Gender and child custody outcomes across 16 years of judicial decisions regarding abuse and parental alienation
Jennifer Harman, Christine Giancarlo, Demosthenes Lorandos & Brian Ludmer.
Highlights
A mother found to have alienated her child(ren) was not more likely to get a decrease in parenting time, but was more likely to lose her case and lose custody than a father who was found to have alienated his children but the effect sizes were small and not meaningfully significant.
25 “abusive” alienated parents were mothers (71.4% of 35), and 10 were fathers (28.6% of 35), so the presumption that “abusive” alienated parents are mostly fathers is not reflected in these data.
Alienating mothers’ claims of abuse against known “abusive” alienated fathers were not being discredited more often than they were for alienating fathers.ze
Alienating mothers were not more likely than chance to lose custody than alienating fathers when the third party who found PA was a custody assessor or GAL.
Alienating fathers made fewer allegations of abuse against alienated mothers (n = 112), but the percentage of substantiated allegations was similar (7.1%) to those of alienating mothers.
Gender of the alienating parent, number of unsubstantiated allegations of abuse, and an interaction of the variables were not predictors of whether an alienated parent lost custody of their children.
Resymé på norsk av Eivind Meland:
I Norge og internasjonalt er det fortsatt sterke krefter som fremholder at å bruke begrepet foreldrefremmedgjøring tjener som påskudd til å undergrave kvinners rett til å beskytte seg selv og sine barn mot vold. En fersk studie med rettsdata fra Canada viser en virkelighet som er annerledes.
https://www.sciencedirect.com/science/article/pii/S0190740923003833
Hovedresultatene i studien var:
1. Bare 10,9% av de 654 anklagene om vold og overgrep mot fremmedgjorte foreldre ble dokumentert og underbygget. Dette betyr at nesten 90% av overgrepspåstandene mot foreldre utsatt for fremmedgjøring ble dokumentert som usanne eller som ubekreftede. Slike påstander har store konsekvenser for dem som urettmessig blir anklaget, konsekvenser som ikke bør undervurderes. Til dags dato er politi og barneverns- og familievernsarbeidere uvitende om dette omfattende lovbruddet. Å forutsette at alle påstander er sanne er søkt, spesielt når vi vet at mange av dem (30% i denne studien) ble gjort direkte etter en rettsavgjørelse eller forslag til avgjørelse om for eksempel å forandre bosted eller omsorg. Slike påstander overser juridiske og administrative aggresjonstaktikker som brukes av noen fremmedgjørende foreldre for å få og opprettholde makt over sine barn og den andre forelderen (Harman et al., 2018).
2. Tjuefem av de anklagede fremmedgjorte foreldre var mødre og 10 var fedre. Antagelsen om at "voldelige" fremmedgjorte foreldre for det meste er fedre, gjenspeiles ikke i disse dataene. Heller ikke kan det dokumenteres, slik Alsalem og andre påstår, at slike saker er vanlige i rettspleien (Alsalem, 2023). Basisraten for denne typen saker var svært lav (7,0 %). Harman & Lorandos (2021) fant også en lav basisrate på 7.9% i 967 amerikanske ankesaker.
3. Alsalem og andres påstander om at mødres anklager om overgrep diskrediteres oftere enn fedres påstander om overgrep, kunne verken Harman og Lorandos (2021) eller denne studien gjenfinne og dokumentere. Vi kan derved fastslå at slike påstander er ideologisk fargede og urettmessige.
4. Gitt at to replikasjonsstudier ikke har funnet støtte for at det er utbredt kjønnsskjevhet i domstolenes behandling av overgrepspåstander, er det grunn til å være bekymret for at kjønnsspesifikke oppfatninger om familievold representerer et rettssikkerhets problem. Et kjønnsskjevt perspektiv på familievold mangler vitenskapelig grunnlag.
Konklusjonene i artikkelen, og dokumentasjonen fra andre artikler er også viktige og nyttige for påvirkningsarbeid som drives av PASG-medlemmer og andre. Forfatterne av denne artikkelen er ikke aktivister, men forskere. Likevel bør godt og saklig påvirkningsarbeid være basert på vitenskapelig dokumentasjon. Vi må engasjere oss saklig for at slik informasjon blir kjent for organisasjoner, institusjoner og enkeltpersoner som trenger å lære om dette.
