Domestic violence laws and women’s unmet need for family planning: Quasi-experimental evidence from Africa

Abstract

Background

Approximately 164 million women report an unmet need for family planning globally. This has far-reaching consequences for the health of women and their children. Women’s exposure to intimate partner violence (IPV) is strongly linked to increased unmet need as IPV likely affects both women’s desire for contraception and their ability to access it. Around 245 million women were subject to physical and/or sexual IPV by an intimate partner in the past twelve months alone, making it the most common form of violence against women. Yet, laws that prohibit domestic violence (DV) are not universal and countries actively debate whether legal provisions are effective in deterring and reducing its harmful impacts. This study examines the impact of DV laws on women’s unmet need for family planning.

Methods

We built new data on DV laws adoption in Africa and used the Demographic and Health Surveys (DHS) data collected between 2000 and 2021 across 23 African countries for outcomes data. Exploiting the staggered adoption of DV laws across the continent, we used a difference-in-differences study design to estimate the impact of DV laws in the treated countries compared to countries without such laws.

Results

We find that DV laws reduced women’s unmet need for family planning by 6.2% points, 95% CI [− 9.2, − 3.2], a 20.5% reduction from the mean. Positive impacts were observed in 6 countries across multiple model specifications. While the impacts were largely broad-based across wealth, age, geography, and education categories, we found no significant impacts for women younger than 20 years of age and those without formal education. These findings were robust to alternative model specifications.

Conclusions

Our findings demonstrate that countries can significantly benefit from prohibiting DV as a fundamental step towards addressing women’s unmet need for family planning and promoting their reproductive health. While fully addressing IPV and women's unmet need may require a range of complementary interventions, especially among marginalized populations, DV laws play a crucial role in improving women's control over their reproductive health.

Background

Freedom from violence and the full realization of reproductive rights are both essential to women’s bodily autonomy, health, and opportunities. Yet the world has far to go in achieving either. According to 2018 WHO global estimates, around 245 million women were subject to physical and/or sexual intimate partner violence (IPV) annually, which remains the most common form of violence against women [1]. Meanwhile, approximately 164 million women report an unmet need for contraception [2]. The consequences of IPV for individual women as well as their children, broader communities, and countries are profound.

IPV can result in extensive and debilitating harm to physical and mental health, with specific consequences including injury, stress-related gastrointestinal disorders, sexually transmitted infections, gynecological problems, chronic pain, depression, anxiety, insomnia, post-traumatic stress disorder, and death, among others [3, 4]. Women experiencing IPV may also face economic obstacles to accessing health care, including barriers to working or lose their jobs as a result of the violence, making it more difficult for them to leave a violent relationship or attain economic security and independence long-term [5]. IPV also has grave intergenerational health consequences, with children’s exposure to IPV associated with higher risks of neurological disorders, chronic stress, sleep problems, frequent headaches, asthma, emotional and behavioral problems, post-traumatic stress and anxiety disorders, poor nutrition outcomes, as well as education disruptions, substance use disorders, and barriers to employment in adulthood [6, 7]. Given its collective toll on health, work, and livelihoods, UN Women has estimated that IPV costs the global economy around $1.5 trillion per year [8].

Among the many consequences of IPV, studies across the globe have found that IPV is associated with adverse reproductive health outcomes, including unmet need for family planning [9,10,11] and unintended pregnancies [12,13,14,15,16,17,18]. Unmet need for contraception is widely recognized as a barrier to women’s reproductive health rights to freely choose the number, timing, spacing, and partners with whom to have children [19]. Unintended pregnancies have profound health implications for mothers and their children. Where reproductive healthcare is limited and abortion laws are restrictive, women facing unintended pregnancies often resort to unsafe abortion methods. Studies have found that complications due to unsafe abortions account for 8% to as many as 13% of maternal deaths each year [20]. Women with unintended pregnancies are vulnerable to increased rates of morbidity and mortality, often exacerbated by poverty, malnutrition, and limited healthcare access [21, 22]. Children born from unintended pregnancies are found to have an elevated risk of infant and child mortality [23,24,25]. Closely spaced pregnancies lead to complications like low birth weight, premature birth, and small for gestational age [26].

