2019-2020 Winter Colloquia

ABSTRACT: Using an online survey experiment on news reports of Hong Kong’s “Occupy Central” movement, the study described in this talk finds the effects of media frames on Chinese netizens’ political attitudes can occur and are conditional on contextual factors as well as individual attributes. More specifically, conservative netizens who were exposed to a news report on the movement with a heated debate significantly prefer economic development and social stability to political freedom than those who read the news only. The framing stimuli, however, can hardly be identified among the liberal netizens. The empirical results show framing effects on Chinese netizens’ political attitudes are consistent, not only on Hong Kong’s issue, but on China’s development strategies in general. This paper highlights the importance of people’s pre-attitudes in framing research of public opinion even with random assignment. The important political implications in the context of Chinese politics are discussed as well.

ABSTRACT: Celebrity involves contests over who will create, define, and appraise those known, in their lifetimes, to more people than can possibly know one another. For individuals to become famous on such a large scale, their images, names, and stories must proliferate, gaining rather than losing interest and resonance as they travel. This talk argues that celebrity does not operate according to a simple metric of supply and demand in which the more images there are of a star, the less value they have. Instead, the more copies, the more celebrity. With each new series of multiples, the celebrity becomes all the more singular. As a result, celebrity culture expands every time a new technology makes it easier and cheaper to produce and distribute copies on multiple scales, in multiple media.

ABSTRACT: These days, it can be difficult to know what Americans really think, and the opinion research industry has been criticized for not accurately capturing all Americans’ views. Some segments of society—such as low-income households, social and political conservatives, youth, rural households and less-educated people—can be harder to reach and routinely are under-represented in polls.

This presentation looks at five AmeriSpeak surveys and analyzes how AmeriSpeak’s extra outreach to these initial nonrespondents, which we call “nonresponse follow-up,” ended up affecting the final survey data.

The AmeriSpeak Panel is the only commercially available household panel in the U.S. that uses professional field interviewers for non-response follow-up during the household recruitment process. Once recruited to the probability-based AmeriSpeak Panel, adult panelists participate in two to three surveys a month. AmeriSpeak supports a broad range of surveys conducted for academics, federal agencies, foundation-supported researchers, non-profit policy organizations, news media, and some commercial marketing and media measurement companies.

In short: Because of NORC’s investment in nonresponse followup, the AmeriSpeak surveys have improved representation of persons who read the news less often , those who have more moderate political opinions, and those who have not thought much about politics or political ideology. The result is that AmeriSpeak’s measurement of the public’s policy preferences reflect this improvement in sample representation. The implication is that polls that do not invest heavily in nonresponse follow-up are more likely to over-state political polarization in our society.

ABSTRACT: Internet technologies and the World Wide Web (WWW) have multiplied the uses of anonymity. Some of these uses are negative, as in the widespread of cyberbullying and hate speech made possible by social network sites such as Yik Yak, Whisper, and Secret. Other uses seem positive, as is the use of anonymity by political dissidents or journalistic sources. As it can be inferred by these contrasting examples, the social effects of anonymity are perplexing and often contradictory, and it is these complex effects that invite a thorough exploration of anonymity. What is anonymity? How should we think about it? What can we learn from anonymity’s contradictory effects? And, how does anonymity’s historicity connect to its contemporary contradictory effects?

ABSTRACT: Surveys and experiments using Amazon’s Mechanical Turk are often used in political science, but research on inattentive respondents on the medium has struggled to find accurate measures of attention. Identifying inattentive respondents correctly is necessary for the internal validity of studies because the inclusion of such respondents in samples may lead to Type 1 errors, introducing sizable correlations when in fact none exist. I introduce a new way to identify inattentive respondents using mouse-tracking technology and show that their inclusion leads to Type 1 and Type 2 error in experiments. Additionally, I show that current methods of identifying inattentive respondents fail to capture their true number and the extent of variance in attention across survey questions. Last, the design-related determinants of inattention, such as the length of prompts, are analyzed and recommendations made for scholars seeking to reduce the number of inattentive respondents in surveys. My results have implications for both the design and implementation of surveys and experiments.

ABSTRACT: Scholars of politics in the US and elsewhere have long debated whether policy preferences influence vote choice. Some claim that policy preferences have no impact on the way people vote, while others argue that they strongly influence voting behavior. However, there is no consensus about the appropriate way to study policy voting. I define policy voting as a change in vote choice due to pre-existing policy preferences. I then identify four models of voting behavior used in studies of the 2016 election. Using data from the 2010-2014 Cooperative Congressional Election Study (CCES), I show that these models give different results due to prior mutual influence between issue preferences and vote choice. I then run simulations and show that cross-sectional models are nearly always biased, overestimating the impact of conventional issues, and underestimating the impact of novel issues. Panel analyses that ignore change have the opposite biases. However, using a panel model of change in vote choice nearly always leads to the right answer. I conclude that scholars seeking to make claims about the impact of policy preferences on voting behavior should think carefully about the model they use.

ABSTRACT: This lecture focuses on the development of the Project Mercury earth station in Zanzibar during the period 1959-1964. To historicize the earth station’s establishment, I adopt a nodal approach and combine archaeological, archival, and phenomenological methods in an effort to bring forth the geopolitical and sociotechnical relations that resulted in the Zanzibar station. My discussion moves from a general description of Mercury’s “worldwide tracking” network, to an analysis of Zanzibari opposition to the station, to a recounting of the building of the station in the midst of this opposition. This earth station, not only contributed to the science of satellite tracking and telemetry, it was an essential node in the first “world wide tracking network” to rely on real-time computing to monitor a manned satellite. What is not as well known, however, is the precarious geopolitical fulcrum upon which the Zanzibar Mercury station’s precise measurements were taken. Given this, I define the station as a contrapuntal node — as a site opposed by local publics — to raise questions about the histories and materialities of other network facilities that have been built against peoples’ will. While network extensions and occupations have been structural to colonial power, Africans’ responses to and involvement in the formation of particular network nodes is much lesser known. These material relations are significant as they helped to shape early global real-time computing networks that became precursors of the internet and world wide web.