Ten years on: a comparative study of women-led civic participation and sustainable sanitation in rural India

Despite the wide use of community-led sanitation programs in low- and middle-income countries, there is still limited evidence on whether behavioral gains remain embedded over long periods and what helps sustain them. This study examined the ten-year legacy of a women-led sanitation champion program in rural Haryana, India. It asked whether outcomes differed between an Intervention Village and a neighboring Comparison Village approximately ten years after sanitation-specific activities ended at the study site, and whether long-term outcomes were better explained by individual sanitation engagement, household decision agency, social norms, or civic participation. Using a comparative case study design with a convergent mixed-methods approach, 131 women were surveyed across two villages, both of which had achieved open defecation-free (ODF) status. Qualitative data were gathered through focus group discussions and key informant interviews. Village-level differences were estimated using covariance-adjusted models, and pathway associations were examined using multivariable regression. The Intervention Village showed large, significant advantages in menstrual waste management, civic action, and community participation in sanitation activities. Children's hygiene did not differ significantly between villages after covariate adjustment. Civic participation was the only significant predictor of menstrual waste management in the multivariable models, suggesting that community-level processes were more strongly associated with this outcome than individual engagement. No predictor domain significantly explained children's hygiene or toilet use and maintenance in the multivariable models. Qualitative findings help explain this pattern. In the Intervention Village, sanitation had become part of a wider social system, supported by women's self-help groups, continuing Panchayat engagement, and active reinforcement of sanitation norms. In the Comparison Village, sanitation was more individually managed and more vulnerable to weak public systems. Women described barriers such as pit-filling costs, water scarcity, and older male resistance to toilet use, along with fewer opportunities for collective action or effective engagement with local governance. Long-term sanitation gains require more than just personal effort and individual behavioral change. They require a multi-pronged approach: strengthening women's roles in local governance, investing in physical infrastructure, and reinforcing the social norms that sustain healthy behaviors.

Community-led total sanitation (CLTS) and related participatory approaches have become a cornerstone of sanitation programming in low- and middle-income countries (Venkataramanan et al., 2018). This field of practice is grounded in the idea that mobilizing communities around collective norm change, rather than subsidizing infrastructure alone, is the pathway to durable open defecation-free environments (Aluoch et al., 2025; Kar and Chambers, 2008). Yet a growing body of evidence reveals a troubling gap between short-term triggering success and long-term behavioral persistence. Studies from Ethiopia, Ghana, Cambodia, and Malawi consistently demonstrate that sanitation gains diverge across communities within one to four years of intervention, with some communities maintaining improvements while others experience rapid slippage (Crocker et al., 2017; Kamwana et al., 2024; Kouassi et al., 2023; Tribbe et al., 2021). What remains largely unexplored is whether sanitation behaviors can become socially embedded as enduring community practices over a decade or more–and if so, through what mechanisms.

This study examines the ten-year legacy of a women-led sanitation champion intervention in two neighboring rural villages in Mahendragarh District, Haryana, India, conducted approximately 10 years after both villages were granted Open Defecation Free (ODF) status under the Swachh Bharat Mission (Sbm.gov.in, 2022). The intervention combined women's hands-on involvement in toilet construction and maintenance with a village-wide program that recruited women from each household as sanitation change champions, seeking to extend behavioral change beyond direct training exposure by fostering collective action, local norm reinforcement, and civic engagement (Bhavani et al., 2016; Sheshadri et al., 2015). By comparing outcomes in the Intervention Village against a demographically matched Comparison Village that received no such program, the study asks whether this form of women-led sanitation change leaves a detectable and differentiated community-level legacy at the ten-year mark.

The following background sections provide a brief overview of the key issues and literature informing this study: sanitation sustainability, community-level mechanisms, women's roles as change agents, menstrual waste management, and social norm diffusion. Together, these sections identify the research gaps that motivate the study objectives and research questions.

The sustainability challenge in community-led sanitation

A consistent finding across the sanitation literature is that initial behavioral gains are easier to achieve than to sustain. Studies spanning Ethiopia, Ghana, Cambodia, and Malawi demonstrate that sanitation outcomes diverge across communities within one to four years of intervention, with sustained gains associated not with triggering events alone but with configurations of active leadership, continued follow-up, collective support, women's engagement, and better-quality infrastructure (Crocker et al., 2017; Kamwana et al., 2024; Tribbe et al., 2021). Even where sanitation awareness remains high, it does not reliably translate into consistent practices such as latrine maintenance or handwashing (Kamwana et al., 2024). What remains largely unexplored is whether sanitation behaviors can persist as enduring community practices over a decade or more – and if so, through what mechanisms.

In this study, sustainability refers to the persistence of sanitation behaviors at the community level beyond the period of active intervention support–including continued latrine use, maintenance, and hygiene practices sustained through shared norms, social expectations, and community processes rather than individual adherence alone. While system-level dimensions such as infrastructure durability and governance also shape sustainability, the primary focus here is community-level behavioral persistence, recognizing that such behaviors are embedded within broader structural conditions.

Understanding how behaviors persist over extended time periods requires examining not just whether they endure, but through what pathways. Much WASH programming implicitly assumes a linear model in which training improves individual knowledge, attitudes, and practices that then self-sustain. Yet emerging evidence suggests that community-level processes – including collective efficacy, community participation, and normatively anchored expectations—may be equally or more important (Dreibelbis et al., 2013). Collective efficacy, defined as a community's shared belief in its capacity for collective action (Sampson et al., 1997), captures whether communities identify common goals, invest shared effort, and persist when collective efforts face setbacks (Hipp, 2016; Vassilev et al., 2019).

Community-level mechanisms

Delea (2019) operationalizes collective efficacy in the WASH context as a group's perceived ability to execute actions related to a common goal—a framing that maps directly onto the sustained, organized sanitation practices examined in this study. Sanitation is, in this sense, a collective good: the health benefits of eliminating open defecation depend on community-wide adoption, and individual household compliance is insufficient for lasting transformation (Dickin et al., 2017; Dery et al., 2020; Ostrom, 1990). Guiteras et al. (2015) found that CLTS effects were strongest in poorer communities, operating through collective sanitation environments rather than household characteristics alone—a pattern consistent with the collective efficacy framing. Yet despite this evidence, few studies empirically test whether collective efficacy and civic participation function independently of individual training exposure, or how they interact with individual-level pathways to produce differential outcomes across neighboring villages.

Women as sanitation change agents

Research on WASH increasingly recognizes women not only as household sanitation managers, but as potential drivers of broader community-level change. Women's everyday responsibilities in managing water, hygiene, and domestic sanitation position them at the front line of behavioral norm transmission, while their lived experiences of sanitation–shaped by concerns around dignity, safety, freedom of movement, and bodily vulnerability–can motivate sustained engagement in sanitation improvement (Caruso et al., 2017; Hirve et al., 2015; Sommer et al., 2015). The WASH-GEM framework argues that sanitation interventions can function as entry points for gender transformation, provided that women's participation moves beyond symbolic involvement to real voice and influence (Carrard et al., 2022), a point reinforced by evidence that sustained WASH outcomes are more likely when women shape norms, mobilize communities, and influence both household and collective practices (MacArthur et al., 2023).

