Back to History

Fields Changed

Registration

Field Before After
Trial Title Gender quotas: Fairness, meritocracy, and participation of women in politics Gender quotas: Fairness, Meritocracy, and Participation of Women in Politics
Abstract Gender quotas in politics are often introduced as measures to increase women’s representation in parliaments and cabinets, trigger societal change, and ultimately promote fairer gender norms. We study one of the first systems of gender quotas, implemented in Tanzania in 1985 as a number of reserved seats for women in parliament. We conduct an online survey to measure preferences for numerical and substantive representation of women in the national parliament and its national committees, and conduct four information experiments to study perceptions of fairness, meritocracy and patronage in relation to the gender quota in Tanzania. The analysis contributes to research on the effect of quotas and affirmative action in politics. Gender quotas in politics are often introduced as measures to increase women’s representation in parliaments and cabinets, trigger societal change, and ultimately promote fairer gender norms. We study one of the first systems of gender quotas, implemented in Tanzania in 1985 as a number of reserved seats for women in parliament. We conduct an online survey to measure preferences for numerical and substantive representation of women in the national parliament and its national committees, and conduct four information experiments to study perceptions of fairness, meritocracy and patronage in relation to the gender quota in Tanzania. The analysis contributes to research on preferences for inequality and redistributive policies (quotas and affirmative action) in politics.
Trial End Date November 15, 2020 December 04, 2020
Last Published October 13, 2020 09:12 AM November 10, 2020 07:36 AM
Intervention (Public) We run our survey both 2 weeks before and 2 weeks after the national election in October 28th to measure changes in perceptions about long term political goals. In the first part of the survey, respondents estimate the current distribution of wealth and gender in Tanzania and, then, indicate what their ideal distributions look like. Our research adopts Rawls' idea of the original position and applies it to principles of gender equality. When asking about ideal distributions, we aim to elicit respondents' preferences freed of their respective characteristics and specific position in society (Rawls (1971)). We combine responses on ideal and estimated distributions with the following 4 treatments: 1 - Survey experiment: Civic action We implement one survey experiment, where respondents are randomly assigned to two treatments in support of participation of women in politics: 1. Some of your fellow students want to send this week a petition letter to the government, requesting action to increase gender equality. Would you sign the petition letter?? 2. Would you donate 5,000 TZS for a female candidate’s campaign this week? 2 - Survey experiment: Gender quota We implement a second survey experiment on the gender quota in politics in Tanzania, where we randomize a question about whether the quota should be kept constant, or increased (with and without peer effects). 1. BASELINE: Do you agree that the number of reserved seats for women should be kept constant at 30% in future parliaments? 2. INCREASE: Do you agree that the number of reserved seats for women should be increased from 30% to 50%? 3. DECREASE: Do you agree that the number of reserved seats for women should be decreased from 30% to 15%? 3 - Survey experiment: Meritocracy We use a third survey experiment to measure the importance of three components of meritocracy: experience, education and effort. We randomize between male and female candidates. Treatment: The next parliamentary elections are coming up at the end of this month. What are the most important qualities for a [female:male] candidate to become Member of Parliament? 4 – Survey experiment: Patronage The last survey experiment measures the influence of patronage on candidate selection. Here, we also randomize between male and female candidate: Treatment: Imagine that after the election you would have to decide between two [female:male] Members of the Parliament. One of them is to become the Speaker of the National Assembly. Assuming that they have equal experience, education, and motivation, which one would you choose as Speaker of the National Assembly? References J. Rawls. A Theory of Justice. Belknap Press of Harvard University Press, Cambridge, Massachussets, 1edition, 1971. ISBN 0-674-88014-5. We conduct two rounds of an online survey, two weeks before and two weeks after the national election on October 28, 2020, to measure changes in perceptions about long-term political goals. In the first part of the survey, respondents estimate the current distribution of wealth and gender in Tanzania and, then, indicate what their ideal distributions look like. Our research adopts Rawls' idea of the original