk. 1991. The Role of Self‐Interest in Social and Political Attitudes. In Mark P. Zanna, ed., Advances in Experimental Social Psychology 24: 1‐91. New York: Academic Press. Sears, David O., and Richard R. Lau. 1983. Inducing Apparently Self‐Interested Political Preferences. American Journal of Political Science 27: 223‐252. 36
Sears, David O., Richard R. Lau, Tom R. Tyler, and Harris M. Allen, Jr. 1980. Self‐Interest vs. Symbolic Politics in Policy Attitudes and Presidential Voting. American Political Science Review 74: 670‐684. Shani, Danielle. 2009. Knowing Your Colors: Information, Partisan Bias, and Political Perceptions. Typescript. Princeton University. Shapiro, Robert Y. and Yaeli Bloch‐Elkon. 2008. Do the Facts Speak for Themselves: Partisan Disagreement as a Challenge to Democratic Competence. Critical Review 20:115‐39 Shaw, Daron R. 1999. The Effect of TV Ads and Candidate Appearances on Statewide Presidential Votes, 1988‐96. American Political Science Review 93:345‐361. Slater, Michael D., Catherine E. Goodall, and Andrew F. Hayes. 2009. Self‐Reported News Attention Does Assess Differential Processing of Media Content: An Experiment on Risk Perceptions Utilizing a Random Sample of U.S. Local Crime and Accident News. Journal of Communication 59: 117‐34. Trenberth, K.E., P.D. Jones, P. Ambenje, R. Bojariu, D. Easterling, A. Klein Tank, D. Parker, F. Rahimzadeh, J.A. Renwick, M. Rusticucci, B. Soden, and P. Zhai. 2007. Observations: Surface and Atmospheric Climate Change. In Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge, UK and New York: Cambridge University Press. Weatherford, M. Stephen. 1983. Economic Voting and the “Symbolic Politics” Argument: A Reinterpretation and Synthesis. American Political Science Review 77: 158‐74. Zaller, John. 1991. Information, Values, and Opinion. American Political Science Review 85:1215‐1237. 37
Zaller, John. 1992. The Nature and Origins of Mass Opinion. New York: Cambridge University Press. 38
APPENDIX TABLE A1. Replication of Table 2, Incorporating Interaction of Educational Attainment and Temperature Ordered probit. DV: Opinion on whether there is “solid evidence” for global warming (scored “no” = 1; “mixed”/”some”/DK/ref = 2; “yes” = 3) PROXY 1: NEWS ATTENTION
LOWEST Departure from normal local temperature (°F), 0.0435* week prior to survey (0.0211) Education 0.0245 (0.0310) Temperature x Education ‐0.0067 (0.0045) Expected percent correctly predicted 58.9% Expected proportional reduction in error 6.5% N 1,140 PROXY 2: OPINIONATION LOWEST Departure from normal local temperature (°F), 0.0188 week prior to survey (0.0168) Education ‐0.0124 (0.0242) Temperature x Education ‐0.0028 (0.0034) Expected percent correctly predicted 60.1% Expected proportional reduction in error 7.8% N 1,931 PROXY 3: STRENGTH OF PARTISANSHIP INDEPENDENTS
Departure from normal local temperature (°F), ‐0.0194 week prior to survey (0.0259) Education ‐0.0792 (0.0425) Temperature x Education 0.0041 (0.0053) Expected percent correctly predicted 58.7% Expected proportional reduction in error 10.1% N 673 MIDDLE 0.0604** (0.0202) ‐0.0238 (0.0297) ‐0.0081* (0.0038) HIGHEST 0.0414* (0.0206) 0.0259 (0.0299) ‐0.0058 (0.0036) 65.0% 13.3% 1,541 66.4% 16.3% 1,556 MIDDLE 0.0580*** (0.0168) 0.0480* (0.0245) ‐0.0080* (0.0031) 64.0% 10.4% 2,056 HIGHEST ‐0.0022 (0.0192) ‐0.0134 (0.0237) 0.0023 (0.0035) 67.4% 18.7% 2,440 LEANING 0.0656*** (0.0196) 0.0536* (0.0258) ‐0.0081* (0.0037) 64.0% 9.3% 1,784 PARTISANS 0.0202 (0.0125) 0.0052 (0.0167) ‐0.0031 (0.0024) 62.3% 11.8% 4,224 Analyses include all control variables included in Table 1, Model III. Coefficients significantly different from zero at +p<.10; *p<.05; **p<.01; ***p<.001 (two‐tailed tests; robust standard errors). Models exclude cases completely determined and those where no valid response was supplied on the education question. 39
TABLE A2. Regression of political knowledge on educational attainment and proxies, 2004 ANES Time Series Study OLS. DV: Political knowledge, scored 0 (lowest) to 6 (highest). (1) coef (se) (2) coef (se) Strength of Partisanship: Leaning 0.323* (0.137) 0.261* (0.126) Strength of Partisanship: Partisan 0.503*** (0.125) 0.394*** (0.115) News Attention 0.0302*** (0.00480) 0.0287*** (0.00451) Ln(Opinionation) 0.816*** (0.0418) 0.631*** (0.0420) 0.305*** (0.0235) Educational attainment 3.448*** (0.150) 1,066 .32 Intercept N Adjusted R2 2.015*** (0.181) 1,066 .41 Coefficients significantly different from zero at +p<.10; *p<.05; **p<.01; ***p<.001 (two‐tailed tests; robust standard errors). Excluded strength of partisanship category is Independent. 40
effect of temperature shift (from 5th to 95th percentile)
on belief that there is "solid evidence" for global warming
FIGURE A1 Replication of Figure 3, Incorporating Interaction of Educational Attainment and Temperature source: predictions generated by models in Table A1 NEWS
ATTENTION
.2
.2
.2
.1
.085
STRENGTH
OF PARTISANSHIP
OPINIONATION
.1
.107
.065
0
0
.047
.035
.137
.1
.104
0
.030
‐.004
‐.1
‐.1
lowest
middle
highest
‐.1
lowest
tercile
middle
tercile
41
highest
inds.
