0
The influence of gender quotas on support for democracy
Bachelor Thesis
17-06-2019
Word count: 7687
Dorine klein Gunnewiek
s1812351
International Relations and Organisations
Department of Political Science
Leiden University
Instructor: Dr. L. Demarest
Abstract
Africa has made many strides towards democratization in recent years, but it has also faced several setbacks, such as armed conflict and coups. One of the most important factors that contributes to successful consolidation of democracy is support for democracy. This thesis will provide an answer to the question whether the implementation of gender quotas leads to a difference in support for democracy in African countries. By means of logistic regression analyses of data collected by the Afrobarometer, I find that the mere presence of gender quotas does not account for the stark differences in support for democracy between countries. However, I do find that when mandatory gender quotas are implemented, they increase support for democracy over time, in contrast to general gender quotas. Yet I also show that, in contrast to popular believe, this is not caused simply because people see an increase in women’s representation. Rather, countries that implement gender quotas might also influence support for democracy in other ways.
1 1. Introduction
In the last two to three decades, the use of gender quotas has seen a steady rise throughout the world (Fallon, Swiss, & Viterna, 2012). Especially in developing nations, gender quotas have become more and more mainstream throughout recent years and in Africa, many states have actively implemented gender quotas (Bauer & Britton, 2006). Countries that have implemented gender quotas include Senegal, that saw the percentage of women in parliament rise from below twenty per cent to above forty per cent, and Rwanda, the only country in the world in which more than half of the parliament is made up of women (Bauer & Burnet, 2013).
A lot of research regarding gender quotas and their influence has been done (e.g. Bauer & Burnet, 2013; Chen, 2010; Kang, 2013; Rosen, 2017). Most research focuses on country-level effects, either on the relationship between gender quotas and regime type or the influence of gender quotas on the degree of women’s representation. This thesis, however, focuses on the influence of gender quotas on the individual attitudes of people concerning support for democracy. The research question that will therefore be answered in this thesis is: “Does the implementation of gender quotas lead to a difference in support for democracy among citizens?” This research question will be tailored to Africa specifically, since many countries in Africa have recently implemented at least one type of gender quotas.
I make use of quantitative analysis by drawing on data from the Afrobarometer project. Afrobarometer collects data on several African issues, such as “democracy, governance [and] economic conditions […] in more than 35 countries” (Afrobarometer, 2019). For data on gender quotas I make use of the Gender Quotas Database from the International Institute for Democracy and Electoral Assistance (IDEA, 2019). The analysis indicates that it is important to distinguish between mandatory gender quotas, which include legislated gender quotas and reserved seats gender quotas, and voluntary party gender quotas. Indeed, results indicate that the implementation of mandatory gender quotas positively influences support for democracy, but that this effect does not apply to a broader definition of gender quotas including voluntary ones.
The next section provides a theoretical framework concerning gender quotas and discusses existing research and its findings. Section 3 will display the different sub-research questions and nine different hypotheses that follow from these questions. In section 4, the research methods will be discussed. Then, in section 5 the results of the analyses will be presented.
2
Lastly, section 6 summarizes the main findings, highlights areas for further research and provides policy recommendations stemming from the findings.
2. Theoretical framework
There are two different ways in which an increase in women’s representation can be achieved: the fast-track and the incremental track (Dahlerup & Freidenvall, 2005). The incremental track is guided by the philosophy that unequal representation is due to the difference in political opportunities offered to women and men. The reasoning is that as soon as you make sure there is equality in political opportunity, equality in representation will follow. On the contrary, when quotas are used, the fast track is implemented, based on the philosophy that the incremental track is not generating results fast enough and thus artificial methods are needed to balance the scales (Tripp & Kang, 2008). This raises the question whether gender quotas actually bring forth these results. As Chen (2010) shows, gender quotas positively influence both the degree of women’s representation in legislatures and policy decisions that are being taken. This is corroborated by evidence from Morocco, where the introduction of reserved seats gender quotas led to an increase in women’s parliamentary representation (Darhour & Dahlerup, 2013).
However, as Rosen (2017) shows, the introduction of gender quotas is not spread equally around the world, but emerges mostly in developing countries. One reason for this could be that developing countries implement gender quotas to climb the ranks of the global hierarchy and to appeal to Western ideas of human rights and democracy (Towns, 2013). Bush (2011) adds that international politics are the main reason why these developing nations seem so keen on implementing gender quotas, even though they do not necessarily seem to support equality between men and women in other aspects of life. She argues that international forces drive the implementation of gender quotas through international democracy assistance, because of “the shared belief that quotas are necessary and appropriate for developing democracies” (Bush, 2011, p. 131). The question is, however, whether this belief is rooted in reality.
2.1. Gender Quotas
Before delving into the empirical research that has been done on gender quotas and democratization, it is important to first establish what exactly gender quotas are. Although motives for the implementation of gender quotas can differ vastly, gender quotas are almost always a way to increase women’s representation in legislatures, by stipulating a minimal number of places on the electoral lists or in legislatures (Tadros, 2010). Gender quotas can be either voluntary or mandatory. The Gender Quota Database from the International Institute for
3
Democracy and Electoral Assistance identifies three types of gender quotas: Legislated gender quotas, reserved seats gender quotas and voluntary party gender quotas (IDEA, 2019). The first two types mean that the law clearly states that a number of places for female candidate have to be reserved. For the first type, they have to be reserved on the electoral lists; for the second type they have to be reserved in the legislating assembly. The third type is a voluntary gender quota, which means that some parties in the country, at least one, have committed themselves to reserving a certain number of seats on their electoral lists for female candidates. In these countries, there are no laws that stipulate a minimal number of places or seats have to be reserved for women.