Married or cohabiting women rarely cite lack of access or cost related barriers as reasons for not using contraception and DHS data have been useful in revealing reasons why women may not seek contraception [27, 28]. A study of DHS data across 52 countries from 2005 to 2014 found that approximately 23% women were not using contraception despite wanting to avoid pregnancy because they or others close to them opposed contraception [27]. Women’s exposure to IPV is strongly linked to increased unmet need for family planning through multiple pathways (see Fig. 1). First, sexual violence, a common manifestation of IPV, directly infringes upon women's reproductive rights, resulting in forced or unprotected sex [29]. Second, abusive men are more likely to engage in contraceptive manipulation and contraceptive sabotage to gain control over their partner’s use of family planning and increase their partner’s dependency [30]. Finally, violent relationships are often characterized by fear and controlling behaviors by partners, undermining women’s ability to negotiate contraceptive use for fear of violence [29, 31].

Fig. 1

Pathways for IPV’s impact on women’s unmet need for family planning. Modified from García-Moreno et al. [31], Moore et al. [29], and Coker [32]

Because IPV is likely to impact both women’s desire for contraception and their ability to access it [33, 34], interventions that address IPV may also have benefits for women’s unmet need for family planning. Yet few causal studies have examined the impact of domestic violence legislations [35, 36], and no causal studies have looked at what works at a national scale to address IPV and its impact on unmet need for contraception in LMICs. Domestic violence legislations and other laws advancing gender equality can both directly support the realization of women’s fundamental rights by deterring discriminatory acts and providing new mechanisms for recourse when rights are violated and can also meaningfully advance gender-equitable norms, which is critical to long-term change [37, 38]. At the same time, legal remedies may be out of reach for the most marginalized women and ensuring that domestic violence laws in particular are both effectively and equitably enforced has been a longstanding challenge across countries. Stigma and family or cultural pressure to avoid reporting, geographic and socioeconomic barriers to courts and lawyers, and inadequate budgets for investigating and prosecuting IPV claims may all hinder effective implementation, especially by women with fewer economic resources. Results of associational studies have been mixed [39, 40].

This study helps to fill these gaps in the literature by using longitudinal data to examine whether national laws prohibiting domestic violence affect rates of unmet need for contraception. This study focuses on Africa where high rates of IPV have been reported [1] and unmet need for contraception remains high despite substantial progress over the past twenty years [2]. There is also substantial variation within the region in IPV prevalence and contraception use [2, 41, 42]. Associational studies across African countries have also found a relationship between exposure to sexual violence, unmet need, and contraception usage [9, 43,44,45], as well as unwanted pregnancies [46, 47]. By rigorously comparing women’s experiences in countries that enacted domestic violence laws with those that did not, this study offers causal evidence into the relationship between IPV and unmet need for family planning and how these impacts vary among women with different socioeconomic backgrounds, while generating actionable evidence about the potential of national laws to help address these dual major challenges.

Methods

We use difference-in-differences (DD), a widely used quasi-experimental research design, to estimate the impact of domestic violence laws on women’s unmet need for family planning. Exploiting the natural variation in the timing of policy adoption across countries, the DD research design identifies policy impact by comparing the trends in outcomes within and across treatment and comparison countries over the study period. Beginning in the early 2000s, many countries across Africa adopted laws prohibiting domestic violence while many others are yet to legally protect women against domestic violence. We use this difference in legal provisions against domestic violence across countries to isolate the impact of domestic violence laws on unmet need from overall trends across the continent.

Policy and outcomes data

The policy data comes from the longitudinal database on domestic violence (DV) laws created by the authors and the WORLD Policy Analysis Center at UCLA. The database covers the period 1990–2021 and uses rigorous techniques developed by WORLD to translate original legislative text into quantitative measures [48], including independent double coding by two policy analysts for each country. This coding process captured how each country's laws regarding domestic violence changed over time, focusing on three key aspects: (i) existence of a specific law against domestic violence, (ii) availability of restraining orders as a legal protection measure, and (iii) classification of domestic violence as a criminal offense. Countries with legislations specifically addressing domestic violence, broader laws prohibiting violence against women, or amendments to penal codes that criminalized domestic violence and established penalties were coded as having legal protections against DV starting from the year the law came into force. Countries with no laws against domestic violence or those lacking mechanisms for victims to seek recourse, such as criminal penalties or restraining orders, were coded as having no legal protections throughout the study period. Notably, no additional or gradual changes in DV laws were observed in the treatment countries during the study period.