leaning
party ID
partisans
MEASUREMENT AND VALIDITY OF PROXIES FOR POLITICAL AWARENESS Construction of news attention index The news attention index is constructed from respondents’ answers to Pew’s questions about how closely they were following particular news stories at the time of their survey. Respondents could indicate they were following a story “very closely” (scored 4); “fairly closely” (3); “not too closely” (2) or “not at all closely” (1). Those who responded don’t know or refused to answer were assigned a score of zero for the question. Each respondent’s scores were summed across all stories to create a raw summary score. In order to make this score comparable across surveys and form splits, we converted the raw score into percentiles. Following are the general language used by Pew to introduce the news attention questions, and the question wording (with Pew’s variable labels) for each of news attention questions in the order in which they were asked. These questions were included in only three of the five surveys in our study (June 2006, August 2006 and January 2007). INTRODUCTORY LANGUAGE USED IN EACH SURVEY “As I read a list of some stories covered by news organizations this past month, tell me if you happened to follow each news story very closely, fairly closely, not too closely, or not at all closely…. First, [INSERT ITEM; RANDOMIZE. OBSERVE FORM SPLITS] [IF NECESSARY “Did you follow [ITEM] very closely, fairly closely, not too closely or not at all closely?”]” JUNE 2006 # of items: 6 Cronbach’s alpha: Form 1: .78 Form 2: .80 ALL RESPONDENTS: Q.5a. News about the current situation in Iraq FORM 1 ONLY: Q.5b.F1 The high price of gasoline these days Q.5c.F1 News about candidates and election campaigns in your state and district Q.5d.F1 The issue of immigration FORM 2 ONLY: Q.5e.F2 Iran’s nuclear research program Q.5f.F2 Reports about the financial links between lobbyists and members of Congress Q.5g.F2 Soccer's World Cup championship being held in Germany ALL RESPONDENTS: Q.5h. The killing of the terrorist leader in Iraq named Zarqawi by U.S. forces Q.5i. The investigation of U.S. Marines who allegedly killed civilians in an Iraqi village 42
AUGUST 2006 # of items: 5 Cronbach’s alpha: Form 1: .77 Form 2: .74 ALL RESPONDENTS: Q.9.a. News about the current situation in Iraq Q.9.c. The military conflict between Israel and Hezbollah forces in Lebanon FORM 1 ONLY: Q.9.d.F1 The high price of gasoline these days Q.9.e.F1 The issue of immigration Q.9.f .F1 Iran’s nuclear research program FORM 2 ONLY: Q.9.g.F2 News about candidates and election campaigns in your state and district Q.9.i.F2 Fidel Castro handing control of Cuba to his brother because of an illness Q.9.j.F2 Actor Mel Gibson’s outburst during a drunk driving arrest [Not included in our index was a question asked about “The hot weather this summer and its impact on the country,” as well as two questions added after the survey was already in the field and thus not asked of a substantial proportion of respondents.] JANUARY 2007 # of items: 5 Cronbach’s alpha: .70 Q.14.a. News about the current situation in Iraq Q.14.b. The execution of Saddam Hussein Q.14.c. The death and memorial services of Gerald Ford Q.14.d. The U.S air strikes on suspected terrorist sites in Somalia Q.14.e. News about the incoming Democratic leaders in Congress [Not included in our index was a question asked about “unusual winter weather.”] 43
Construction of opinionation index The opinionation index is constructed by selecting all survey questions having to do with public affairs and counting the number of times each respondent supplied a “refused” or “don’t know” response. Not included were the news attention questions used to create the news attention index; questions about vote choice; survey questions about the environment or global warming; or any items that might invoke a social desirability bias and thus make respondents reluctant to offer an opinion. In order to make this score comparable across surveys and form splits, we converted it into percentiles. Following are lists of questions from each survey (with Pew’s variable labels) used to create the opinionation index. For purposes of constructing the index, respondents who volunteered “don’t know” or refused to answer the question were scored one; those who provided a response were scored zero. Questions with labels “F1” and “F2” appeared only on Form 1 and Form 2 of the surveys, respectively. JUNE 2006 # of items: Form 1: 32 Form 2: 33 Cronbach’s alpha: Form 1: .85 Form 2: .87 Q.1 Do you approve or disapprove of the way George W. Bush is handling his job as president? Q.2 Do you approve or disapprove of the way George W. Bush is handling [INSERT ITEM, RANDOMIZE; OBSERVE FORM SPLITS] How about [NEXT ITEM]? a. The economy b.
