2017-2018 Spring Colloquia

ABSTRACT: The focus of the seminar will be the development and use of policymaking simulations as an online survey questionnaire. In a policymaking simulation respondents are briefed on a policy issue, presented policy options, evaluate pro and con arguments for the policy options, and finally make their recommendations, often in an interactive context that requires that they be aware of trade-offs. Policymaking simulations can be used for policy topics on which the average citizen does not have enough information to give a meaningful response. Time permitting, I will also share some fresh data on how voters feel about being consulted in this way.

ABSTRACT: We study political entry before elections for new village councils in Pakistan. We randomize one-on-one and public conversations to encourage people to run for office, and show that, together, they increase candidacy five-fold. We also randomize whether personal benefits of political office – such as gaining respect and status – or social benefits – such as the ability to help others – are highlighted during conversations. We find that social benefits boost candidacy, while personal benefits reduce it, but only when benefits are discussed in public. Finally, spending in villages where social benefits were highlighted is more aligned with citizen preferences a year after elections.

ABSTRACT: Attitudes serve multiple functions for people, such as object appraisal (Katz, 1960) and value expression (e.g., Eagly & Chaiken, 1993). In addition to these types of functions, attitudes can also serve the function of providing self-expressiveness and self-definition. In this paper, we show that attitudes toward polarizing targets (here, polarizing products) are perceived by people to be particularly self-expressive. As a result, people prefer polarizing products when they seek greater self-definition and when they consume products for self-expressive purposes.

ABSTRACT: Donald Trump’s unorthodox presidential candidacy and his unexpected Electoral College win set researchers scurrying for potential explanations. These efforts gradually cohered into two master narratives, one emphasizing economic vulnerability and its associated community dysfunctions, and the other, a variety of cultural grievances. While each has its own merits, largely overlooked in these accounts are the potential political drivers of Trump’s appeal. We contend that the seeds from which Trump’s eventual victory sprouted were sown in soil increasingly enriched by Americans’ growing disenchantment with the quality of their political representation. To substantiate our claim, we first document the trends in various indicators of representation. We find growing disconnection from people’s own representatives in Congress, increasing misgivings about the ability of the two major parties to represent the American people, and mounting displeasure with the performance of Congress. These indicators, in turn, feed into political discontent about the federal government and public officials more generally. Through a comparison with earlier 21st century presidential elections, we show that Trump uniquely capitalized on this large reservoir of public sentiment to broaden his coalition. In fact, we find that the importance of political discontent rivals that of cultural and economic variables, exposing the limitations of other leading explanations to fully account for the 2016 election result.

ABSTRACT: Traditionally, public attitudes towards government policies have been measured via surveys of probability samples from the populations of interest. Recently, however, an alternative method has been proposed that minimizes the use of survey data while generating assessments of attitudes in small geographic areas that would be extraordinarily expensive to survey, such as congressional districts: multilevel regression and post-stratification (MRP).
Generally in MRP, data from a national survey are used to build a regression predicting respondents’ attitudes on a policy issue using demographics along with other predictors that can be known for all congressional districts. Then, characteristics of each congressional district and the parameters of the first-stage regression equation are used to predict the attitudes of people in each district.
In principle, this method sounds fabulous – producing an ocean of survey results from a thimble full of data for almost no cost. But is it too good to be true?
This presentation will evaluate the success of MRP by comparing its estimates about public opinion on global warming with direct measurements of those opinions through high quality probability sample surveys at a state-level. MRP estimates will also be compared to estimates from a new technique called “Longitudinal Aggregation and Disaggregation” (LAD).
LAD begins with a set of probability sample surveys conducted over time on a single population asking the same questions repeatedly in multiple waves. Founded on the assumption of “parallel publics”, the data are stacked and analyzed using an equation that models sources of change over time and methodological factors involved in each data collection and non-independence of observations to generate predictions of disaggregated public opinion at a single point in time.

ABSTRACT: This presentation investigates when and how lay perceptions about places accurately correspond to the characteristics of its visitors. The location of 26 participants was continuously tracked for over four weeks using a smartphone application, yielding a total of 77,306 GPS location samples. Trained raters predicted the personality traits of a typical visitor to the 434 most frequented places. In addition, each location was coded for several psychological criteria (e.g., liveliness of atmosphere, perceived safety of environment), generating 34,720 ratings. While personality impressions yielded moderate consensus (mean ICC=.34), they demonstrated little accuracy when compared to the ground truth, the actual personality traits of the average visitor (only significant for neuroticism, r = .11). A Brunswik Lens Model analysis revealed which location cues raters relied on when forming impressions of a place (mean R-squared across all traits = .55), and showed that these cues had little validity (mean R-squared across all traits = .05). We discuss what factors might explain this pattern of moderate consensus but low accuracy and the social implications of this erroneous, shared reality.

ABSTRACT: The efficient collection of data about individual behaviors in real life, has constituted a central challenge in the social sciences, for a long time. The usage of smartphones and other consumer electronics for data collection in the field, represents one approach to improve this. After an short introduction to mobile sensing, I will present first results from my own research project with a focus on big five personality traits and individual behavior. Finally, I will conclude with an outlook on new research ideas and discuss remaining challenges.

ABSTRACT: In recent years, governments have used various tactics to influence elections in other countries, but we know surprisingly little about how citizens respond to foreign electoral interference. Under what conditions would citizens (dis)approve of foreign involvement in their elections? To what extent does foreign electoral intervention sap confidence in democracy? How does election interference affect relations between countries? We use survey experiments to answer these fundamental questions. Our experiments, administered to more than 3,700 adults in the United States, reveal that even modest forms of electoral intervention provoke public ire, polarize citizens along partisan lines, and undermine faith in democratic institutions. Nonetheless, Americans are unwilling to retaliate harshly against electoral interference, especially when the interference benefits their party. Our findings suggest that electoral interference can be an effective tactic for manipulating and/or weakening an adversary, without running the risks associated with conventional military intervention.

ABSTRACT: In recent years, terrorist attacks in Western countries have reignited debates concerning the trade-off between national security and democratic freedoms. While associations between terrorist attacks, political ideology, and political attitudes have gained significant attention in scholarly work, the role of political elites in shaping these associations has not yet been examined. The present study examines the extent to which party cues can exacerbate – or mitigate – the negative effects we so often witness in the aftermath of terrorist violence. To do so, a survey experiment was disseminated to 4,700 U.S. adults. Participants were randomly assigned to a treatment condition – using a highly realistic priming technique consisting of actual breaking news coverage of recent terrorist attacks – or a control condition. Subsequently, participants read one out of five vignettes containing various degrees of polarized and non-polarized party cues, and indicated their support for harsh security measures. Results suggest that, when primed with terrorism, support for harsh security increases in aggregate. However, party cues exert a notable impact on this association. On the one hand, polarized party cues increase the gap between partisans on policy support, and do so relative to the degree of elite polarization. On the other hand, unified elite cues supporting or opposing harsh security measures mitigate these rifts. Since political elites play a large role in communicating group norms, these findings suggest that they might be particularly influential under conditions of societal threat – for better or for worse.