The outcomes data came from the Demographic and Health Surveys (DHS) harmonized by the Integrated Public Use Microdata Series (IPUMS). In cases where a harmonized version wasn't available, a researcher at our center harmonized the data from the original DHS survey following the IPUMS codebook. This approach ensured all surveys used the same methods. The DHS surveys target women aged 15–49 and gather in-depth information on health and demographics across over 90 low- and middle-income countries at regular intervals. The repeated cross-sectional DHS data are ideal for making comparisons across many countries over time due to the standardized way these surveys are conducted and their consistent measurement techniques. For our analysis, we specifically focused on African countries with at least two DHS surveys containing family planning data collected between 2000 and 2021. We focused on all married or cohabiting women born after 1975. These restrictions allowed us to examine changes in unmet need for family planning within each country over time.

We merged the domestic violence legislative database and the outcomes data collected by the DHS for countries that met our inclusion criteria in Africa. This results in an analytical sample pooled data from 69 DHS surveys across 23 countries (see Table 1). We dropped surveys conducted within two years of DV laws’ adoption because the time elapsed would have been insufficient for implementation and to determine treatment effects. The treatment group comprised 11 countries that have data on women surveyed before and after the adoption of DV laws. The comparison group included 12 countries that either lacked domestic violence laws entirely or implemented them after the most recent available national survey. We used only pre-reform data in Nigeria and included them as part of the comparison group because Nigeria’s DV law was not national.

Table 1 Treatment and comparison countries as per changes in domestic violence laws and DHS survey availability

Measures

Outcome variable

We used a dichotomous variable indicating women’s unmet need for family planning (yes/no) using DHS’s revised definition of unmet need [49]. It is the sum of unmet need for spacing (women who wanted to delay having another child) and limiting (women who did not want another child) among women of reproductive age who were married or cohabiting, fecund, and/or sexually active. We constructed a binary variable (see Additional file 1 for details) using this information by coding all women reporting an unmet need for spacing or limiting as 1 and all other women who reported using a contraceptive, having no unmet need, or having a desire for birth within the next 2 years as 0. We excluded women who were infecund, menopausal, or sexually inactive because they had no need for family planning. This is a standard definition of unmet need used across studies in Africa using DHS data [9, 50, 51].

Primary independent variable: Exposure to DV laws

We considered women legally protected against domestic violence if they were surveyed after DV laws were in force. The legal protections provided women with access to recourse either in the form of restraining orders or criminal penalties against the perpetrator. All women who were surveyed before countries passed DV laws constituted the unexposed category. Table 2 shows the sample sizes and the prevalence of unmet need for family planning among married or cohabiting women in exposed and unexposed categories for each country in the sample. Among a total of 367,365 married or cohabiting women of reproductive age, 123,199 women were exposed to DV laws. In the treatment group, unmet need was 30.3% among unexposed women and 21.3% among women exposed to DV laws. Unmet need was 22.4% in the comparison group over the study period.

Table 2 Sample size and sample characteristics of women currently married or living with a partner in the treated and comparison countries

Control variables

We included as control variables women- and household-level characteristics that are identified as key determinants of unmet need for family planning, particularly in Africa [51, 52]. Women’s characteristics included her age in 5-year intervals, total years of education, and age at first cohabitation. Household characteristics comprised total number of members in the household, wealth status in quintiles, and urban/rural location. We also included country-level urbanization rates over time to account for changes in access to family planning services over time.

Statistical analysis

We used a two-way fixed effects (TWFE) model that allows for multiple groups to change over multiple periods in a difference-in-differences (DD) study setup. It is widely used to estimate the effect of laws and policies [53,54,55]. TWFE models estimate weighted average treatment effects across countries and time periods while controlling for country-specific and time-specific differences. Accordingly, we use the following equation:

$${Probit(Y}_{ict})={\beta }_{0}+{\beta }_{1} {DV\_law}_{ict}+ \delta {X}_{ict}+\xi {X}_{ct}+{\gamma }_{c}+ {\delta }_{t}+{e}_{ict}$$