The situation in Iraq
c.F1 Terrorist threats d.F1 The environment e.F1 The nation’s immigration policy f.F2 The nation’s foreign policy g.F2 Energy policy h.F2 The issue of global warming Q.3 Do you approve or disapprove of the job the Republican leaders in Congress are doing? Q.4 Do you approve or disapprove of the job the Democratic leaders in Congress are doing? Q.18 Compared to recent Congresses, would you say THIS Congress has accomplished more, accomplished less, or accomplished about the same amount? Q.20 As I name some issues that are in the news these days, please tell me how important the issue is to you personally. First, is [INSERT ITEM, RANDOMIZE – OBSERVE FORM SPLITS] an issue that is very important, somewhat important, not too important, or not at all important to you? … How about [NEXT ITEM]? [IF NECESSARY “Is [ITEM] very important, somewhat important, not too important or not at all important to you personally?] a. The nation’s economy b. The situation in Iraq c. Immigration d. Gay marriage 44
Q.50 Q.51 Q.52 Q.53 Q.54 Q.55 Q.56 e. Global warming f.F1 Terrorism and homeland security g.F1 A Constitutional amendment to ban flag burning h.F1 The federal budget deficit i.F1 Education j.F1 The job situation k.F1 Environmental policy l.F1 Eliminating the inheritance tax m.F2 Government surveillance programs in the U.S. n.F2 Abortion o.F2 Health care p.F2 Taxes q.F2 Social Security r.F2 Energy policy s.F2 Increasing the minimum wage Do you think the U.S. made the right decision or the wrong decision in using military force against Iraq? How well is the U.S. military effort in Iraq going? Do you think the U.S. should keep military troops in Iraq until the situation has stabilized, or do you think the U.S. should bring its troops home as soon as possible? Do you think the war in Iraq has helped the war on terrorism, or has it hurt the war on terrorism? Do you think the U.S. should or should not set a timetable for when troops will be withdrawn from Iraq? Regardless of what you think about the original decision to use military force in Iraq, do you now believe that the United States will definitely succeed, probably succeed, probably fail, or definitely fail in establishing a stable democratic government in Iraq? As I read a few specific things about Iraq, tell me if you think we are making progress or losing ground in each area. First, are we making progress or losing ground in [INSERT ITEM; RANDOMIZE OBSERVE FORM SPLITS]? How about [NEXT ITEM]? [IF NECESSARY: Do you think we are making progress or losing ground in this area?] Training Iraqi security forces so they can replace U.S. troops Reducing the number of civilian casualties there Preventing terrorists from using Iraq as a base for attacks against the U.S. and its a.F1 b.F1 c.F1 allies d.F1 Establishing democracy in Iraq e.F2 Defeating the insurgents militarily f.F2 Preventing a civil war between various religious and ethnic groups g.F2 Rebuilding roads, power plants and other services in Iraq ASK ALL: Q.57 As you may know, the terrorist leader of al Qaeda in Iraq named Zarqawi was killed last week by U.S. forces. Will Zarqawi’s death make the situation in Iraq better, worse, or not have much of an effect? IDEO In general, would you describe your political views as… Very conservative, Conservative, Moderate, Liberal, or Very liberal? 45
JULY 2006 [NOTE: DUE TO LACK OF QUESTIONS AVAILABLE ON FORM 1, OPINIONATION INDEX WAS CONSTRUCTED ONLY FOR FORM 2 RESPONDENTS .] # of items: Form 2: 13 Cronbach’s alpha: Form 2: .70 Q.1 Do you approve or disapprove of the way George W. Bush is handling his job as president? Q.2 All in all, are you satisfied or dissatisfied with the way things are going in this country today? Q.3 Recently, there have been news reports that the federal government has been secretly examining the bank records of American citizens who might have ties to terrorist groups…how much have you heard about this? Q.5F2 In reporting that the federal government has been secretly examining the bank records of American citizens who might have ties to terrorist groups, do you believe that news organizations have told the American people something they should know about, or was it something they didn’t need to know about.? Q.7F2 I’d like to get your opinion of some groups and organizations in the news. Is your overall opinion of... [INSERT ITEM; RANDOMIZE AND ROTATE ITEMS IN BLOCKS, FIRST a THRU c, THEN d THRU f] very favorable, mostly favorable, mostly UNfavorable, or very unfavorable? a.F2 The Christian conservative movement b.F2 The environmentalist movement c.F2 The Supreme Court d.F2 The Republican Party e.F2 The Democratic Party f.F2 The United Nations Q.18 Now thinking about Iraq, do you think the U.S. made the right decision or the wrong decision in using military force against Iraq? Q.70dF2 Do you think of yourself as a member of the religious right political movement, or not? IDEO In general, would you describe your political views as… Very conservative, Conservative, Moderate, Liberal, or Very liberal? AUGUST 2006 # of items: Form 1: 44 Form 2: 46 Cronbach’s alpha: Form 1: .91 Form 2: .88 Q.1 Do you approve or disapprove of the way George W. Bush is handling his job as president? Q.2F1 Do you approve or disapprove of the way George W. Bush is handling [INSERT ITEM, RANDOMIZE; OBSERVE FORM SPLITS] How about [NEXT ITEM]? a.F1 The economy b.F1 The situation in Iraq c.F1 Terrorist threats d.F1 The nation’s foreign policy 46