In the Indian context, gendered power dynamics constrain this potential. Male household heads control financial and infrastructural sanitation decisions, while women bear responsibility for maintenance but are excluded from governance (Routray et al., 2017). Women may feel more confident in household sanitation decisions than in community-level leadership, with structural and normative barriers limiting broader participation even where willingness exists (Doma et al., 2023). Yet most studies examine women's empowerment at the individual or household level. Whether women-led sanitation interventions produce community-level norm transformations extending beyond direct participants, and whether the long-term effects operate primarily through individual training exposure or through broader village-level shifts in collective efficacy and institutionally embedded expectations, remains under-investigated (Dery et al., 2020; Gressel et al., 2020).

Menstrual waste management as a collective practice

Menstrual waste management represents a particularly underdeveloped dimension of sanitation research. Elledge et al. (2018) identified menstrual waste disposal as a marginalized dimension of WASH systems, noting that the convergence of sociocultural stigma, gendered norms, and infrastructural deficits produces environmental hazards and exacerbates psychosocial distress. A systematic review of 82 publications across 26 countries found that disposal and washing behaviors are driven by infrastructure availability, privacy concerns, shame, and fear, with menstruators engaging in secretive practices shaped by cultural expectations of concealment (Robinson and Barrington, 2021). Critically, the presence of improved sanitation facilities does not necessarily dictate usage; in rural Odisha, many women still prefer managing menstruation in private domestic spaces, suggesting that physical infrastructure alone is insufficient to shift practices deeply rooted in social and personal preferences (Behera et al., 2022).

Existing scholarship rarely conceptualizes menstrual waste management as a collectively negotiated social practice that is deeply integrated into broader sanitation systems. Limited attention has been paid to how disposal practices emerge, stabilize, and persist at the community level, or whether the pathways sustaining menstrual practices differ from those sustaining toilet use or children's hygiene. This gap matters for assessing women-led intervention legacies, because menstrual practices are among the most norm-sensitive and community-mediated dimensions of sanitation behavior.

Social norms, diffusion, and long-term behavioral embedding

Social norms—understood as socially conditioned preferences shaped by beliefs about what others do and what others consider appropriate—play a foundational role in sanitation behavior (Bicchieri, 2016; Gauri et al., 2023). In many South Asian and African contexts, inconsistent latrine use persists despite available facilities, attributed to prevailing norms tolerating or even preferring open defecation (Bicchieri, 2016; Coffey et al., 2015, 2016). Social Impact Theory (Latané, 1981) and related models suggest that as more community members practice and approve of a behavior, pressure on non-compliers intensifies. Through norm internalization, widely practiced behaviors become morally embedded, making non-compliance subject to both external sanctions such as gossip and disapproval, and internal ones such as guilt and shame (Bicchieri, 2016; Morris et al., 2015). Perceived descriptive norms have been identified as key predictors of toilet adoption and maintenance in Indian contexts (Novotný et al., 2017), and diffusion research suggests that behaviors spread as individuals observe and conform to the practices of credible others (Dearing, 2009).

This body of work suggests that sanitation behaviors may persist not because individuals were trained once, but because behaviors have become socially expected, collectively reinforced, and normatively anchored across the community. Evidence that configurations of leadership, collective support, and women's engagement distinguish sustained from eroded gains (Crocker et al., 2017; Tribbe et al., 2021) is consistent with this logic. Yet the pathways through which behaviors become institutionalized over extended time horizons, and whether these pathways operate differently across distinct sanitation outcome domains, remain empirically under-specified.

Study objectives and research questions

Taken together, the literature points to five interrelated gaps that this study addresses. First, evidence on sanitation sustainability beyond four years post-intervention is scarce, leaving open whether behaviors durably persist or gradually erode over decade-length horizons. Second, community-level mechanisms, particularly collective efficacy and civic participation, have been described conceptually but rarely tested as independent pathways alongside individual training exposure. Third, menstrual waste management continues to be treated as an individual hygiene practice rather than a collectively negotiated behavior, despite its norm-sensitive and community-mediated character. Fourth, the community-level legacy of women-led sanitation interventions remains under-theorized, with most evidence drawn from individual or household-level effects within shorter time frames. Fifth, comparative evidence between neighboring intervention and non-intervention communities over extended time periods is scarce, limiting understanding of how intervention effects diffuse or differentiate spatially.

This study addresses these gaps through a comparative case study design, examining an Intervention Village and a neighboring Comparison Village in Mahendragarh District, Haryana, approximately ten years after a women-led sanitation champion program and the granting of ODF status to both villages. This design enables theoretically grounded examination of whether observed differences reflect intervention history rather than regional trends, and supports explanations of sustainability mechanisms that are intended to be theoretically generative rather than statistically generalizable (Flyvbjerg, 2006; Seawright and Gerring, 2008). The study constitutes one site within a larger multi-site evaluation of collective efficacy ten years post-intervention.

Two linked research questions guide the analysis:

RQ1: Do sanitation-related outcomes differ between the intervention and Comparison Village at the ten-year mark?

RQ2: Which pathway domains – individual sanitation engagement, household sanitation decision agency, social norms, and civic participation, help explain long-term sanitation outcomes, and do these explanatory patterns differ across outcome domains.

Study objectives and hypotheses

This study addresses two primary research questions. RQ1 examines whether sanitation-related outcomes differ between an Intervention Village and a neighboring Comparison Village approximately ten years after a women-led sanitation champion program. These outcomes include children's hygiene practices, menstrual waste management, toilet use consistency, sanitation action at the household level, participation in sanitation activities at the community level, and civic action. RQ2 examines which pathway domains with respect to individual sanitation engagement, household decision agency, social norms, and civic participation – help explain long-term sanitation practices, and whether the explanatory profile differs across outcome domains.

The following hypotheses were specified prior to analysis and are organized by research question. Not all outcomes were linked to separate directional hypotheses. Toilet use consistency was included in the RQ1 analyses as an additional outcome indicator but was not tied to a separate a priori directional hypothesis.

RQ1: Village-Level Outcome Differences (H1a–H1c)

H1a. The Intervention Village is expected to demonstrate substantially better menstrual waste management and environmental care outcomes than the Comparison Village, reflecting a broader village-level legacy of women-led collective sanitation change that extends beyond direct training exposure.

H1b. The Intervention Village is expected to report higher community participation in sanitation activities, civic action, and sanitation engagement than the Comparison Village, reflecting the persistence of collective sanitation culture and women's public sanitation roles beyond the intervention period.

H1c. Children's hygiene practices—representing inter-generational transmission of sanitation behaviors—are expected to show attenuated or non-significant village-level differences, as these outcomes are more likely to have converged over time through regional sanitation trends, habituation, and household-level socialization processes common to both villages.

RQ2: Pathway Domains and Sanitation Outcomes (H2a–H2d)

H2a. Different sanitation outcomes are expected to persist through distinct long-term pathways rather than a single common mechanism, with the profile of significant predictor domains differing meaningfully across outcome types.

