Intimate partner violence (IPV) has significant implications for mothers’ and children’s health and wellbeing after separation (Hardesty et al., 2012, 2019). For some women, IPV continues after they separate, compromising their safety and long-term adjustment (Logan & Walker, 2004). Risk of homicide and severe violence also increases after separation and divorce (Campbell et al., 2003). More frequent contact with the abuser due to joint custody or visitation (vs. less contact with sole custody or restricted visitation) can create opportunities for violence (Fleury-Steiner et al., 2016). Controlling and intrusive behaviors also have been found to continue after separation even when physical violence declines (Hayes, 2012), especially among mothers (Davies et al., 2009). Considering the evidence, IPV is believed to preclude the potential for cooperative joint custody after divorce (Pruett & DiFonzo, 2014).

Although studies have examined IPV in custody cases, most focus on contested cases (i.e., when parents’ custody requests are inconsistent), especially those involving protracted litigation and third-party intervention (e.g., court-referred mediation, custody evaluation; Holtzworth-Munroe et al., 2021; Logan et al., 2002). The extent of IPV among cases – contested or not – that settle without such intervention or without courts being made aware of past violence is unknown. In general, divorcing parents are encouraged to negotiate agreements and avoid litigation, including those with IPV (Holtzworth-Munroe et al., 2021). Judges can reject or modify proposed settlements if they fail to meet children’s best interests, which may include evidence of IPV. In practice, courts generally approve proposed agreements (Emery, 2012), reserving time and resources for cases with protracted litigation. The assumption may be that IPV, if present, does not pose a substantial risk among parents who do not contest custody. Yet, abused mothers encounter numerous barriers (e.g., fear) to contesting custody (Laing, 2017). This is especially true among women who have experienced coercive controlling violence (CCV) versus situational couple violence (SCV; Hardesty & Ganong, 2006) because they live in chronic states of fear (Logan, 2017) and are at higher risk of postseparation abuse (Hardesty et al., 2017). Lack of awareness of IPV and its relevance contributes to custody agreements that compromise the safety and wellbeing of victims and their children (Stark et al., 2019).

A gap exists in the literature as to when and how courts are made aware of IPV and how IPV relates to custody outcomes, especially among samples with diverse IPV and divorce process experiences. Although we would not expect all experiences or forms of violence to be equally likely to be documented or to influence custody outcomes, there is a critical need to understand how and when IPV does appear in divorce cases and its impact on custody. Thus, the purpose of this paper is to examine the process of divorce in a general sample of mothers who self-reported different types of IPV during their marriage. First, we examine if and how self-reported IPV is documented in court records, either in or outside divorce cases. Second, we explore whether the association between IPV (self-reported or court documented) and final custody agreements varies by whether the case was contested or whether the case involved third-party intervention.

Background

In all 50 states, custody decisions in family courts are guided by the “best interest of the child” standard, which encourages child-centered custody agreements that support ongoing contact between children and both parents (Frankel, 2007). This standard is based upon literature indicating benefits to children when parents cooperate after divorce and minimize children’s exposure to conflict (Amato, 2010). Thus, many states have presumptions in favor of joint custody. When parents contest custody, courts face the complex task of evaluating what custody agreements best meet children’s needs. Because the needs in each family are unique, courts are encouraged to take an individualized, case-by-case approach (Pruett & DiFonzo, 2014).

In cases with IPV, individualized approaches ideally permit decisions that prioritize safety (Rossi et al., 2016). Indeed, researchers and advocates urge courts to prioritize safety in agreements, such as limiting abusers’ access to children and protecting mothers when risk persists (Davis et al., 2011). Courts are also urged to consider differences in IPV types. For example, CCV (i.e., violence that occurs as part of a larger pattern of dominance and control) has been linked with greater risks for postseparation violence and control compared to SCV (e.g., violence that arises in the context of conflict; Elizabeth, 2017; Hardesty et al., 2017, 2019). Pence and Paymar’s (1993) power and control wheel visually depicts physical and sexual violence in the outer rim, holding together the spokes of the wheel (i.e., the coercive control tactics). As Johnson (2008, p. 9) explains,

. . . the diagram does not include the violence as just another means of control, another spoke in the wheel…When violence is added to such a pattern of power and control, the abuse becomes much more than the sum of its parts. The ostensibly nonviolent tactics that accompany that violence take on a new, powerful, and frightening meaning, controlling the victim not only through their own specific constraints, but also through their association with the general knowledge that her partner will do anything to maintain control of the relationship, even attack her physically.

This theoretical framework informed Johnson’s conceptualization of CCV as involving at least one act of physical or sexual violence accompanied by a pervasive pattern of domination and control. Thus, different approaches to custody agreements depending on type of IPV may be warranted (Hardesty et al., 2012).