2.2. Gender quotas and democratization
Several scholars argue that the implementation of gender quotas leads to a higher degree of women represented in national legislatures (Tripp & Kang, 2008; Messing-Mathie, 2011; Freidenvall, 2005). This seems evident, but it shows that gender quotas are not only implemented as a symbolic measure, but also have a real tangible effect. A larger percentage of women in parliament is thought to have a positive influence on politics and law-making, since research shows the increase of female policy makers negatively impacts corruption rates (Kumar Jha & Sarangi, 2018). Although other research opposes these results, attributing the lower rates of corruption to culture instead of gender (Debski, Jetter, Mösle, & Stadelmann, 2018), this is a widely shared belief that is often mentioned as one of the reasons behind the implementation of gender quotas (Towns, 2013). This reduction of corruption is part of the main reason why the implementation of gender quotas would lead to a process of democratization: Gender quotas will lead to more women in legislatures, which will lead to policies that better represent the population (Bush, 2011). This improves the quality of democracy. Mi Yung Yoon (2013) provides empirical evidence for this claim. She uses qualitative data, newspaper clippings and interviews that she conducted with Tanzanian members of parliament to show that gender quotas in Tanzania have had a positive influence on the country’s path towards democracy. It has helped broaden discourse in parliament and has ensured women’s issues are more prominently present in parliament. These results are corroborated by evidence from Niger and Uganda (Kang, 2013; Wang, 2013).
However, there has not been a lot of research done regarding the effect of gender quotas on the transition to democracy. Most research focuses on the reverse relationship: Whether regime type influences women’s representation in legislatures. For example, Tripp and Kang (2008)
4
do not find a strong relation between the two, arguing that regime type is irrelevant to the level of women’s participation. Bauer and Burnet (2013) show similar results, arguing that the differences between the cases of Rwanda and Botswana show that a country does not necessarily have to be democratic to implement successful gender quotas. This thesis is aimed at addressing the gap concerning gender quotas and the transition to democracy, by researching the relationship between gender quotas and support for democracy. Support for democracy is one of the most essential factors during the transition to and consolidation of a stable democracy. If it is weak, then consequently democracy will be weak as well (Grassi, 2011). If the population of a country does not actively support democracy, it is much easier for a democratic regime to be overthrown, as the chance of political protest breaking out is much lower. Therefore, this thesis focuses on the research question:
Does the implementation of gender quotas lead to a difference in support for democracy among citizens?
This thesis will specifically focus on Africa, since many African countries have implemented gender quotas throughout the last decades. Furthermore, since becoming independent from their colonizers, many African states have made strides towards democratization, but also faced several setbacks, such as armed conflict and coups. Therefore, researching whether gender quotas lead to a difference in support for democracy could lead to important policy implications for strengthening African democracies.
3. Sub-research questions and hypotheses
To lay down a solid foundation to research whether gender quotas influence support for democracy over time, this thesis will first answer the following sub-research question (R1):
Is there a significant difference in citizens’ support for democracy between countries with gender quotas and countries without gender quotas?
The degree of support for democracy is dependent on many factors, such as cultural factors and previous experiences with democratic rule (Mattes, 2018). Because of this, there might be huge differences in support for democracy between countries, that cannot be attributed to the presence of gender quotas. This expectation applies to both the restricted definition of gender quotas, in which only mandatory quotas are included, as well as the broad definition, in which all types of gender quotas are included. Therefore, the first two hypotheses are:
H1: “There is no significant difference in support for democracy between countries with gender quotas in a general sense and countries without these gender quotas.”
5
H2:“The presence of mandatory gender quotas will not lead to a significant difference in support of democracy between countries with gender quotas and countries without gender quotas.”
To further test whether gender quotas influence the support for democracy in general, part of the research will specifically focus on gender, to study whether there is a significant difference between men and women. The second sub-research question (R2) that will be answered in this
thesis therefore is:
Does the presence of gender quotas lead to a difference in women’s support for democracy between countries with gender quotas and countries without gender quotas?
Even though research shows that political attitudes of women are influenced by gender quotas (Messing-Mathie, 2011), this cannot be expected to have a stronger influence on support for democracy than factors such as culture and previous experience with democratic rule. Again, this expectation applies to the broad definition as well as the restricted definition. Therefore, the next two hypotheses are:
H3:“The presence of gender quotas in general will not lead to a significant difference in women’s support for democracy between countries with gender quotas and countries without gender quotas.”
H4: “The presence of mandatory gender quotas will not lead to a significant difference in women’s support for democracy between countries with gender quotas and countries without gender quotas.”
However, while it is important to question whether gender quotas are associated with higher levels of support for democracy, it is also important to analyse whether gender quotas increase support for democracy over time. The third sub-research question (R3) that this thesis will
therefore answer is:
Does the presence of gender quotas lead to a difference in support for democracy over time in countries with gender quotas?
Andrea Messing-Mathie (2011) argues that the presence of gender quotas not only leads to a higher percentage of women in legislatures, but that it also directly influences individuals in society. Drawing on data from the Afrobarometer and the World Values Survey, she argues that the presence of gender quotas in Africa has led to a higher degree of political participation of
6
women. It has also changed their political attitudes. For example, women in countries with gender quotas report higher political interest and political trust. It should be noted though, that the change in political participation and attitudes only happened when there was a visible change in the representation of women. This can be explained by the fact that women are encouraged to intensify their political participation and political attitudes, if they see other women in action. Therefore, gender quotas will only have an effect on political participation and political attitudes, if women actually see a change in the degree of representation of women.