H2b. Children's hygiene practices are expected to be weakly explained by the measured pathway domains, as these outcomes are more likely sustained through habituation, inter-generational modeling, and routinized practice—mechanisms not adequately captured by cross-sectional self-report measures of engagement and participation.

H2c. Menstrual waste disposal and environmental care are expected to be more strongly associated with civic participation than with individual sanitation engagement or household-level agency, reflecting their character as socially embedded and collectively negotiated practices shaped by community-level normative and institutional context.

H2d. The bivariate association between individual sanitation engagement and menstrual environment outcomes is expected to attenuate—and potentially reverse—once civic participation is modeled simultaneously, as civic participation represents a more proximal and socially embedded mechanism for this outcome domain.

Table 1 summarizes all hypotheses, the specific prediction for each, and the theoretical rationale.

Study design

This study employed a comparative case study design with a convergent mixed-methods approach, examining the long-term sustainability of sanitation-related practices approximately ten years after a women-led sanitation intervention. The comparative case study framework was selected because the research questions required both comparison of measurable sanitation outcomes across two purposively selected village contexts and explanation of the social processes through which those outcomes were sustained or weakened over time (Yin, 2018). In convergent mixed-methods research, quantitative and qualitative strands are collected within the same study and integrated during interpretation to illuminate complementary dimensions of the same phenomenon (Creswell and Plano Clark, 2018).

This study constitutes one site within a larger, multi-site evaluation of collective efficacy ten years post-intervention. As a single-site comparative case study, findings are intended to be theoretically generative rather than statistically generalizable (Flyvbjerg, 2006). Comparisons between villages are interpreted as associative rather than causal, acknowledging that village-level confounding cannot be fully ruled out in a two-village design. The design logic rests on the selection of villages that are geographically proximate, share cultural and administrative context, and have a known difference in their intervention history.

Study setting

The study was conducted in Mahendragarh District, Haryana, India, across two rural villages: one Intervention Village and one Comparison Village, located in the same administrative block and embedded in similar caste and social structures. To protect participant confidentiality, identifying details including village names have been anonymized.

The intervention was delivered under the Women's Empowerment: Sanitation (WE: Sanitation) program (Bhavani et al., 2016; Sheshadri et al., 2015), a women's empowerment and community-based sanitation initiative implemented across 21 states in India with support in part from the United Nations Democracy Fund. The broader multi-state program operated from 2015 to 2018. Across participating states, program records indicate that 188 women completed toilet-construction and maintenance training, more than 300 household toilets were constructed, approximately 290 sanitation ambassadors were trained, and approximately 3,500 household change champions were engaged in participating communities.

At the study site, active implementation occurred from 2015 to 2016 and comprised two linked phases. The first was a vocational toilet-building phase in which women self-selected into a computerized vocational education and training course; eligibility was restricted to women without a household toilet who were willing to build toilets collectively for one another, and participants learned to construct and maintain their own facilities alongside a life-skills component addressing hygiene and upkeep. In the study village, approximately 11 women completed toilet-construction and maintenance training and collectively constructed approximately 27 household toilets. The second phase responded to the observation that training individual women to build toilets was insufficient to produce community-wide change and adopted a tiered mobilization model: interested women were trained as sanitation ambassadors to design and lead awareness activities, coordinate clean-up drives, and mobilize other women; ambassadors in turn recruited household change champions to carry sanitation information into their families and participate in community events; and the wider community engaged by attending these events. Approximately 20 sanitation ambassadors and 200 household change champions were engaged in the study village, representing about 10 percent of village households. The program comprised toilet construction and maintenance training, hygiene and life-skills education, household and community awareness activities, collective clean-up campaigns, menstrual health discussions, and general waste-management awareness; it did not include a dedicated municipal waste-collection service, menstrual-waste incinerators, formal menstrual-product distribution, or a formal Panchayat governance-strengthening component. In this way, the intervention sought to extend behavioral change beyond the direct training cohort by fostering collective action, local norm reinforcement, and civic engagement around sanitation and hygiene.

Following completion of the sanitation-specific intervention at the study site in 2016, no formal sanitation refresher program, dedicated sanitation-monitoring system, infrastructure-maintenance arrangement, or sanitation-specific external funding was maintained by the implementing organization. Qualitative accounts indicated that self-help group (SHG) members active during and following the intervention continued to meet and undertook some sanitation-related collective activities on their own initiative. A locally recruited facilitator remained in contact with the implementing organization; however, subsequent organizational engagement focused primarily on unrelated vocational activities and, later, on facilitating access to government welfare schemes (such as Aadhaar enrolment and ration-card updating) rather than on sanitation delivery. No formal Panchayat-led monitoring or sanitation-maintenance mechanism was identified after the intervention, although participants described approaching the Panchayat on an ad hoc basis regarding waste and sanitation issues. No additional sanitation-focused NGO or community-based organization program was identified during the follow-up period. The study therefore examines outcomes approximately ten years after sanitation-specific activities ended at the study site, in the absence of continuous sanitation-specific external delivery, while acknowledging that limited contact with the implementing organization continued.

Both villages had achieved ODF Plus Model status at the time of data collection, reflecting a certification standard that encompasses solid and liquid waste management arrangements, visual cleanliness, and ODF Plus information, education, and communication messaging (Department of Drinking Water and Sanitation, Ministry of Jal Shakti, Government of India, 2026). This administrative designation was treated as a program-status indicator rather than as independent verification that open defecation had been eliminated. This shared status was a deliberate design feature, allowing analysis to focus on the quality and social embedding of sanitation-related behaviors rather than differences in toilet access alone–the more theoretically relevant question at a ten-year horizon.

Sampling and participant selection

Sampling in both villages was conducted using stratified random sampling to improve demographic comparability while preserving the intervention/comparison logic. The sample was restricted to female participants to capture the women-led community sanitation perspective. In each village, strata were constructed based on caste and community composition, and respondents were then randomly selected within strata. This approach was used to ensure that the comparison reflected similar social composition across the two village contexts while allowing assessment of differences associated with intervention history.

Preliminary equivalence testing confirmed that the resulting samples did not differ significantly on community category (p = 0.149) or marital status (p = 0.090). Where significant between-village differences were identified—on age, education, earning status, and bank account ownership—these were adjusted for in outcome analyses. Households were sampled from the village household lists maintained by the respective Gram Panchayats, which served as the sampling frame in each village. Of 140 households approached, 131 women consented and completed the survey across the two villages, yielding a response rate of 93.6%. Item-level missingness varied across outcome variables, ranging from 3.1% for civic action and sanitation action to 22.1% for menstrual waste management, reflecting higher non-response rates on sensitive items relating to menstrual practices. Intermediate missingness was observed for children's hygiene (16.8%) and toilet use consistency (11.5%), with community participation showing low missingness (3.8%). Cases with missing values on focal covariates or outcomes were excluded listwise from the ANCOVA and regression models, with the analytic sample size for each model reported in Table 2. Data collection was conducted between January and February 2026, with survey administration taking approximately 45 minutes per respondent.