A challenge, however, is that courts may not be made aware of IPV in divorce cases (Kernic et al., 2005), or when made aware, may minimize or ignore its relevance to custody (Pruett & DiFonzo, 2014). A comparison of contested cases with substantiated IPV (n = 44) and without IPV (n = 36) found little difference between these groups in the custody evaluation process or custody outcomes (Logan et al., 2002). The same study also found that documentation of IPV (e.g., orders of protection) available in case files was often excluded from custody evaluation reports. A study of mediated agreements found higher levels of violence predicted increased probability of joint (vs. sole mother) legal custody and more child time with fathers. Those with and without IPV did not differ in physical custody outcomes (Rossi et al., 2015). Fathers’ history of protective orders in contested cases involving a clinical evaluation was unrelated to custody decisions, although a record of arrests was correlated with sole mother custody (Raub et al., 2013). Thus, the limited available evidence appears to show few, if any, consistent differences in custody outcomes for cases with and without a history of IPV, even with some type of third-party intervention (e.g., custody evaluation). When evidence of IPV is overlooked, the safety of mothers and their children after separation is at stake (Watson & Ancis, 2013). Moreover, none of these studies differentiates between IPV types.

Although IPV is common among divorce cases that involve judicial intervention (e.g., trial; Beck et al., 2009), most divorcing parents settle out of court. Voluntary participation in alternative dispute resolution methods, such as mediation, is encouraged as a way to avoid adversarial litigation. Indeed, divorce law can be seen as empowering parents to negotiate agreements on their own (Emery, 2012). These divorce cases also include those with IPV, and, like litigated cases, there is evidence that IPV goes undetected or is not adequately considered (Rossi et al., 2015). However, family courts lack systematic approaches for identifying IPV in divorce cases, differentiating between IPV types, and assessing postseparation risk. Victims also may be reluctant to disclose IPV or may be at a disadvantage due to power differences (Watson & Ancis, 2013). For example, abused mothers experiencing CCV have reported feeling compelled to agree to joint custody out of fear for their own and their children’s physical safety and the prospect of lengthy court battles they could not afford (Hardesty & Ganong, 2006). Such fears may be reinforced by attorneys who discourage victims from bringing IPV into divorce cases because doing so may result in worse custody outcomes (Saunders et al., 2012). It is also possible that some abused mothers incorrectly assume that IPV documented elsewhere (e.g., in protective order or criminal records) is known to family court judges (Araji & Bosek, 2010).

We posit that IPV documentation in divorce cases is important even when custody is not contested because victims may return to court to modify agreements due to continued violence or abuse (Shipley, 2011). Furthermore, without documented IPV, judges are unlikely to intervene or request additional information when they review agreements negotiated out of court. This is especially concerning for mothers who experience CCV and are too afraid to legally challenge highly controlling abusers. Thus, we examined if and how self-reported IPV is documented in divorce cases and how it relates to custody outcomes. Using data collected from in-person interviews with divorcing mothers (referred to as self-reports) and data extracted from publicly available divorce, civil protective order, and criminal cases (referred to individually as cases or collectively as court records), we explored the following research questions:

  1. 1.

    Are divorcing mothers’ self-reports of IPV documented in court records, either outside or within the divorce case?

    1. a.

      Are self-reports of IPV documented with civil protective orders or criminal records?

    2. b.

      Are self-reports of IPV or associated civil protective orders or criminal records documented in the divorce case file (i.e., does IPV make it into the divorce case)?

  2. 2.

    Do custody outcomes differ by whether IPV is self-reported or documented in the divorce case?

  3. 3.

    Does the association between self-reported or documented IPV and custody outcomes differ by whether custody was contested or whether there was third-party intervention?

Method

This study involved data collected as part of two federally funded projects on IPV, divorce, and custody. The initial phase involved self-reported data from mothers early in the divorce process. The second phase involved administrative divorce, civil protective order, and criminal court records. IRB approval was obtained for both studies and numerous precautions were taken to ensure women’s safety and privacy (see Hardesty et al., 2017).

Mothers’ Self-Report Data

Sample

The self-report data came from mothers who had recently been named in a divorce filing. Data collection occurred between July 2009 and November 2013 in a large Midwest county. In the initial phase, mothers were asked to participate in a 1-year longitudinal study about the divorce process, focusing specifically on coparenting relationships and health outcomes for mothers and children.

Of the approximately 1,200 divorce petitions filed during the sampling timeframe that also involved minor children, we obtained a randomly selected sample of 190 mothers, which allowed for adequate power (> 0.80) to detect small effects (d = 0.20). The sample consisted of women 18 and older who (1) were named in a divorce filing within the last 12 weeks; (2) had at least one child under age 18 with their former partner; (3) had at least 25% time with their child(ren), whether formally or informally arranged; and (4) could understand and speak English (see Hardesty et al., 2017 for a detailed description of recruitment and responses).

Mothers in the sample were primarily White (80%) with 12% Black/African American, 4% Asian/Asian American, 3% Latinx or Hispanic, and 2% other. They were between the ages of 21 and 55 years (M = 36.00, SD = 7.28), and had between one and four children (M = 1.75, SD = 0.75). Their marital length ranged from a few months to 27 years (M = 9.60, SD = 5.89) and they had been physically separated for under a month to nearly 3 years (M = 8.00 months, SD = 6.76). The majority (81.6%) of divorces were not yet finalized at baseline. On average, mothers and former partners attended college but level of education ranged from less than high school to advanced graduate degrees and was normally distributed. Mothers were employed full-time (57%; n = 108), part-time (19%; n = 37), or unemployed (24%; n = 45) at the baseline interview.

