Is It Gender, Ideology or Resources? Individual-Level Determinants of Preferential Voting for Male or Female Candidates

Abstract

This article examines the role of gender in determining preferential votes for electoral candidates in the 2009 Belgian regional elections. Specifically, we examine how far votes for male or female candidates can be explained through explicit gender-based motives versus being based on other non-gendered grounds. Our findings show that while at least half of the voters express a gender preference and this is typically in favour of male candidates, the determinants of these preferences are not predominantly motivated by gender-based concerns. More important factors are voters’ access to political resources and party affiliation. Politically disengaged and right-wing voters display a clear preference for male candidates. The findings are important in showing that institutional factors only go part of the way to explaining gender imbalances in parliamentary representation. In particular, individuals’ political outlook, rather than explicit gender preference, plays a significant role in determining whether voters support a male or female candidate.

Acknowledgements

The authors would like to thank Petra Meier for her excellent comments and highly valued contributions to earlier versions of this manuscript, as well as the three anonymous reviewers and the editor Rachel Gibson for their insightful suggestions.

Supplementary Material

Supplemental data for this article can be accessed at http://dx.doi.org/10.1080/17457289.2015.1008495.

Notes

1. Gender ideology is also different from “gender stereotypes”. Stereotypes are “logical guesses about who is better able to represent you” and relate to candidates’ traits, issue beliefs, and issue competencies (Sanbonmatsu 2002, 21). They influence voters’ preferences and to a lesser extent also vote choice for male/female candidates (Lefkofridi, Giger, and Holli 2014). Gender ideology measures voters’ general ideas about the overall organization of gender relations and shapes – indirectly – how voters evaluate candidates in politics.

2. We thank one of the anonymous reviewers for this insight.

3. Belgium has three regions: Flanders, Wallonia and the Brussels Capital Region. We focus on the first two regions because they are the largest regions and because the Regional Election Survey did not include Brussels.

4. Can you indicate on a scale from 0 to 10 to what extent you personally trust the following institutions: the judiciary, the police, the media, the political parties, the Flemish government, the Flemish parliament, the Federal government, the Federal parliament, the social movements, the politicians.

5. Can you indicate how often you use the following media: reading a newspaper, watching the news on TV, looking at a news website and listening to the radio (never, less than one day/week, one or two days a week, three to four days a week, five days a week, six to seven days a week).

6. Specific formulation: Can you indicate to what extent you have used the following: website of the political party, website of a politician, printed magazines and folders of a party, advertising of parties in the media, direct contact with party members or party activists? Answering categories: never, seldom, sometimes, often.

7. The data were weighted according to province, age, gender, education, occupation and party choice to correct for possible sampling effects.

8. Logistic regression generally requires a ratio of at least 1:50 in parameters to be estimated to observed cases (Burns and Burns 2009, 570). However, the number of cases required is increased when (as in our data) the number of cases in each category of the dependent variable are unequal (the number of respondents voting only for women (N = 119) is much lower than those voting only for men (N = 345), or for men and women (N = 376)) and also when categorical independent variables are included in the model (Spicer 2005).

9. Table 4 does not distinguish between voters who voted only for the top-list candidate and voters who voted for multiple candidates on the list. However, we have to take into account that voting for a top-list candidate might not constitute a real “gendered” choice made by voters. After all, it is parties who decide which candidate is selected at the top of the list. In order to control for this potential bias in our dependent variable, we reran the analyses of Table 4 adding a control dummy measuring whether the respondent voted only for the top-list candidate or for multiple candidates. We present these results in an online appendix (see http://dx.doi.org/10.1080/17457289.2015.1008495). The results did not change any of our main findings (although the models become clearly oversaturated and unstable due to the low N). Voting for the top-list candidate therefore also constitutes a gendered choice, rooted in voters’ resources and attitudes, and not simply the “accidental” result of a decision made by parties.

Additional information

Funding

Silvia Erzeel would like to thank the Belgian Fund for Scientific Research (F.R.S.-FNRS) for its financial support. Didier Caluwaerts wishes to thank the Research Foundation Flanders (FWO) for its financial support. Both authors also thank the PARTIREP consortium and the Belgian Science Policy (BELSPO) for their support.