As mentioned above, this thesis specifically focuses on support for democracy. Messing-Mathie (2011) did not study this variable, but she did find that the implementation of gender quotas led to higher scores on political trust and political interest and a higher percentage of women participating in politics. These variables are closely related to support for democracy, since high levels of political participation, political trust and political interest are instrumental to a functioning democracy (van der Meer, 2017). The explanation that would follow from this is that the implementation of gender quotas will lead to a higher percentage of the female population that thinks women are just as capable holding political positions as men, because they actually see women holding these positions. Consequently, the support for democracy under women might increase as well, since democracy is a system in which the entire population is represented by elected officials. Therefore, if people acknowledge women are just as capable as men in governing the country, they might also acknowledge that democracy is their preferred mode of governing, since it ensures all voices in society are heard.
Even though Messing-Mathie (2011) only looks at women in her research, the expectation in this thesis is that these findings will hold for the entire population, but only when using the restricted definition in which only mandatory gender quotas are included. The expectation is that using the broad definition would not lead to any significant results, because the population might not see a significant change in representation, which would mean that they would not have any reason to change the way in which they think about democracy. Therefore, the fifth hypothesis is:
H5:“The implementation of gender quotas in general will not increase support for democracy over time.”
7
The sixth hypothesis applies to the restricted definition and is thus:
H6: “The implementation of mandatory gender quotas increases support for democracy over time."
As mentioned above, the expectation is that the support for democracy will increase as a result of the implementation of gender quotas, because gender quotas influence the degree of women’s representation. The seventh hypothesis is therefore:
H7: “The implementation of mandatory gender quotas increases support for democracy over time, because it influences the degree of women’s representation.”
However, since gender is an important aspect in this research, it’s essential to factor it in. Therefore, this thesis will also study whether there’s a difference between men’s support for democracy and women’s support for democracy as a result of the implementation of gender quotas. This leads to the fourth sub-research question (R4):
Does the implementation of gender quotas have a different impact on women’s support for democracy than men’s support for democracy?
Using the broad definition of gender quotas means that gender quotas are not mandatory in all countries. This means that even though some parties may have voluntary party quotas, the degree of women’s representation might not actually change significantly. If women do not see a difference in representation, they will not feel more represented and the expectation that results is that their support for democracy will not increase. Therefore, the eighth hypothesis is: H8: “The presence of gender quotas in general will not positively influence the degree of women’s support for democracy in contrast to men’s support for democracy.”
On the contrary, a difference is expected between women’s support for democracy and men’s support for democracy when using the restricted definition. The expectation is that men, in contrast to women, will be indifferent to the issue, since they will not feel more represented by the inclusion of women in politics. Their support for democracy may even decline, because gender quotas could be seen as an artificial method used to promote the inclusion of women in politics, while they are not actually being chosen (Kittilson, 2016). Thus, men might feel they are not accurately being represented. Therefore, the ninth hypothesis is:
8
H9: “The presence of mandatory gender quotas will positively influence the degree of women’s support for democracy, but it will not positively influence the degree of men’s support for democracy.”
4. Research methods
As mentioned above, in this thesis the relationship between the presence of gender quotas and support for democracy in African countries will be analysed. To do this, data from the Afrobarometer, the World Development Indicators and the Gender Quotas Database from the International Institute for Democracy and Electoral Assistance (IDEA, 2019) are used.
The independent variable is the presence of gender quotas. This research tests both a broad definition of gender quotas as well as a restricted definition. In the broad definition, a country that has any of the three types of gender quotas, either legislated gender quotas, reserved seats gender quotas or voluntary party gender quotas, will be considered a country with gender quotas. In the restricted definition, only a country with either the legislated gender quota or reserved seats gender quota will be considered a country with gender quotas. So countries that only have voluntary party gender quotas are not included in the restricted definition. This is done, to see whether this affects the results, because these countries do not have any laws that stipulate a minimal number of places or seats have to be reserved for women. These quotas are hence no guarantee that the entire country is exposed to the results, it could just be a marginal party that has voluntarily decided to reserve a number of seats for female candidates.
Furthermore, research done in South Africa has shown that parties with voluntary quotas might actually copy the strategy of opposing parties without gender quotas if they think they might benefit from this (Muriaas & Kayuni, 2013). Because these opposing parties are parties without gender quotas, these strategies often counteract gender quotas, keeping powerful people, often men, in place. The internal regulations of parties with gender quotas might therefore stipulate a minimal number of places reserved for women, but the reality might prove different and the priority of winning votes might prevail over gender parity. To investigate this, both the restricted and the broad definition will be tested to see whether there are any significant differences. Table 1 below shows all the countries in the dataset and whether they have implemented gender quotas using the broad definition and the restricted definition respectively.
