The impact of an entertainment education TV series, ‘C’est la vie!’ on Gender-based Violence and Sexual and Reproductive Health

The TV series C’est la vie! is now in it’s third season of production and is specifically designed to address issues related to adolescents’ and women’s rights. The plot revolves around everyday life in a maternal health clinic in Senegal and characters are based on extensive formative research. Themes presented in the series focus on GBV (IPV, forced and early marriage, sexual abuse, female genital mutilation and cutting (FGM/C), illegal abortion), on sexual and reproductive health (family planning, use of contraceptives, HIV and sexually transmitted infections), on maternal and child health (prenatal and postnatal care, quality of health care, traditional medicine), and more generally on gender equality, couple communication, and female autonomy. C’est la Vie! is developed and produced by the Réseau African pour l’Education à la Santé (RAES), a Senegalese non-governmental organization with support from UN partners.

The intervention being evaluated is Season 1 and 2 of the TV series C’est la vie!. Due to interruption of implementation by COVID-19, the intervention has two distinct implementation modalities. From December 2019 to March 2020, the intervention was screened through regular film clubs in each village. Season 1 of C’est la vie! is composed of 26 episodes (25 minutes for each episode). Each village screening showed three episodes in a row and took place every other week. Film versions of C’est la vie! season 1 are translated to Wolof and Pular, the dominate languages in study villages. In addition to the TV series, study arm 2 includes “Pedagogical kits,” which have been developed by the RAES social behavior change communication (SBCC) team, in collaboration with United Nations Agencies. The objective of the pedagogical kits is to strengthen the impacts of the TV series by stimulating personal reflections and collective debates. The kits are composed of post-projection discussion guides that accompany each episode and workshop guides that are composed of seven themes.

Implementation was carried out by MobiCiné, through mobile units that visited each village on a rotating basis. Mobile units are cars carrying projectors and screens. Each mobile unit staff includes a screening technician and a communication specialist for monitoring attendance and leading the SBCC component (Pedagogical kit including the post-screening discussions and workshops). Implementation of the film clubs was tested via a pilot study which was approved by the Senegalese ethics committee, Comité National d’Éthique pour la Recherche en Santé (CNERS), in March 2019, and carried out over a two week period in six villages over the month of April 2019. The pilot study was implemented with the objective of determining implementation feasibility, optimization and test intervention components prior to development of the present impact evaluation.

After the disruption of the intervention due to COVID-19, the implementation will take place remotely, via podcasts transmitted through 10-20 min segments (season 1 and 2 of C’est la vie!). These will be accessed via a free hotline by study women in both treatment arm 1 and 2, and run from June to September 2020. The intervention will continue to be overseen by MobiCiné, and post-discussion groups will be facilitated via one-to-one calls with communication specialists or in groups of up to 10 women via the call platform (treatment arm 2 only). C’est la vie! has already been converted to podcasts for radio dissemination in Wolof. A placebo podcast will also be administered on the control arm with content unrelated to the C’est la vie! themes. The podcast intervention will be rolled out in 60% of each treatment arm in order to disentangle impacts of the film clubs only versus film clubs plus additional podcast extended treatment. For the continued treatment groups, the intervention is expected to end with a last in-person ‘movie night’ in September or October 2020 to thank women for participation (pending the spread and containment of COVID-19). Similar to the original randomization, continued treatment villages will be selected randomly, stratified on region.

2019-12-02

2020-10-30

Knowledge, attitudes and practices of C’est la Vie! themes, including GBV, sexual and reproductive health, and maternal and child health.

• Knowledge: Regarding GBV (including IPV, FGM/C and early marriage), sexual and reproductive health (including family planning and HIV/STIs), and maternal and child health;
• Attitudes and personal norms: Around GBV (including IPV, non-partner violence, FGM/C, and early marriage), sexual and reproductive health (including family planning), and gender equitable attitudes;
• Practices and behaviors: Regarding GBV (including IPV, non-partner violence, FGM/C of daughters and intentions, early marriage), sexual and reproductive health (including family planning utilization, HIV/STI testing and proxy indicators), and maternal and child health service utilization.

Moderators and facilitating factors, including intra-household decision-making power and self-efficacy, time allocation, communication, labor force participation, social cohesion/capital, emotional wellbeing, aspirations and community social norms.