Referanser
Harman, J. J., Kruk, E., & Hines, D. A. (2018, Dec). Parental alienating behaviors: An unacknowledged form of family violence. Psychol Bull, 144(12), 1275-1299. https://doi.org/10.1037/bul0000175
Abstract
There have been legislative efforts to control how child custody decisions are handled in family courts where allegations of abuse and of parental alienation (PA) are levied. The “findings” reported to support such legislation have been based on one unreviewed study with identified methodological issues (Harman & Lorandos, 2021). We tested six pre-registered hypotheses to determine whether there is empirical support for the “research findings” used to support these laws. Five-hundred PA cases were sequentially selected from 4,889 Canadian trial court decisions. Independent coders who were blind to the hypotheses coded all cases for details about custody and allegations of abuse. We failed to find support for the “findings” that have been used to support legislative changes. For example, this study focused only upon cases where PA was determined to actually have occurred in at least one of the children in the family. It differs from Harman & Lorandos (2021) in that this study found that alienating mothers’ claims of abuse against known “abusive” alienated fathers were not being discredited more often than they were for alienating fathers. The negative impact of failing to base legislation on a comprehensive consideration of the full scope of scientific evidence available (e.g., Kayden’s Law in the reauthorized Violence Against Women’s Act, 2022) is discussed.
1. Introduction
Trial courts in English-speaking jurisdictions have been dealing with parents alienating their children from their other parent for more than two hundred years (Lorandos, 2020a, Joshi, 2021). Despite the qualitative and quantitative scientific evidence that has been accumulating on parental alienation (PA) over the last 77 years (Harman, Warshak, et al., 2022), some scholars have claimed there is controversy regarding the scientific validity and reliability of the PA construct (Meier, 2013, Mercer and Drew, 2022, Saini et al., 2016). A number of nonscientists and parent advocates have also insisted across various non-scholarly publications that PA is not admissible as a scientifically accepted construct in court under Frye (U.S.), Daubert (U. S.), and Mohan (Canada) standards; for examples see Bruch, 2002, Dalton et al., 2006, Hoult, 2006; and Milchman, 2019. The reiteration of such “conclusions” conflates the actual scientific findings of peer-reviewed studies, 40% of which have been generated since 2016 (Harman, Warshak, Lorandos & Florian (2022). Yet, a review of over 3,500 American appellate cases tested this “inadmissibility hypothesis,” finding that in 1,181 appellate decisions published through 2018, PA was determined to be “material to the proceedings, probative of important facts, relevant to the court’s deliberations, admissible, and worthy of discussion” (Lorandos, 2020b, p. 3).
Another recent debate regarding PA is related to the strategic use of abuse allegations by a parent to alienate children from their other parent. The new claim is that courts have failed families because they consistently “discredit” mothers’ (e.g., Death et al., 2019, Meier et al., 2019, Sheehy and Boyd, 2020) or parents’ (Webb et al., 2021) claims of abuse, and that trial court judges have struggled with claims that abuse allegations should always be believed when made by a parent who asserts they are trying to “protect” their children (Dallam & Silberg, 2016). Underlying these arguments are assumptions that there are pervasive gender biases in the court that are harming mothers and children (e.g., Zaccour, 2022), as reflected in claims that PA was only invented as a legal defense for abusive fathers to evade abuse allegations made against them (e.g., Meier et al., 2019). And there is no mention of fathers as potential victims of abuse in either Zaccour (2022) or Meier et al. (2019). Written reports and public presentations by individuals who assert these arguments have influenced law and public policy. For example, in the March 2022 reauthorization of the Violence Against Women Act, in TITLE XV—KEEPING CHILDREN SAFE FROM FAMILY VIOLENCE, also called ‘‘Kayden’s Law’’, U.S. Senate Bill S 3623, Section 1502 described “FINDINGS” which contained three illustrative paragraphs:
PP (6). Empirical research indicates that courts regularly discount allegations of child physical and sexual abuse when those allegations are raised in child custody cases. Courts believed less than 1⁄4 of claims that a father has committed child physical or sexual abuse. With respect to cases in which an allegedly abusive parent claimed the mother ‘‘alienated’’ the child, courts believed only 1 out of 51 claims of sexual molestation by a father.