9
Table 1: Countries used in dataset and presence of gender quotas using the broad and restricted definition respectively
Countries Presence of gender quotas (GQ) in R4
Presence of GQ in R5
Presence of GQ in R6
Benin No/No No/No No/No
Botswana Yes/No Yes/No Yes/No
Burkina Faso No/No Yes/Yes Yes/Yes
Cape Verde No/No Yes/Yes Yes/Yes
Ghana No/No No/No No/No
Kenya No/No Yes/Yes Yes/Yes
Lesotho Yes/Yes Yes/Yes Yes /Yes
Liberia No/No No/No Yes/Yes
Madagascar No/No No/No No/No
Malawi Yes/No Yes/No Yes/No
Mali No/No No/No No/No
Mozambique Yes/No Yes/No Yes/No
Namibia No/No No/No Yes/No
Nigeria No/No No/No No/No
Senegal No/No Yes/Yes Yes/Yes
South Africa No/No Yes/No Yes/No
Tanzania Yes/Yes Yes/Yes Yes/Yes
Uganda Yes/Yes Yes/Yes Yes/Yes
Zambia No/No No/No No/No
Zimbabwe No/No No/No Yes/Yes
The dependent variable is support for democracy. This variable is available in the Afrobarometer. To be able to measure whether support for democracy has changed as a result of implementation of gender quotas, the datasets of Round 4, Round 5 and Round 6 of the Afrobarometer had to be merged. Round 4 was completed in 2008, Round 5 in 2011 to 2013 and Round 6 in 2016. Round 4 was chosen as the first round, because almost all countries available in the Afrobarometer that have implemented gender quotas have implemented these between Round 4 and Round 5 (using the restricted definition). So for most countries, Round 4 is before gender quotas were implemented and Round 5 and Round 6 are after. To be able to work with the same data in all rounds, all countries from Round 5 and Round 6 that were not available in Round 4 had to be deleted. The countries that were available in all three rounds are listed in Table 1 above.
The data in the Afrobarometer is sampled by using a “clustered, stratified, multi-stage, area probability sample” (Afrobarometer, 2019). Primary Sampling Units, or country localities, are randomly selected across the country. For each locality a random starting point is selected from which interviewers start a random walk procedure. Then, households and individual
10
respondents are randomly selected by the interviewers. The total N for this dataset was 97293 people that responded over the three rounds, with 27713 people in Round 4, 34809 people in Round 5 and 34771 people in Round 6.
The dependent variable support for democracy was originally an ordinal variable in which people could answer the question “Which of these statements is closest to your own opinion?” with either “Democracy is preferable to any other kind of government,” “In some
circumstances, a non-democratic government can be preferable,” “For someone like me, it doesn’t matter what kind of government we have” or “I don’t know” (Afrobarometer, 2019).
This was recoded to a dichotomous variable, in which “Democracy is preferable to any other
kind of government” was given value one and all other answers were given value zero. Value
one then was named Explicit support for democracy and value zero was named No explicit
support for democracy.
To answer all four sub-questions, four different independent variables concerning gender quotas were constructed. These were:
1. Presence of gender quotas using the broad definition. Any country with any of the three types of gender quotas was awarded a one, all other countries were awarded a zero. 2. Presence of gender quotas using the restricted definition. Any country with either
legislated gender quotas or reserved seats gender quotas was awarded a one, all other countries were awarded a zero.
3. Change in presence of gender quotas using the broad definition. Any country that didn’t have gender quotas in Round 4, but implemented any of the three types in Round 5 and Round 6 was awarded a one, all other countries were awarded a zero.
4. Change in presence of gender quotas using the restricted definition. Any country that didn’t have gender quotas in Round 4, but implemented either legislated gender quotas or reserved seats gender quotas was awarded a one, all other countries were awarded a zero.
To test whether a possible change in women’s support for democracy differed from a possible change in men’s support for democracy, an interaction-effect variable was calculated for each of the independent variables above, multiplying them by the independent variable female, in which females were awarded a one and men a zero.
11
Lastly, to test whether gender quotas influence support for democracy as a result of their influence on women’s representation, the variable percentage of women in parliament was used, in which the percentage of women in parliament was noted. This was used to test whether a change in women’s representation could be the reason why gender quotas could possibly influence support for democracy. The descriptive statistics for all of the variables mentioned above can be found in Table 2 below.
Table 2: Descriptive statistics for independent variables
Minimum Maximum N Mean Standard
Deviation Dichotomous variable democracy 0 1 97293 0,76 0,425 Presence of Gender Quotas using broad definition 0 1 97293 0,56 0,497 Presence of Gender Quotas using restricted definition 0 1 97293 0,33 0,471 Change in presence of Gender Quotas using broad definition 0 1 97293 0,38 0,486 Change in presence of Gender Quotas using restricted definition 0 1 97293 0,27 0,440 Percentage of women in parliament 6,7 43,3 97293 21,28 11,913
To control for other factors that could also explain the change in support for democracy, several control variables were included. Individual control variables include age, gender, media use, political interest, access to basic necessities, education and religion. Gender is a dichotomous variable in which female was awarded a one and male a zero. Media use is measured by how
12
often people get their news from newspapers. Political interest is measured by how often people discuss political matters and access to basic necessities is measured by how often people have gone without food in the last twelve months. Media use as well as political interest and access to basic necessities were all recoded in such a way that zero means people have never engaged in these activities and one means people have engaged in these activities. The control variable education was condensed and divided in four categories: no formal education, primary education, secondary education and post-secondary education. Religion was recoded and divided in three categories: Muslim, Christian and other religion. Country-level control variables include Freedom House and the natural logarithm of GDP per capita adjusted for power purchasing parity, since a change in both of these scores might cause people to adjust their attitudes regarding support for democracy.
The influence of the presence of gender quotas on support for democracy is tested by means of logistic regression analyses. Nine different models are run to test all hypotheses. In all models the dependent variable is support for democracy; the independent variable is one of the variables concerning gender quotas described above. As this research uses data from the same countries taken in different rounds, the observations are correlated. Hence, clustered standard errors were used, in which the data was clustered by country. To control for survey round effects, survey round dummies were added. In this way, possible general changes in support for democracy are taken into account. To account for over- or underrepresentation in the data, weights were included.