(1)

to estimate the impact of DV laws on women’s unmet need for family planning. ${Y}_{ict}$ was a binary variable indicating whether woman i surveyed in country c in year t had an unmet need for family planning and ${DV\_law}_{ict}$ captured exposure to a DV law (it takes the value 1 for all women surveyed after the adoption of DV laws in treated countries and 0 otherwise). The main coefficient of interest was ${\beta }_{1}$ that estimates the effect of prohibiting domestic violence in treated countries compared to countries without DV laws. ${X}_{ict}$ represented individual- and household-level characteristics. The model accounted for time-invariant differences among countries as well as shared secular time trends over the study period across countries using country (${\gamma }_{c}$) and year (${\delta }_{t}$) fixed effects. Finally, ${X}_{ct}$ controlled for potential time-varying country-level confounders.

We used probit models because they are appropriate to analyze binary outcome variables. To estimate the output in terms of predicted probabilities, we calculated average marginal effects using the post-estimation commands, and report differences in the probabilities of unmet need. All models accounted for individual-level de-normalized sampling weights as recommended by the DHS for carrying-out multi-survey pooled analysis [56] and incorporated robust standard errors clustered at the country level.

Apart from estimating the pooled average effect of DV laws across all treatment countries, we also estimated the impact of DV laws in each treatment country individually. To examine country-specific impacts, we compared each treatment country at a time to the comparison group using the same estimation strategy. We could not carry out separate analysis for Benin and Uganda due to lack of overlapping data in the treatment and comparison groups in either the pre- or the post-reform period.

All statistical analyses were conducted using STATA 14.2. The study did not require IRB approval because it used publicly available data that are not individually identifiable.

Model assumptions

The validity of TWFE DD estimates rests upon the parallel trends assumption that the trends in unmet need would have remained similar across the treatment and comparison groups in the absence of DV laws. We tested this assumption using the event-study model. The event-study design uses a set of dummy variables to capture the impact of an event in each time period. The data are centered at the time of policy adoption (t = 0) and a set of leads and lags comprise the periods before and after the event. We tested the parallel trends assumption by examining all coefficient estimates during the pre-reform period with reference to the last unexposed group of women (t =− 1). The post-reform coefficient estimates indicate how treatment effects vary over time after the adoption of DV laws.

Additional analyses

While we looked at the overall impact of DV laws across all treatment countries and each one of them individually as well, the effects of these laws may vary across women depending on their socioeconomic backgrounds or other characteristics that determine access and use of family planning services. We report the heterogeneity of treatment effects across location, wealth, education, age, timing of marriage, and fertility preferences. Using linear probability models (LPM), we interacted policy exposure with each of these characteristics to determine differential policy effects. We report both the average effect of DV laws on each subcategory as well as the differential effect of the subcategories with respect to the reference omitted category to test statistical significance.

We conduct a range of sensitivity analyses to test the robustness of our findings. All models presented control for individual-level covariates and country-level urbanization over time. While religious beliefs may impact women’s unmet need, we do not include religion as a covariate in the main analysis due to data quality and availability issues. Similarly, we do not include partner’s education in the main analysis due to sample attrition. We also consider two macroeconomic indicators that, while not perfect, indicate access to and affordability of family planning services in a country. GDP per capita growth indicates the overall economic situation but could potentially be endogenous if DV laws enable more women to join the labor force and increase their incomes. Domestic governmental health expenditure (as a % of GDP) indicates public investments in overall healthcare, not limiting to family planning services alone. We present analyses including these variables as sensitivity checks.

Additionally, the main pooled analyses comprise 11 treatment countries including Benin and Uganda that were not individually estimable due to lack of overlapping data with the comparison countries. We dropped these countries from the pooled sample to test the sensitivity of the pooled treatment effects to exclusion of countries with insufficient data for country-specific analysis.

TWFE DD models are an extension of the simple 2-group × 2-time period DD models wherein the coefficient is a weighted average of treatment effects across all groups and times (Goodman-Bacon 2018) [57, 58]. Sometimes when the treatment timing varies across groups, negative weights may occur if already-treated units act as controls and changes in their treatment effects over time get subtracted from the DD estimate. We examined whether negative weights arise in our analysis and whether the treatment effects are sensitive to these occurrences.