e.F1 Health care policy f.F1 Education g.F1 The conflict in the Middle East between Israel and Hezbollah Q.3 Right now, which is more important for President Bush to focus on… domestic policy or foreign policy? Q.12 What one issue would you most like to hear the candidates in your state or district talk about this fall? Q.17aF1 Do you happen to know the name of the person who will soon be taking over as the anchor of the CBS Evening News later this summer? {new} Q.17bF1 Do you happen to know the name of the Democratic senator from Connecticut who recently lost in the state’s primary election? {new} Q.17cF1 Do you happen to know the name of the Hollywood actor who made anti‐Semitic remarks during a recent drunk driving arrest? {new} Q.18F2 I’d like to get your general impression of some news figures. Even if it is not someone you watch regularly, we just want to get your overall impression. First… [INSERT ITEM; RANDOMIZE] What ONE WORD best describes your impression of [REPEAT NAME ONLY]? Just the one best word that describes [him/her]. (OPEN‐
END. PROBE ONCE IF RESPONDENT ANSWERS “DON’T KNOW”) How about [NEXT ITEM]? What ONE WORD best describes your impression of [REPEAT NAME ONLY]? a.F2 Katie Couric [PRONOUNCE: KUR‐rik], who will become the anchor of the CBS Evening News later this summer b.F2 Brian Williams, the anchor of the NBC Nightly News c.F2 Charles Gibson, who recently became the anchor of the ABC World News Q.19 Do you think the U.S. made the right decision or the wrong decision in using military force against Iraq? {7­06 Religion} Q.20 How well is the U.S. military effort in Iraq going? [READ IN ORDER] {6­06} Q.21 Do you think the U.S. should keep military troops in Iraq until the situation has stabilized, or do you think the U.S. should bring its troops home as soon as possible? {6­06} Q.24F1 Regardless of what you think about the original decision to use military force in Iraq, do you now believe that the United States will definitely succeed, probably succeed, probably fail, or definitely fail in establishing a stable democratic government in Iraq? {6­06} Q.25F2 Regardless of what you think about the original decision to use military force in Iraq, do you now believe that the United States will definitely succeed, probably succeed, probably fail, or definitely fail in achieving its goals in Iraq? {new} Q.26 As I read a few specific things about Iraq, tell me if you think we are making progress or losing ground in each area. First, are we making progress or losing ground in [INSERT ITEM; RANDOMIZE OBSERVE FORM SPLITS]? How about [NEXT ITEM]? a.F1 Training Iraqi security forces so they can replace U.S. troops b.F1 Reducing the number of civilian casualties there c.F1 Preventing terrorists from using Iraq as a base for attacks against the U.S. and its allies d.F1 Establishing democracy in Iraq e.F2 Defeating the insurgents militarily f.F2 Preventing a civil war between various religious and ethnic groups g.F2 Rebuilding roads, power plants and other services in Iraq Q.27 Do you think the U.S. should or should not set a timetable for when troops will be withdrawn from Iraq? {6­06} 47
Q.28 In the dispute between Israel and the Palestinians, which side do you sympathize with more, Israel or the Palestinians? {7­06} Q.31F1 Compared with the past, would you say the U.S. is MORE respected by other countries these days, LESS respected by other countries, or AS respected as it has been in the past? Q.33F2 Do you think each of the following is a major reason, a minor reason, or not much of a reason why people around the world are unhappy with the U.S.? First, [INSERT ITEM; RANDOMIZE. OBSERVE FORM SPLITS]. a.F2 U.S. support for Israel b.F2 The U.S.‐led war on terrorism c.F2 America’s wealth and power d.F2 The Iraq war Q.36 Compared to the Japanese attack on Pearl Harbor in 1941, would you say that the September 11th terrorist attacks were more serious, not as serious, or about equal? Q.39 Do you think that the September 11th terrorist attacks were the start of a major conflict between the people of America and Europe versus the people of Islam, or is it only a conflict with a small, radical group? Q.42 Do you happen to recall in what year the September 11th terrorist attacks occurred? [IF “YES” PROBE: Which year?] Q.43F1 Overall, do you think the ability of terrorists to launch another major attack on the U.S. is greater, the same, or less than it was at the time of the September 11th terrorist attacks? Q.44F1 So far, there has not been another terrorist attack in America since 2001. Is this mostly because [READ AND ROTATE] The government is doing a good job protecting the country OR America is a difficult target for terrorists OR America has been lucky so far? Q.45F2 In general, how well do you think the U.S. government is doing