Family Functioning in Humanitarian Contexts: Correlates of the Feminist-Grounded Family Functioning Scale among Men and Women in the Eastern Democratic Republic of Congo

In humanitarian contexts, family functioning may have a positive influence on outcomes of mental health and wellbeing for both children and adults. This study sought to adapt and explore an existing family functioning measure for use in eastern Democratic Republic of Congo. Cross-sectional data were collected from 394 respondents (196 men and 198 women) in North Kivu, DRC. The Feminist-Grounded Family Functioning Scale was developed through qualitative research and cognitive testing which conceptualized a healthy and functioning family as one without gender and power hierarchies. Generalized linear mixed models stratified by gender assessed the association between family functioning and demographics, conflict experiences, and correlates of interest including gender attitudes, acceptance of harsh discipline, positive parenting, and power sharing. In the final adjusted models, family functioning among women was associated with positive parenting (β = 0.32, p = 0.009) and power sharing (β = 0.24, p < 0.001), but attitudes around gender and harsh discipline did not retain their significant association. Similarly, men’s family functioning was significantly associated with positive parenting (β = 0.41, p = 0.011) and power sharing (β = 0.16, p = 0.014). Results suggest that family functioning is positively associated with behaviors indicative of non-violent and equitable relationships between family members, but that behaviors might be attenuating the influence of attitudes on family functioning, as the significant association between attitudes and family functioning in the univariate models was not retained in the full models accounting for both behaviors and attitudes. Further research is needed to capture the context-specific nuances of family functioning and to understand the influence of gender and power hierarchies for families in conflict settings. Measures to assess family functioning such as the index used in this study that are highly contextualized and which consider gender and power hierarchies within the family are essential for understanding how to best support families in conflict settings. Family functioning was significantly associated with behaviors indicative of non-violent and equitable relationships between family members, including positive parenting and power sharing between couples. Family functioning was negatively associated with the presence of inequitable attitudes related to gender and acceptance of harsh discipline; however, this significant association was not retained in the full model, demonstrating that behaviors may attenuate the influence of attitudes on family functioning. Conflict and displacement had a significant influence on family functioning, with both negatively impacting family functioning in unadjusted models. Measures to assess family functioning such as the index used in this study that are highly contextualized and which consider gender and power hierarchies within the family are essential for understanding how to best support families in conflict settings. Family functioning was significantly associated with behaviors indicative of non-violent and equitable relationships between family members, including positive parenting and power sharing between couples. Family functioning was negatively associated with the presence of inequitable attitudes related to gender and acceptance of harsh discipline; however, this significant association was not retained in the full model, demonstrating that behaviors may attenuate the influence of attitudes on family functioning. Conflict and displacement had a significant influence on family functioning, with both negatively impacting family functioning in unadjusted models.