Finally, it is important to note that the model is strongly influenced by the analysis conducted by Eifert et al. (2010) on elections and ethnic identity in Africa. Yet, while Eifert et al. (2010) also use country-fixed effects, this approach led to estimation problems in the analyses used in this paper. This can be due to the fact that while they use an interval-ratio independent variable, gender quotas is a dichotomous variable. However, by including level of democracy and development, two of the most potentially influential control variables are added to the models. This supports the results found.
5. Results
In Graph 1 below, the mean scores of support for democracy are shown for each country in the dataset. The higher the bar, the higher the percentage of people that explicitly preferred democracy over any other regime type. As can be seen, there does not seem to be a clear pattern
13
in change in support for democracy over time. Some countries, like Senegal and Burkina Faso, show a clear increase, while other countries, such as Nigeria and Mozambique, first show an increase in Round 5, before falling again in Round 6. There are also stark contrasts between countries, with the mean scores of support for democracy ranging from 48,5 per cent in Lesotho in Round 4 to 92,5 per cent in Senegal in Round 6. Nevertheless, this does not necessarily mean that there are no effects. Important individual and country-level control variables could obscure the impact of gender quotas. This could happen, for example, if support for democracy goes up for women but down for men. Hence it remains interesting to study whether the presence of gender quotas influences support for democracy across countries and over time.
First, a regression model was run to answer whether the presence of gender quotas in general leads to a difference in support for democracy between countries with gender quotas and countries without gender quotas. The dependent variable is support for democracy and the independent variable is the presence of gender quotas using the broad definition. As can be seen in table 3 below, model 1 shows that there is no significant relation between the presence of gender quotas in general and the support for democracy, Wald F(1;19) = 0,378, p = 0,546. This means that H1 “There is no significant difference in support for democracy between countries with gender quotas in a general sense and countries without these gender quotas.” is supported.
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 SUPP O R T FO R DE MO CR A CY COUNTRIES IN DATASET
Graph 1: Change in support for democracy for each country
over Round 4, Round 5 and Round 6
14
A second regression analysis was run to answer whether the presence of mandatory gender quotas leads to a difference in support for democracy between countries with gender quotas and countries without gender quotas. In this analysis, support for democracy is again the dependent variable and presence of gender quotas using the restricted definition is the independent variable. Model 2 shows that there is no significant relation between the presence of gender quotas in general and the support for democracy, Wald F(1;19) = 0,476, p = 0,498. This means that H2 “The presence of mandatory gender quotas will not lead to a significant difference in support of democracy between countries with gender quotas and countries without gender quotas” is supported.
Then, an interaction-effect was added to both of the regression analyses to test whether there was a difference in support for democracy between men and women. In model 3 the interaction-effect that was added multiplied the variable female and the variable presence of gender quotas using the broad definition. As can be seen in model 3, the presence of gender quotas in general did not lead to a significant difference in women’s support for democracy, Wald F(1;19) = 0,503, p = 0,487. H3 “The presence of gender quotas in general will not lead to a significant difference in women’s support for democracy between countries with gender quotas and countries without gender quotas” is supported.
In model 4, the interaction-effect that was added multiplied the variable female and the variable presence of gender quotas using the restricted definition. The model shows that there is no significant difference in women’s support for democracy in countries with mandatory gender quotas, Wald F(1;19) = 0,421, p = 0,524, indicating that H4 “The presence of mandatory gender quotas will not lead to a significant difference in women’s support for democracy between countries with gender quotas and countries without gender quotas” is supported.
Regarding the control variables: In each and every model the variables gender, discussed
politics and gone without food are all significant. Regarding the education variables, secondary education and post-secondary education are significant, while primary education is not. As can
been seen in models 1 to 4, the odds that women would support democracy are lower than men. The odds that people who discuss politics would support democracy are higher than people who do not discuss politics and the odds that people who have gone without food in the last twelve months would support democracy are lower than people who have not gone without food. Lastly, the odds that someone with either secondary or post-secondary education would support
15
democracy are higher than someone with no formal education. The odds that someone with primary education would support democracy also seem higher than someone with no formal education, but these results are not significant.
16
Table 3: Logistic regression analysis of the probability of support for democracy between countries with gender quotas and countries without
Model 1 Model 2 Model 3 Model 4
(Constant) 3,415 (1,363) 3,151 (1,301) 3,365 (1,358) 3,135 (1,301) Presence of general gender quotas 1,158
(0,239)
1,191 (0,241)
Presence of mandatory gender quotas 1,237
(0,0308)
1,264 (0,307)
Inter-action effect GQ and female 0,948
(0,075) 0,959 (0,064) Freedom House 0,932 (0,082) 0,929 (0,083) 0,932 (0,082) 0,929 (0,083) Ln of GDP (PPP) per capita 0,964 (0,160) 0,977 (0,154) 0,964 (0,160) 0,977 (0,154) Age 1,005 (0,003) 1,005 (0,003) 1,005 (0,003) 1,005 (0,003) Gender (Ref. = male)
Female 0,834*** (0,042) 0,835*** (0,043) 0,856** (0,044) 0,845** (0,058) Used Newspaper 0,997 (0,069) 1,004 (0,068) 0,997 (0,069) 1,003 (0,068) Discussed politics 1,435*** (0,044) 1,433*** (0,044) 1,435*** (0,044) 1,433*** (0,044)
Gone without food 0,812**
(0,068) 0,815* (0,072) 0,812** (0,068) 0,815* (0,072) Education (Ref. = No formal education)
Primary education 1,021 (0,087) 1,025 (0,081) 1,022 (0,087) 1,026 (0,080) Secondary education 1,243* (0,082) 1,248* (0,082) 1,245* (0,083) 1,249* (0,082) Post-secondary education 1,480** (0,105) 1,477** (0,103) 1,483** (0,106) 1,478** (0,103) Religion (Ref. = Christian)
Muslim 1,133 (0,207) 1,113 (0,205) 1,133 (0,207) 1,113 (0,205) Other Religion 0,891 (0,083) 0,901 (0,071) 0,891 (0,083) 0,901 (0,071) Cox and Snell’s R2
Nagelkerke’s R2 N 0,017 0,025 97293 0,018 0,026 97293 0,017 0,026 97293 0,018 0,026 97293
Note: odds ratios with standard errors (clustered at the country level) between brackets.
