A policy-oriented electorate: evaluations of candidates and parties in the Obama elections compared to the 1952-1980 period.

Author:Wattenberg, Martin P.

It is often said after an election that the people have spoken and scholars regularly debate what they have said. This article replicates Miller and Wattenberg's (1985) framework for analyzing what Americans said about the candidates and parties, using the transcripts from the 2008 and 2012 American National Election Studies (ANES). (1) We revisit the question of how voters frame their discussion of candidates and parties employing data from the Obama era. Our findings reveal remarkable changes since the time of the publication of The American Voter (Campbell et al. 1960). Whereas American voters once focused on results produced by the government, in the Obama elections they were much more inclined to try to guide future policies.

First, we describe the primary research question examined in this project and the expectations we have of the data. Next, we explain our sources of data, outlining the advantages and disadvantages of using open-ended data and the coding scheme. We then present our findings regarding candidate and party evaluations. Our emphasis in these sections is on the proportion of different types of evaluations, and we also assess whether respondents were considering candidates and parties in retrospective or prospective terms. We find that respondents' evaluations of both candidates and parties were far more policy oriented in the Obama elections of 2008 and 2012 than they were in the 1952-80 period. Finally, we examine and present voters' rationales and investigate who is most oriented toward policy arguments.

Research Question and Expectations

There have been many changes in the political landscape since 1980, and we investigate whether evaluations of candidates and parties during the Obama elections show change as well. Contributing primarily to the voting and campaign literature, our findings will show whether rationales are the same or distinct from previous elections. We hypothesize policy evaluations to play a greater role in 2008 and 2012 than in the years Miller and Wattenberg coded for a variety of reasons.

To begin, the composition of the electorate has changed significantly since the 1980s due to generational replacement. Of the voting age population in 2012, 56% were not old enough to have been eligible to vote back in 1980. Newer entrants into the electorate have been socialized in a much more polarized environment with regard to policy, as will be discussed below. In addition, they have acquired far more education than the people they have replaced. The greater level of education attained by today's electorate gives more people the ability to absorb advanced concepts, such as policy stands, as opposed to making a simple assessment of whether the nation is better or worse off (Abrajano 2005; Dalton 2008, chap. 1).

Furthermore, political elites have made learning about policies easier for all voters by sorting themselves into partisan camps that now clearly differ according to many policy stands. While there is much disagreement as to whether or not the electorate has become more ideologically polarized (Abramowitz 2010; Fiorina 2010), there is consensus that party elites have sorted themselves to present clear and consistent differences between the presidential nominees, as well as most congressional candidates (Hetherington 2001; Levendusky 2009). Whereas clearly conservative nominees, such as Barry Goldwater in 1964, or clearly liberal nominees, such as George McGovern in 1972, were once anomalies, now the electorate regularly chooses between polar opposites. In addition, Layman and Carsey (2002) show that clear conflict between the parties now occurs on multiple fronts, encompassing social welfare, racial, and cultural policies. Because choices in American politics are clearer now than they were in the period Miller and Wattenberg analyzed, we expect the public to give policy considerations greater weight in the Obama elections.

A final reason we expect policy evaluations to have increased since the 1952-80 period is that the information environment in which political campaigns take place has changed markedly. Consider, for example, that many policy differences are now prominently presented in policy-oriented political commercials, which have increasingly come to dominate modern political campaigns (West 2014). Although the vast increase in campaign spending on political commercials is often criticized as poisoning the political environment with a spate of charges and countercharges, one positive side to more advertising is that the electorate has more opportunities to be exposed to policy arguments. Research over the years has shown that people do indeed learn a significant amount about public policy stands from TV commercials (Brians and Wattenberg 1996; Patterson and McClure 1976). Gilens, Vavreck, and Cohen (2007) demonstrate that the American public's greater focus on issues can be traced to the high level of policy content in paid political commercials. In addition to advertising, television programming choices are now larger than ever, and some scholars have argued that nontraditional sources of political news, such as soft news (Baum 2005) and late night comedy shows (Parkin 2010), have policy effects on certain segments of the public. Further, beyond television, in recent elections the Internet has made policy information widely available to large swaths of the electorate (Tolbert and McNeal 2003). The 2008 Obama campaign, in particular, was able to mobilize youth voters in great numbers using their digital strategy (Harfoush 2009). Given the changes in accessible campaign information, in volume, medium, and strategy, we expect the electorate to exhibit a greater propensity toward policy arguments made during the 2008 and 2012 elections than in the 1952-80 period.

Sources of Data: Advantages and Disadvantages of Open-Ended Questions

In order to test our hypothesis that more Americans are now evaluating candidates and parties in policy terms, we decided to replicate the Miller and Wattenberg (1985) coding scheme for analyzing the open-ended questions from the ANES. An excellent way to directly ascertain the nature of the messages people send on Election Day is simply to ask a representative sample of citizens what was on their minds as they evaluated the candidates and parties. A set of eight open-ended questions in the ANES have been asked in every presidential election dating back to 1952. This question set asks what people like and dislike about the Democratic and Republican presidential nominees and their parties. Each survey essentially involves short conversations with a random sample of the electorate that year; people are free to say anything they want, unconstrained by the limited number of choices provided by closed-ended questions. A great benefit of the ANES open-ended questions is that they allow respondents to identify what is most important to them about the candidates and the parties, regardless of whether such considerations might occur to the designers of the survey. Furthermore, people can put these considerations into their own words, explaining their own reasoning for why such factors are important to them.

However, this survey technique does have some limitations. For example, some respondents say little or nothing to the open-ended questions, either because they are unable to express themselves or are reluctant to talk openly. Other respondents simply repeat what they have heard recently from friends or encountered that day in the media. Additionally, the possibility remains that some respondents will provide what they perceive to be socially acceptable answers, rather than what is foremost on their minds. Consider, for instance, that many may have been reluctant to say they were voting against Barack Obama because of his race or because of concerns that he was a Muslim (though we were startled by how many respondents did indeed say things like this).

For the purposes of our analysis here, there is little reason to suspect that such problems with open-ended questions are too different from one cross-section to another. Hence, for assessing whether voters now are thinking in similar or different frameworks from past years, these data are virtually ideal. It is exceedingly difficult to write closed-ended questions that will be relevant across many campaigns; open-ended questions resolve this difficulty by leaving it to the respondents themselves to define what is most relevant.

Reading the verbatim transcripts of what respondents have said to the ANES open-ended yields many insights into what people were thinking when they cast their votes. Unfortunately, all but a few of the scholarly analyses of these data since 1952 have been limited to analyzing the responses coded by the ANES, in which each individual comment is classified into one of several hundred generic categories (e.g., general assessment of economic policy; social security/pensions; foreign policies more clear-cut/decisive). Reading the transcripts as opposed to using coded responses is akin to the advantage a teacher gets from reading a student's short essay as opposed to a few multiple-choice questions. Multiple-choice answers are necessarily limited to a broad brush of generic responses, whereas reading a few sentences can reveal one's reasons for thinking a certain way, as each segment builds on the others to provide insight into one's framework of analysis. Yet, just as with exams from a large lecture class, reading a set of essays is far more time consuming than checking multiple-choice responses.

For our analysis of the 2008 ANES open-ended questions, we read how 2,323 randomly chosen citizens responded to a series of the eight questions about the presidential candidates and the political parties. The mean number of open-ended questions answered was 4.04; therefore, we coded about 9,380 responses to the 2008 ANES survey. In order to make this project more...

To continue reading