Introduction
While definitions vary across contexts, family functioning has generally focused on the social and instrumental elements of family relationships that can lead to overall improved wellbeing of its members (Lebow & Stroud, 2012). Family functioning can capture multiple levels of relationships, with families broadly defined as a social group connected by relationships through marriage, kinship, adoption, or attachment, including extended systems beyond the immediate family (Taliep et al., 2014). Constructs underpinning family functioning have generally included elements of communication, adaptability and responsiveness, cohesion, problem solving, and growth, as a few examples (Currie, 2009;Dai & Wang, 2015;Lewandowski et al., 2010;Miller et al., 2000;Olson, 2000;Pritchett et al., 2011;Sherriff et al., 2010;Skinner et al., 2000;Smilkstein, 1978;Taliep et al., 2014). Given its proximity to both individual outcomes of mental health and psychosocial wellbeing, as well as interpersonal relationships and household power dynamics (Bogels & Brechman-Toussaint, 2006;Khasakhala et al., 2013;Repetti et al., 2002;Sheidow et al., 2013), family functioning is a crucial concept to define and measure in order to develop a better understanding of the complex dynamics between all individuals within a household and the influence of social interventions on these dynamics.
In settings requiring humanitarian interventions, recent literature has demonstrated the ways in which family functioning may play an important role in promoting resilience in the presence of daily stressors and improving mental health and wellbeing outcomes for both children and caregivers (Betancourt et al., 2010;Orpinas et al., 2014;A. L. Sim et al., 2021;Sriskandarajah et al., 2015). Despite the important nature of this complex construct, few studies have sought to measure and examine family functioning in low and middle-income countries, and even fewer studies have measured family functioning in conflict-affected settings such as the eastern region of the Democratic Republic of Congo (DRC) (examples of such research include Amone-P' Olak et al., 2013;El-Khani et al., 2020;Haar et al., 2020;Puffer et al., 2017). One notable study amongst Burmese migrants in Thailand developed a 20-item scale from qualitative data focused on three subscales: family cohesion, family communication, and negative family interactions (Puffer et al., 2017). Recent additional diagnostic work has also been undertaken to validate measures of family functioning in Kenya (Puffer et al., 2021) as well as South Africa (Taliep et al., 2014).
To date, studies examining family functioning have explored its association with factors including child and caregiver mental health and wellbeing (Bogels & Brechman-Toussaint, 2006;Boudreault-Bouchard et al., 2013;Putnick et al., 2015), as well as outcomes of family and community violence (Blackwell et al., Under Review); Lansford et al., 2005;Laskey & Cartwright-Hatton, 2009); however, there is a dearth of evidence exploring family functioning as the primary outcome and dependent variable. In particular, as asset-based programming models seek to support families in humanitarian settings, better understanding of the pathways and mechanisms that may help to improve family functioning is needed to tailor programming to best support families. For instance, recent qualitative research in eastern DRC identified potential shared risk factors and drivers of violence in the home including gender inequitable attitudes, acceptance of harsh discipline, and gendered and age hierarchies (Falb et al., 2022). Findings from the study posit that when families have gender equitable attitudes, do not accept harsh discipline practices, and have equitable power sharing, family functioning may improve, and violence may be reduced.
Through this qualitative research, as well as adaptation and cognitive testing of existing family functioning measures as part of this study, a healthy and functioning family in DRC was conceptualized as one where there are not gender and power hierarchies. However, while previous scales to measure family functioning have been geographically contextualized and validated, they have rarely included items to specifically assess dynamics of gender or age (Repetti et al., 2002) or other power dynamics that exist within the household. In addition, no previous studies have stratified their analyses on family functioning by gender, sex, or age to assess differences between male and female caregivers or children, with most assessments of family functioning focused on female caregivers. Recent studies have found differences between child and caregiver reporting of family functioning and other outcomes of interpersonal dynamics within the household, such as communication and parenting behaviors. For example, in a study examining the impact of a family-based intervention delivered to displaced Burmese families in Thailand, caregivers who received the intervention reported a significant decrease in harsh discipline (β = −0.39, p < 0.001), but children did not (β = −0.09, p = 0.455) (Puffer et al., 2017). Therefore, the lack of consideration of power and gender in previous family functioning scales and the lack of disaggregation of these measures in analyses highlights a critical gap in evidence with regard to family dynamics. It is essential that power and gender are considered in measures of family functioning used in the eastern DRC in order to fully understand relationships and wellbeing within the home, and how these differ for different family members dependent on their gender, age, and disability level (mild, moderate, and severe functional difficulty with vision, hearing, remembering, self-care, and communication), among other factors. This study seeks to address this gap by using an intersectional feminist approach to examine perception of family functioning in the eastern DRC and how this relates to other norms and attitudes around gender and power. Intersectionality is an analytical framework for understanding how aspects of a person's social and political identities combine to create different modes of discrimination and privilege (Crenshaw, 1991). This lens can be applied to the exploration of violence, as it links gender inequality and cycles of abuse within the family while also considering the influence of other socioeconomic factors (Fulu et al., 2017). This approach emphasizes how both intimate partner violence (IPV) and child abuse are normalized by the patriarchal family structure which emphasizes power and control and reinforces women's subordination to men and children's subordination to their parents (Namy et al., 2017). The framework explored by Namy et al., (2017) illustrates how women and children are blamed for violating social expectations around their identities and behaviors and validating violence as a legitimate form of control.
In order to address these gaps in evidence and improve understanding of the mechanisms that influence family functioning, the present study sought to adapt a measure of family functioning for use in eastern DRC which took into considerations the dynamics of power within the family. The purpose of this paper is to assess the psychometric properties of the Feminist-Grounded Family Functioning Scale adapted for the study and to explore potential correlates of this construct with an intersectional feminist lens to deepen understanding of the factors that influence family functioning.

Study Design and Data Collection
The present analyses utilize baseline, cross-sectional data collected as part of an ongoing cluster randomized controlled trial in North Kivu, DRC. The purpose of the trial is to evaluate the effectiveness of the Safe at Home program, a family-based intervention which aims to improve family functioning and reduce outcomes of IPV and child abuse (Clinical Trials #NCT04163549).

