![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
ABSTRACT
Across Africa, it has been found that women participate in politics less than men, undermining prospects for gender equality and shared development. Despite theoretical and practical reasons for concern over the “gender gap”, we lack insight into its variation across countries, particularly in Africa, where studies accept women’s lower rate of participation as a uniform background condition. I address this lacuna using four rounds of Afrobarometer data from 31 countries to identify predictors of country-level variation in the gender gap. Drawing from literature on gender gaps elsewhere, I identify institutional and structural variables thought to influence women’s political engagement relative to men’s and evaluate these hypotheses quantitatively at the cross-national level, finding that country-level gaps vary predictably based on the percentage of female legislative representatives, the duration of democracy, and French colonial history. Finally, I use the least-likely case of Senegal to explore how the process to increase women’s legislative representation coincided with a dramatic increase in women’s participation.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 According to the index of political participation developed in this study from Afrobarometer Round 6 data.
2 I exclude Rounds 1–2 because key questions were absent or worded differently than in later rounds.
3 Notable exceptions include Coffe and Bolzendahl (Citation2011) and Isaksson, Kotsadam, and Nerman (Citation2014), using Rounds 3 and 4 of the Afrobarometer, respectively, but these analyses require updates with more recent data.
4 This analysis includes Afrobarometer rounds 4–6, as the question regarding group membership was worded differently in Round 3. Respondents who answered “don’t know” or failed to respond were dropped from the analysis.
5 Standard SES predictors per Verba and Nie (Citation1972).
6 Due to the low reliability of self-reported income in African counties, other scholars (Bratton 2006; MacLean Citation2011) using the Afrobarometer prefer to use such an index of “lived experiences” of poverty.
7 Per Brady, Verba, and Scholzman (Citation1995).
8 Previous analyses have shown that religious observance may also influence political participation. However, these effects are highly contextual, including distinctions between sects that are not available in the Afrobarometer, and have contradictory predictions. Without the ability to provide more fine-grained, contextual analysis for the possible political implications of different forms of religious observance in each country context, I excluded this indicator from analysis.
9 This analysis excludes Round 3 due to different wording on the question for ‘group membership’.
10 This binary distinction is an oversimplification, but captures different attitudes towards colonization – even if the reality on the ground resulted in similar practice, as Herbst (Citation2000) argues.
11 Rounds One and Two were excluded because they are missing key questions.
12 From the World Bank.
13 The Afrobarometer is a nationally representative random sample, so this measure should reflect women’s labor force participation. The results are robust to alternative specification of this variable using the World Bank’s estimate of female labor force participation.
14 From Interparliamentary Union.
15 Many countries with voluntary quotas have similar outcomes as those with mandatory quotas. Several countries have quotas restricted only to municipal elections. I excluded those due to the assumption that local representatives are unlikely to generate the symbolic effects that the theory implies.
16 Following Tripp’s Citation2008 coding. She did not code Senegal or Cote d’Ivoire; I coded Senegal as 0 due to the limited nature of the Casamance conflict, and Cote d’Ivoire 1 due to its major civil conflict (ending as of 2007).
17 Greater than 50% of the population reporting Muslim affiliation, as measured by Pew. The results are robust to including the percentage of Muslims in the population rather than the binary variable. I prefer the binary variable because majority/minority status likely has a larger impact on politics than the overall percentage.
18 Round 6 includes all countries displayed in . Round 5 omits Sao Tome and Principe and Gabon. Round 4 further omits Mauritius, Sierra Leone, Niger, Togo, Burundi, Cameroon, Cote d’Ivoire, Guinea, and Swaziland. Round 3 further omits Burkina Faso and Liberia.
19 Estimating these models together through Seemingly Unrelated Regression (SUR) creates more efficient estimates with smaller standard errors and larger coefficients. Using this modeling strategy, the main results are the same, and some additional coefficients gain significance. I focus my analysis here to the results that were robust to both GLS and SUR modeling.