Individual Submission Summary

Direct link:


In my Humble Non-Opinion: Confidence Drives the Gender Gap in Opinion Reporting

Sat, August 31, 4:00 to 5:30pm, Hilton, Tenleytown East


Why do women offer “don’t know” responses more often than men when asked for opinions on political issues? There has been some important and widely-cited work in political science on the ways in which “don’t know” responses in public opinion surveys can systematically misrepresent how respondents truly feel. For example, Berinsky (2004) showed that those who hold “unpopular” (e.g., bigoted) views tend to respond “don’t know” to avoid revealing opinions they know are widely reviled. Past research has also shown that women have lower levels of political knowledge. This gap is sometimes attributed to lower access to education, opportunities to develop opinions through discussion with peers and other important resources. This knowledge gap suggests one obvious reason why women might be more likely to offer non-opinion responses on public opinion surveys. Here, I show both that the causes of women’s non-opinion responding are wider ranging than has previously been acknowledged and that the consequences of this gender gap are much farther reaching—resulting in systematic underrepresentation of the views and priorities of women—fully half of the American electorate.

In this paper, I show first, using data from the 2012 ANES, that there is a statistically significant and substantively large gender gap in opinion reporting that is robust even when including controls for myriad demographic and resource variables. This suggests that informational and educational differences by themselves are unlikely to be causing this hesitance among women when considering whether to offer their opinions.

I theorize that the gender gap in non-opinion responding is caused by men’s and women’s different levels of comfort expressing their opinions to others, which may result from long-standing gender norms and expectations. From a young age, boys are rewarded for confidence and boldness, while girls are rewarded for demure behavior and chastised for boldness. I argue that women have a higher informational threshold that must be crossed in order for them to be confident enough in their competence for offering an opinion. Consistent with this theory, I find that women require more information about a new policy than men do before they are willing to offer an opinion about it. I designed a survey vignette that gave respondents one piece of information about a real but obscure new policy proposal they were unlikely to know about. This policy would replace the current gas tax with a per-mile driving tax. After each piece of information, respondents could opt to report opinions, or could ask for more information before reporting their opinions. Female participants requested almost 40% more information than men did before they were willing to offer an opinion about it.

Finally, to conclusively rule out resource- or information-based explanations for the gender gap in non-opinion responding, I conducted a within-subjects survey experiment in which participants were asked two types of questions: (1) ones identical to the issue questions on ANES, and (2) questions about real-sounding but fictitious policies (e.g., “Do you favor or oppose the Interest Rate Restructuring Program (IRRP) currently under consideration by the Federal Reserve?”). The logic of this design is that fictitious issues reverse the expected relationship between knowledge and non-opinion responding: the more a person knows, the more likely they are to recognize that an issue is not real and decline to offer an opinion. If differences in information or related resources explain the gender gap in non-opinion responding, that gap should be eliminated or reversed for fictitious issues. By contrast, if my theory is correct and the gender gap is caused by psychological differences—women feeling less confident in the amount of information they have and men feeling pressure to appear decisive—the size and direction of the gap should be similar, whether questions are about real or fake issues. Not only were men significantly more likely than women to offer opinions about the fictitious issues, the size of the gap was larger for those issues than it was for most real ones.

This work demonstrates that there is more to the broader story of gender differences in political participation than just material and informational resources. Moreover, the fact that differences in confidence reduce women’s participation even in a form as unthreatening as survey taking suggests that these psychological mechanisms might also be driving participation gaps in political activities that require more assertiveness (e.g., volunteering on campaigns, running for office). I conclude by discussing ideas for simple, low-cost interventions that might ameliorate this serious problem.