Participants
Baseline data were collected from 394 participants (203 adult couple dyads: 196 men and 198 women) from four communities in North Kivu, DRC between November 2019 and January 2020. Participants were 18 years of age or older, part of a monogamous couple, had at least one child between ages 6-12 years, spoke Swahili, Kinyarwanda, or French, and were participating in the Safe at Home program. Both partners from all eligible couples were invited to participate in the study. The response rate for those eligible participants who agreed to partake in the research study was 96.6% for men and 97.5% for women.
Female respondents were 32.00 years of age on average (SD = 8.71; range 19-59) and male respondents were 35.88 years of age on average (SD = 8.71; range 21-59) on average (Table 1). Overall, women in the sample had a lower level of educational attainment than men, with only 15.15% of women attending secondary school or above compared to 43.40% of men (p < 0.001). The majority of women (60.10%) reported a mild disability (mild functional difficulty with vision, hearing, mobility, remembering, selfcare, and/or communication; see more in the Measures subsection), compared to less than half of men (42.35%) (p < 0.001). Conflict and displacement experiences were high among all respondents, who reported being displaced three or more times during their lifetime on average and having experienced at least three different forms of conflictrelated traumatic events.

Procedure
The baseline survey was administered by gender-matched enumerators who read each question aloud to the participant and recorded their response in a tablet using the data collection software SurveyCTO. Enumerators were members of nearby communities within the region hired externally for the research team. They received an intensive training on research ethics for sensitive research on violence including concepts of confidentiality, privacy, and informed consent, as well as principles of gender-based violence, violence against children, and referral for protection and psychosocial support services. All survey tools were translated and back translated into French from English, then translated into Swahili and Kinyarwanda.
The study protocol followed global ethical guidance on research on violence (Ellsberg et al., 2005;Garcia-Moreno et al., 2001;World Health Organization, 2007). Informed consent was obtained for all study participants, with substantial effort made to explain that participants' decisions to participate would have no influence on current or future assistance from the International Rescue Committee IRC or any other organization, including their ability to participate in the Safe at Home program. All informed consent procedures and interviews were conducted in a private space where confidentiality could be maintained. All study respondents were provided with information for follow-up services. The study protocol and tools were approved by the IRC Institutional Review Board (WPE 1.00.014) and by the DRC Ministry of Public Health's Ethics Committee (Comité National d'Ethique de la Santé).

Family Functioning
The primary outcome variable of interest for the study is family functioning. The present analyses define this variable using the Feminist-Grounded Family Functioning Scale, a 26-item family functioning scale adapted from the scale developed by Taliep et al., (2014) for use in South Africa. The original 32-item scale was developed based on a literature review of existing instruments which might have applicability to the South African context and expert panel review using the Delphi method across six domains of family functioning: family cohesion, affective responsiveness, communication, problem solving, boundaries, and roles. Responses were captured using a 4-point Likert scale (strongly agree, agree, disagree, or strongly disagree). The scale was then adapted for this study through translation to French, Swahili, and Kinyarwanda, content validation, and psychometric testing to achieve the final 26-item index used in this paper (Table 2).
To ensure the applicability of the family functioning measure to the study context, the research team reviewed definitions of family functioning and conducted free-listing and item generation as well as field testing with IRC program staff and members of the study population in North Kivu, DRC. One question from the original scale was removed, and seven new items were added to ensure contextually appropriate and relevant concepts of family functioning. Item generation reflected discussions on discrimination and gender inequality as key issues in family dysfunction which were missing from the externally developed family functioning measure. For this reason, the new items developed for the context centered on power dynamics within the home with regard to gender and age. They included items on gender roles within the family, as well as the consultation of both partners and children in family decision-making and the perception that different household members could express their points of view. Thirty-eight items (31 items from the a priori scale and seven locally generated items) were included in the baseline questionnaire. The 26-item Feminist-Grounded Family Functioning Scale was then determined through psychometric testing, which is described in the Results section. The measure used for analyses is a continuous score summing the number of statements with which a respondent agreed (possible range 0-78), a higher score denoting better family functioning.

Attitudes Regarding Gender and Discipline
Gender inequitable attitudes were assessed using a continuous measure (range: 0-22) calculated by summing the score for 11 statements with which a respondent agreed (2), partially agreed (1), or disagreed (0) (Pulerwitz & Barker, 2008). For example, "There are times when a woman deserves to be beaten", or "A woman should tolerate violence to keep her family together". Higher scores denote more agreement with gender inequitable attitudes. Internal consistency was sufficient with a Chronbach's alpha of 0.82.
Acceptance of harsh discipline against women and children was adapted from the Multiple Indicator Cluster Surveys (MICS) (UNICEF, 2013). Responses to 19 binary items were summed to form a continuous measure (range: 0-19). Higher scores denote higher acceptance of harsh discipline. Example items include: "It is OK for a husband to beat or hit his wife if she goes out without telling him" and "It is OK for a parent to beat or hit a child if the child argues with the parent". Internal consistency was high with a Cronbach's alpha of 0.92.