Procedure

In the initial phase, in-person interviews were conducted at 3-month intervals, across 5 time points. In the current study, only information obtained from the baseline interview was used. Mothers received $35 for participation in the baseline interview. Interview questions primarily focused on the parents’ relationship during marriage and after separation.

Variables and Measurement

Self-Reported IPV

We used a combination of measures to classify IPV. We used the Revised Conflict Tactics Scale (CTS2; Straus et al., 1996) and Goodman et al.’s (2003) modified version of the sexual coercion subscale of the CTS2 to measure self-reported marital violence. We used a subset of items to index direct physical violence: 11 physical violence items (e.g., hit, choked, slammed against wall) plus two sexual coercion items that involved physical force (e.g., used force to have sex). Mothers indicated whether each of the 13 acts occurred during the marriage. We then created a dichotomous indicator of presence versus absence of marital violence reflecting whether mothers had experienced at least one act of physical violence during their marriage, which is consistent with Johnson (2008) that any act of physical or sexual violence provides victims with the knowledge that it could happen again. We used the Dominance-Isolation Subscale of the Psychological Maltreatment of Women Inventory – Short Form (PMWI; Tolman, 1992) to assess coercive control during marriage. Mothers rated how often they experienced seven acts (e.g., “he monitored my time and made me account for my whereabouts”) during the year prior to separation on a scale from 1 (never) to 5 (always). Coercive control tactics can vary over time and tend to increase in response to potentially losing one’s partner if the relationship ends. Therefore, we measured the control tactics used in the last year because they reflect the most salient context for negotiating divorce and custody agreements. Responses were summed, with higher scores indicating a higher frequency of coercive controlling behaviors (M = 16.05; SD = 7.51; Range = 7–35; α = 0.85).

Based on Pence and Paymar (1993) and Johnson (2008), any experience of physical or sexual violence in marriage is relevant to the experiences of coercive control in the last year of the relationship. Thus, measuring coercive control tactics in the last year of marriage and combining it with ever having experienced violence is consistent with this conceptualization of CCV. Using procedures from Hardesty, Crossman, et al. (2015), we then categorized mothers into three mutually exclusive IPV groups based on presence versus absence of physical violence and low versus high coercive control during marriage. Mothers who reported at least one act of violence and had a PMWI score of 18 or less were categorized as having experienced SCV. Mothers who reported at least one act of violence and had a PMWI score of 19 or higher were categorized as having experienced CCV. The remaining mothers were classified as having experienced no violence (see Hardesty et al., 2015). A total of eight (~ 4%) mothers met the cut-off for coercive control but reported no physical or sexual violence and were included in the no violence group because nonviolent coercive control was not (and is still not) criminalized as domestic violence in the county where the data were collected.

Demographic Covariates

Marital length was numerically measured in years using the date of marriage and date of physical separation reported by the mothers. We controlled for other demographic factors that have documented associations with custody outcomes, including mothers’ and fathers’ education level (1 = less than high school, 9 = doctoral degree), race (0 = White, 1 = not White), and gross monthly income (measured continuously).

Court Records Data

Sample

Using a case review approach, we collected court records data to examine factors associated with custody outcomes in divorces with and without IPV (based on self-reports collected in the prior study). We collected these data from publicly available divorce, civil protective order, and criminal cases associated with the petitioner and respondent named in the divorce filing. Project staff reviewed the entire case files, which typically contain copies of all documents in chronological order.

Divorce Records

The 190 divorce cases correspond to the 190 mothers from whom we also had self-report data.

Civil Protective Order Records

Civil protective orders filed between the parents named in the divorce were identified on the Circuit Clerk’s website using full names. We included only protective orders filed by and granted to mothers against fathers at any time during the relationship or after separation but prior to final custody determinations. These criteria resulted in a sample of 37 mothers who filed and were granted a protective order.

Criminal Records

Criminal records of fathers were identified using full names on the Circuit Clerk’s website. We did not have access to criminal records outside the focal county because systems across counties and states are independent. We included only fathers’ criminal records because these data were used to identify whether mothers’ self-reports of IPV were documented in court records and whether this documentation influenced custody outcomes. We only included charges relevant to IPV (i.e., domestic battery, violating a protective order, interfering with IPV reporting) and excluded those filed after custody determinations. Using these criteria, we identified 34 fathers with criminal records related to IPV at some point during the relationship or after separation but before final custody decisions.