in reducing the threat of terrorism? Q.46F2 What concerns you more about the government’s anti‐terrorism policies? [READ AND ROTATE] That they have gone too far in restricting the average person’s civil liberties ‐‐OR‐‐That they have not gone far enough to adequately protect the country Q.51 How would you rate the job your LOCAL government is doing preparing for the possibility of a terrorist attack – would you say excellent, good, only fair, or poor? Q.52 All in all, which do you think would have a greater effect in reducing the threat of terrorist attacks on the United States? Increasing America’s military presence overseas– OR Decreasing America’s military presence overseas Q.55 As I read from a list tell me how important each of the following is as a way to reduce terrorism in the future. (First), [INSERT ITEM; RANDOMIZE; OBSERVE FORM SPLITS] is this very important, fairly important, not too important, or not at all important as a way to reduce terrorism in the future? (Next…[NEXT ITEM]) {8­
02} a.F1 Decrease American dependence on oil imported from the Middle East b.F1 Encourage more democracy in Mideast countries like Egypt and Saudi Arabia c.F1 Take military action to wipe out facilities of countries attempting to build nuclear weapons d.F2 Increase defense spending to maintain our military preparedness e.F2 Attempt to reduce poverty with foreign aid as a way of combating terrorism in poor Mideast countries 48
f.F2 Not get involved in trying to solve the problems of other countries Over the past few weeks, there has been a significant escalation of violence between Israel and the Hezbollah group in Lebanon. From what you’ve read and heard, who do you think is most responsible for the outbreak of the recent violence? Q.63 And who do you think is most responsible for the civilian casualties that have occurred since the fighting began? Q.64 What should the United States do in this conflict? Should the United States government publicly support Israel, should it say or do nothing, or should it publicly criticize Israel? Q.65F1 Do you approve or disapprove of the military action Israel has taken in Lebanon? Q.66F2 What do you think about the way Israel is responding in the current conflict with Hezbollah militants in Lebanon? Has Israel gone too far, not gone far enough, or has Israel's response been about right? Q.67 From what you’ve heard or read, what’s your best understanding of Hezbollah? Q.68 In terms of resolving this conflict, do you think the United States should be more involved than it is now, less involved than it is now, or is it as involved as it should be? Q.70 How much influence do you think the U.S. can have on the way Israel is dealing with the current situation? Q.71 Do you think news organizations are giving too much, too little, or the right amount of coverage to the situation in the Middle East? Q.72 In your opinion, are the news organizations you are familiar with being fair or unfair in their coverage of the conflict between Israel and the Hezbollah group in Lebanon? IDEO In general, would you describe your political views as… Very conservative, Conservative, Moderate, Liberal, or Very liberal? JANUARY 2007 # of items: Form 1: 47 Form 2: 46 Cronbach’s alpha: Form 1: .89 Form 2: .91 Q.1 All in all, are you satisfied or dissatisfied with the way things are going in this country today? Q.2 Do you approve or disapprove of the way George W. Bush is handling his job as president? Q.3 Do you approve or disapprove of the job the Democratic leaders in Congress are doing? Q.4 What do you think is the most important problem facing the country today? Q.8F1 In the long run, do you think George W. Bush will be a successful or unsuccessful president, or do you think it is too early to tell? Q.9F2 In the long run, do you think the accomplishments of the Bush Administration will outweigh its failures, or will the failures outweigh the accomplishments? Q.10aF1 Now thinking about George W. Bush’s upcoming State of the Union address... Do you consider this year’s State of the Union address to be MORE important than past years’, LESS important, or about as important as past years’? Q.11F1 Right now, which is more important for President Bush to focus on… domestic policy or foreign policy? {8­06} Q.62 49
Q.12F2 Right now, which is more important for Democratic leaders in Congress to focus on… domestic policy or foreign policy? {new} Q.13 Thinking about the Democratic and Republican parties, would you say there is a great deal of difference in what they stand for, a fair amount of difference, or hardly any difference at all? Q.18 I’d like to ask you some questions about priorities for President Bush and Congress this year. As I read from a list, tell me if you think the item that I read should be a top priority, important but lower priority, not too important or should it not be done? (First,) should (INSERT ITEM; RANDOMIZE; OBSERVE FORM DIFFERENCES) be a top priority, important but lower priority, not too important, or should it not be done? (What about (INSERT ITEM)?) a.F1 Improving the job situation {94,97 to 06} b.F1 Reducing the budget deficit {94,97, 02,03,04,05,06} c.F1 Reducing crime {94,97 to 06} d.F1 Reducing federal income taxes for the middle class {94,97 to 06} e.F1 Protecting the environment {97 to 06} f.F1 Reducing health care costs {new; compare to “regulating HMO’s” ??