position and applies it to principles of gender equality. When asking about ideal distributions, we aim to elicit respondents' preferences freed of their respective characteristics and specific position in society (Rawls (1971)). We combine responses on ideal and estimated distributions with the following 4 treatments: 1 - Survey experiment: Civic action We implement one survey experiment, where respondents are randomly assigned to two treatments (A, B) that describe the extent to which one is actively willing to take action to support equal representation between women and men. This week, some of your fellow students are going to request action from the government to increase gender equality: 1. (A) To show your support, would you sign their petition letter? 2. (B) To show your support, would you donate them 5,000 TZS? 2 - Survey experiment: Gender quota We implement a second survey experiment on the gender quota in politics, where we randomly ask respondents about their support on one of three different policy options (A, B, C). 1. (A) BASELINE: Do you agree that the number of reserved seats for women should be kept constant at 30% in future parliaments? 2. (B) INCREASE: Do you agree that the number of reserved seats for women should be increased from 30% to 50% in future parliaments? 3. (C) DECREASE: Do you agree that the number of reserved seats for women should be decreased from 30% to 15% in future parliaments? 3 - Survey experiment: Meritocracy We use a third survey experiment to measure the importance of three components of meritocracy: experience, education and effort. We randomize between male and female candidates. Treatment: The next parliamentary elections are coming up at the end of this month. How important are the following qualities for a [(A)female; (B)male] candidate to become Member of Parliament? 1 1. Candidate has a university degree. 2. Candidate has previous experience as Member of Parliament. 3. Candidate put a lot of effort into campaigning for the election. 4 – Survey experiment: Patronage The last survey experiment measures the influence of patronage (family relations) on candidate selection. Here, we randomize between two male and two female (A, B) candidates. Treatment: Imagine that after the election you would have to decide between two [(A)female; (B)male] Members of the Parliament. One of them will become the Speaker of the National Assembly. Assuming that they have equal experience, education, and motivation, which one would you choose as Speaker of the National Assembly? References J. Rawls. A Theory of Justice. Belknap Press of Harvard University Press, Cambridge, Massachussets, 1edition, 1971. ISBN 0-674-88014-5.
Intervention End Date November 15, 2020 December 04, 2020
Primary Outcomes (End Points) (A) Differences in estimated vs. ideal distribution in number of seats in parliament and committee (B) Survey experiments: 1. Support to take action on equal representation between women and men on a 5 point likert scale from ‘very much agree’ to ‘very much disagree’. 2. Agreement with proposed quota policy – scaled from ‘absolutely not’ to ‘absolutely yes’ 3. Scale of importance for each element of merit – education, experience, effort – on a scale from 1- very important to 10- not important at all. 4. Choice of candidate as Speaker when the two candidates only differ in whether or not they have family relations to other Members of Parliament. (A) Differences in estimated vs. ideal number of seats for women in parliament and committee (B) Differences in actual (objective) vs. estimated/ideal (subjective) number of seats for women in parliament and committee (C) Survey experiments: 1. Civic action: Support to take action on a 5-Point Likert Scale from ‘very much agree’ to ‘very much disagree’. 2. Gender quota: Agreement with proposed quota policy on a 5-Point Likert Scale from ‘very much agree’ to ‘very much disagree’ 3. Meritocracy: Scale of importance for each element of merit – education, experience, effort – on a scale from 1- Not at all important to 10- Very important. 4. Patronage: Choice of candidate as Speaker when the two candidates only differ in whether or not they have family relations to other Members of Parliament. (1- The candidate with no family relations; 2- The candidate with family relations)