Procedure

All court records data were located using a publicly available online database through the Circuit Clerk’s office. Project staff used structured data extraction forms to collect relevant data from the online records as well as from publicly available hard copy files stored at the Circuit Clerk’s office. Training was conducted to ensure the reliability of data extraction, using actual records to ensure that data extraction and coding were consistent across research staff. Any inconsistencies were resolved through additional training and/or revision of the coding scheme to reflect nuances in the case records. All data corresponding to the extraction form were entered into statistical software by two researchers. The two entries were later compared by a third researcher who eliminated any discrepancies by returning to the original records. At the time of the original data extraction, 22 of the 190 divorce cases did not have a final custody order so our team revisited those cases two years after the initial extraction timeline. To maintain consistency across cases, we conducted a recheck process for all divorce and protective order cases. The recheck process involved identifying any new court activity (e.g., additional filing of petitions or motions after divorce) since the initial extraction and updating the extraction form and data entries for analyses. A total of 10% (n = 19) of divorce cases were updated accordingly, and three civil protective order cases were added to the final sample. All updated or newly added cases went through the same process of extraction, entry, and cross-comparison to ensure reliability and validity. Data collection took approximately 3,720 person hours in total.

Variables and Measurement

Custody Decisions

We extracted data on custody decisions from the final divorce decree contained in the divorce records. Specifically, we collected information on both legal and physical custody decisions and whether custody was joint, sole, or split, reflecting the legal terminology of the state where the data were collected. Legal custody referred to a parent’s right to make significant decisions (e.g., education, health, religion, and extra-curricular activities) for the minor child. Physical custody referred to a parent’s right to have the minor child under their physical care. In this paper, we present findings for physical custody, but all analyses were replicated for legal custody and are available upon request. Custody was measured dichotomously (0 = mother sole custody; 1 = joint custody). Cases in which fathers were awarded sole custody (n = 5) or custody was split (i.e., different arrangements for different children; n = 2) were excluded due to small sample size. An additional 22 cases were excluded because they had no final custody decision.

Court-Documented IPV

Court-documented IPV was measured dichotomously (i.e., documented/not documented) based on whether and where it was documented. IPV was considered documented outside the divorce case if fathers had criminal records or civil protective orders against them. Specifically, we coded whether there was a (a) court-granted civil protective order filed by the mother against the father before final custody decisions, and/or (b) prior criminal charges against the father that were relevant to IPV (i.e., domestic battery, violating a protective order, interfering with domestic violence reporting) before final custody decisions. Each of these indicators was coded dichotomously (0 = not documented; 1 = documented).

IPV was coded as documented in the divorce case if the divorce case file contained a reference to any indication of IPV perpetrated by the father either raised by a party or noted by the court. The primary forms of documentation in the divorce case file were: (a) mother filing for divorce under grounds of physical and/or mental cruelty, (b) reference to a court-granted civil protective order against the father, (c) reference to a father’s IPV-related criminal arrest or charge, and/or (d) a direct reference to any type of physical violence perpetrated by the father against the mother. Each of these indicators was coded dichotomously (0 = not documented; 1 = documented). For the purposes of the logistic regression analyses (described below) we also computed a single dichotomous variable combining all the above indicators (0 = no documentation; 1 = any form of documentation).

Contested Cases

Cases were coded as contested (n = 70) if the respondent’s custody request differed from the petitioner’s request. Cases were coded as uncontested (n = 80) if parties agreed or the respondent did not challenge the petitioner’s custody request. We coded as missing cases where the respondent was in default (i.e., did not make an appearance and could not be located; n = 14), did not file a response to the petition (n = 19), or if the data were missing (n = 12).

Third-Party Intervention

For the purposes of this study, we coded cases as involving third-party intervention if they had a bench trial or any outside intervention that may have informed custody decisions. Specifically, we coded dichotomously (0 = no; 1 = yes) whether the case went to trial (n = 7) or whether the court ordered a custody evaluation (n = 5), mediation (n = 15), or appointed a guardian ad litem (n = 14) for the resolution of custody disputes. Cases in which any of these four variables were coded “yes,” were labeled third-party intervention (n = 24).

Plan of Analysis

We used a number of different analytic strategies depending upon the nature and type of variables included in each research question using SPSS Version 27. We tested our first research question (1a and 1b) about whether self-reported IPV was documented in civil protective order, criminal, or divorce court records and whether this differed by IPV type using nonparametric chi-square tests (and accompanying Phi coefficients), applying the Yates correction in analyses that did not have adequate sample size per cell (n = 10). To test our second and third research questions examining the association between self-reported IPV types or court documented IPV and custody outcomes (and the moderating roles of contested custody and third-party intervention), we employed binary logistic regression analyses. To handle missing data, we first performed Little’s MCAR test to determine whether our data were missing completely at random. The test was not significant (χ2 [161] = 120.196; p = 0.99) indicating that our data were missing completely at random. Therefore we used multiple imputation to handle missing data in the logistic regression analyses. Dependent variables were not imputed so our final analytic sample size was 161 due to missing data on custody outcomes. Analyses were interpreted by converting regression estimates into odds ratios (by means of exponentiation) that yielded the likelihood of sole (vs. joint) custody as a function of each factor. We used Wald tests to determine the statistical significance of each variable in the logistic regression models. We regressed the custody outcome on self-reported IPV group, whether IPV was documented (using the summary variable) as well as the interactions testing differences by whether custody was contested or whether there was third-party intervention. We ran these models in three steps. In the first step, we entered demographic covariates (i.e., marital length, race, income, and education of both the mother and the father). In the second step, we entered the main effects of IPV type, documentation of IPV, whether the case was contested, and whether there was third-party intervention. In the final step we added the interactions of whether the case was contested, and whether there was third-party intervention with IPV type and court documentation.