} g.F1 Increasing the minimum wage {00,01,04,05,06} h.F1 Making the recent federal income tax cuts permanent {03 mod,05 verbatim,06 mod} i.F1 Defending the country from future terrorist attacks {02,03,04,05,06} j.F1 Dealing with global trade issues {00 to 06} k.F1 Dealing with the issue of illegal immigration {new} l.F2 Improving the educational system {97 to 06} m.F2 Taking steps to make the Social Security system financially sound {97 to 06} n.F2 Taking steps to make the Medicare system financially sound {97 to 06} o.F2 Dealing with the problems of poor and needy people {97 to 06} p.F2 Dealing with the moral breakdown in the country {97 to 06} q.F2 Providing health insurance to the uninsured {00 to 06} r.F2 Strengthening the U.S. military {01 to 06} s.F2 Strengthening the nation’s economy {00 to 06} t.F2 Dealing with the nation’s energy problem {02,03,04,05,06} u.F2 Making it tougher for illegal immigrants to enter the U.S. {06} v.F2 Reducing the influence of lobbyists and special interest groups in Washington {new} w.F2 Dealing with global warming {new} Q.19F1 What’s your view… Do you think the country is more politically divided these days than in the past, or not? Q.21 Are Democratic leaders in Congress making an effort to reach out to Republicans to work out solutions to policy problems, or don’t you think so? Q.22 Is the Bush administration making an effort to reach out to the Democratic leaders in Congress to work out solutions to policy problems, or don’t you think so? Q.23 In general, do the Democratic leaders in Congress right now take positions that are more liberal than yours, more conservative than yours, or similar to yours? Q.24 Do you think President Bush generally takes positions that are more liberal than yours, more conservative than yours, or similar to yours? Q.25 As I read some characteristics associated with political leaders in Washington, please tell me how much you like or dislike each. We’ll use a scale from 4 to 1 where “4” represents something you like a lot and “1” represents something you dislike a lot. On this scale, how would you rate political leaders who [INSERT ITEM; 50
RANDOMIZE ITEMS a. THRU c. FOLLOWED BY RANDOMIZED ITEMS d. THRU f.]? How about political leaders who [INSERT ITEM]? a. Take conservative positions on nearly all issues b. Take liberal positions on nearly all issues c. Take a mix of liberal and conservative positions d. Are willing to compromise e. Stick to their positions, even if unpopular f. Often side with members of the other party Q.26 All things considered, which do you admire most? [READ AND ROTATE]: Political leaders who make compromises OR Political leaders who stick to their positions Q.31 Generally, do you think Democratic leaders in Congress will be successful or unsuccessful in getting their programs passed into law? Q.32 This year should the Democratic leaders in Washington... Try as best they can to work with George W. Bush to accomplish things, even if it means disappointing some groups of Democratic supporters OR Should they stand up to George W. Bush on issues that are important to Democratic supporters, even if it means less gets done in Washington? Q.33 This year should the Republican leaders in Washington... Try as best they can to work with Democratic leaders to accomplish things, even if it means disappointing some groups of Republican supporters OR Should they stand up to the Democrats on issues that are important to Republican supporters, even if it means less gets done in Washington? Q.34 Do you think relations between Republicans and Democrats in Washington will get better in the coming year, get worse, or stay about the same as they are now? Q.35 Please tell me which political party, the Republicans or the Democrats, comes closer to your view on each issue I name. First, which party's position comes closer to your views on… [INSERT ITEM; RANDOMIZE]? And which party comes closer to your view on [NEXT ITEM]? a. Abortion policy b. The environment c. Iraq d. Illegal immigration e. Federal taxes Q.37 Do you think the U.S. made the right decision or the wrong decision in using military force against Iraq? Q.38 How well is the U.S. military effort in Iraq going? Q.39 Do you think the U.S. should keep military troops in Iraq until the situation has stabilized, or do you think the U.S. should bring its troops home as soon as possible? Q.42 Do you think George W. Bush has a clear plan for bringing the situation in Iraq to a successful conclusion, or don’t you think so? Q.43 Do you think Democratic leaders in Congress have a clear plan for how to deal with the situation in Iraq, or don’t you think so? Q.44 Do you think the U.S. should or should not set a timetable for when troops will be withdrawn from Iraq? IDEO In general, would you describe your political views as… Very conservative, Conservative, Moderate, Liberal, or Very liberal? 51
