Abstract
Child marriage, defined as marriage before the age of 18, is a human rights concern and a form of gender-based violence. We co-produced a series of radio dramas with residents of Ibadan, Nigeria that addressed common justifications for child marriage and used a difference-in-differences approach to estimate the impact of the series on knowledge of and support for the practice. The drama series led to reductions in support for child marriage, though most respondents reported that their religion encourages marriage before the age of 18. Targeted, co-produced radio programming could be a promising, cost-effective means of addressing child marriage.
Introduction
Child marriage, defined by the United Nations (UN) as marriage before the age of 18, is a human rights and population health concern. The practice is strongly correlated with a wide range of negative social, economic, and health outcomes, and those correlations persist across varied geographic and sociocultural contexts (Fan & Koski, 2022).
Eliminating child marriage is a high priority for multinational organizations and national governments. The UN calls for the elimination of child marriage by the year 2030 as a means of advancing progress toward gender equality, and many nations have adopted legislation that prohibits marriage before the age of 18 in recent years (Government of the United Kingdom, 2023; Nierenberg, 2024; United Nations, n.d.). Nigerian government officials have pledged to eliminate the practice and have made doing so a domestic priority (Indira et al., 2023). Marriage before the age of 18 is now illegal throughout much—but not all—of the country (Government of Nigeria, 2003; UNICEF, 2021). Despite these efforts, Nigeria remains home to the largest number of child brides in Africa and has one of the highest rates of child marriage in the world (UNFPA & UNICEF, 2018).
There are stark differences in the frequency of child marriage between regions and ethnic groups in Nigeria. Child marriage is most common in the North Central and North East regions of the country, where the prevalence is greater than 70% in some states (UNFPA & UNICEF, 2018). The states where child marriage is most common are those that have resisted adopting laws against the practice. These geographic patterns in child marriage are correlated with ethnic patterns in the practice: girls from the Hausa ethnic group are most likely to be married as children, and Hausa persons most commonly reside in the northern regions of Nigeria. A recent study based on a nationally representative sample reported that 55% of Hausa girls between 15 and 19 years of age were married, a prevalence more than ten times as high as that among other major ethnic groups in the country (Mobolaji et al., 2020).
Child marriage is typically assumed to result from decision-making processes among adults and is not usually thought of as a decision made by the children themselves, or at least not one made independently (Horii, 2020a, 2020b; Lokot et al., 2021). For example, parents or other adult family members may arrange the marriage of young girls to mitigate financial hardship, particularly where marriage is accompanied by a financial transaction such as bride price.
For these reasons, our study focused on changing perceptions of child marriage among Hausa adults. We tested whether a series of radio dramas co-produced with Hausa persons residing in Ibadan and addressing motivations for child marriage that they identified could reduce support for the practice. Further information on the rationale for these choices is included in the published protocol (Omobowale et al., 2023).
Methods
Study Setting
This study was conducted in two urban centers in southwestern Nigeria. Ibadan, the capital of Oyo state, served as the intervention site. Akure, the capital of neighboring Ondo state, was the control site. These cities are comparable in geographic, demographic, and socioeconomic characteristics, and child marriage is illegal in both. Both cities are home to substantial populations of Hausa persons who have migrated from northern regions of the country.
Development and Airing of Radio Dramas
The intervention tested in this study was a series of ten radio dramas co-developed with persons of Hausa ethnicity residing in Ibadan. Prior to initiating this study, qualitative research was conducted with them to learn why child marriage persists. Participants in that research identified several motivations for the practice that were then addressed directly through the story lines scripted for the radio dramas, as described in Table 1. For example, participants in the focus groups shared concerns regarding the potential for pre-marital sexual activity and pregnancy among girls. One perceived solution to that concern was to arrange for girls to be married. Episodes 7 and 8 of the series addressed that issue (Table 1). In Episode 7, the parents of a girl who began menstruating at 10 years of age were initially considering marriage for her but were persuaded to delay that event. Episode 8 stressed that sexual maturation, including sexual interest, is a normal stage of human development that requires guidance and counseling from parents instead of marriage.
Outline of Topics Related to Child Marriage Addressed in Each of the Radio Dramas in the Series.
A Local Advisory Committee (LAC) composed of persons of Hausa ethnicity residing in Ibadan helped develop story lines for the dramas, which were then professionally scripted and produced with assistance from the Biomedical Communications Centre at the University of Ibadan College of Medicine. The dramas aired in the Hausa language on three radio stations in Ibadan, KING FM, Pensioners FM, and Lagelu FM, between May and August 2023. Each episode was 15 minutes in duration, including a 12-minute dramatic play and 3 minutes for introductory and closing material. One episode of the series aired each week on each radio station to provide a range of potential listening times. To encourage listenership, small prizes in the form of cell phone credit were awarded after each episode to the first listeners who correctly responded to two questions about the content of the drama via text message. A more detailed description of co-development of the radio dramas and community engagement in that process is available in a separate publication (Omobowale et al., 2024).