PP (7). Empirical research shows that alleged or known abusive parents are often granted custody or unprotected parenting time by courts. Approximately 1⁄3 of parents alleged to have committed child abuse took primary custody from the protective parent reporting the abuse, placing children at ongoing risk.
PP (9). Scientifically unsound theories that treat abuse allegations of mothers as likely false attempts to undermine fathers are frequently applied in family court to minimize or deny reports of abuse of parents and children. (Violence Against Women’s Act, 2022, pp. 306–307).
The conclusory “findings” in these three paragraphs can be traced directly to Meier et al. (2019), an unreviewed study without a literature review that was published in an internet-archived university paper series. Several scientifically peer-reviewed studies, however, have failed to find support for the findings reported in the Meier et al. (2019) paper. For example,Bala et al., 2010 did not find statistically significant gender differences in loss of custody between alienating mothers and fathers in Canadian court decisions. Likewise, a randomly selected sample of divorcing mothers in a U.S. county found it more likely for mothers to receive sole custody of their children if they made an allegation of abuse against the father than if no allegations were made (Ogolsky et al., 2022). Another study of over 1,000 U.S. appellate court cases did not identify any case that lent support to the statement that “courts regularly discount allegations of child physical and sexual abuse when those allegations are raised in child custody cases” (Violence Against Women Act, 2022, p. 306), or that alleged or known abusive parents were often granted custody or unprotected parenting time by courts (Lorandos, 2020b).
Harman & Lorandos (2021) conducted a thorough review of the Meier et al. (2019) study, and noted over 30 methodological and statistical issues, many of which were due to a lack of details about how the authors conducted their study. Although Harman & Lorandos (2021) were unable to assess the scientific merits of the Meier et al. (2019) study due this lack of transparency, they tested six pre-registered hypotheses based on the conclusions reported in Meier et al. (2019) using open science research practices.1 They failed to replicate any of Meier et al.’s conclusions. There were no statistically significant gender differences in whether alienating parents lost custody or parenting time, and contrary to what Meier et al. (2019) reported, parents who had findings of abuse against them were not likely to get custody of their children when they claimed to have been alienated by the other parent. What is of great concern is that Meier et al.’s (2019) conclusions and the empirically unsupported opinions of other critics of PA scholarship described above are foundational to the “findings” in Kayden’s Law (Violence Against Women’s Act, 2022, p. 304), while other scientific evidence is entirely omitted.
1.1. The current study
Data driven science evolves with scientists of good will working to question and replicate data. This study is a second attempt to test the assertions made in the Meier et al. (2019) report that lacked support in Lorandos (2020b) and Harman & Lorandos (2021) but this time using trial level cases. The samples of PA cases used in all three prior studies were drawn from U.S. appellate cases. However, appellate cases as a sample source pose a limitation because appellate reports typically provide less detailed insight into the facts of the case (appeals are not “re-trials”) and deference is frequently afforded to the trier of fact. In addition, appellate-level case samples may not be typical of trial-level decisions because the trial decision might be so well-grounded in the evidence that no appeal is made, or the parties cannot afford the cost of an appeal. Another conundrum is that U.S. Court systems do not require trial level court opinions to be published, so collecting a generalizable sample of trial court opinions that reflect what occurs at that level across the U.S. is not possible. To find readily available trial court opinions we turned to Canada, where a large number of court decisions for such cases are publicly available on free public (e.g., CanLII) and commercially licensed databases (e.g., WestLaw Canada, Lexis Nexis).
We pre-registered thirteen separate hypotheses and analytic models on OSF (https://osf.io/3ngqm/?view_only=f3ebcbfc511548919f10536616b0803a), and the current project tests six of them2:
H1: When a mother is found to be undermining the father’s paternal rights and alienating their child(ren), she is more likely to get a decrease in parenting time, lose custody of her children, and lose her case than a father.
H1a: H1 results will be statistically significant even when the alienated parent is found to be abusive.
H2: When mothers claim intrafamilial abuse in family court and the father is found3 to be alienated from their children by her, her reports of abuse will be determined by the court to be unfounded more often than if the father claimed abuse and the mother was found to have committed PA.