APRIL 2008 # of items: Form 1: 31 Form 2: 32 Cronbach’s alpha: Form 1: .86 Form 2: .89 Q.1 Do you approve or disapprove of the way George W. Bush is handling his job as president? Q.2 Would you say you are looking forward to the presidential election this fall or not? Q.3 Now I’d like your views on some people. As I read some names, please tell me if you have a favorable or unfavorable opinion of each person. (First, INSERT NAME) would you say your overall opinion of… [INSERT ITEM; RANDOMIZE] is very favorable, mostly favorable, mostly UNfavorable, or very unfavorable? How about (NEXT NAME)? a. George W. Bush b. John McCain c. Hillary Clinton d. Barack Obama Q.17 Now a different kind of question. Please tell me if you think of each person I name as very conservative, conservative, moderate, liberal, or very liberal. (First,) [INSERT NAME; RANDOMIZE]. From what you’ve read and heard, would you say [NAME] is very conservative, conservative, moderate, liberal, or very liberal? How about [NEXT NAME]? a. George W. Bush b. Hillary Clinton c. Barack Obama d. John McCain Q.21 Now, I’d like your opinion of some organizations and institutions. (First,) is your overall opinion of… [INSERT ITEM; RANDOMIZE; OBSERVE FORM SPLITS] very favorable, mostly favorable, mostly UNfavorable, or very unfavorable? a.F1 The federal government in Washington b.F1 Your STATE government c.F1 Your LOCAL government d.F2 The news media e.F2 The military f.F2 Business corporations g.F2 The Supreme Court Q.22 How would you rate economic conditions in this country today… as excellent, good, only fair, or poor? Q.27 Would you favor or oppose a law that banned the sale of handguns? Q.28 What do you think is more important – to protect the right of Americans to own guns, OR to control gun ownership? Q.36 In general, do you think that free trade agreements, like NAFTA and the policies of the World Trade Organization, have been a good thing or a bad thing for the United States? Q.38 As I read some of the possible ways free trade agreements affect America, tell me what your impression is. First [INSERT ITEM; RANDOMIZE]? a. Do free trade agreements make the price of products sold in the U.S. higher, lower or not make a difference? 52
b. Do free trade agreements make the wages of American workers higher, lower, or not make a difference? c. Do free trade agreements create jobs in the U.S., lead to job losses, or not make a difference? d. Do free trade agreements make the American economy grow, slow the economy down, or not make a difference? e. Are free trade agreements good for the people of developing countries, bad for the people of developing countries, or don’t they make a difference? Q.39 Do you think the U.S. made the right decision or the wrong decision in using military force against Iraq? Q.40 How well is the U.S. military effort in Iraq going? Q.41 Do you think the U.S. should keep military troops in Iraq until the situation has stabilized, or do you think the U.S. should bring its troops home as soon as possible? Q.44 Regardless of what you think about the original decision to use military force in Iraq, do you now believe that the United States will definitely succeed, probably succeed, probably fail, or definitely fail in achieving its goals in Iraq? Q.45 As I read a few specific things about Iraq, tell me if you think we are making progress or losing ground in each area. First, are we making progress or losing ground in [INSERT ITEM; RANDOMIZE; OBSERVE FORM SPLITS]? How about [NEXT ITEM]? a.F1 Training Iraqi security forces so they can replace U.S. troops b.F1 Reducing the number of civilian casualties there c.F1 Preventing terrorists from using Iraq as a base for attacks against the U.S. and its allies d.F1 Establishing democracy in Iraq e.F2 Defeating the insurgents militarily f.F2 Preventing a civil war between various religious and ethnic groups g.F2 Rebuilding roads, power plants and other services in Iraq h.F2 Getting Iraqi political leaders to work together ASK FORM A ONLY (DIFFERENT RANDOMIZATION FROM FORM 1): Q.46FA If the U.S. withdraws its troops from Iraq while the country remains unstable, would it make a terrorist attack in the United States more likely, less likely, or not make a difference? ASK FORM B ONLY (DIFFERENT RANDOMIZATION FROM FORM 2): Q.47FB If U.S. troops remain in Iraq for many years, would it make a terrorist attack in the U.S. more likely, less likely, or not make a difference? IDEO In general, would you describe your political views as… Very conservative, Conservative, Moderate, Liberal, or Very liberal? Survey questions used to measure strength of partisanship PARTY In politics TODAY, do you consider yourself a Republican, Democrat, or Independent? 1 Republican [code as: PARTISAN.] 2 Democrat [code as: PARTISAN.] 3 Independent 4 No preference (VOL.) 53
5 9 Other party (VOL.) Don't know/Refused (VOL.) IF ANSWERED 3, 4, 5 OR 9 IN PARTY, ASK: PARTYLN As of today do you lean more to the Republican Party or more to the Democratic Party? 1 Republican [code as: LEANING PARTISAN.] 2 Democrat [code as: LEANING PARTISAN.] 9 Other/Don't know/Refused (VOL.) [code as: INDEPENDENT.] 54
