Targeting men, women or both? Changing outcomes for adolescent girls

Our interventions took place in Shikarpur and Larkana districts of Sindh Province, and Lodhran and Muzaffargarh districts of Punjab province in Pakistan. Joint with Oxfam Pakistan, our local implementing partners Indus Resource Centre (IRC) and Bedari prepared a movie of a theatre show that addressed child marriage and girls’ education. IRC and Bedari then visited each intervention village and organized a movie screening that was followed by a group discussion. In each village, during a subsequent visit, IRC and Bedari organized another structured discussion about the content of the movie. After each session, our research team conducted a small evaluation exit survey with randomly chosen participants.

2018-12-28

2019-06-30

Girls’ education (attitudes, beliefs, norms, behaviors)
Marriage (attitudes, beliefs, norms, behaviors)

Violence against women and girls (attitudes, beliefs, norms, behaviors)

We conduct a randomized control trial in 177 villages of Sindh and Punjab provinces in Pakistan to investigate how edutainment interventions affect attitudes, beliefs, norms, and behavior about child marriage, education for girls, and violence against women and girls (VAWG). The villages are randomly assigned into four treatment groups to enable a rigorous evaluation of the impact of the programme:

T1. targeting the intervention at men and boys only,
T2. targeting the intervention at women and girls only,
T3. targeting the intervention at both genders, and
T4. a control group with no intervention until after the end-line survey.

First, we selected 80 villages in Sindh Province and 99 villages in Punjab Province for potential inclusion in the intervention to be implemented by IRC and Bedari, respectively. We collected village-level baseline data on key village characteristics and conducted a household listing exercise in each village to obtain a census of eligible households comprising at least one unmarried adolescent aged 14 to 17 years. We randomly selected 10 households per village from this list: 5 households with an adolescent boy and 5 households with an adolescent girl.

Next, a baseline household survey collected data amongst others on key demographic and socio-economic indicators, as well as on attitudes, beliefs, norms, and behaviors about child marriage, girls’ education, and VAWG. Through lab-in-the-field experiments played with the three respondents per household (male and female caregiver, and adolescent child), we elicited their individual beliefs and their preferences for investments in the adolescent child.

After baseline data collection, villages were randomly assigned to one of the four treatment groups. Edutainment activities (movie screening and follow-up discussions) were conducted across both provinces from 28th December 2018 to 30th June 2019. The interventions were monitored through exit surveys and attendance lists.

Since the start of the baseline activities until completion of the endline data collection, monthly village-level marriage data and school enrolment and attendance data are collected through focal persons, village focus group discussions, school principals and local government administrations.

The endline survey is conducted among the initially selected households in both Sindh and Punjab Province from the beginning of October 2019 until the end of June 2020. The endline survey includes all baseline components, as well as additional modules on experiences with the intervention, returns to education, and adolescent marriages.

After the baseline survey, we randomly assigned villages to one of the four treatment groups: ‘men and boys only’ edutainment intervention (“T1” villages), ‘women and girls only’ edutainment intervention (“T2” villages), edutainment intervention targeted at ‘both genders’ (“T3” villages) and a control group (“T4” villages) where no intervention activities were offered. We used Mahalanobis distance matching techniques to randomly assign villages to one of the four treatment arms, stratified by province.

The unit of treatment randomization was the village-level.

Yes

Planned number of units: 179 villages (80 villages in Sindh Province and 99 villages in Punjab Province). Due to logistical issues, two villages in Punjab were dropped from the study. Our final number of clusters is hence 177 villages (80 villages in Sindh Province, with 20 villages assigned to each of the four treatment arms, and 97 villages in Punjab Province of which 25 villages in the “women and girls only” treatment arm and 24 villages in each of the other three arms).

Planned number of observations: 1,770 households (10 households in each of the 177 villages – with 3 respondents per household: adult male, adult female, and either an unmarried adolescent boy or an unmarried adolescent girl). Due to their small size, some villages did not comprise sufficient households with adolescent teenagers, leading to a final sample size of 1,716 households.

Treatment arm 1 (men and boys only): 44 villages
Treatment arm 2 (women and girls only): 45 villages
Treatment arm 3 (both genders): 44 villages
Control arm (no intervention): 44 villages