We also re-ran the pooled analysis using the logit model that is widely used to study dichotomous outcome variables. Lastly, as a final sensitivity check, we estimated the model on a subset of currently married or cohabiting women who report no desire for births within the next two years i.e. women who indicate a demand for family planning. Because demand for FP may itself be susceptible to sample selection bias, the main results follow the standard practice of unmet need expressed as a percentage based on women who are married or in a union.

Results

Effects of domestic violence laws on women’s unmet need

Table 3 reports the marginal effects (difference in predicted probabilities in absolute percentage points) from the probit TWFE model shown in Eq. (1). These estimates measure the impact of prohibiting domestic violence on the probabilities of unmet need for family planning when compared to countries without such laws. Column (1) shows the estimates from the null model with only country and year fixed effects. Column (2) builds on the null model by controlling for individual-level confounders and finally, column (3) additionally controls for country-level urbanization levels that varies over time. We find that prohibiting domestic violence reduced women’s unmet need for family planning across all specifications. DV laws reduced the probability of unmet need by 6.2 percentage points overall, a 20.5% reduction from the mean.

Table 3 Impact on unmet need for family planning among married or cohabiting women of reproductive age

Table 3 also reports the impact of domestic violence laws on each of the treatment countries individually. We find consistent declines in 4 countries—Ethiopia, Malawi, Rwanda, and Zambia. Ethiopia witnessed greatest reduction of 14.9 percentage points, followed closely by Rwanda with 12.4 percentage points due to passage of domestic violence laws. The probability of unmet need declined by 8.7 percentage points in Malawi and by 6.5 percentage points in Zambia. Lesotho and Namibia show reductions in unmet need by around 10 percentage points across the null model and the model controlling for individual-level characteristics. These impacts are sensitive to the inclusion of country-level urbanization rates. Ghana, Madagascar, and Zimbabwe do not show any impact of domestic violence laws on women’s unmet need for family planning. The statistically non-significant results may also be due to insufficient sample sizes before or after policy implementation in many of these countries.

Examining model assumptions

The fundamental model assumption is that the trends in unmet need would have been similar across treatment and comparison countries in the absence of DV laws. To test parallel trends before the adoption of DV laws and homogenous treatment effects over time, we present the event-study plot for the pooled analysis [59] in Fig. 2. Figure 2 provides evidence that there were no differential trends in the probability of unmet need between treatment and comparison countries before the adoption of DV laws. It also indicates homogenous treatment effects over time except a slightly greater impact during the last time interval.

Fig. 2

Event-study estimates of the impact of DV laws on the probability of unmet need. Notes: Each time point represents two-year intervals to improve estimation. The graph shows the difference in the probability of unmet need between treatment and comparison countries. Time is centered at the last untreated group. Point estimates of the lag and lead terms are presented along with their 95% confidence intervals. The omitted base category is taken as 1 year prior to the reform, indicated by the solid vertical line. The pre-reform period shows no evidence of divergent pre-trends, and the post-reform period is suggestive of homogenous treatment effects over time

In addition to the parallel trends assumption, another significant challenge to the validity of DD estimates concerns their ability to isolate the impact of the policy from other concurrent changes that might have occurred during the study period. To test the possibility of other concurrent changes impacting access to family planning, we run the model on a group of women who would not be expected to be impacted by DV laws: women who were not married or living with a partner at the time of survey. If the observed change among married or cohabiting women is solely due to DV laws, then the placebo group shouldn’t show any significant change because DV laws do not apply to this group. We find that DV laws had no impact (1.7, 95% CI [− 9.2, 12.6]) on the unmet need for family planning among these women. These findings strengthen the argument that the decline in unmet need followed from the passage of DV laws and not any other policy changes (such as those increasing access to family planning services) or other macroeconomic factors that would have impacted the placebo group as well.

Heterogeneity analyses

In Table 4, we report the heterogeneity of treatment effects across location, wealth, education, age, timing of marriage, and fertility preferences. In Panel A, we find that DV laws impacted women across urban and rural areas. While the impact was substantially higher among women residing in urban areas at 10.4 percentage points than those residing in rural areas at 6.7 percentage points, the difference was not statistically significant. Panel B shows that the magnitude of the impact of DV laws increased significantly with women’s wealth background. The impact was the lowest among the poorest women at 4.5 percentage points and the highest among the richest women at 11.4 percentage points. Panel C highlights the vulnerability of women without any education as opposed to women with at least primary education. DV laws had no impact on women who were not educated whereas the impact grew with women’s education level from 8.0 percentage points for women with primary education to 11.6 percentage points for women with higher education. All differences were statistically significant when compared to women with no education.