Results

To test whether mothers’ self-reports of IPV by type were documented in civil protective order or criminal records (RQ1a), we cross-tabulated self-reports of IPV (no violence, SCV, or CCV) with both orders of protection (presence vs. absence) and father’s criminal record (presence vs. absence). Approximately 45% (n = 85) reported no violence, 30% were categorized as having experienced SCV (n = 57), and 25% (n = 48) were categorized as having experienced CCV. A total of 37 mothers (19.4%) were granted civil orders of protection from their partners. Among fathers, 34 (17.4%) had criminal records with charges related to IPV. The majority of these fathers had a single criminal record although the number of records ranged from 1–9. Chi-square analyses demonstrated a significant association between IPV type and the presence of a protective order (χ2 [1, N = 190] = 13.98, p = 0.001; φ = 0.27). Among mothers who self-reported no violence in marriage, 7% had been granted a protective order whereas 14% of mothers who reported SCV and 31% of mothers who reported CCV had been granted a protective order. There was also a significant association between IPV type and having a partner with a criminal record (χ2 [1, N = 190] = 14.00, p = 0.001; φ = 0.27). Among mothers who reported no violence in marriage, 12% had a partner with a criminal record whereas 28% of mothers who reported SCV and 40% of mothers who reported CCV had a partner with a criminal record.

Research question 1b asked whether self-reports of IPV were documented in divorce cases. Direct references to IPV in divorce records were found in only 8 of the 190 cases we reviewed (seven of which were in cases in which the mother also self-reported IPV; three were classified as SCV and four were classified as CCV). Given the small percentage of direct references to IPV in divorce records, we examined grounds for divorce (e.g., whether mental or physical cruelty was used instead of irreconcilable differences) and whether previous protective orders or criminal records were mentioned in the divorce case. Original and final grounds for divorce were not always the same due to settlement terms and negotiations throughout the divorce process. Only 1% of mothers filed for divorce under physical cruelty grounds, and none used physical cruelty as the final grounds. Mental cruelty was used as the original grounds in 52.3% of divorce cases but in only 24.6% of cases as the final grounds. Irreconcilable differences was used as original grounds in 71.3% of divorce cases and in 71.8% of cases as final grounds. (The total percentage of original grounds exceeds 100% because 24.6% of mothers originally filed under both mental cruelty and irreconcilable differences grounds, most of whom dropped the mental cruelty grounds during the divorce process). Roughly half (n = 18) of the 37 protective orders granted against fathers were mentioned in the divorce case. Mentions of fathers’ criminal records showed up in only four of the 34 divorce cases in which the father had a previous criminal record related to IPV.

Chi-square tests showed that self-reported IPV was significantly associated with filing for divorce under mental cruelty grounds (χ2 [1, N = 190] = 11.43, p = 0.003; φ = 0.25). In terms of percentages, 41% of women who reported no violence, 58% who reported SCV, and 71% who reported CCV used mental cruelty as their original grounds for divorce. There was also a significant association between self-reported IPV and mental cruelty as the final grounds for divorce (χ2 [1, N = 190] = 17.60, p < 0.001; φ = 0.32). Roughly 15% of women who reported no violence, 30% who reported SCV, and 100% who reported CCV used mental cruelty as their final grounds for divorce. There was no significant association between self-reports of IPV and irreconcilable differences as original grounds (χ2 [1, N = 190] = 4.62, p = 0.10; φ = 0.16). There was, however, a significant association between self-reported IPV and using irreconcilable differences as final grounds (χ2 [1, N = 190] = 8.28, p = 0.016; φ = 0.22). Nearly 87% of women who reported no violence, 76% who reported SCV, and 64% who reported CCV used irreconcilable differences as final grounds for divorce. Mothers’ self-reports of IPV were not significantly associated with documentation of a protective order (χ2 [1, N = 36] = 1.23, p = 0.54; φ = 0.19) or a father’s criminal record (χ2 [1, N = 45] = 3.57, p = 0.17; φ = 0.28) in the divorce case.

Results from the binary logistic regression testing research questions 2 and 3 are presented in Table 1. The full model accounted for 28% of the variability in custody outcomes. We were first interested in whether self-reported IPV types or court documented IPV were related to custody outcomes. Mothers who experienced SCV (OR = 0.25, p < 0.001) or CCV (OR = 0.42, p < 0.01) were significantly more likely to receive sole custody as compared to women who reported no violence. Although not significant in the second model prior to entering the interactions, documentation of IPV in the divorce case was also associated with a higher likelihood of receiving sole custody (OR = 0.65, p < 0.05) in the final model.