Positive Behaviors
Correlate behaviors of family functioning were assessed, including equitable power sharing and positive parenting.  Puffer et al., (2015), which asks respondents how often they do nine different parenting behaviors: very frequently (3) to almost never (0). Responses were summed to form a continuous measure (range: 0-27), with higher scores denoting more use of positive parenting behaviors. Example items include: "How often did you tell your child that you care for / love them?" and "How often are you able to give your child guidance by setting rules?" Internal consistency was determined to be sufficient as a secondary outcome with a Chronbach's alpha of 0.59.

Demographic Characteristics
Demographic measures included a continuous measure of age in years, categorical measure of education level and disability level (The Washington Group on Disability Statistics, 2016), and continuous measures of conflict experiences including number of times displaced and number of conflict-related traumatic events (Global Women's Institute, and WHO (2017)). Disability level was measured using a series of six questions about level of difficulty with six core functions (vision, hearing, mobility, cognition (remembering), self-care, and communication) using a Likert scale rating. The summary variable is coded according to the Washington Group on Disability Statistics guidelines based on any functional difficulty, with respondents having no disability (no difficulty), mild disability (some difficulty), or moderate and/or severe disability (a lot of difficulty and/or cannot do at all) (The Washington Group on Disability Statistics, 2016).

Analysis
Exploratory and confirmatory analyses of the family functioning measure included assessment of missingness and variable distributions and central tendency overall and by gender. Items with a high degree of missingness were considered for removal from the scale. Exploratory factor analysis (EFA) was used to examine the factor structure of the family functioning items and a polychoric correlation matrix and principal axis factoring were used to account for the ordinal nature of the indicators. To select the number of factors for inclusion in the EFA and confirmatory factor analysis (CFA), eigenvalues over 1.00, percentage of variance explained, examination of scree plots, and dimensionality tests were considered. To assess internal consistency of each scale as determined from the EFAs, Chronbach's alpha was calculated and items were considered to meaningfully load on a factor if it was ≥0.5 (Knekta et al., 2019). CFA assessed dimensionality separately by gender and in a joint module using tests including Room Mean Squared Error of Approximation (RMSEA) (≤0.06), Tucker Lewis Index (TLI) (<0.09), Comparative Fit Index (CFI) (≥0.95), and Standardized Root Mean Square Residual (SRMR) (≤0.08) (Shi et al., 2019). Revisions to scales were guided by factor loadings (<0.30) and uniqueness (>0.75), as well as content validity. Tests for suitability and adequacy using Bartlett's test of sphericity (<0.05) and the Kaiser-Meyer-Olkin (KMO) test (>0.80) (Shrestha, 2021) were then used to shorten the family functioning scale (refined based on the EFA and CFA results) to improve its utility in humanitarian contexts.
Descriptive statistics and models were constructed for the male and female samples separately. Demographic data and correlates of interest were initially assessed for their relationship with the family functioning index using crosstabulations and Pearson correlation coefficients.
Once the dimensions of the family functioning index were defined, unadjusted, gender-stratified generalized linear mixed models accounting for clustering at the village level were used to assess potential associations between family functioning and demographics, conflict experiences, and correlates of interest. Those variables that were associated with family functioning at the p < 0.05 level in unadjusted models were retained in the final adjusted models. Regression diagnostics were conducted for all gender-stratified unadjusted and adjusted models and visually assessed for collinearity, normality, and heteroscedasticity. Complete case analyses were implemented for missing data in all models. Analyses were conducted using SAS v9.4.