***p < 0,001, **p < 0,01, *p < 0,05. Model 1 with presence of gender quotas in general; model 2 with presence of mandatory gender quotas; model 3 with presence of gender quota in general and added inter-action effect; model 4 with presence of mandatory gender quotas and added inter-action effect. All models include survey round controls and weights.
17
To test whether gender quotas influence support for democracy over time, five more regression analyses were run. First, a regression analysis was run to study whether the implementation of gender quotas in general would influence the support for democracy over time. In this analysis, the dependent variable was support for democracy and the independent variable was whether gender quotas in general have been implemented. The results of this regression analysis can be found in model 5 of Table 4 below. This model shows that the implementation of gender quotas in general does not lead to a significant change in support for democracy over time, Wald
F(1;19) = 1,283, p = 0,271, hence H5 “The implementation of gender quotas in general will not increase support for democracy over time” is supported.
A sixth regression analysis was run to test whether mandatory gender quotas influence support for democracy over time. Support for democracy was again the dependent variable in this analysis and the implementation of mandatory gender quotas was the independent variable. The results of this regression analysis can be found in model 6. It shows that the implementation of mandatory gender quotas significantly increases the support for democracy, Wald F(1;19) = 13,358, p = 0,002. Therefore, the null hypothesis can be rejected and H6 “The implementation of mandatory gender quotas increases support for democracy over time" is supported. The odds
that people will support democracy after gender quotas have been implemented are 1,799 times higher than in countries where gender quotas have not been implemented.
An explanation for this result was that gender quotas influence the degree of women’s representation, which in turn would influence the support for democracy. To test whether this holds, a seventh regression analysis was run with the added variable percentage of women in
parliament. In this analysis, support for democracy was again the dependent variable and the
percentage of women in parliament and support for democracy were the independent variables. Model 7 shows that the implementation of mandatory gender quotas still significantly increases the support for democracy, Wald F(1;19) = 11,552, p = 0,003. The change in the percentage of women in parliament, however, does not significantly influence support for democracy, Wald
F(1;19) = 0,441, p = 0,515. Therefore, H7 “The implementation of mandatory gender quotas increases support for democracy over time, because it influences the degree of women’s representation” is not supported.
18
Then, an interaction-effect was also added to both of the regression analyses in model 5 and model 6 to test whether the change in support for democracy over time differed between men and women. In model 8, the interaction-effect that was added multiplied the variable female and the variable that measured implementation of gender quotas in general. As can be seen in this model, the implementation of gender quotas in general did not lead to a significant difference in women’s support for democracy, Wald F(1;19) = 0,311, p = 0,583. Therefore, H8 “The presence of gender quotas in general will not positively influence the degree of women’s support for democracy in contrast to men’s support for democracy” is supported.
In model 9, the interaction-effect that was added multiplied the variable female and the variable that measured implementation of mandatory gender quotas. Model 9 shows that there is no significant difference in the effect of gender quotas on women’s support for democracy relative to men in countries with mandatory gender quotas, Wald F(1;19) = 1,112, p = 0,305. The null hypothesis is therefore not rejected and H9 “The presence of mandatory gender quotas will positively influence the degree of women’s support for democracy, but it will not positively influence the degree of men’s support for democracy” is not supported.
Table 4 also lists all the control variables. Just like in table 3, in each and every model the variables gender, discussed politics and gone without food are all significant. Of the education categories, secondary education and post-secondary education are significant, while primary
education is not. Using table 4, it can be deducted that the odds that women would support
democracy are lower than men. The odds that people who discuss politics support democracy are higher than people who do not discuss politics and the odds that people who have gone without food in the last twelve months will support democracy are lower than people who have not been in such a situation. The odds that someone with secondary or post-secondary education would support democracy are higher than someone with no formal education. Added to this, the odds that someone with primary education would support democracy seem higher than someone with no formal education, but these results are not significant.