For the qualitative component, purposive sampling captured variation in age, household role, sanitation experience, and community participation, including women with and without direct intervention exposure, household heads, younger women, and community leaders (Creswell and Plano Clark, 2018). Key informant interviews were conducted with former and current Sarpanches (village leaders) and SHG leaders. The total quantitative sample comprised 131 respondents.

Data collection

Prior to main data collection, the survey instrument was piloted with a sample of female respondents sharing a similar demographic profile to the target villages. Each question was appraised by trained field investigators across four dimensions: clarity of the question as administered, difficulty experienced by the respondent in answering, respondent comfort level, and overall need for revision before main data collection. This appraisal combined structured field ratings with qualitative observation of respondent behavior during administration, enabling systematic identification of questions that required rephrasing, reordering, or contextual adaptation.

The quantitative component consisted of a structured household survey administered in an interviewer-administered format to improve comprehension and standardize administration across participants with variable literacy (Groves et al., 2011). Data collection was carried out by trained surveyors, fluent in Hindi (the respondents' local language) with no prior association with the intervention program to minimize social desirability bias (Grimm, 2010). Respondent households were mapped using Google My Maps (Google, 2026) to verify field coverage.

The qualitative component drew primarily on focus group discussions (FGDs) and key informant interviews (KIIs), supplemented by contemporaneous field notes, survey recordings, and daily fieldwork debrief reports. These secondary materials were used to contextualize how sanitation-related practices were maintained, normalized, negotiated, or weakened over time (Hennink, 2013). Qualitative interactions were conducted in Hindi, audio-recorded with consent, transcribed verbatim, and translated into English for analysis; where analytically relevant, excerpts are presented in the text with the original transliterated Hindi followed by the English translation.

Given the sensitivity of menstrual topics, female field investigators administered the menstrual waste management questions in the Comparison Village to support respondent comfort and disclosure. The menstrual waste management item concerned the respondent's own current or most recent disposal practice; post-menopausal or non-menstruating respondents were recorded as not applicable rather than as missing where this could be ascertained.

Outcome domains

The outcomes selected for investigation represent stages of the life course as opposed to a single unified measure of sanitation (Dreibelbis et al., 2013). For RQ1, five outcomes were examined: (1) children's hygiene practices, representing inter-generational transmission; (2) menstrual waste management, representing middle-late stage of women's socially embedded disposal practices; (3) toilet use and maintenance, representing sustained defecation practice and facility upkeep; (4) community participation in sanitation activities; and (5) civic action around sanitation and environmental hygiene. For RQ2, pathway analyses focused on the first three behavioral outcomes.

The menstrual waste management outcome was treated with particular importance. The outcome was conceptualized as a matter of both personal hygiene, and environmental and public health implications. It was operationalized as a five-point ordinal scale reflecting this focus. The range comprised open disposal (1-highest environmental externality: uncontained dispersal and water pollution), burning (2-toxic air emissions), burial (3-soil and groundwater contamination), disposal of plastic-containing products in general solid waste (4- long-term plastic degradation burden), cloth reuse (5-lowest solid-waste burden). The scoring captures the relative environmental externality of each practice, consistent with the study's framing of menstrual waste management as a collectively consequential sanitation outcome.

It should be noted that this scale is a pragmatic environmental-externality index rather than a validated uni-dimensional scale and does not constitute a health-risk hierarchy. Cloth reuse, while ranked highest on environmental grounds, may carry infection risk where washing and drying conditions are inadequate—a tension acknowledged in the menstrual health literature (Sumpter and Torondel, 2013; Torondel et al., 2018). Although the scale is ordinal, treating ordered scales with five or more categories as approximately continuous is an accepted practice in health and social science research that typically produces results comparable to ordinal-specific methods (Norman, 2010). The scale is therefore analyzed as approximately continuous in ANCOVA and regression models. To confirm that this specification does not alter the substantive conclusions, ordinal logistic regression sensitivity analyses were estimated for the village-effect model and the mutually adjusted pathway model; these sensitivity analyses are presented in the Supplementary Table. Because ordinal logistic regression produces coefficients on the log-odds scale, these are not directly comparable in magnitude to the linear-model coefficients reported in the primary analyses; interpretation focuses on direction, statistical significance, and consistency of conclusions across specifications.

Qualitative interviews and field records indicated that disposable commercial sanitary pads were the most reported menstrual material in both villages, while reusable cloth was reported less frequently and, where used, handled more discreetly, consistent with stigma surrounding menstruation. Product use was not quantified in the household survey, and menstrual cups and commercially produced reusable pads were not reported. No menstrual-specific collection service, incinerator, dedicated disposal facility, product-distribution program, or continuing community-level menstrual hygiene awareness program was identified in either village. A general solid-waste collection vehicle served some locations, but access was inconsistent and it was not designed specifically for menstrual waste; some Intervention Village residents reported occasionally using it, whereas participants in the Comparison Village generally described no accessible collection option. Retail availability and affordability of menstrual products were not systematically assessed. The limited availability of shared disposal systems meant that menstrual waste management remained largely individualized unless collective structures organized an alternative.

Predictor domains and measurement

RQ2 explored four predictor domains: (1) individual sanitation engagement, (2) household sanitation decision agency, (3) social norms, and (4) civic participation. These domains were grounded in established frameworks of WASH behavior change, including the RANAS model (Mosler, 2012) and the Theory of Planned Behavior (Ajzen, 1991), and align with pathways identified in the sanitation sustainability literature (Crocker et al., 2017; Chambers and Myers, 2016).

Predictor domains were operationalized as theoretically derived domain indices using a formative measurement approach, in which items represent conceptually related but empirically distinct components of a broader pathway (Bollen and Bauldry, 2011). High internal consistency was therefore not treated as a validity criterion (Diamantopoulos and Winklhofer, 2001).

Quantitative analysis

All quantitative analyses were conducted in R (R Core Team, 2023). Descriptive statistics were generated for all study variables, and village-level demographic differences were examined prior to outcome analysis.

For RQ1, adjusted village comparisons were estimated using analysis of covariance (ANCOVA) models for each outcome (Maxwell et al., 2017). Village type was entered as the focal predictor, and age, education, earning status, and bank account ownership were included as covariates because preliminary analyses indicated between-village imbalance on these characteristics. Homogeneity of regression slopes was assessed for each outcome model by entering covariate-by-village interaction terms (age and village, earning status and village, bank account ownership and village, and education and village where applicable) and testing their joint significance using Type III sums of squares.

Education was included as a main effect only, as sparse education-by-village cells precluded reliable interaction estimation. For hygiene, menstrual waste management, and community participation, no interaction term reached conventional significance (all p >0.05), indicating that the assumption of parallel regression slopes was satisfied. For civic action and sanitation action, a significant interaction between age and village was observed (p = 0.039); however, the village effect for these outcomes remained large and highly significant regardless (civic action: β = 4.205, p < 0.001; sanitation action: β = 0.320, p < 0.001), and this mild violation was not considered to substantively alter the conclusions.

Normality of model residuals was assessed by inspection of Q–Q plots and the Shapiro–Wilk test, and homoscedasticity was assessed using Levene's test; where minor departures from these assumptions were observed, heteroscedasticity-consistent (HC3) standard errors were applied as a robustness check and produced substantively identical conclusions. Unadjusted means and standard deviations are reported alongside covariate-adjusted estimated marginal means and 95% confidence intervals. Effect sizes are reported as partial η 2 and Cohen's d where appropriate (Cohen, 1988).