Table 1 Regression of Custody Outcomes on Type, Contested Custody, Third-party Intervention, and Demographic Factors
Full size table

We were then interested in whether the association between self-reported or documented IPV was moderated by whether the case was contested or whether there was third-party intervention. There was a significant main effect such that when cases were contested mothers were more likely to be granted sole custody (OR = 0.29, p < 0.001). The effect for mothers who self-reported SCV (vs. no violence) was qualified by a significant interaction with whether custody was contested (OR = 2.80, p < 0.01). As shown in Fig. 1, mothers who self-reported experiencing SCV were equally likely to get sole custody when outcomes were contested (OR = 0.08) or uncontested (OR = 0.12). Mothers who reported no violence (vs. SCV), however, were more likely to get sole custody when custody was contested (OR = 0.17) versus uncontested (OR = 0.68).

Fig. 1
figure 1

Interaction between type of IPV (SCV vs. no violence) and contested custody predicting custody outcome

Full size image

The main effect for mothers who self-reported CCV (vs. no violence) was qualified by significant interactions with contested custody (OR = 3.02, p < 0.01) and whether the case involved third-party intervention (OR = 0.17, p < 0.01). As shown in Fig. 2, mothers who self-reported CCV were equally likely to get sole custody when outcomes were contested (OR = 0.11) or uncontested (OR = 0.09). Mothers who reported no violence (vs. CCV), however, were more likely to get sole custody when custody was contested (OR = 0.22) versus uncontested (OR = 0.53). In addition, mothers who self-reported CCV were more likely to get sole custody when the case involved third-party intervention (OR = 0.09) and more likely to get joint custody when it did not (OR = 1.31). Mothers who reported no violence were more likely to get sole custody when the case involved third-party intervention (OR = 0.22) versus when it did not (OR = 0.53; see Fig. 3). Taken together, we found that third-party intervention and contesting custody moderated the association between self-reported IPV and custody outcomes depending upon the type of IPV.

Fig. 2
figure 2

Interaction between type of IPV (CCV vs. no violence) and contested custody predicting custody outcome

Full size image
Fig. 3
figure 3

Interaction between type of violence (CCV vs. no violence) and third-party intervention predicting custody outcome

Full size image

Although not part of our central research questions, several covariates were significantly associated with custody outcomes. Fathers who were white versus nonwhite (OR = 2.09, p < 0.001) and had more education (OR = 1.28, p < 0.001) had increased odds of joint custody. Mothers who were white versus nonwhite (OR = 0.57, p < 0.001) were significantly more likely to be granted sole custody. Mothers reporting higher income (OR = 1.18, p < 0.001) were significantly more likely to be granted joint custody. Moreover, we also explored the length of time from divorce filing to final judgement for mothers who requested sole versus joint custody. Our data showed that mothers who received sole custody (Mean = 32.30; SD = 40.85 months) spent 10.15 months longer in the divorce process than mothers who received joint custody (Mean = 22.15; SD = 24.14 months), t (162) = 2.15, p = 0.03.

Discussion

IPV is common among divorces requiring judicial intervention, but most divorcing parents settle custody agreements out of court (Beck et al., 2009). Our findings demonstrate that many of these cases also involve IPV, which can go undetected or unaddressed. However, not all divorcing parents with past violence present the same dynamics and risks that warrant documentation or consideration in custody determinations. Thus, we examined whether custody outcomes differed by IPV type, documentation, whether the case was contested, and whether there was third-party intervention. Overall, the mothers in our sample who self-reported IPV were more likely to get sole custody than those who reported no violence, which differs from past findings documenting minimization of IPV when detected in family courts (e.g., Kernic et al., 2005). However, our study is unique in that we obtained abused and nonabused mothers’ self-report data from a general sample early in the divorce process and then related court records data in the years following. Using a general (vs. clinical or help-seeking) sample as well as both self-reports and administrative records, we were able to obtain data on a range of experiences with IPV and the divorce process. This provided us with a more detailed picture of when and how self-reported IPV “shows up” in divorce cases and its association with custody outcomes.

Our first research question asked whether divorcing mothers’ self-reports of IPV were documented in court records, either outside of or within the divorce case. We found that self-reports of IPV were generally not documented in divorce cases. Direct references to IPV in the divorce case file (e.g., witness testimony) almost never happened. Of the 104 women who self-reported IPV during marriage, only 7 had a direct reference to abuse in their case file. This may mean inclusion of IPV in divorce records relies on other forms of documentation to be corroborated and noted, such as protective orders or criminal records. There was evidence that self-reports of IPV, especially CCV, were associated with having sought civil (i.e., protective orders) or other legal help (i.e., arrests, criminal charges). However, only half (18 of 37) of protective orders were documented in divorce case files and even fewer (4 of 34) of the criminal records. There are various reasons why this might be the case. Mothers may not have disclosed IPV during divorce to avoid protracted disputes or retaliation (Hardesty & Ganong, 2006). Indeed, abusers have been found to manipulate proceedings by distorting facts to disparage a victim’s character and portray them as an unfit parent (Elizabeth, 2017; Watson & Ancis, 2013). Inadequate training of legal practitioners and biases that mothers make false allegations to gain advantage can work against mothers (Saunders et al., 2011; Silberg & Dallam, 2019). Furthermore, in the absence of trauma-informed practices, legal practitioners can misinterpret abused mothers’ presentation in court as hostile or uncooperative (e.g., Hardesty, Crossman, et al., 2015; Hardesty, Hans, et al., 2015; Stark et al., 2019). Thus, attorneys may also discourage mothers from alleging IPV because doing so could negatively affect custody outcomes (Meier & Dickson, 2017). It is also possible mothers disclosed IPV but it was not documented in the case file because it lacked corroboration or was deemed irrelevant.