19
Table 4: Logistic regression analysis of the probability of support for democracy over time in countries with gender quotas (odds ratios)
Model 5 Model 6 Model 7 Model 8 Model 9
(Constant) 4,633 (1,417) 2,664 (1,202) 3,079 (1,085) 4,681 (1,426) 2,644 (1,202) Presence of general gender quotas 1,276
(0,215)
1,246 (0,226) Presence of mandatory gender quotas 1,799**
(0,161)
1,792** (0,172)
1,865** (0,163)
Interaction-effect GQ and female 1,047
(0,082)
0,934 (0,064)
Percentage of women in parliament 0,995
(0,008) Freedom House 0,918 (0,078) 0,898 (0,075) 0,893 (0,078) 0,918 (0,078) 0,898 (0,075) Ln of GDP (PPP) per capita 0,924 (0,174) 0,991 (0,145) 0,983 (0,131) 0,924 (0,174) 0,991 (0,145) Age 1,005 (0,003) 1,005 (0,003) 1,005 (0,003) 1,005 (0,003) 1,005 (0,003) Gender (Ref. = male)
Female 0,834** (0,044) 0,833** (0,043) 0,835** (0,043) 0,820** (0,054) 0,847** (0,055) Used Newspaper 1,000 (0,059) 1,006 (0,060) 1,019 (0,062) 1,000 (0,059) 1,005 (0,060) Discussed politics 1,457*** (0,047) 1,467*** (0,037) 1,471*** (0,036) 1,457*** (0,046) 1,466*** (0,038) Gone without food 0,808**
(0,063) 0,815** (0,059) 0,813** (0,058) 0,808** (0,063) 0,815** (0,059) Education (Ref. = No formal education)
Primary education 1,072 (0,076) 1,117 (0,072) 1,145 (0,071) 1,072 (0,076) 1,117 (0,075) Secondary education 1,250* (0,080) 1,276** (0,073) 1,300** (0,069) 1,249* (0,080) 1,275** (0,063) Post-secondary edu. 1,489** (0,102) 1,491*** (0,094) 1,504** (0,093) 1,488** (0,102) 1,491*** (0,054) Religion (Ref. = Christian)
Muslim 1,105 (0,183 1,073 (0,144) 1,071 (0,148) 1,113 (0,180) 1,072 (0,137) Other Religion 0,894 (0,089) 0,905 (0,092) 0,897 (0,084) 0,892 (0,090) 0,904 (0,094) Cox and Snell’s R2
Nagelkerke’s R2 N 0,018 0,028 97293 0,027 0,040 97293 0,027 0,041 97293 0,018 0,028 97293 0,027 0,040 97293
Note: odds ratios with standard errors (clustered at the country level) between brackets.
***p < 0,001, **p < 0,01, *p < 0,05. Model 5 with implementation of gender quotas in general; model 6 with implementation of mandatory gender quotas; model 7 with implementation of mandatory gender quotas and percentage of women in parliament; model 8 with implementation of gender quotas in general and added inter-action effect; model 9 with implementation of mandatory gender quotas and added inter-action effect; All models include survey round controls and weights.
20 6. Conclusion and discussion
In this thesis, the research question “Does the implementation of gender quotas lead to a difference in support for democracy among citizens” was analysed. A distinction was made between a broad definition of gender quotas, in which legislated gender quotas, reserved seats gender quotas as well as voluntary party gender quotas were used and a restricted definition of gender quotas, in which only legislated gender quotas and reserved seats gender quotas were used.
To answer the question, nine different regression analyses were run. The results show that the difference in support for democracy between countries with gender quotas and countries without gender quotas cannot be explained by the presence of gender quotas, no matter which definition is used. However, when measuring the influence of gender quotas on support for democracy over time, the results show a clear difference between mandatory gender quotas and gender quotas in general. Though the relationship between the implementation of gender quotas in general and support for democracy is positive, this result is not significant; only the implementation of mandatory gender quotas led to significant results. When mandatory gender quotas were implemented, the odds that someone would support democracy were 1,799 times higher than in countries where no mandatory gender quotas were implemented.
The underlying assumption that was made was that this result is caused by the influence of gender quotas on the degree of women’s representation. The assumption was that an increase in the degree of women’s representation would lead to a higher percentage of people that feel represented by the legislature, which would in turn lead to a higher degree of support for democracy. However, the results show that an increase in the percentage of women in parliament does not lead to a significant change in support for democracy. Thus, though gender quotas significantly increase support for democracy, this is not because they influence women’s representation. An explanation for this unexpected result could be that support for democracy of the population is not increased because they feel more represented by increased representation in the legislature, but because they might feel more represented already just by the symbolism of the implementation of gender quotas. Of course, judging by the small explanatory power of this model as evidenced by the Nagelkerke’s R2 of 0,040, there are certainly other possible variables that contribute to the increase in support for democracy as well.
The results also showed that there was no clear difference between women’s support for democracy and men’s support for democracy. Even when the implementation of mandatory gender quotas showed a clear positive influence on support for democracy, there was no
21
significant difference between women’s support for democracy and men’s support for democracy. So, deducting from this, as there was no clear difference between men and women, but the entire population’s support for democracy did increase, both women’s support for democracy and men’s support for democracy increase when mandatory gender quotas are implemented.
As shown above, there is a clear difference between mandatory gender quotas and gender quotas in general. This could have severe policy implications, since it shows that mandatory gender quotas actually garner results, in contrast to voluntary party gender quotas. They are not purely symbolic, though symbolism could be the reason why people’s attitudes towards support for democracy change. This research strengthens the view that the fast-track view to achieve equal representation in politics not only symbolically influences parliament, but also significantly changes individual attitudes of the population concerning support for democracy. Furthermore, this research could have policy implications for countries during the transition to democracy. Support for democracy is an essential part of the stable transition to democracy, without it, democracy is almost guaranteed to fail (Grassi, 2011). This research shows that the implementation of gender quotas positively influences support for democracy and therefore could help stabilise the transition to democracy.
