CEU Electronic Theses and Dissertations, 2017
| Author | Means, Jason Donovan |
|---|---|
| Title | What Do We Actually Know? Correcting for Differential Item Functioning with a Survey Experiment |
| Summary | Despite being one of the best-documented features of contemporary U.S. politics, scholars are still in disagreement on how to address the extent to which voters are uninformed about politics. A large portion of the electorate is comprised of low-information voters—a fact that is dissonant to most characterizations of democratic citizenship. This study seeks to discern whether the electorate, specifically African Americans, are actually uninformed about politics or whether their general level of political knowledge is misrepresented due to knowledge batteries biased in favor of majority (White) Americans. Before testing for differential item functioning, a methodological problem causing respondents to answer survey questions in vastly different ways from one another, a theoretical argument against disenfranchising low-information voters is presented, and is followed by an overview of the literature on political knowledge and its measurement. This study’s findings remain inconclusive. Though its treatment has an effect, the extent of that effect is uncertain. Multiple regression analysis on factors influencing respondents’ level of political knowledge shows a statistically significant relationship between race and knowledge, however, only among control-group respondents. Further research is needed to assess whether item bias may be reduced through altering the construction of surveys measuring political knowledge. |
| Supervisor | Toka, Gabor |
| Department | Political Science MA |
| Full text | https://www.etd.ceu.edu/2017/means_jason.pdf |
Visit the CEU Library.
© 2007-2021, Central European University