Experimental Design (Public) We set up an online survey and sent out via e-mail. The survey takes about 15 min to fill in, respondents are compensated both in the first as well as in the second round. Within the survey we ask for their estimates of the distributions of wealth and political power between men and women, their ideal versions of these distributions, demographics, gender norms, interest in politics, views on gender quotas, and peers' influence. Moreover, we conduct the four survey experiments described above. Finally, we include questions to compute the social desirability index by Crowne and Marlowe (1960). References: D. P. Crowne and D. Marlowe. A New Scale of Social Desirability Independent of Psychopathology. Journal of Consulting Psychology, 24(4):349–354, 1960. doi:10.1037/h0047358. We conduct two online surveys (sent out via email) with students at the Faculty of Social Sciences in Dar es Salaam. Students are compensated for their participation and each survey takes about 15 minutes to complete. The survey covers topics on students' estimated and ideal distributions of wealth and political power between men and women, gender norms, political behavior, views on gender quotas, and peers' influence. Moreover, we conduct four survey experiments described above to uncover possible mechanisms for observed differences between actual, estimated, and ideal distributions of political power between men and women. The survey contains further a social desirability index, relying on previous research by Crowne and Marlowe (1960), who developed a social desirability scale for surveys. We use the short form of their module developed by Reynolds (1982), which includes unrealistic personality traits, that respondents have to 'agree' or 'disagree' with. References: D. P. Crowne and D. Marlowe. A New Scale of Social Desirability Independent of Psychopathology. Journal of Consulting Psychology, 24(4):349–354, 1960. doi:10.1037/h0047358. W.M. Reynolds, Development of Reliable and Valid Short Forms of the Marlowe‐Crowne Social Desirability Scale. Journal of Clinical Psychology, 38: 119-125, 1982. https://doi.org/10.1002/1097-4679(198201)38:1<119::AID-JCLP2270380118>3.0.CO;2-I.
Randomization Method Randomization to each treatment in office by a computer. Randomization to each treatment group in office by a computer.
Planned Number of Observations 492 individuals ca. 500 individuals
Sample size (or number of clusters) by treatment arms In each treatment arm there are 246 students for survey experiments but the gender quota support experiment. There, we have 3 treatment arms, i.e. 164 respondents in each. For each treatment arm, we expect ca. 250 students (2 treatment groups), except of for the gender quota experiment, where we have 3 treatment arms, i.e. ca. 166 respondents in each.
Power calculation: Minimum Detectable Effect Size for Main Outcomes A power analysis was conducted and the sample chosen, using a minimum detectable effect size for the average treatment effect of 0.4 (for survey experiment on gender quota, only survey experiment with three treatment groups). A power analysis was conducted and the sample chosen, using a minimum detectable effect size for the average treatment effect of 0.4 (for the survey experiment on gender quotas, which is the only survey experiment with three treatment groups).
Additional Keyword(s) inequality, substantive representation inequality, substantive representation, quotas
Secondary Outcomes (End Points) TREATMENT EFFECTS FROM SURVEY EXPERIMENTS 1. Survey Experiment Civic Action: Treatment is to increase the cost of civic action in support of the gender quota: Probability(Take Action | High Cost) – Probability(Take Action | Low Cost) 2. Survey Experiment Gender Quota: Questions vary for whether there is support to increase, decrease, or keep the quota constant. 3. Survey Experiment Merit: Treatment is to indicate importance of merit for either a male or a female candidate: Importance Effort | Female – Importance Effort | Male Importance Education | Female – Importance Education | Male Importance Experience | Female – Importance Experience | Male Rank Merit Elements | Female – Rank Merit Elements | Male 4. Survey Experiment Patronage: Treatment is to choose from two male or two female candidates either one with or one without family relations. Probability (Punish for Family Relations | Female) – Probability (Punish for Family Relations | Male) HETEROGENEOUS EFFECTS: by self-identification, demographics, gender norms, support of gender quotas, interest in politics, peer effects. TREATMENT EFFECTS FROM SURVEY EXPERIMENTS 1. Survey Experiment Civic Action: Treatment is to increase the cost of civic action in support of the gender quota: Probability(Take Action | High Cost) – Probability(Take Action | Low Cost) 2. Survey Experiment Gender Quota: Questions vary for whether there is support to increase, decrease, or keep the quota constant. 3. Survey Experiment Merit: Treatment is to indicate importance of merit for either a male or a female candidate: Importance Effort | Female – Importance Effort | Male Importance Education | Female – Importance Education | Male Importance Experience | Female – Importance Experience | Male 4. Survey Experiment Patronage: Treatment is to choose from two male or two female candidates either one with or one without family relations. Probability (Punish for Family Relations | Female) – Probability (Punish for Family Relations | Male) HETEROGENEOUS EFFECTS: by self-identification, demographics, gender norms, support of gender quotas, interest in politics, peer effects.
Back to top

Other Primary Investigators

Field Before After
Affiliation Copenhagen Business School University of Copenhagen/Copenhagen Business School
Back to top