Regrettably, this research encountered some limitations that should be taken into account when assessing the results. First, this research was unable to incorporate country-fixed effects, to control for differences within countries, because there was not enough variation within the data. Because of this, the results are not as robust as they could have been and this should be taken into account. Furthermore, this research was limited by the amount of data available in the Afrobarometer, especially in Round 4, in which data from only twenty countries was collected. Of these countries, only six countries (using the restricted definition) did not have gender quotas in Round 4, but implemented them in either Round 5 or Round 6. This is quite a small sample to generate results from and one should therefore be careful with generalising these results to a broader population.
Further research could concentrate on expanding the sample, thus making results more reliable and robust. Furthermore, as this research was only focused on African data, further research might take other regions into account or even other types of countries, instead of the developing countries that this research mostly focused on. Other research might focus on clarifying the causal mechanism that influences support for democracy when gender quotas are
22
implemented; What is the underlying mechanism that causes the increase in support for democracy? Could this be symbolism or are there any other factors in play?
23
Bibliography
Afrobarometer. (2019). About Afrobarometer. Retrieved May 30, 2019, from Afrobarometer: http://www.afrobarometer.org/about
Afrobarometer. (2019). Merged Round 6 Codebook. Retrieved May 13, 2019, from
Afrobarometer: http://afrobarometer.org/data/merged-round-6-codebook-36-countries-2016
Afrobarometer. (2019). Sampling principles and weighting. Retrieved May 7, 2019, from Afrobarometer: http://www.afrobarometer.org/surveys-and-methods/sampling-principles
Bauer, G., & Britton, H. E. (2006). Women in African Parliaments: A Continental Shift? In G. Bauer, & H. E. Britton, Women in African Parliaments (pp. 1-30). London: Lynne Rienner Publisher.
Bauer, G., & Burnet, J. E. (2013). Gender quotas, democracy, and women's representation in Africa: Some insights from democratic Botswana and autocratic Rwanda. Women's
Studies International Forum, 41, 103-112.
Bush, S. S. (2011). International Politics and the Spread of Quota for Women in Legislatures.
International Organization, 65(1), 103-137.
Chen, L.-J. (2010). Do Gender Quotas Influence Women's Representation and Policies? The
European Journal of Comparative Economics, 7(1), 13-60.
Dahlerup, D., & Freidenvall, L. (2005). Quotas as a 'fast track' to equal representation for women. International Feminist Journal of Politics, 7(1), 26-48.
Darhour, H., & Dahlerup, D. (2013). Sustainable representation of women through gender quotas: A decade's experience in Morocco. Women's Studies International Forum, 41, 132-142.
Debski, J., Jetter, M., Mösle, S., & Stadelmann, D. (2018). Gender and corruption: The neglected role of culture. European Journal of Political Economy, 55, 526-537. Eifert, B., Miguel, E., & Posner, D. N. (2010). Political Competition and Ethnic Identification
in Africa. American Journal of Political Science, 54(2), 494-510.
Fallon, K. M., Swiss, L., & Viterna, J. (2012). Resolving the Democracy Paradox:
Democratization and Women's Legislative Representation in Developing Nations, 1975 to 2009. American Sociological Review, 77(3), 380-408.
24
Freidenvall, D. D. (2005). Quotas as a 'fast track' to equal representation for women.
International Feminist Journal of Politics, 7(1), 26-48.
Grassi, D. (2011). Democratic consolidation. In B. Badie, D. Berg-Schlosser, & L. Morlino,
International Encyclopedia of Political Science, Vol. 1 (pp. 614-620). Thousand Oaks:
SAGE Publications.
IDEA. (2019). Gender Quotas Database. Retrieved April 19, 2019, from www.idea.int: https://www.idea.int/data-tools/data/gender-quotas
Kang, A. (2013). The effect of gender quota laws on the election of women: Lessons from Niger. Women's Studies International Foru, 41, 94-102.
Kittilson, M. C. (2016). Gender and Political Behaviour. Oxford Research Encyclopedia of
Politics, 1-17.
Kumar Jha, C., & Sarangi, S. (2018). Women and corrpution: What positions must they hold to make a difference. Journal of Economic Behavior and Organization, 151, 219-233. Mattes, R. (2018). Support for Democracy. Oxford Research Encyclopedia of Politics, 1-22. Messing-Mathie, A. (2011). After the Election: Fast-track Gender Quotas and the Political
Engagement of Women in Sub-Sharan Africa. Ann Arbor: ProQuest LLC.
Muriaas, R. L., & Kayuni, H. M. (2013). The reversed contagion effect: Explaining the unevenness of women's representation across South African municipalities. Women's
Studies International Forum, 41, 150-159.
Rosen, J. (2017). Gender quotas for women in national politics: A comparative analysis across development thresholds. Social Science Research, 66, 82-101.
Tadros, M. (2010). Introduction: Quotas - Add Women and Stir? IDS Bulletin, 41(5), 1-10. Towns, A. E. (2013). Women and States: Norms and Hierarchies in International Society.
Cambridge: Cambridge University Press.
Tripp, A. M., & Kang, A. (2008). The Global Impact of Quotas: On the Fast Track to Increased Female Legislative Representation. Comparative Political Studies, 41(3), 338-361.
van der Meer, T. W. (2017). Political Trust and the "Crisis of Democracy". Oxford Research
25
Wang, V. (2013). Women changing policy outcomes: Learning from pro-women legislation in the Ugandan Parliament. Women's Studies International Forum, 41, 113-121. Yoon, M. Y. (2013). Special seats for women in parliament and democratization: The case of