For RQ2, associations between the four predictor domains and the three sanitation outcomes were examined using regression models. First, single-predictor models were estimated to characterize the bivariate association between each predictor domain and each outcome. Second, a multivariable model including all four predictor domains simultaneously was estimated for each outcome to examine their independent contributions. Regression coefficients, standard errors, p-values, and R 2 values are reported. Multi-co-linearity was assessed using variance inflation factors (O'brien, 2007). For the menstrual waste management outcome, which is measured on a five-point ordinal scale, ordinal regression models were estimated as sensitivity analyses alongside the primary ANCOVA and linear regression models. Statistical significance was assessed at p < 0.05.

Qualitative analysis

Qualitative data were analyzed using the Framework Method (Gale et al., 2013), selected for its suitability for applied, multidisciplinary health research requiring structured comparison across cases, themes, and data sources. An initial analytic framework was developed from the study questions and pathway domains–individual engagement, household agency, social norms, and civic participation–alongside sanitation sustainability themes identified during fieldwork, including normalization of toilet use, menstrual waste management, inter-generational learning, and institutional support. Analysis combined deductive and inductive coding (Fereday and Muir-Cochrane, 2006), with codes grouped into higher-order categories and charted into matrices organized by village type, participant type, and theme, enabling within-case and cross-case comparison (Ritchie et al., 2013).

Daily fieldwork debrief reports, written by field investigators at the end of each day of data collection, were used as a secondary qualitative source to help interpret and support patterns identified in the FGD and KII data. Observations derived from daily fieldwork debrief reports are distinguished from participant-generated FGD and KII data throughout the Results and Discussion sections. Where findings draw on debrief records, this is indicated inline with the phrase ‘field debrief records indicate' or ‘field debrief records suggest,' so that readers can assess the evidential basis of each claim accordingly.

Integration

Integration occurred at the interpretation stage following a convergent mixed-methods strategy (Creswell and Plano Clark, 2018). Quantitative analyses identified whether outcomes differed between villages and which pathway domains were associated with each outcome. Qualitative findings were then used to explain the mechanisms underlying those patterns (Fetters et al., 2013). Where quantitative models showed limited explanatory power, qualitative data were used to identify unmeasured mechanisms and contextual factors that help account for those patterns. Divergence between quantitative and qualitative findings was treated as analytically informative rather than as a validity threat.

Justification of the comparative case study design

Three features support the interpretive validity of the two-village comparison. First, the villages are geographically proximate, share cultural and administrative context, have a known difference in intervention history, reducing plausible alternative explanations for observed differences–consistent with the logic of comparative case selection in implementation research (Seawright and Gerring, 2008; Yin, 2018). Second, equivalence testing confirmed that key demographic characteristics were either non-significantly different or explicitly adjusted for outcome analyses, as reported at the beginning of the Results section. Third, the pattern of findings is more consistent with an intervention-specific mechanism than with undifferentiated village-level difference: large effects emerged for collectively organized outcomes–menstrual waste management, civic action, and community participation–while individual hygiene showed no adjusted village difference. This differentiated pattern strengthens the comparison by showing that the village differences were not uniform across all outcomes, but varied in meaningful ways across domains. As one site within a larger multi-site evaluation, findings are intended to be theoretically generative; replication across additional sites will test whether patterns are consistent across contexts.

Ethical considerations

Ethical approval was obtained from the authors' Institutional Human Ethics Committee (IHEC) prior to data collection. All participants provided informed consent prior to participation in accordance with the Declaration of Helsinki (Association et al., 2013). Participation was voluntary, and respondents were free to withdraw at any time without consequence. Confidentiality was maintained through anonymized coding of participants.

Before examining village differences in sanitation outcomes, we assessed demographic balance between the Intervention Village (n = 65) and the Comparison Village (n = 66). The study population in the Intervention Village was significantly older (mean 43.89 vs. 37.62 years, p = 0.004), had lower levels of higher education (6.2% vs. 35.3%, p = 0.005), and had a higher proportion of participants actively earning (43.1% vs. 24.6%, p = 0.041) than the Comparison Village, whereas community composition and marital status were broadly similar (p > 0.05). The villages differed significantly in age, education, earning status, and bank account ownership (all p < 0.05; standardized mean differences = 0.40–0.82); adjusted village comparisons were therefore estimated using ANCOVA models controlling for these characteristics. Full results are presented in Table 2.

Menstrual waste management (H1a)

The largest and most robust village difference was observed for menstrual waste management. After demographic adjustment, respondents in the Intervention Village reported substantially higher menstrual waste management scores than those in the Comparison Village (β = 1.862, SE = 0.303, p < 0.001, partial η 2 = 0.300), with covariate-adjusted estimated marginal means of 3.72 (95% CI [3.01, 4.43]) and 1.90 (95% CI [1.39, 2.42]) for the Intervention and Comparison Villages respectively. Village membership explained 30% of the residual variance in menstrual waste management outcomes, indicating a large effect. Demographic adjustment did not attenuate this difference—the adjusted village contrast (β = 1.862) was slightly larger than the unadjusted mean gap (3.26 − 1.71 = 1.55 scale points), indicating that demographic differences between villages were not driving the intervention-village advantage and that adjustment for age, education, earning status, and bank account ownership marginally sharpened rather than reduced the estimated effect. The 95% confidence interval for the village contrast was [1.268, 2.456], excluding zero. An ordinal-regression sensitivity analysis treating menstrual waste management as a five-category ordered response produced substantively identical conclusions, with a comparably large and highly significant village effect (Supplementary Table, Panel A). H1a was strongly supported.

Qualitative findings help explain this divergence. In the Intervention Village, women described menstrual waste disposal as increasingly organized, with waste collected by truck or burned collectively at the village outskirts. Toilet use was also described as normalized and morally valued; as one respondent stated, “Everyone uses it; no one goes outside” (“ sab karein hain, baahar koi naa jaye”). In the Comparison Village, by contrast, menstrual waste was described as being routinely discarded in open areas or forested spaces in the absence of collection systems. One respondent explained, “I dump [the menstrual waste] in the forest” (“ [periods ke kachre] bani mein … jangal”), while open defecation persisted as a socially embedded routine, as another respondent noted, “it has been 30 years of defecating like this” (“ tees saal ho gayi, latrine karta”). Toilet access alone did not determine disposal practices; village-level normative context played a more important role.

Community participation, sanitation action, and civic action (H1b)

Large and significant village differences were observed for community participation (β = 1.125, SE = 0.210, p < 0.001, partial η 2 = 0.204) and sanitation action (β = 0.320, SE = 0.057, p < 0.001, partial η 2 = 0.219). The near-floor mean for sanitation action in the Comparison Village (M = 0.06, SD = 0.17) indicates that collective sanitation action was effectively absent there, in contrast to the Intervention Village (M = 0.40, SD = 0.33). Civic action also showed a significant village difference in the predicted direction (β = 4.205, SE = 2.042, p = 0.042, partial η 2 = 0.039), although the effect was smaller and showed greater variability in the Comparison Village. The 95% confidence intervals for the adjusted village contrasts were [0.713, 1.537] for community participation, [0.208, 0.432] for sanitation action, and [0.203, 8.207] for civic action; all three intervals excluded zero. H1b was therefore supported across all three collective engagement outcomes.

