2014-2015 Fall Colloquia

ABSTRACT: A great deal of research suggests that the persuasiveness of messages depends on the credibility of the source. Yet little research has explored the persuasiveness of the messages delivered by highly-credited experts for their recognized expertise on certain domains but the messages concern issues that are outside the legitimate domains of their expertise; under such circumstances it is plausible for the impact of their statements to decline because people might ascribe them less credibility. The study reported in this paper explored this hypothesis with regard to two disjoint domains of expertise – scientific and policymaking expertise, through an experiment embedded in a survey of a national sample of American adults whereby respondents were randomly assigned to watch a video whereby natural scientists talking about science, or the same natural scientists talking about science plus giving policy prescriptions, yielding three main findings. First, natural scientists crossing the line by advocating policies (beyond their domain of expertise) in addition to discussing science (within their domain of expertise) reduced public endorsement of policies and reduced the desire for actions to deal with the threat of global warming, thus constituting reduced persuasion. Second, the reduction in persuasion was moderated by self-efficacy (using educational attainment as a proxy measure) and partisanship, specifically, the reduced persuasiveness of messages by natural scientists crossing the line were more pronounced among people low in self-efficacy (less educated people) than high efficacy people, and manifested among partisans—people who identified themselves Democrats or Republicans—than nonpartisans— people who identified themselves neither Democrats nor Republicans. Third, the negative effect of natural scientists crossing the line on message persuasiveness was mediated in party by trust in scientists’ statements about the environment, and watching one specific scientist crossing the line on one specific issue caused a general decline in trust in scientists about the environment, suggesting people’s tendency of over- generalization. Thus, when an expert communicates to the public as an expert from his/her recognized domain of expertise, his/her crossing the line by making assertions that go beyond the domain of his/her expertise ascribed by the public may reduce the persuasion of his/her own communication and may also cause a negative externality by damaging public trust in all experts’ communication.

ABSTRACT: In the present research, the interactive influence of ingroup status and self-esteem on attitudes toward outgroups were investigated in three experiments. Four specific hypotheses were tested in the present research: 1) that when their self-esteem was threatened, members of a high status group would increase negative evaluations of domain-relevant low status outgroups; 2) that when their self-esteem was threatened, members of a high status group would increase their positive evaluations of domain-relevant high status outgroups; 3) that when their self-esteem was threatened, members of a high status group would increase their positive evaluations, or processing fluency, of a domain-relevant low status group if that group was associated with positive characteristics that complement the high status group’s identity and 4a) that when their self-esteem was threatened, members of a high status group would increase their positive evaluations of outgroups that are viewed as promoting the status quo, independent of outgroup status, and 4b) that they would increase their negative evaluations of outgroups that are seen as challenging the status quo, independent of outgroup status.

Results from Experiments 1 and 2 supported the first three hypotheses. In both experiments, heterosexual men took an ostensible test of psychological strength. One group was threatened with false feedback informing them that they scored below most men. A control group was told they would receive feedback later in the study. In Experiment 1, threatened heterosexual men increased their positive evaluations of domain-relevant high status groups while also increasing their negative evaluations of domain-relevant low status groups. In Experiment 1, threatened heterosexual men also increased their positive evaluations of a low status group, heterosexual women; a group that complements heterosexual male identity. In Experiment 2, threatened heterosexual men were found to have increased process fluency for women dressed in swimsuits while process fluency decreased for men dressed in swimsuits. Both of these findings suggest that high status group members use low status group members for self-protection when those low status groups complement a high status identity.

Experiment 3 was designed to test the fourth hypothesis. However, self-esteem effects failed to emerge in this experiment. Differences in administration of the attitudes measures may have influenced the results. One suggestive effect to emerge was that low status group members were found to endorse system-challenging groups far more than high status group members. No differences were found in endorsements of system-promoting groups. Implications for intergroup relations research and status legitimizing ideology are discussed.

ABSTRACT: This project involves analyzing pre-election polls for the US Senate, House, Gubernatorial, and Presidential elections conducted over the 2008, 2010 and 2012 elections.  Specifically we analyzed how several methodological choices: time between the poll and election day, time in field, mode of data collection, sample size, voter type, inclusion of leaning voters in poll results, election type, election year, and firm partisanship influenced pre-election polling accuracy. Accuracy was measured in two ways: absolute difference between margin of victory in the poll and actual election and average error between poll and actual results. Through highlighting both effective and ineffective methods of conducting pre-election polling and evaluating the biases inherent with firm partisanship, we hope to increase the understanding of what produces accurate pre-election polling and the public’s ability to understand and use pre-election polling results.

ABSTRACT: Recent surveys show that Americans are increasingly distrustful of public opinion polls.  But why and when do Americans reject a poll’s conclusions?  Building off of past research showing that people are critical of scientific research when its conclusions support a view with which they disagree (e.g., Lord, Lepper, & Ross, 1979), we explore whether people see polls as less trustworthy and researchers as more biased when they disagree with the views supported by polls.  We explore whether providing additional information about polls (e.g., information about a poll’s sample) can reduce biased evaluation of opinion polls.