Results
Psychometric Properties of the Feminist-Grounded Family Functioning Scale Table 2 includes all items from the final 26-item Feminist-Grounded Family Functioning Scale and the frequency and percentage of men and women reporting agreement with each item. The a priori scale was assessed prior to exploring psychometrics for an adapted scale. Internal consistency of the original 32-item scale was lower than desired for the primary outcome (Chronbach's alpha = 0.73) and the six subscales had low internal consistency (Cronbach's alpha <0.50). Confirmatory factor analyses analyzed for men's and women's data separately indicated poor fit for the a priori model: for women, dimensionality tests did not meet standards for good fit (Root Mean Squared Error of Approximation (RMSEA) 0.09, Tucker Lewis Index (TLI) 0.46, Comparative Fit Index (CFI) 0.51, Standardized Root Mean Square Residual (SRMR) 0.14) and for men the model was not estimable. A joint model which included participant gender as an explanatory variable yielded slightly better results but did not consistently indicate good fit across tests of dimensionality (RMSEA 0.08, TFI 0.66, CFI 0.70, SRMR 0.10) and the direction of correlations was inconsistent. These findings, in combination with contextual experts' determination that additional items were needed to capture family functioning in the study communities and poor factor loadings for almost one-third of items in the original South Africa sample (internal communication), underscored the utility of a locally relevant measure which included testing of the seven items developed in-country for this study population.
Bartlett's test of sphericity (p < 0.001) and the Kaiser-Meyer-Olkin (KMO) test (0.83 overall; 0.70 for women, 0.75 for men) indicated suitability and adequacy, respectively, of the 38 items for factor analysis. Principal component analysis yielded 12 components with Eigenvalues >1.00 but scree plots and parallel analysis suggested a oneor two-factor solution; since contextual discussions posited family functioning as one joint concept, authors explored a one-factor model with polychoric correlation and principal axis factoring. Twelve items with both low factor loadings (<0.30) and high uniqueness (>0.75) were dropped from the index, with exception of those locally developed items which were retained regardless of their loading or uniqueness. Bartlett's test of sphericity (p < 0.001 overall and for men and women separately) and KMO test (0.87 overall; 0.77 for women, 0.83 for men) indicated suitability and adequacy of the remaining 26 items for factor analysis. The adapted scale used in this study is 26 items, with good internal consistency (Cronbach's alpha 0.86 overall, 0.84 for women, 0.86 for men). Measures of dimensionality demonstrated better fit than the a priori scale (Root Mean Squared Error of Approximation (RMSEA) 0.07, Tucker Lewis Index (TLI) 0.87, Comparative Fit Index (CFI) 0.88, Standardized Root Mean Square Residual (SRMR) 0.07). Analyses report the scale as one single continuous latent construct with standard deviations (possible range 0-78, with 78 indicating best possible family functioning; observed range 21-70).

Correlates of Family Functioning
While family functioning measures are highly contextual and external validity is difficult to achieve, the approach to developing in the index does strengthen the validity of the construct. The index was adapted and developed in collaboration with DRC child protection experts and validated through cognitive testing with community members of neighboring villages in eastern DRC, ensuring the face validity of the construct for the context. Construct validity was then assessed and strengthened through exploratory analyses examining the correlation of the measure with similar and dissimilar correlates.
Overall, women reported lower family functioning, with a mean score of 47.99 (SD = 5.55) compared to a mean score of 51.08 (SD = 6.57) among men (p < 0.001). With regard to negative attitudes, women reported higher acceptance of harsh discipline with a score of 10.01 (SD = 5.41) ( Table 1) and higher gender inequitable attitudes with a score of 17.02 (SD = 4.11). In comparison, men reported an acceptance of harsh discipline score of 5.60 on average (SD = 4.95, p < 0.001) and gender inequitable attitudes score of 13.09 (SD = 5.64, p < 0.001). When examining positive behaviors, women reported a mean power sharing score of 86.12 (SD = 6.61), and men reported slightly more power sharing between partners with a mean score of 90.99 (SD = 8.12, p < 0.001). Similarly, women reported a mean positive parenting score of 14.51 (SD = 3.19) and men reported a slightly higher mean score of 16.44 (SD = 3.21, p < 0.001), denoting higher use of positive parenting behaviors.

Unadjusted Associations between Family Functioning, Hypothesized Correlate Variables, and Demographics
In unadjusted linear regressions adjusting for clustering at the site level (Table 3), family functioning score among women was most strongly positively associated with positive parenting behaviors (β = 0.51, p < 0.001), as well as power sharing between partners (β = 0.26, p < 0.001). Acceptance of harsh discipline was negatively associated with family functioning (β = −0.20, p = 0.007). Conflict-related traumatic experiences and displacement were also significantly associated with average family functioning score among women. For both, a one-unit increase in displacement or conflict experiences resulted in a decrease in average family functioning score (number of times displaced: β = −0.78, p = 0.004; conflict-related traumatic events experienced: β = −0.71, p = 0.005). Demographic variables of age, level of education, and disability status were not significantly associated with average family functioning score for women and therefore were dropped from the full adjusted models.