Qualitative findings illustrate the mechanisms through which these differences were sustained. Respondents from the Intervention Village described involvement in, “ gully cleaning… waste bin placement… collective sanitation activities” and reported that they “approach Panchayat … raise demands … apply collective pressure” – indicating that sanitation-related concerns were addressed through institutional channels and collective negotiation, rather than only through individual household practice. In the Comparison Village, comparable forms of organized collective sanitation action and engagement with local governance were largely absent from respondent accounts.

Children's hygiene practices (H1c)

The adjusted village difference for children's hygiene practices approached but did not reach conventional significance (β = 0.051, SE = 0.028, p = 0.069, partial η 2 = 0.034), with covariate-adjusted estimated marginal means of 0.79 (95% CI [0.75, 0.84]) and 0.74 (95% CI [0.70, 0.78]) for the Intervention and Comparison Villages respectively. The effect was small and in the predicted direction, with the Intervention Village scoring higher than the Comparison Village; the 95% confidence interval for the village contrast was [−0.004, 0.106], just crossing zero. This pattern supports the hypothesis that differences between villages are less pronounced for children's hygiene than for outcomes organized at the collective level. The result was consistent with H1c, although the analysis was not designed as a formal equivalence test and cannot establish that the villages had converged; the small, non-significant adjusted difference is suggestive of partial convergence rather than a strong, enduring intervention legacy for this outcome.

Toilet use consistency

The village difference for toilet use consistency did not reach significance after adjustment (β = 0.115, SE = 0.067, p = 0.090, partial η 2 = 0.028), with covariate-adjusted estimated marginal means of 4.08 (95% CI [3.98, 4.18]) and 3.96 (95% CI [3.87, 4.06]) for the Intervention and Comparison Villages respectively; the 95% confidence interval for the village contrast was [−0.016, 0.246]. However, interpretation of this finding is limited by a ceiling effect in the Intervention Village, where all respondents reported maximum consistent toilet use (M = 4.00, SD = 0.00), eliminating the respondent-level variance on which a village contrast depends. This result should therefore not be interpreted as evidence of equivalence between villages, but as analytically constrained by lack of variance in the intervention setting.

Taken together, the RQ1 findings reveal a three-cluster pattern of village differences 10 years post-intervention. Outcomes most closely tied to collective processes and social embedding—menstrual waste management, community participation, and sanitation action—showed large and highly significant village effects (partial η 2 = 0.204–0.300), with adjusted marginal mean differences ranging from 0.32 scale points (sanitation action) to 1.82 scale points (menstrual waste management). Children's hygiene showed a small directional effect consistent with convergence (partial η 2 = 0.034, p = 0.069), while toilet use consistency was limited by a ceiling-constrained measurement pattern that precluded meaningful inference. This differentiated pattern supports the broader proposition that the long-term legacy of women-led sanitation change operates most durably through collectively organized and normatively embedded processes rather than through individual behavioral outcomes alone.

RQ2: predictor domains and long-term sanitation outcomes

To address RQ2, associations between four theoretically derived predictor domains—individual sanitation engagement, household decision agency, social norms, and civic participation—and three outcome composites were examined: children's hygiene (inter-generational transmission), menstrual waste disposal and environmental care, and toilet use and maintenance. Results from the final multivariable model, with all four predictors entered simultaneously, are presented in Table 3.

Results revealed a strongly differentiated pattern of pathway associations across outcome domains, consistent with H2a. No single predictor explained all three outcomes, and the profile of significant associations differed substantially by outcome domain.

Children's hygiene (H2b)

The full model explained negligible variance in children's hygiene (adjusted R 2 = 0.026) and no predictor reached significance (individual engagement: β = 0.151, 95% CI [−0.165, 0.467], p = 0.352; household agency: β = −0.022, 95% CI [−0.095, 0.051], p = 0.550; social norms: β = 0.045, 95% CI [−0.041, 0.131], p = 0.300; civic participation: β = −0.104, 95% CI [−0.480, 0.272], p = 0.589). This pattern is consistent with the hypothesis that inter-generational hygiene transmission is sustained through habituation, modeling, and routinized practice—mechanisms not adequately captured by cross-sectional self-report measures of engagement and participation. In the Intervention Village, hygiene had become self-regulating; the Village Head described it as “ once the habit is formed, there is no need to explain again” —suggesting that engagement pathways had already been absorbed into routinized practice beyond the reach of the measured constructs.

Menstrual waste management (H2c, H2d)

The full model explained 11.3% of variance in menstrual waste management outcomes (adjusted R 2 = 0.113), the strongest explanatory model across the three outcomes. Civic participation was the sole significant predictor (β = 2.758, SE = 1.296, 95% CI [0.218, 5.298], p = 0.036), with a large positive coefficient indicating that women with greater civic participation reported substantially better menstrual disposal and environmental care practices. The remaining predictors were not significant (individual engagement: β = −1.240, 95% CI [−3.445, 0.965], p = 0.274; household agency: β = −0.379, 95% CI [−0.932, 0.174], p = 0.185; social norms: β = 0.353, 95% CI [−0.319, 1.025], p = 0.311). The ordinal-regression sensitivity analysis produced the same pattern of associations, with civic participation remaining the only significant pathway predictor after mutual adjustment (Supplementary Table, Panel B). H2c was supported: menstrual waste management was more strongly associated with civic participation than with individual sanitation engagement, household agency, or social norms.

Individual sanitation engagement, which showed a significant positive bivariate association with menstrual outcomes (β = 1.09, p = 0.015, 95% CI: [0.218, 1.962]), reversed sign in the full model (β = −1.240, p = 0.274). H2d was supported: this suppression pattern—in which the bivariate association between individual engagement and menstrual outcomes attenuated and reversed, once civic participation was modeled simultaneously. This is consistent with shared variance among the predictor domains, in which the bivariate individual-engagement association overlaps with civic participation; the cross-sectional design cannot, however, establish mediation or causal ordering.

Although collective burning represented a more organized practice than uncontrolled open dumping, it remained environmentally undesirable and was assigned a relatively low score on the environmental-externality index. Qualitative evidence contextualizes this pattern. Women in the Intervention Village who described the highest civic participation also described the most organized menstrual disposal practices–collectively managed burning, use of collection vehicles, and shared accountability for environmental cleanliness. In the Comparison Village, the absence of collective structures left disposal as an individualized and externalized practice, normalized through spatial displacement in the absence of any waste management infrastructure.

