Suchergebnisse

AbstractThis article introduces a framework for evaluating methods of combatting information search in online surveys. Three empirical studies based on the framework suggest that search is a serious but manageable problem. Search frequency varies substantially according to question content, ranging from 2% to 30% in batteries of general political knowledge questions. Deterrence works: a pledge not to cheat reduces search by half. Detection also works: web browser paradata identify 70% to 85% of search, while 60% to 85% of search on knowledge questions is undertaken by respondents who correctly answer "catch" questions about obscure Supreme Court cases. Detection and deterrence are complementary: deterrence reduces search ex ante, while detection quantifies success and provides ex post options for dealing with undeterred search. In combination, the three methods tested (pledge, paradata, and catch) deter or detect more than 90% of search, leaving search to affect about 0.5% of the remaining observations.

AbstractSurvey data are commonly cited as evidence of widespread misperceptions and misinformed beliefs. This paper shows that surveys generally fail to identify the firm, deep, steadfast, confidently held beliefs described in leading accounts. Instead, even those who report 100% certain belief in falsehoods about well-studied topics like climate change, vaccine side effects, and the COVID-19 death toll exhibit substantial response instability over time. Similar levels of response stability are observed among those who report 100% certain belief in benign, politically uncontested falsehoods—for example, that electrons are larger than atoms and that lasers work by focusing sound waves. As opposed to firmly held misperceptions, claims to be highly certain of incorrect answers are best interpreted as "miseducated" guesses based on mistaken inferential reasoning. Those reporting middling and low levels of certainty are best viewed as making close-to-blind guesses. These findings recast existing evidence as to the prevalence, predictors, correction, and consequences of misperceptions and misinformed beliefs.

AbstractThe most familiar approach to handling respondent uncertainty about survey questions, the "don't know" (DK) response, has an inconvenient feature: it records no direct information about respondents' confidence in their answers. This article shows that DK responses contain more information than meets the eye. Using 161 survey questions from 11 surveys conducted between 1993 and 2019, the analysis demonstrates that the percentage of respondents saying DK can explain about half the variation in average confidence among respondents who provided an answer (i.e., those who did not say DK). This "we don't know" means "they're not sure" heuristic is especially reliable for relative comparisons between similar questions. For example, in a battery of favorability questions featuring two recent presidents and 10 contenders for the 2020 Democratic presidential nomination, the percentage DK explains more than 90 percent of the variation in average confidence among those who offered an opinion. To provide a deeper intuition, the article draws upon a threshold model of DK responding, which is consistent with a range of existing research on the properties of DK responses.

Crises and disasters give voters an opportunity to observe the incumbent's response and reward or punish them for successes and failures. Yet, even when voters perceive events similarly, they tend to attribute responsibility selectively, disproportionately crediting their party for positive developments and blaming opponents for negative developments. We examine selective attribution during the COVID-19 pandemic in the United States, reporting three key findings. First, selective attribution rapidly emerged during the first weeks of the pandemic, a time in which Democrats and Republicans were otherwise updating their perceptions and behavior in parallel. Second, selective attribution is caused by individual-level changes in perceptions of the pandemic. Third, existing research has been too quick to explain selective attribution in terms of partisan-motivated reasoning. We find stronger evidence for an explanation rooted in beliefs about presidential competence. This recasts selective attribution's implications for democratic accountability.

Surveys often ask respondents how information or events changed their attitudes. Does [information X] make you more or less supportive of [policy Y]? Does [scandal X] make you more or less likely to vote for [politician Y]? We show that this type of question (the change format) exhibits poor measurement properties, in large part because subjects engage in response substitution. When asked how their attitudes changed, people often report the level of their attitudes rather than the change in them. As an alternative, we propose the counterfactual format, which asks subjects what their attitude would have been in the counterfactual world in which they did not know the treatment information. Using a series of experiments embedded in four studies, we show that the counterfactual format greatly reduces bias relative to the change format.

Is support for democracy in the United States robust enough to deter undemocratic behavior by elected politicians? We develop a model of the public as a democratic check and evaluate it using two empirical strategies: an original, nationally representative candidate-choice experiment in which some politicians take positions that violate key democratic principles, and a natural experiment that occurred during Montana's 2017 special election for the U.S. House. Our research design allows us to infer Americans' willingness to trade-off democratic principles for other valid but potentially conflicting considerations such as political ideology, partisan loyalty, and policy preferences. We find the U.S. public's viability as a democratic check to be strikingly limited: only a small fraction of Americans prioritize democratic principles in their electoral choices, and their tendency to do so is decreasing in several measures of polarization, including the strength of partisanship, policy extremism, and candidate platform divergence. Our findings echo classic arguments about the importance of political moderation and cross-cutting cleavages for democratic stability and highlight the dangers that polarization represents for democracy.

Abstract
In today's competitive information environment, clicks are the currency of the digital media landscape. Clickbait journalism attempts to entice attention with provocative and sensational headlines, but what are the implications when public opinion polls are the hook? Does the use of survey clickbait—news stories that make misleading claims about public opinion—have implications for perceptions of the public, journalists, or the polling industry? In two survey experiments conducted in the United States, we find that exposure to apolitical survey clickbait that makes exaggerated claims about the incompetence of the American public undermines perceptions of their capacity for democratic citizenship. At the same time, we find no evidence that this type of survey clickbait damages the reputations of the media or polling industry, suggesting that the media may have perverse incentives to use low-quality polls or to misrepresent polling results to drive traffic.

Abstract
This study adds to the analogic perspective-taking literature by examining whether an online perspective-taking intervention affects both antisemitic attitudes and behaviors – in particular, engagement with antisemitic websites. Subjects who were randomly assigned to the treatment viewed a 90-s video of a college student describing an experience with antisemitism and reflected on its similarity to their own experiences. In a survey, treated subjects reported greater feelings of sympathy (+29 p.p.), more positive feelings toward Jews, a greater sense that Jews are discriminated against, and more support for policy solutions (+2–4 p.p.). However, these effects did not persist after 14 days. Examining our subjects' web browsing data, we find a 5% reduction in time spent viewing antisemitic content during the posttreatment period and some limited, suggestive evidence of effects on the number of site visits. These findings provide the first evidence that perspective-taking interventions may affect online browsing behavior.