ABSTRACT: Immigration is dramatically changing the United States’ racial makeup. According to Census estimates, the U.S will be a majority-minority nation by 2043, and Latinos are already the largest non-White group, surpassing Blacks. Immigration is also increasing diversity within racial groups: about one-third of Latinos, two-thirds of Asians, and one-tenth of Blacks in the United States are foreign-born. Existing scholarship focuses on either the status of immigrants or the status of racial groups, but not the interaction between them, ignoring both the reality of today’s diversity and key debates on the relative assimilation trajectories of different groups. My dissertation uses a series of survey experiments to explore how Americans understand this shifting landscape and to assess the implications for immigrant assimilation. I will be presenting two components of the larger dissertation project.

The first survey experiment examines whether native-born White Americans hold different stereotypes about native-born Whites, Blacks, Latinos, and Asians relative to their stereotypes about foreign-born Whites, Blacks, Latinos and Asians. I find that native-born White Americans hold distinct stereotypes about native and foreign-born members of racial groups. For example, respondents rate U.S.-born Latinos as equivalent to Whites on several stereotype items, while rating Latino immigrants significantly more negatively. I also find that both the magnitude and the direction of the immigrant-native differential varies by racial group, such that the effect of immigrant status is most negative for Latinos, while Black immigrants are rated more positively than U.S.-born Blacks.

In the second survey experiment I build off of these findings by examining whether differences in affect (i.e., overall stereotypes) can help us understand the assimilation trajectories of immigrant groups. In sociology, a commonly accepted definition of assimilation is the decline of social boundaries between groups (Alba and Nee 2003). I measure the extent to which native-born White Americans perceive stronger social boundaries between themselves and immigrant members of racial groups, compared to the social boundaries they perceive between themselves and native-born members of racial groups. I use a conjoint design to develop measures of perceived social boundaries and test several mechanisms that may be driving differences in the assimilation trajectories of racial groups.

ABSTRACT: Survey experiments like vignette and conjoint analysis are nowadays widely used in the social sciences to elicit stated preferences and study how humans make multidimensional choices. Yet, there exists almost no research on the external validity of these methods that examines whether what respondents say they would do when making hypothetical choices accurately captures what they actually do when making similar choices in real world situations. This study compares the results from conjoint and vignette analysis on which immigrant attributes generate support for naturalization to closely corresponding behavioral data from Switzerland, where some municipalities used referendums to decide on the citizenship applications of foreign residents. Using a representative sample from the same population and the official descriptions of applicant characteristics that voters received before each referendum as a behavioral benchmark, we find that the effects of the applicant attributes estimated from the survey experiments perform remarkably well in recovering the effects of the same attributes in the behavioral benchmark. We also find some differences in the relative performance of the different designs. Overall, the paired conjoint design comes closest to the behavioral benchmark; on average its estimates are within 3 percentage points of the effects in the behavioral benchmark.

ABSTRACT: A decades-long consensus about what makes a high-quality survey in the real world has sharply deteriorated. That much is clear. But what’s next? Is there a chance for agreement around new terminology and a “fit for purpose” approach to survey research? I’ve listed below a series of articles that spotlight some of the thorniest trade-offs we now confront in public polling. I look forward to discussing these big, important questions in seminar.

ABSTRACT: Many pre-election polls identify the preferences of people labeled “likely voters”. One challenge in pre-election polling is the fact that often a substantially greater proportion of respondents report that they will vote than the proportion of the population who actually end up voting. Consequently, researchers wish to identify the subset of respondents who are truly likely voters, and different organizations and academic researchers use different approaches. Our study evaluates the effectiveness of a variety of different methods for identifying likely voters. We use data from the American National Election Studies (ANES) 2008 Time Series Study, which administered many measures that could be used to identify likely voters during pre-election interviews. Then, these same respondents were interviewed post-election and were asked whether they in fact voted. We attempt to identify the optimal combination of pre-election reports with which to effectively separate people who did vote from those who did not. We use the ANES vote intention question as a starting point and supplement it with a variety of non-demographic and demographic measures that are known to predict voter turnout. We explore several varieties of two basic, but methodologically distinct approaches: (1) dividing respondents into two categories: voters and non-voters via a discriminant function approach, and (2) assigning a probability of turnout to each respondent and then comparing the effectiveness of various cut points along the probability dimension for separating voters from non-voters. We assess model accuracy using post-election reports of turnout and estimates of candidate vote share along with external benchmarks of the demographic composition of the 2008 electorate. The result is evidence pointing to the methods that can be most effective in identifying which survey respondents will vote on Election Day. This evidence will be of value to all researchers who conduct pre-election polls or who interpret such data.