Toilet use and maintenance

The full model explained 5.7% of variance in toilet use and maintenance (adjusted R 2 = 0.057) and no predictor reached significance (individual engagement: β = −0.050, 95% CI [−0.203, 0.103], p = 0.523; household agency: β = −0.012, 95% CI [−0.047, 0.023], p = 0.498; social norms: β = 0.026, 95% CI [−0.015, 0.067], p = 0.223; civic participation: β = −0.032, 95% CI [−0.216, 0.152], p = 0.734). Bivariate associations observed for individual engagement and civic participation did not persist after mutual adjustment, indicating confounding rather than independent effects.

Qualitative evidence from the Comparison Village attributes toilet disuse not to disengagement but to structural barriers—pit-filling costs, water scarcity, and elder male resistance—that no measured pathway domain adequately captured. Together, the RQ2 findings demonstrate the pathway heterogeneity predicted by H2a: different sanitation outcomes are sustained through different mechanisms, with civic participation the only pathway domain that retained a statistically significant association with menstrual waste management after mutual adjustment.

This study analyzed the ten-year legacy of a women-led sanitation intervention by comparing outcomes between an Intervention Village and a matched Comparison Village in rural India. Results are interpreted in relation to the six pre-specified hypotheses and to existing literature on WASH sustainability, collective efficacy, and women's agency.

Collective organizations and the persistence of sanitation practices

The primary finding is not that the Intervention Village outperformed the Comparison Village on all outcomes, but that it did so specifically on the collectively organized ones (H1a, H1b): menstrual waste management, civic action, community participation, and sanitation action. In the Intervention Village, sanitation operated as part of a normative and institutional fabric sustained through women's civic engagement, SHG-organized collective action, and Panchayat-level accountability; in the Comparison Village, it remained domestically contained and institutionally tenuous.

This pattern maps onto the RANAS framework (Mosler, 2012), in which durable behavior change requires sustained activation of norms, perceived ability, and self-regulation rather than awareness alone. The Intervention Village showed stronger community-level norms, perceived collective ability, and self-regulatory capacity alongside individual attitude shifts—consistent with literature framing collective efficacy as a pathway through which intervention exposure links to sustained outcomes (Bandura and Wessels, 1997; Carrard et al., 2022), and with evidence that women's participation in WASH strengthens collective action beyond the intervention period.

The modest explanatory power of the RQ2 models (R 2 = 0.026 to 0.113) may itself be theoretically informative rather than solely a measurement limitation. A decade after the intervention, behaviors initially supported through active engagement and collective reinforcement may have transitioned into habituated routines no longer meaningfully linked to the constructs through which they were originally produced—a process for which cross-sectional self-report measures are poorly suited. The ceiling effect for toilet use consistency in the Intervention Village (M = 4.00, SD = 0.00) is consistent with this reading: complete normalization eliminates the respondent-level variance on which predictor-outcome associations depend. Future designs addressing sustainability over extended horizons should consider longitudinal approaches capable of tracking this transition from motivated to habituated behavior.

Qualitative evidence provides critical insight into how this embedding occurred. In the Intervention Village, sanitation practices were not only internalized at the household level but actively produced and maintained through collective processes. Women described ongoing engagement through self-help groups (SHGs), including organizing meetings, conducting awareness activities, monitoring practices, and interacting with local governance institutions. As one SHG leader explained, “we have to talk to them, show them videos, tell them that this will cause pollution,” indicating that behavioral change was continuously reinforced through collective communication and persuasion. The Intervention Village Sarpanch noted that “earlier women did not come to meetings… now 90–95% participation is of women… they have started speaking,” reflecting a transformation from passive participation to active civic engagement. Over time, these actively maintained practices appear to have transitioned into self-regulating routines, as the Intervention Village Head observed: “once the habit is formed, there is no need to explain again.” This trajectory—from deliberate collective reinforcement to habituated community norm—is further illustrated by a KII with the SHG leader:

For instance, if someone is burning plastic waste, we approach them individually and tell them, “We will come and speak to you about this tomorrow.” We just tell them once or twice verbally. If they still don't listen, we show them a video explaining the consequences—how burning plastic can cause respiratory diseases or even lead to cancer. By showing such videos repeatedly, four or five times, we try to make them understand and realize that what they are doing is actually wrong.

These accounts suggest that sanitation norms were not passively adopted but actively constructed through women-led civic processes, and that this active construction eventually gave way to self-sustaining collective regulation. In the Comparison Village, by contrast, such collective structures were largely absent. Respondents described the absence of organized groups, limited participation, and an inability to act collectively: “I cannot be the only one to raise an objection” (“ Akeli toh main objection utha nahi sakti”), indicating weak civic infrastructure and constrained individual efficacy.

The role of self-help groups (SHGs) merits closer attention. SHGs were considerably more active in the Intervention Village than in the Comparison Village, consistent with an institutional legacy of the program, though pre-existing organizational differences cannot be fully excluded. In this reading, SHG activity may have functioned as a mediating pathway through which the intervention's effects on community participation, norm reinforcement, and menstrual waste management were sustained over time. Field debrief accounts also indicate that safety concerns, including harassment of women defecating in open areas, were cited as early motivators for toilet adoption, suggesting that dignity and physical safety operated alongside collective norm processes rather than being fully reducible to them. Taken together, the findings are consistent with collective organizational capacity being associated with whether sanitation gains persist over decade-long horizons, though confirmation in larger multi-site samples would be required before stronger inferential claims can be made.

Education, household agency, and the limits of individual-level explanations

The qualitative evidence reveals an important distinction between household-level agency and the more public, institutionally supported forms of agency that were visible in the Intervention Village. Women in both villages described themselves as responsible for domestic sanitation decisions. The difference lay in whether that agency extended into collective action, Panchayat engagement, and norm enforcement—a finding consistent with a growing literature noting the gap between household-level decision-making and community-level influence in WASH (Gressel et al., 2020; Routray et al., 2017; Doma et al., 2023).

While prior research has suggested that women's agency is often stronger within the household than in community-level governance spaces (Doma et al., 2023), the present findings complicate this distinction. The Comparison Village exhibited higher levels of formal education yet weaker collectively organized sanitation outcomes, challenging the assumption that increased knowledge and awareness lead to improved sanitation practices. Sanitation work in the Comparison Village was described primarily as routine household labor—“either sister-in-law does it, or I do the household cleaning”—with community-level concerns mediated through male household members: “I tell my husband to tell the Sarpanch.” Women's agency was further limited by restricted economic roles, limited awareness of legal rights, and caste-based divisions that inhibited collective organization. Field debrief records further suggested that several women ward heads functioned as proxy representatives, with male relatives attending Gram Sabha meetings on their behalf. A law graduate from within the Comparison Village itself shared her observations as follows: “villagers are not that aware about their rights… without such awareness, they do not question institutions or mobilize for change.” That this assessment came from a member of the community with legal training lends weight: it reflects an insider account of the structural and informational constraints that limit collective mobilization in that setting.

In the Intervention Village, by contrast, women not only participated in intervention activities but continued to act as community mobilizers, educators, and monitors long after the intervention period, despite comparatively lower levels of formal education. As one respondent described, “SHG members talk to others, show videos, and explain the harms,” indicating ongoing involvement in norm reinforcement. This pattern—stronger collective outcomes at lower education levels—suggests that the translation of household agency into public and collective forms is shaped less by individual-level factors such as education than by the presence of enabling institutional structures, collective platforms, and sustained community mobilization (Dreibelbis et al., 2013). In the context of sanitation, where behaviors are inherently collective and interdependent, community-level processes may be more strongly associated with long-term sustainability than individual-level cognitive factors.

