Suchergebnisse

Media coverage of State of the Union proposals increases public knowledge on presidential policy initiatives, especially among individuals who follow news coverage. These estimates are based on within‐survey/within‐subjects comparisons of answers to factual questions for respondents who are simultaneously unexposed and exposed to media coverage on the same issue. In this powerful but underused design, individuals serve as counterfactuals for themselves, holding constant all relevant observed and unobserved characteristics. The findings are based on statistical analyses of data from four national surveys conducted since the late 1990s covering a range of topics from health care to Social Security reform. Watching or listening to the address directly does not appreciably affect knowledge once control variables are added for whether individuals follow news coverage of the speech, the amount of news coverage, and the interaction of these two factors.

Abstract
Social scientists employ survey methods to explore the contours of human behavior. Today there are more opportunities to collect survey data than at any time in recent history. Yet sample quality varies dramatically due in part to the availability of nonprobability samples (NPSs) from commercial survey organizations. While these kinds of surveys have advantages in terms of cost and accessibility, the proprietary nature of the data can be problematic. In this synthesis, we describe situations in which researchers typically employ NPSs and consider whether these data are fit for purpose. Next, we discuss use cases that are not widespread but may be appropriate for these data. We conclude that potential utility of NPSs will remain out of reach unless scholars confront the tension between the operation of online survey organizations and the goals of transparent research.