Items used in 2004 ANES Time Series Study analysis Political knowledge Range: [0, 6] # of items: 6 Cronbach’s alpha: .73 • One point each for correct recognition of offices held by: Dennis Hastert (V045162)
Dick Cheney (V045163)
Tony Blair (V045164)
William Rehnquist (V045165)
• One point for correctly placing Democratic Party as no more conservative than Republican Party on liberal/conservative placement scales (V043090, V043091) • Up to one point for ANES interviewer’s observed level of respondent’s political awareness (V045303). Scored very low = 0; fairly low = .25; average = .5; fairly high = .75; very high = 1. Strength of partisanship Derived from V043114, V043115: • Generally speaking, do you usually think of yourself as a REPUBLICAN, a DEMOCRAT, an INDEPENDENT, or what? (V043114) 1. Republican (code as PARTISAN) 2. Democrat (code as PARTISAN) 3. Independent 4. Other party 5. No preference 8. Don't know (code as INDEPENDENT) 9. Refused (code as INDEPENDENT) If response is 3, 4, or 5, ask: • Do you think of yourself as CLOSER to the Republican Party or to the Democratic party? (V043115) 1. Closer to Republican (code as LEANING) 3. Neither [VOL] (code as INDEPENDENT) 5. Closer to Democratic (code as LEANING) 8. Don't know (code as INDEPENDENT) News attention 55
Range: [0, 35] # of items: 5 Cronbach’s alpha: .54 • Up to seven points each for reported number of days in the past week that respondent reports s/he has: Watched national news on TV (V043014) Watched local TV news in the late afternoon or early evening (V043016) Watched local news on TV in the late evening (V043017) Read a daily newspaper (V043019) Read a daily online newspaper (V043020) Opinionation Range: [‐3.87, 0] # of items: 51 Cronbach’s alpha: .91 Calculate X, the # of items to which respondent answered “don’t know” or refused from the list of survey items below. Opinionation score is then calculated as the negative of the natural log of X+1, or – ln(X+1). Liberal/conservative self‐placement ‐7‐point scale Liberal/conservative Placement ‐ GW Bush Liberal/conservative Placement ‐ Kerry Liberal/conservative Placement ‐ Nader Liberal/conservative Placement ‐ Dem Party Liberal/conservative Placement ‐ Rep party Will national economy be better or worse in next 12 mo Interventionism by diplomacy/military: self‐placement Which party better: handling nations economy Which party better: handle war on terrorism Which party better: handle keeping out of war During last year, U.S. position in world weaker/strong Country would be better off if we just stayed home War in Afghanistan worth the cost Approve Bush handling of war in Iraq Was Iraq war worth the cost Iraq war increased or decreased threat of terrorism Spending and Services ‐ 7‐point scale self‐placement Spending and Services Placement: GW Bush Spending and Services Placement: Kerry Spending and Services Placement: Dem party Spending and Services Placement: Rep party Defense spending ‐ 7‐point scale self‐placement Defense spending scale: GW Bush placement Defense spending scale: Kerry placement Defense spending scale: Dem party placement Defense spending scale: Rep party placement Govt/private medical insurance scale: self‐placement Job and Good Standard of Living ‐scale self‐placement 56
(V043085) (V043087) (V043088) (V043089) (V043090) (V043091) (V043099) (V043107) (V043109) (V043110) (V043111) (V043112) (V043113) (V043131) (V043132) (V043134) (V043135) (V043136) (V043138) (V043139) (V043140) (V043141) (V043142) (V043144) (V043145) (V043146) (V043147) (V043150) (V043152) Job and Good Standard of Living ‐ GW Bush placement Job and Good Standard of Living ‐ Kerry placement Job and Good Standard of Living ‐ Dem party placement Job and Good Standard of Living ‐ Rep party placement Federal Budget Spending: building/repairing highways Federal Budget Spending: Social Security Federal Budget Spending: public schools Federal Budget Spending: science and technology Federal Budget Spending: dealing with crime Federal Budget Spending: welfare programs Federal Budget Spending: child care Federal Budget Spending: foreign aid Federal Budget Spending: aid to the poor Fed Budget Spending: border sec to prevent illeg imm Federal Budget Spending: war on terrorism Do rich pay right amount of taxes Do poor pay right amount of taxes Environment vs. jobs tradeoff scale ‐ self‐placement Environment vs. jobs tradeoff scale ‐ GW Bush placemt Environment vs. jobs tradeoff scale ‐ Kerry placement National economy better/worse in last year Does R favor/oppose tax cuts Pres. Bush initiated Educational attainment Range: [0, 7] Derived from V043254: 0. NA/DK number of grades; no HS diploma 1. 8 grades or less and no diploma 2. 9‐11 grades, no further schooling 3. High school diploma or equivalency 4. More than 12 years of schooling, no college 5. Junior or community college level degree 6. BA level degrees; 17+ years, no advanced degree 7. Advanced degree, including LLB 57
(V043154) (V043155) (V043156) (V043157) (V043164) (V043165) (V043166) (V043167) (V043168) (V043169) (V043170) (V043171) (V043172) (V043173) (V043174) (V043176) (V043177) (V043182) (V043184) (V043185) (V043097) (V043148)