In light of few direct references to IPV, an examination of the original grounds for divorce was revealing. Physical cruelty grounds were almost never used even though they offer a clear indicator of the presence of IPV. Mental cruelty grounds were more common, especially among mothers who self-reported CCV. Coercive control or psychological abuse may be more salient motivations for seeking divorce than physical violence (Jacobson & Gottman, 1998). Importantly, when mental cruelty grounds were used originally, they were reverted to irreconcilable differences by the final judgment almost half the time, although less often for mothers who self-reported IPV, especially CCV. The exact reasons this happens are unknown, but divorce grounds are likely a point of negotiation in the divorce process. An initial filing for divorce on the grounds of mental cruelty could reflect mothers’ sense of empowerment, which may dissipate when family court processes are invalidating and harmful (Gutowski & Goodman, 2020; Khaw et al., 2021). However, of those who experienced CCV, 100% retained mental cruelty as final grounds. Access to no-fault divorces may render some of the specifics obsolete, but the broader pattern remains important. Filing for divorce under mental cruelty grounds may be the closest proxy to IPV disclosure at the onset of the divorce process in a general sample and should signal the need to consider IPV. This also raises concerns about removing divorce grounds in favor of no-fault divorces if that is one of the few indicators of IPV, especially CCV, in the absence of routine screening for IPV in family courts (Stark et al., 2019).

Our second research question asked whether IPV was associated with custody outcomes. Consistent with Davis et al. (2011), we conceptualized custody outcomes as safer or more protective when they restricted physical access, which we operationalized as sole versus joint physical custody. When mothers who self-reported IPV used mental cruelty as their final grounds, they were more likely to be awarded sole custody. Thus, when IPV “shows up” as grounds for divorce and is sustained in the final judgment, it matters for custody outcomes; this is further supported by the main effect of documentation in our regression model. Nonetheless, even when IPV is not documented in any clear way in divorce cases, self-reported IPV still relates to greater odds of sole custody, but the mechanisms are not clear. Some IPV disclosures may not be documented in the divorce case file (e.g., confidential mediation reports) but influence outcomes nonetheless, as speculated by Kernic et al. (2005). Other issues pertinent to custody that can co-occur with IPV, such as substance abuse, mental health issues, or child abuse (Smith Stover & Spink, 2012), also likely influence outcomes even when IPV is not known to the courts.

Our final research question asked if contested custody or third-party intervention moderated associations between IPV and custody outcomes. We found that contesting custody and third-party intervention moderated the effects of IPV on custody outcomes depending on the type of IPV. When mothers self-report IPV, especially CCV, they were more likely than other mothers to contest custody, but contesting custody did not change their likelihood of receiving sole custody. Instead, contesting custody increased the likelihood of sole custody only for mothers who reported no violence. There are significant barriers to contesting custody (e.g., financial and time constraints); thus, there are likely unique circumstances for mothers with no violence who contest that increase their likelihood of sole custody but were not the focus of the current study.

For mothers who experienced CCV, third-party intervention maximized their likelihood of obtaining sole custody. In contrast, third-party intervention seemed to provide no additional benefit to mothers who experience SCV. This suggests that third-party intervention is a useful tool for detecting risks associated with CCV or for empowering these mothers to negotiate safer agreements. Furthermore, custody evaluation reports or involvement of guardians ad litem may provide courts with information about abuse dynamics that pose a risk after separation and warrant safer outcomes (Stark et al., 2019). This is important because courts disincentivize litigation by encouraging parents to settle agreements on their own. By discouraging formal legal processes for securing protective agreements or denying equal access to the means to litigate, abused women and their children, especially those with fewer resources, could remain at greater risk. At the same time, this finding should be interpreted with caution as third-party interventions are not always appropriate or helpful. For example, divorce mediation in cases with IPV, especially CCV, may not be appropriate without comprehensive screening and extensive training for mediators (e.g., Applegate et al., 2021). Likewise, studies show that custody evaluators may minimize IPV and suggest unsafe parenting arrangements (Haselschwerdt et al., 2011; Saunders et al., 2012). Moreover, we do not know if the current finding is driven by any one type of intervention or if third-party intervention might signal the presence of other risk factors (e.g., child abuse) that help explain its impact for mothers who experienced CCV.