Adjusted Associations between Family Functioning, Hypothesized Correlate Variables, and Demographics
In the final adjusted model accounting for significantly associated demographics and clustering at the site level, women's family functioning score retained its significant association with positive behaviors (Table 4). A one-unit increase in both positive parenting and power sharing resulted in an increase in average family functioning score (Positive parenting: β = 0.32, p = 0.009; Power sharing: β = 0.24, p < 0.001). However, the association between average family functioning score and attitudes was no longer significant in the final adjusted model, demonstrating that behaviors might be attenuating the influence of attitudes on family functioning. Among demographics included in the full model, displacement remained negatively associated with average family functioning score (β = −0.65, p = 0.011), but women's experience of conflict was no longer significantly associated with family functioning.

Unadjusted Associations between Family Functioning, Hypothesized Correlate Variables, and Demographics
In unadjusted linear regressions accounting for clustering at the site level (Table 3), men's family functioning score was strongly correlated with reported behaviors including positive parenting behaviors (β = 0.54, p = <0.001) and power sharing between partners (β = 0.23, p < 0.001). For men, acceptance of harsh discipline (β = −0.24, p = 0.011) and gender inequitable attitudes (β = −0.22, p = 0.009) were both negatively associated with average family functioning score. Men's experience of displacement was most strongly associated with average family functioning score (β = −1.08, p = 0.014), but neither conflict experiences nor other demographic variables were significantly associated with the family functioning outcome for men.

Adjusted Associations between Family Functioning, Hypothesized Correlate Variables, and Demographics
In the final model adjusting for significantly associated demographics and clustering at the site level (Table 4), men's family functioning score remained significantly associated with positive behaviors. An increase in positive parenting score resulted in an increase in average family functioning score for men (β = 0.41, p = 0.011), as did an increase in power sharing score (β = 0.16, p = 0.014). Average family functioning score did not retain a significant association with attitudes including acceptance of harsh discipline and gender inequitable attitudes, similarly suggesting that behaviors could be attenuating the influence of men's attitudes on family functioning. No demographics retained their significance in the men's adjusted model.

Discussion
Measuring family functioning is an important advancement in understanding how to best support families to thrive, particularly during periods of instability and conflict. This study advances the literature by demonstrating the need for contextually relevant measures and providing psychometric assessment of a family functioning scale piloted in eastern DRC. Qualitative formative research and confirmatory factor analysis revealed that the concept of family functioning may differ across contexts: while the a priori Family Functioning Scale was developed in response to the need for better measures of family functioning in the Global South (Taliep et al., 2014), the scale had poor fit among men and women living in North Kivu, DRC. Qualitative formative research (Falb et al., 2022) and discussions with IRC DRC program staff on local conceptions of high functioning, happy families revealed contextually relevant concepts of discriminatory gender relations, child rights, and control of resources which were not adequately captured in the original scale and are often not included in wellvalidated measures of family functioning used in Western contexts (e.g. the Family Assessment Device by Miller et al., 2000). The adapted scale presented here is an outcome for a holistic family-focused intervention that seeks to improve well-being in a context marked by decades of political conflict, relevant for the North Kivu context, and is not meant to provide clinical categorization or diagnosis. The authors consider this adapted scale to be a starting point for more substantive qualitative research focused on understanding family functioning as a gender and power construct within the North Kivu context and other contexts in the Global South. Inclusion of power dynamics within the home also provides a pathway for assessing intersecting forms of privilege and disadvantage, such as disability and age, in addition to gender.
Unadjusted linear regressions demonstrated convergent validity of the family functioning measure, as people exemplifying characteristics with higher levels of wellbeing also reported higher levels of family functioning. The findings of the analyses demonstrated that in family functioning is positively associated with behaviors indicative of use of non-violent and equitable relationships between family members, such as positive parenting and power sharing between couples. Family functioning was also negatively associated with the presence of inequitable attitudes related to gender and acceptance of harsh discipline, with family functioning worsening as negative attitudes increased. However, in the adjusted models, while the association between behaviors and family functioning was retained, attitudes were no longer significantly associated with family functioning. This dynamic between behaviors, attitudes, and functioning at the family level warrants further investigation, as it could demonstrate that behaviors are mediators and may attenuate the influence of attitudes. This potential relationship between correlates does support the hypotheses from the formative research (Falb et al., 2022) by demonstrating that behaviors may be the mechanism by which attitudes can influence family functioning.