A further interpretive consideration concerns the higher earning rate in the Intervention Village (43.1% vs. 24.6%, p = 0.041). Economic autonomy is a plausible facilitator of civic participation (Guiteras et al., 2015), raising the possibility that the civic participation pathway for MWM is partially confounded by this difference. Three checks speak against this. First, earning status was not significant when added as a covariate to the RQ2 pathway model, and the civic participation coefficient remained the sole significant predictor (β = 2.758, p = 0.036). Second, the ordinal sensitivity analyses (Supplementary Table, Panel A) show that earning status does not independently account for the village MWM difference once village assignment is in the model (β = 0.758, 95% CI [−0.097, 1.629], p = 0.084). Third, earning status was already a covariate in all RQ1 ANCOVA models, so the reported village effects reflect adjustment for this difference. The observed associations are therefore not reducible to between-village differences in economic autonomy, though cross-sectional inference limits causal attribution (Section 4.4).

The role of public systems in sustaining sanitation

The qualitative evidence from the Comparison Village underscores that household motivation and individual awareness are insufficient to sustain sanitation practices when public systems fail. Toilet reversion occurred when septic pits filled and could not be affordably emptied; hygiene practices deteriorated when waste collection vehicles ceased to operate; and menstrual disposal remained externalized in the absence of collection infrastructure. These findings are consistent with evidence from Kamwana et al. (2024) that sanitation gains are vulnerable to infrastructural withdrawal and reinforce the argument that intervention sustainability cannot be separated from the ongoing functioning of local governance and service delivery systems. Respondents in the Comparison Village explicitly linked ODF reversion to water unavailability, reinforcing that behavioral gains remain contingent on reliable infrastructure functioning.

ODF certification and behavioral reality: a gap with policy implications

A related observation merits attention at the policy level. The Comparison Village had achieved ODF status under the Swachh Bharat Mission, yet qualitative evidence indicates that toilet use remained incomplete and situationally contingent a decade after certification – with respondents describing alternation between toilet use and open defecation depending on water availability and convenience. Field debrief records also suggested that some residents described open defecation as culturally and physically preferable to toilet use, associating it with fresh air, morning routine, and social interaction. These accounts indicate that long-standing habits and socially valued routines persisted despite the period of Swachh Bharat enforcement.

Debrief reports further noted that some respondents described enforcement as intense but short-lived, with portable facilities installed and later removed without explanation, after which earlier practices resumed. The absence of a statistically significant village difference on toilet use consistency should not obscure this finding: that open defecation persists as a practiced and socially normalized behavior in a formally certified ODF village, ten years after certification, raises important questions about the relationship between administrative status and behavioral reality in large-scale national sanitation programs. Swachh Bharat Mission certification appears to emphasize infrastructure access and community-level verification at a point in time more than the longer-term depth of behavioral embedding or the presence of the collective and institutional conditions through which toilet use is sustained over extended periods.

This gap between certified status and behavioral reality points to a structural limitation in how ODF certification is currently designed. Certification frameworks verify infrastructure access and open defecation-free conditions at a point in time; they were not designed to assess the depth of social embedding through which toilet use is sustained, nor the collective and institutional conditions that determine whether certified gains persist. Although ODF certification remains an important development milestone, follow-up assessment frameworks attentive to normative anchoring and collective norm maintenance may more reliably predict long-term behavioral persistence. This has implications beyond India: national sanitation programs that rely on point-in-time certification may underestimate the share of certified communities where behavioral gains remain fragile. Further research on the long-term behavioral trajectories of Swachh Bharat-certified villages—particularly those that did not receive complementary community mobilization programming—would clarify what large-scale policy interventions leave behind.

Limitations

Comparative case study design does not allow the observed village differences to be attributed causally to the intervention. Although the two villages are geographically close, share administrative and cultural context, and differ primarily in intervention history, unmeasured pre-existing differences cannot be fully excluded—including differences in Panchayat functioning, variation in the strength of self-help groups before the program began, and caste-compositional factors that may have influenced collective action capacity. The pattern of findings makes a broad village-level explanation less likely, as differences were concentrated in collectively organized outcomes while children's hygiene showed relative convergence, but alternative explanations cannot be ruled out entirely. The observed differences should therefore be interpreted as intervention-associated rather than intervention-caused.

The cross-sectional nature of the data means that temporal ordering among the RQ2 predictor domains and outcomes cannot be established; stronger sanitation outcomes may reinforce community participation over time rather than the reverse. The modest explanatory power of the RQ2 models further suggests that the measured pathway domains capture only part of the variation in outcomes, with habituation, structural conditions, and longer-term socialization also likely to matter.

Several qualitative findings draw on daily fieldwork debrief reports prepared by field investigators rather than on primary FGD and KII data. These records provide contextually grounded observations that help interpret the primary qualitative findings, but they are investigator-generated and should be read as contextualizing evidence rather than independently verified participant accounts. Debrief data also suggested a decline in sanitation-specific SHG activity over time within the Intervention Village, though the present design cannot determine whether this reflects erosion of the intervention's legacy or maturation into self-regulating community norms. As one site within a larger ongoing multi-village evaluation, the present findings are intended to be theoretically generative rather than statistically generalizable; replication across additional sites will test whether similar patterns hold in other contexts.

Ten years after a women-led sanitation intervention, the Intervention Village demonstrated large and robust advantages in menstrual waste management, civic action, community participation, and sanitation action (H1a, H1b supported), but not in children's hygiene practices, which showed a small, non-significant adjusted village difference consistent with the proposed convergence hypothesis. Pathway analyses revealed that civic participation was the sole independent predictor of menstrual waste management outcomes (H2a-H2d all supported), while children's hygiene and toilet maintenance remained unexplained by the measured pathway domains. Together, the results indicate that the long-term legacy of women-led sanitation change may operate primarily through the consolidation of collective norms, civic engagement, and institutional participation rather than through sustained individual training effects. Menstrual waste management appears to be a particularly sensitive indicator of this legacy, shaped more by community-level civic participation than by individual or household-level processes.

More broadly, sanitation outcomes appear to follow different long-term trajectories, and the durability of early gains depends on the social and institutional conditions through which behavior is maintained—not only on individual awareness or infrastructure access. Where collective structures are absent, even compliance achieved through enforcement in formally certified communities may be insufficient to produce lasting behavioral change. Dignity and safety concerns may serve as important motivators for initial adoption, but without the reinforcing scaffolding of collective norms, women's public-facing agency, and accountable local systems, such gains remain fragile. Sustaining sanitation over the long term may depend on more than individual behavior change; collective organizational capacity, functional infrastructure, and governance conditions appear to be associated with whether sanitation practices endure. Point-in-time certification frameworks, while valuable as programmatic milestones, may benefit from complementary follow-up mechanisms that attend not only to infrastructure access but also to the depth of normative anchoring through which sanitation practices are sustained.