Data Collection
The radio dramas were intended to reach an audience of adult members of the Hausa community in Ibadan. Therefore, to be eligible for inclusion in this study, potential participants needed to be at least 40 years of age, identify as Hausa, and reside in one of the sampled neighborhoods in Ibadan or Akure. Information on sampling procedures is included in the published study protocol (Omobowale et al., 2023).
Baseline data were collected through in-person surveys conducted in Ibadan and Akure in February 2023, before the radio dramas aired. All participants provided verbal informed consent at the beginning of the survey. Participants were asked to provide demographic information and to answer questions regarding their knowledge of and opinions and perspectives on child marriage. Specifically, they were asked whether certain behaviors or circumstances that aligned with the story lines in the radio dramas, like sexual activity or pregnancy, justified child marriage. They were also asked what they believed the “ideal” age for marriage was among girls and boys, whether they were willing to delay the marriage of their own child(ren) until after the age of 18, whether they would object to the marriage of a person in their family before the age of 18, and whether there were any potential harms resulting from child marriage. All outcomes of interest were identified prior to data collection. They were also asked to provide contact information so they could be reached to participate in the post-intervention survey.
The same individuals were contacted to participate in the post-intervention survey in September 2023. Many participants from Akure were unavailable for in-person surveys because they were tending their farms at the outskirts of the city, as it was the start of planting season. The logistics of following up in person across such a wide geographic range were not feasible given the study's budget. As a result, post-intervention surveys with participants from Akure were conducted via telephone.
Participants were asked the same questions regarding their knowledge of and opinions on child marriage, which allowed us to compare those responses to reports at baseline. The post-intervention survey also included two questions intended to assess coverage of the radio dramas. The first was general: in the past 2 months have you heard any radio shows about child marriage? The second referred to the drama series using its Hausa title, Burina.
Impact Evaluation
We measured the effect of the radio dramas on all outcomes using a difference-in-differences approach that compared differences in the frequency of the outcomes over time and between Ibadan and Akure. This approach is advantageous because it accounts for time-invariant traits that differ between the cities, as well as any pre-existing trends in the outcomes. Given that our study involved only two locations and two time points, we used a straightforward modeling approach represented by the following equation.
In the model above, the measured outcome (Y) for individual i (Yi) is a function of the city they reside in
This research was reviewed and approved by Institutional Review Boards at the University of Ibadan and McGill University.
Results
Sample Characteristics
We recruited 1596 persons at baseline: 802 from Ibadan and 794 from Akure. Of those, 1546 (97%) completed the post-intervention survey. Only 50 persons were lost to follow-up (36 from Ibadan and 14 from Akure) resulting in a total of 766 complete observations in Ibadan and 780 in Akure. The 50 individuals who could not be reached to complete the post-intervention survey were dropped from the analyses.
Characteristics of samples from both cities are described in Table 2. Men comprised 73% of people recruited from Ibadan and 82% of those recruited from Akure. Almost all respondents were married and identified as Muslim. Roughly half of respondents in both cities, 44% in Ibadan and 55% in Akure, reported having received Islamic education. This refers to schooling based exclusively or nearly exclusively on the Qur’an, the central religious text of Islam. The majority of respondents reported that their state of ancestral origin was in the North West region of the country. All respondents had immigrated to Ibadan or Akure, though immigration tended to be more recent among respondents living in Akure.
Sample Characteristics.
Coverage of the Intervention
Sixty percent of participants in Ibadan reported hearing radio programs on child marriage. There may have been some minor spillover of the intervention as 48 respondents from Akure (6%) also reported they had heard the dramas. No other radio programs on child marriage were airing at the same time as our intervention.
Effect on Justifications for Child Marriage
The distribution of responses to questions regarding justifications for child marriage in both cities before and after airing of the radio dramas is shown in Table 3, along with difference-in-differences estimates of the impact of the dramas on each of these justifications. Most respondents believed there were justifications for child marriage: even after the intervention, only one third (32%) of respondents from Ibadan said there were no justifications and only 10% from Akure indicated there were none. Most people agreed that religion encourages child marriage. At baseline, 87% of respondents in both cities reported they believed religion encourages marriage before the age of 18. Those percentages fell after the radio dramas aired. The decline was 18 percentage points greater in Ibadan compared to Akure, but more than half of respondents in both cities (53% in Ibadan and 71% in Akure) reported that religion encourages child marriage even after the series aired (Table 3).