Family Functioning as a Protective Factor against Conflict Experiences in Humanitarian Settings
While there is limited evidence examining the influence of correlates on family functioning as the dependent variable, the positive relationship between behaviors and family functioning is consistent with other literature from humanitarian and conflict settings. Recent studies evaluating family-based programs among conflict-affected populations in the West Bank, Afghanistan, and Thailand have found that interventions targeting behaviors such as parenting practices and decision-making at the family level improved  family functioning overall Puffer et al., 2017). Qualitative findings from Thailand also confirmed that respondents perceived an increase in equitable decision-making involving both their partners and their children and linked this with improvements in family functioning (A. Sim et al., 2014). However, there is no existing evidence examining the relationship between discriminatory and harmful attitudes and measures of family functioning, and there remains a dearth of evidence that specifically considers power dynamics within the home as they relate to family functioning, emphasizing the strength of this index as it integrates these factors and highlights the need for future research to examine how underlying patriarchal attitudes and norms may affect power dynamics within the home. Related to this potential influence of gender norms on family functioning, the present analyses also found that men reported significantly higher family functioning than women overall. Most studies that measure the construct of family functioning only measure it as reported by the primary caregiver, and most often the female caregiver Haar et al., 2020;Puffer et al., 2017). Among those few studies that measure family functioning as reported by both men and women, none disaggregate the outcome by gender. The original family functioning score developed by Taliep et al. (2014) is largely derived from the Family Assessment Device (FAD), which was intended to be asked of the primary caregiver, which in the DRC and other contexts is often the female caregiver (Asghar et al., 2017;Digolo et al., 2019). IRC program staff from the DRC emphasized that women spend more time caring for the children and doing housework while men are often gone for most of the day for income generating activities. Therefore, the difference in perception between men and women with regards to family functioning may largely be due to the difference in gender norms around family roles within eastern DRC. Women as the primary caretakers may perceive worse functioning because they shoulder the burden of ensuring functioning within the home through managing daily activities associated with child-rearing and social and emotional development, and the needs of their intimate partners (Falb et al., 2022;Niu et al., 2018). Women may also have greater awareness of negative interactions between family members than men since gender norms often dictate that they spend more time with those family members in the home. For example, other findings from this study demonstrate that women report higher levels of perpetration of violent discipline against children than men (Blackwell et al., Under Review).
Conflict and displacement also had a significant influence on family functioning overall, with higher conflict and displacement experiences negatively impacting family functioning in unadjusted models and displacement retaining its negative association with family functioning in adjusted models. This is consistent with other literature from conflict and humanitarian settings that have found that conflict and displacement impact family functioning both directly and indirectly. Direct conflict experiences such as displacement and witnessing or experiencing conflictrelated violence impacts the mental and physical health of both caregivers and children (Amone-P'Olak et al., 2013;A. Sim et al., 2018). In addition, the disruption of social networks, the breakdown of economic structures and rule of law, and increased stressors caused this breakdown exacerbates the risk factors of family violence (Ellsberg et al., 2020;. Consistent with other findings, conflict and displacement also had a stronger influence on family functioning for women than it did for men in the present analyses.