Outcomes expects to be affected by participation in the intervention, including benefits of gathering in groups and increased leisure time – but not key themes of C’est la Vie! – including time allocation, closer personal relationships with friends and family, social support and trust, emotional wellbeing, intra-household decision-making power and self-efficacy, and labor force participation.

Due to the large number of outcomes and indicators, for each theme, we will group outcomes into indices at the level of knowledge, attitudes and behavior. We will construct equally weighted averages of z-scores for index aggregation and also report outcomes for individual indicators. More details are given in our pre-analysis plan.

Due to concerns around ethics of asking sensitive behaviors remotely, the intermediate phone survey will assess only primary outcomes of knowledge and attitudes, alongside select secondary outcomes. The full set of behavioral outcomes will be assessed at endline when in-person surveys are possible. See pre-analysis plan for further details.

In total, 120 rural villages across two regions of Senegal – Kaolack and Kolda – are randomly assigned to one of three arms:

• Arm 1: C'est la vie! season 1 and 2 transmitted through film clubs and podcasts
• Arm 2: Same as Arm 1 plus “pedagogical kits”;
• Arm 3: a TV series "placebo" (on a subject not related to GBV or gender issues) screened through film clubs and podcasts (control group).

Randomization is stratified on region, and at the village level as opposed to the household level because it is likely that individuals will discuss the TV series C’est la vie! with other community members; thus, even individuals who are not directly invited to the film clubs may be exposed to the messages, and they cannot be considered ‘untreated’. In each intervention arm, 34 young adult women aged 14-34 per village, living up to 2 km radius to the film club location (village primary school) were invited to attend the sessions. Each woman was allowed to bring one guest. The second stage of randomization assigns women to use their guest pass on either: 1) their husband, partner, boyfriend/male peer, or brother or 2) a female friend, sister or neighbor. The second stage randomization is at the individual level across all three treatment arms and takes the form of a soft nudge (suggestion) at the time of invite, as there is no strict monitoring or consequence if the women does not bring the type of person suggested.

Prior to the first screening and baseline survey, a census was conducted in 160 villages to identify 34 eligible women to participate in the study. Of the villages with at least 34 eligible females, 120 were randomly selected to participate in the study and randomized into one of the three treatment arms. Women are eligible if they 1) meet the age criteria, 2) speak and understand wolof or pular, 3) live within a 2km radius of a primary school.

In office, by a computer.

First level (treatment arm): Villages
Second level (sex of second invitee): Individual

Yes

120 villages

4,080 women (1,360 per study arm)

Equal numbers of clusters (40) per treatment arm, stratified on region (20 per region per arm).

Power calculations for this study drew on data from the 2017 Senegalese Demographic and Health Survey (DHS). We conducted power calculations on three primary outcomes related to behaviors and attitudes of IPV and sexual and reproductive health among women – modern contraceptive use, attitudes justifying IPV and 12-month experience of IPV. Desired statistical power was set to 80 percent, and desired significance level was set to 0.05. Means and intra-cluster correlations were estimated from the 2017 DHS, using the rural population and accounting for survey weights. We allow for 0.10 correlation with baseline characteristics. The desired minimum detectable impacts were set to 6 percentage points for contraceptive use and IPV, and 10 percentage points for attitudes. Given the low prevalence of IPV in the last 12 months and modern contraceptive use, 6 percentage points are large effect sizes, however, these are in the range found by studies examining impacts of social behavior change communication and mass media on both outcomes. For contraceptive use and attitudes towards IPV we need a minimum of 28 women per village by endline. For IPV, we are not powered to detect impacts across each intervention arm, but we are powered to detect impacts if we combine intervention arm 1 and arm 2. The needed sample for IPV when arm 1 and arm 2 are combined is 23 married (partnered) women per village at endline. Assuming 5 percent attrition by endline, this means a sample of 24 married (partnered) women per village at baseline. If we assume approximately 70 percent of women 14-34 years are partnered, then this would mean we need a sample of approximately 34 women per village at baseline – 24 of which will likely be partnered and 10 without a partner. The total sample of women is thus 4,080 across the 120 villages. We note that due to geographical variation and differences in recall periods, among others, power calculations are not able to provide exact predictions of statistical power. However, given a sample design, they can provide useful guidance in designing the sample.