Table 4 Heterogeneity in the impact of DV laws on women’s unmet need for family planning across different groups of women

Similarly, Panel D highlights the vulnerability of another group of women who are under 20 years of age. While DV laws had significant effects on women across different age groups ages 20 and older, they had no impact on women aged 15–19 years. In Panel E, we examine whether the impact varied across women who married before the passage of DV laws compared to those who married afterwards. The results show that DV laws impacted women regardless of whether they got married before or after the adoption of DV laws. Lastly, the impact of DV laws may vary depending on the difference in fertility preferences between the respondent and the partner. While DV laws reduced unmet need for family planning across the board, the impact was higher among women whose partners wanted the same or fewer children at about 9.5.-8.5 percentage points compared to the impact of 6.1 percentage points among women whose partners wanted more children than them.

Sensitivity analyses

Table 5 presents a range of sensitivity analyses that we conducted to test the robustness of our findings. In Panels A and B, we see that controlling for partner’s level of education or woman’s religious beliefs in addition to previously included individual and country-level covariates does not change our results. Panel C shows that our main findings are also not sensitive to controlling for overall macroeconomic indicators such as GDP per capita growth and domestic government health expenditures. Panel D excludes countries with insufficient sample sizes for individual country-specific impact estimation. The pooled treatment effects remain stable.

Table 5 Sensitivity analyses to test the robustness of the impact of DV laws on women’s unmet need for family planning

We examined whether negative weights arise in our analysis and tested the sensitivity of our treatment effects to these occurrences. We found only three country-year cases with negative weights (see Additional file 2). To assess their severity, we removed the data with negative weights from the analytical sample and re-ran the pooled analysis. The result in panel E is the same as our main result, suggesting that the presence of few negative weights did not impact the treatment effects. Lastly, Panel G shows that the coefficient estimate is larger in size among women with a demand for FP. DV laws reduced the overall unmet need by 11.6 percentage points, a 21.3% reduction from the mean, among women with a demand for FP.

Discussion

This study uses a quasi-experimental design to evaluate the impact of national laws prohibiting domestic violence on women’s unmet need for family planning. We find that domestic violence laws reduced the probability of unmet need by 6.2 percentage points overall, a 20.5% reduction from the mean. As the first study using causal methods to measure how domestic violence laws affect unmet need for family planning, this research makes an important contribution to the literature. While prior research shows that experience of IPV is associated with higher probabilities of unmet need for family planning, our study demonstrates that laws prohibiting IPV have broader implications for the realization of women’s sexual and reproductive health and rights. These results are robust to different model specifications and assumption checks.

Despite the recognition that the effectiveness of DV laws can vary significantly depending on factors such as specific legal provisions, enforcement, support systems for survivors, and the broader socio-cultural context, this study found strong evidence that DV laws were associated with a reduction in women's unmet need for family planning across 11 countries in Africa. Underreporting and other obstacles may mean that the observed impact reflects partial implementation and likely represents a conservative estimate. Addressing these barriers, such as improving access to support services, enhancing survivor safety, and challenging societal norms that perpetuate IPV, could lead to a more significant and sustained reduction in IPV prevalence and its associated harms.

This observed reduction likely reflects a combination of normative shifts and law enforcement. Researchers have found that even weakly enforced DV laws can contribute to positive social change by transforming women’s attitudes toward violence and raising awareness about protecting their rights [38, 60]. DV laws may reduce the prevalence of IPV [35], which in turn reduces barriers to women meeting their family planning needs [9, 43]. Previous literature also demonstrated that DV laws improve women's household decision-making power regarding healthcare and finances [36], which is likely to affect women’s ability to access and use family planning.

Among the 11 treatment countries in our sample, the impacts of domestic violence laws on unmet need were greatest in Ethiopia (14.9 percentage points), Malawi (12.4 percentage points), Rwanda (8.7 percentage points), and Zambia (6.5 percentage points). No significant effects were found in Ghana, Madagascar, and Zimbabwe, but these non-significant results may reflect relatively small sample sizes. Differences in policy design aspects and quality of enforcement could help explain the variation in impacts across countries. Future studies should examine the impact of policy design decisions in DV laws, including what constitutes violence, who is covered, whether it’s a criminal offence and its associated penalties, and access to restraining orders and institutional support systems.