Limitations
These findings must be interpreted with limitations in mind. Firstly, the analyses used one continuous summary variable which did not allow for nuanced analyses by subscale to reflect the conceptual domains that make up family functioning. Family functioning is often conceptualized as a set of factors or domains; while the present analyses conceptualize family functioning as one factor, this construct was validated through exploratory and confirmatory factor analyses. Further work is needed to unpack whether those domains are relevant or not in North Kivu. The scale which was adapted for this study relies heavily on the FAD, which has inconsistent factor structure across studies (Lebow & Stroud, 2012). While our analysis includes power-sharing between partners and our adapted scale demonstrated better fit than the original scale, the adapted scale did not meet the appropriate thresholds for Comparative Fit Index or Tucker Lewis Index. The use of one continuous measure may obscure more nuanced analyses on the influence of factors on specific aspects of the family functioning construct within this context, including dynamics between different members of the household or specific domains of functioning. As such, further research is needed to improve conceptualization of family functioning with regards to differential power by gender and age within the home.
Secondly, the discordance in reporting on family functioning between women and men demonstrate that concepts of family functioning may differ not only by context but also by gender, age, and other factors. Gender equity in decision-making and other gender norms are not captured in most family functioning indices but were included in the final scale used in these analyses. There may be differences in how those questions are perceived and interpreted between men and women. In addition, most family functioning indices were developed in high-income countries with women as the intended targeted primary caregiver, and therefore different approaches to conceptualizing and measuring family functioning for men may need to be considered. This measure may also be subject to under-or overreporting due to social desirability bias. The relatively lower reporting of functioning among women could also be further evidence of the validation of the scale, given that women are also reporting higher prevalence of IPV and child abuse than men (Blackwell et al., Under Review). While this paper did not account for dyadic interdependence, the use of gender-stratified analyses enabled examination of the construct of family functioning at the individual level to deepen theoretical understanding of how this construct might differ for women and men.
Thirdly, while we cannot assess criterion validity, this may not be an appropriate approach for family functioning scales given the lack of a gold standard measure and the contextual nature of the construct. Nevertheless, the approach to developing and finalizing the family functioning index used in the present analyses ensured face validity and construct validity of the construct. In addition, the index operated as hypothesized with regard to its correlation to the positive behavioral constructs of power sharing and positive parenting. Power sharing was positively correlated with family functioning (Pearson Correlation Coefficient: Women = 0.22, p < 0.001; Men = 0.28, p < 0.001), as an increase in power sharing resulted in a strong significant increase in family functioning. Similarly, positive parenting was positively correlated with family functioning (Pearson Correlation Coefficient: Women = 0.30, p < 0.001; Men = 0.28, p < 0.001) and an increase in positive parenting resulted in a significant increase in family functioning. These correlations demonstrated convergent validity with the Feminist-Grounded Family Functioning Score. Family functioning was also lower for women who reported higher levels of violence in the home, demonstrating discriminant validity (Falb et al., In Press).
Finally, while family functioning was assessed for both partners within a couple, the voices and perspectives of children as well as other family members within the home are absent from the data. Measures of family functioning such as the FAD (Miller et al., 2000) have been administered to both adults and children. Recent studies in conflict settings have shown meaningful differences between caregiver and child dyads with regard to reporting of family functioning and other outcomes (for example, Puffer et al., 2017). Future analyses should include these diverse perspectives using the same indices or through other quantitative and qualitative measures, as differing child perspectives are critical to inform effective, evidence-based family programming in the DRC and globally. Similarly, perspectives of other caregivers in the family such as grandparents, older siblings, and other relatives could be included in future research to diversify perspectives within the household, given the frequent occurrence of multigenerational households within the context.

Implications and Future Research
Further research is needed to develop family-level assessments that can capture the complex nuances of family functioning, including dyadic-and family-level relationships and differing perspectives. In addition, such an approach should consider the specific situation of families in conflict settings, who may be geographically separated or experiencing other risks of conflict and displacement that make standardized measures less relevant. Additionally, approaches to measuring parenting behaviors and family cohesion from other disciplines could be incorporated into future studies. For example, a study in Kenya adapted a tool for observational learning on problem solving to reduce the variability and challenges common in self-report measurement (Giusto et al., 2019).
Despite the need for further research and testing of measures, the largest correlates of family functioning amongst couples centered on positive behaviors between parents and children, and between intimate partners. Interventions may generate greater improvement in family wellbeing by jointly targeting gender equality and positive parenting behaviors.
The Feminist-Grounded Family Functioning Scale adds an important dimension to theoretical frameworks and operationalization of family functioning, as it brings together family level outcomes with a feminist lens of gender and power dynamics. Historically, violence against women and girls (VAWG) practitioners have been skeptical of utilizing family functioning as an outcome because existing constructs masked IPV or gender dynamics. However, this scale specifically calls out the importance of these dimensions by including them within the overall construct. The Feminist-Grounded Family Functioning Scale could therefore be adapted for evaluating other feminist-informed violence prevention or family strengthening interventions.
Context-specific measures of family functioning are needed to accurately and adequately assess outcomes of trials that aim to influence positive change at the household level, despite the benefits of having a standardized measure of such a complex construct. While the approach used to develop the Feminist-Grounded Family Functioning Scale for this study produced a strong and reliable measure, we would not expect the same parameters if the index was used in settings outside of the eastern DRC, and for that reason cannot make claims on the generalizability of the findings. However, the same procedures could be applied to adapt a family functioning measure in other contexts to see how both the measure and its functioning would change. Future research on family functioning should consider the wellbeing of families to reduce violence and improve gender and power hierarchies within the home. This study therefore serves as a launchpad for replicating the adaptation process and developing stronger family functioning outcomes that are not only geographically contextualized but also apply an intersectional feminist lens to consider gender and power dynamics within the home. Such conceptualizations of family functioning are essential for understanding how to best support families in conflict settings and developing and evaluating programs to meet their needs.

Compliance with Ethical Standards
Conflict of Interest The authors are employed by the agency that is implementing the program that is being evaluated. The authors report no financial conflicts of interest.

Ethics Approval
The study received ethical approval from IRC's institutional review board (WPE 1.00.014) and from the Comité National d'Ethique de la Santé (CNES) Direction Provinciale du Sud-Kivu.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.