Our data also showed that mothers who received sole custody spent 10.15 months longer in the divorce process than mothers who received joint custody. This finding supports the notion that mothers may be discouraged both by the legal process itself and by personal constraints to avoid lengthy and expensive proceedings. Sole custody awards take more time and financial resources and require prolonged engagement with abusive former partners. As Gutowski and Goodman (2020) found, even when mothers obtain the outcomes they want in family courts, they face numerous obstacles to being heard and can feel victimized by the process itself.

Seeking safety and protection from abuse often involves difficult tradeoffs (Thomas et al., 2015). Future studies must explore how such tradeoffs disproportionally affect or prevent some women from receiving protection. For example, in the current study, demographic characteristics (included as covariates) such as being white and educated helped men obtain joint custody yet being white with lower income helped women obtain sole custody. Race, class, and immigration status, among others, influence women’s access and ability to pursue and secure protective agreements (Maldonado, 2017); this remains a crucial next step. Future research is also needed to identify the myriad ways mothers and children remain accessible and at risk via visitation arrangements with abusers. National data on the rate at which IPV perpetrators are granted sole or joint physical custody are not available. However, even when abusers are denied physical custody due to risks associated with IPV, courts often still award unsupervised visitation, and fail to incorporate provisions to ensure safety (Davis et al., 2011; Holtzworth-Munroe et al., 2021).

Limitations

Study findings should be considered in the context of several limitations. Publicly available records present unique challenges, including the potential for incomplete or missing information, errors in interpreting information, and variation in depth and quality (Raub et al., 2013). We mitigated the potential for these issues by using a structured extraction guide to collect the same information across cases, consulting with legal scholars to interpret case information, and returning to the court records for rechecks to fill in any missing or incomplete information. A limitation of the recheck process, however, was that we resolved any discrepancies between coders and were unable to report interrater reliability. Publicly available court records also often lack information about mental health diagnoses for parents or children and information on parental substance abuse, both of which can influence custody outcomes (see Raub et al., 2013). We also did not collect information on child abuse because it is not part of public record.

A major strength of our study – self-report data – also has some limitations. Mothers self-reported their experiences with IPV; however, we did not collect self-report data on their motivations for seeking particular custody arrangements or for contesting custody. We also did not have self-report data from fathers. Although both women and men experience IPV, women are more likely to experience severe violence in the context of coercive control (Hardesty & Ogolsky, 2020), which was a primary focus of the self-report study. Future studies should consider the perspectives and experiences of both parents during marriage and divorce as well as prior to marriage, as patterns of IPV often begin during dating relationships. Furthermore, all mothers in the sample were married to/divorcing men. At the time the mothers in our sample married, same sex marriage was not legal. Thus, none of the divorce cases included same sex marriages. Further work is needed to understand divorce processes among LGBTQ + couples. Our sample was also primarily white, which further limits generalizability. Moreover, most of these cases were opened nearly a decade prior to writing this article during which time some of laws for handling such cases have changed. As the study ages, we encourage careful consideration of the current legal context to understand how our findings apply.

Although our approach to collecting self-report and administrative data provided us with in-depth and complex data about individual cases, we caution generalizing our results from one county to other populations and contexts. Using data from one county allowed us to control for legal and social effects that can influence custody decisions and provided us with access to the whole “population” of divorces in the given years, but future studies should replicate our findings in a larger, more diverse sample involving multiple states. This is especially true of criminal records given that records in one county or state do not show up in other county databases nor can we be certain that judges refer to these cases directly. Moreover, although IPV scholars generally agree there are different types or contexts of IPV, approaches to conceptualizing and measuring IPV types continue to be the subject of debate (see Hardesty & Ogolsky, 2020). Thus, although our findings underscore important experiences with divorce and custody decisions based on violence experiences, we lack systematic approaches within family courts to identifying and addressing such variations. We continue to need rigorous studies that expand our understanding of IPV types and their implications for family court settings.

Conclusion

Our findings concur with existing literature demonstrating that family courts are often not explicitly made aware of a history of IPV, despite documentation in other civil and criminal cases, which underscores important practical implications. Routine screening for IPV by professionals in family court is needed but rarely required. Reliance on voluntary self-disclosure instead of formal documentation can result in courts not being made aware of IPV. Our findings support Stark et al.’s (2019) recommendations for routine screening using validated tools to identify IPV coupled with a review of civil and criminal court records. When IPV is detected, comprehensive assessments of risk, including high levels of coercive control, should follow (see Stark et al., 2019, for detailed recommendations). Routine screening also must be paired with mandatory training of judges and other family court practitioners. A total of 39 states including the state from which these data were collected do not currently require IPV training for judges (Stark et al., 2019). Family courts must be able to recognize and understand the dynamics of IPV, its effects on victims and children, and potential risks after separation to ensure safe and protective custody outcomes. Thus, legal professionals should encourage rather than discourage the use of the legal system in cases with IPV to promote the health and safety of victims and their children.