Respondents’ Reports of Agreement With Various Justifications for Child Marriage and the Effect of the Radio Drama Series on Those Responses, Based on Difference-in-Differences Models.
Interestingly, the percentage of people who reported that child marriage is justified if a girl has sex or becomes pregnant increased between the baseline and post-intervention surveys in both cities. Support for all the rationales we asked about remains prevalent.
Effect of the Intervention on Perceptions of the Ideal Age for Marriage
Respondents were asked what they believed the “ideal age” for marriage to be during both baseline and post-intervention surveys. The shift in the distribution of reported ideal ages in both cities is illustrated in Figure 1. The distribution remained essentially unchanged in Akure but shifted toward later ages in Ibadan on the post-intervention survey. Difference-in-differences estimates indicate the dramas led to a 15-percentage-point reduction in the proportion of study participants who reported an ideal age of marriage less than 18 for girls (Table 4). Very few participants reported that it was ideal for boys to marry before the age of 18 at baseline and we found no indication the intervention affected that perception (Table 4).
Distribution of reported ideal ages of marriage for girls in Akure and Ibadan. Grey bars represent the distribution at baseline. Black bars represent the distribution reported in the post-intervention survey.
Estimated Effect of the Radio Drama Series on Perceived Ideal age for Marriage, Support for the Practice, and Knowledge of Harms, Based on Difference-in-Differences Models.
Results of triple-differences comparisons did not indicate the magnitude of the effect on ideal age at marriage differed by gender, but there was some indication the dramas may have had a larger effect among respondents who had Islamic education compared to those who had some non-religious education (i.e., those who attended primary or secondary school or a higher level). The proportion of people who reported an ideal age of marriage less than 18 fell by an additional 13 percentage points among people with Islamic education only, compared to those with non-religious education (95% CI 0.00, 0.27). The confidence interval is wide but is primarily consistent with a greater reduction among those with Islamic education.
Effects on Support for Child Marriage
Our intervention also decreased reported support for child marriage based on two indicators. First, respondents were asked whether they would be willing to delay the marriage of their child until after the age of 18. At baseline, 50% of respondents in Ibadan reported a willingness to do so and that proportion rose to 66% after the series aired. In comparison, the same proportion remained stable at 42% in Akure, resulting in a difference-in-differences estimate that indicates the dramas resulted in a 16-percentage-point increase in that outcome (Table 3). Second, we also asked participants, “Would you object if someone in your family wanted to get married before the age of 18?” As shown in Table 3, the increase in willingness to object was 30 percentage points higher in Ibadan compared to Akure.
There was no indication that the magnitude of either of these estimates related to support for child marriage differed by the gender of the respondent, but education does seem to change the magnitude of the effects according to triple-differences comparisons. The effect on willingness to delay the marriage of one's own child was 17 percentage points higher among those with Islamic education compared to those with non-religious education (95% CI 0.31, 0.03). The effect on objection to the marriage of a child in one's family was 9 percentage points greater among respondents with Islamic education (95% CI −0.04, 0.22).
Effect on Perceptions of Harm
At baseline, fewer than half of respondents in Ibadan (41%) and Akure (32%) indicated they believed there were harms to marrying as a child. However, those percentages roughly doubled in both cities after the series aired (Table 4). Part of the magnitude of these increases is likely due to a subtle change in the survey questionnaire: at baseline, participants were only permitted to respond yes or no to the question about harms, but at endline they could also indicate that they didn’t know. When measuring the effect of the intervention on this outcome, we dropped individuals (n = 267, or 17% of all respondents at follow-up) who reported that they did not know whether there were harms on the post-intervention survey for comparability purposes, but this necessarily led to a smaller denominator that affected percentages. As a sensitivity analysis, we coded all don’t know responses as though participants reported there were no harms. This resulted in a slight increase in the magnitude of the estimated effect of the dramas, from a 13-percentage point increase to a 16-percentage-point increase in those reporting that child marriage can be harmful.
Discussion
Ending child marriage is a priority for the Nigerian government (Indira et al., 2023; UNICEF, 2024). Prior interventions that have proven effective at reducing the frequency of child marriage at large scale in other settings, such as cash transfer programs, are often prohibitively costly or politically infeasible for governments (Malhotra & Elnakib, 2021). Thus, there is a pressing need for effective means of preventing child marriage that are readily scalable and financially feasible, particularly for governments in countries like Nigeria where the rate of child marriage is high and monetary resources to address it compete with many other urgent needs. This is especially true shortly following the COVID-19 pandemic and at a time of rampant inflation in Nigeria that creates even more competition for financial resources.