This study also provides new insights into whether domestic violence laws are reaching women equitably. Women across demographics experienced reductions in unmet need, though the magnitude of impact varied for women from different backgrounds. For example, impacts among the richest women were over twice as large as impacts among the poorest women, as measured by wealth quintiles. Similarly, women with higher education experienced a greater reduction in unmet need (11.6 percentage points) than women with only a primary education (8.0 percentage points). The only groups for whom we found no significant effects were women with no formal education and the youngest women (ages 15–19).

In demonstrating the role of domestic violence laws in reducing unmet need for contraception, this study highlights the benefits of laws to the urgent priorities of women’s health, autonomy, and fundamental rights. At the same time, it provides further evidence that the ability of women to decide freely about when, how, with whom, and how many children to have depends on their socioeconomic conditions, communities, and political surroundings. To fully realize reproductive justice, future research should examine whether knowledge of laws, implementation barriers, or other obstacles are driving the variation in these impacts across socioeconomic status and age, among other demographic characteristics. Given the justice system may have limitations [61] in adequately addressing this issue for the most vulnerable and marginalized women, it is critical to identify strategies for more effectively reaching those least served by DV laws.

Importantly, we recognize that prohibiting domestic violence is only one aspect of addressing women’s unmet need for family planning and is only one part of a comprehensive approach to addressing IPV. Future research should consider other factors such as the supply and ease of access to contraceptives and accurate knowledge and awareness of family planning services which are also critical to reducing women’s unmet need for family planning. Furthermore, addressing the vulnerability to IPV and the need for family planning services among unmarried or non-cohabiting girls and women requires further attention.

A common limitation of studies using the DD approach is that despite effectively controlling for secular time trends and fixed country-level characteristics, it's still possible that impact estimates capture the combined effect of the policy and unobserved changes happening within countries over time. We addressed this concern by presenting the results for the placebo group of women who were not married or in union. Our results show that DV laws did not impact unmet need among these women, while they did reduce unmet need among women who were married or cohabiting. The lack of a similar reduction in the placebo group suggests that the relationship between DV laws and unmet need is unlikely to be confounded by other coinciding changes that are likely to impact all women regardless of marital status. If other unobserved factors, like family planning services or campaigns, were significantly influential, we'd expect some effect in the placebo group, unless those factors exclusively targeted married women. We also show in the main results and sensitivity checks that these impacts are robust to the inclusion of other relevant time-varying country-level characteristics such as urbanization levels, GDP per capita growth rate, and domestic government health expenditures.

The study reports the average effect of these laws on women's unmet need across all countries in the sample. It also reports the impact in each country individually. However, the study design does not consider the varying durations of historical exposure to these laws across countries. Although the event-study plot (Fig. 2) suggests that, on average, the treatment effects remain consistent over time, variations in the duration of exposure to these laws across countries may explain the observed differences in their impact. Additionally, we used the most widely available and used definition of women’s unmet need available in the DHS surveys. However, several studies point to its limitations and have proposed new measurements that consider women’s fertility intentions and agency in different ways [62,63,64]. Future research examining how public policies affect different measures of women’s unmet need may provide further insights into the mechanism affected.

Another potential limitation is the presence of negative weights in the Two-Way Fixed Effects (TWFE) model [65]. However, the robustness checks we conducted suggest these negative weights do not have a significant impact on the overall findings. Additionally, the country-specific analyses avoid potential biases that can arise when previously treated countries are included as comparisons because we only used one treatment country at a time to estimate these impacts [66].

Conclusion

IPV and barriers to women’s reproductive autonomy are closely linked. This study demonstrates that domestic violence legislation matters to reducing unmet need for contraception. While fully addressing both IPV and unmet need for contraception will require a wide range of legal, policy, and community interventions, laws addressing domestic violence represent one fundamental step that all countries can take. The potential benefits for women’s health, autonomy, and opportunities—and the intergenerational and societal harms that will result if the current scale of IPV and unmet need for contraception persist—underscore the urgency of action.