The radio programming intervention tested in this study was designed to address that need. Radio has the potential to reach large audiences at low cost. Crafting messages that reflect the perspectives of target audiences has the added benefit of addressing the root causes of child marriage, something that cash transfer programs are typically unable to do.
Our results show that radio programming is an effective way of reaching the targeted audience. Sixty percent of respondents in Ibadan reported having heard radio dramas about child marriage over the course of our study. Given that there were no other radio-based projects related to child marriage ongoing at that time, we are confident that statistic reflects the reach of the radio dramas. Our results also indicate that the dramas were effective at changing people's support for the practice and their knowledge of the potential harms of child marriage.
Our results reflect the complexity of motivations for child marriage. Few people reported there were no justifications for child marriage yet nearly half—even at baseline—also reported there were harms associated with it. That is consistent with the challenging circumstances that often lead to child marriage, such as poverty and conformity with social norms. For example, having insufficient financial means to provide for a girl child may be one reason a family would support her marriage, even if they are concerned that marrying may result in a different spectrum of challenges. A family may also arrange the marriage of a young girl out of concern that failure to do so might subject her to greater harm because of violating social norms.
Our results highlight the importance of engaging with religious beliefs regarding child marriage. This issue was raised during qualitative research we conducted prior to this study and led to the inclusion of an Islamic cleric on the LAC that helped guide this research. The cleric's perspective that Islam does not encourage child marriage was expressed in one of the dramas (Table 1). At baseline, nearly nine of ten respondents in both cities reported that religion encourages child marriage. Although that proportion declined, more than half of respondents in Ibadan reported the same perspective after the series aired. None of the interventions included in three systematic reviews of efforts to prevent child marriage addressed religious aspects of this practice directly (Kalamar et al., 2016; Lee-Rife et al., 2012; Malhotra & Elnakib, 2021).
Interestingly, we also found evidence that the dramas had a larger effect on the perspectives of people who reported having an Islamic education compared to those who reported any level of non-religious schooling. The reasons for that are not clear. However, the proportion of participants who reported hearing the dramas was not markedly different between those groups, so we are reasonably confident that differences in exposure do not explain these heterogenous effects.
Limitations
Nigeria experienced massive inflation while our project was being conducted. We initially budgeted to air each episode three times per week on each of three stations. By the time the dramas had been produced and were ready to air, inflation made that impossible and our budget only permitted airing each episode once per week on each station. One of the stations, KING FM, struggled to manage inflated costs, leading to inconsistent airing time for our program. Reported listenership remained high despite these difficulties.
We did not anticipate that the timing of our follow-up survey would coincide with a planting season that meant many participants in Akure, the control site, were unavailable to respond to survey questions in person. As a result, we had to collect endline data from respondents in Akure over the phone. Response rates remained very high because we collected thorough contact information at baseline. However, if responses to survey questions were systematically different when delivered over the phone rather than in person, this could be a source of misclassification that would bias our results. In-person and telephone responses might differ if the questions asked for sensitive information. None of the information we collected seemed particularly sensitive among respondents, as demonstrated by a high level of willingness to report support for child marriage, for example. Therefore, we do not have reason to believe this unanticipated change in mode of data collection is cause for concern.
We anticipated that recruiting women into the study would be challenging due to social norms in Hausa culture that result in women spending most of their time in their homes and often needing permission from their husbands to participate in research. To mitigate that, we ensured that female data collectors were available to survey female participants. Even so, only 27% of respondents from Ibadan and 18% from Akure were women. We were underpowered to examine differences in the impact of the dramas by gender as a result.
Measuring respondents’ intended behaviors is not the same as measuring the effect of this intervention on the actual incidence of child marriage. People do not always act in the ways they say they will or in the ways they expect they would. Moreover, social desirability bias is to be expected when studying an intervention like this and would likely result in exaggerated reports of lack of support for the practice. Funding for this project was limited to a one-year timeline, which was not sufficient to measure the effect of the dramas on the incidence of child marriage. However, the consistency and magnitude of the effects suggests that targeted, co-produced radio programming may be a promising way forward. Similar radio and storytelling interventions to address child marriage in other African and South Asian settings have shown similarly promising results, though few have been rigorously evaluated (Green et al., 2023; Udengwu et al., 2024). We hope these collective results motivate funding for longer-term projects that can evaluate whether such interventions go beyond influencing knowledge and intentions and affect the frequency of child marriage.
Footnotes
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this research was provided by Grand Challenges Canada through the Stars in Global Health Program [Grant number ST-POC-2206-53579].
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.