H3: Mothers will have a decrease in parenting time or lose all custody more often than fathers when a guardian ad litem (GAL) or custody evaluator/assessor is involved in the case.
H4: When a mother claims that both child abuse and sexual abuse occurred and one or both were substantiated,4 she is more likely to be penalized by getting a decrease in parenting time or lose all custody than fathers making the same claims.
H5: The greater the number of unsubstantiated allegations of abuse that a mother makes, the more likely it is for the father to have a decrease in parenting time or lose all custody.5
H6: Fathers are more likely to be an alienated parent than mothers.2,6
2. Method
Harman and Lorandos (2021) sampled U.S. appellate court cases where PA was either alleged or found to have occurred by a court-appointed third party (e.g., custody evaluator or assessor), or the court itself determined PA to have occurred based on the evidence presented. The Harman & Lorandos (2021) study sample was purposefully selected to have equal proportions of mothers and fathers who were found or alleged to have alienated their child(ren) from their other parent. The current study differs from the Harman & Lorandos (2021) study in that we were only interested in cases where PA was determined to actually have occurred in at least one of the children in the family (merely alleged cases were not included). We did not purposively select proportionate cases of alienating mothers and fathers—we selected all founded PA cases regardless of the gender of the alienating parent. We also added mental health providers (such as family therapists) as another party who could make this determination due to their close and often extended involvement with the families and the courts. Insofar as any finding by a licensed MHP had to be material, probative, relevant and admissible to find its way into a Court’s written opinion, this analysis was less restrictive than Lorandos, 2020a, Lorandos, 2020b or Harman and Lorandos (2021). These parties did not have to agree with each other—the inclusion criterion was only that at least one of these authorities determined PA to have occurred.
The project was determined to be exempt from Institutional Review Board approval, given that the data are publicly available, published court decisions. On May 12, 2020, the WestlawNext Canada FamilySource database was used to search for cases in which PA was raised as an issue in the case. The search terms alienat! /s mother! Father! Son! Daughter! Parent! Child! were used, and this search resulted in the identification of 4,889 cases. The cases were all downloaded as Word documents into a shared folder where they were screened for inclusion by two research assistants (RAs) who were blind to the study’s hypotheses. Our initial pre-registered research plan entailed the selection of 1,000 cases; however, given the amount of detail provided in each case (fewer missing data than in appellate cases), we determined that 500 cases would provide enough power to statistically test the proposed hypotheses. The RAs read through cases sequentially, starting from the 2020 search date and working back in time. Cases published in French were translated using Google Translate (https://translate.google.com/) by the RAs in this first step to determine whether they were to be included in the database. Trial level cases that involved a determination of PA having occurred were identified and the electronic files were re-labeled with a number assigned to each separate case. Cases were excluded if there were no findings of PA, but did appear in our search (e.g., the judge only cited prior case law regarding the issue, but no finding had been made for the family). Some families had multiple court appearances, and so to prevent the violation of the assumption of independence for our statistical models, only the most recent court decision was selected for inclusion. This process continued until 500 cases were selected (regardless of the gender of the alienating parent), and these cases spanned 16 years between 2004 and 2020.
Eighteen independent RAs, who were blind to the study’s hypotheses thoroughly read all cases (two per case) and coded them using two or more fillable pdf coding forms (available on the OSF project page7). Cases published in French were coded by one English speaking RA through the use of Google Translate, and a second RA who had reading knowledge of or was fluent in French. There were five RAs who worked on the French cases as the second coder. In order to ensure that the RAs did not have prior knowledge of the study’s hypotheses that could inadvertently bias their sample selection or coding, each independently provided a list of hypotheses they believed were being tested after completing their training and before they started coding. Then, after completing their coding tasks, each RA again provided a list of the hypotheses they believed were being tested (these pre-post guesses are also archived on the OSF project page). None of the coders correctly presumed all of the pre-registered hypotheses; there were two coders (of 18) who correctly guessed one (H6) of the six pre-registered hypotheses, but coders were not matched to the same cases as other coders. Therefore, it was unlikely that their correct guess on one of six hypotheses significantly impacted the data that were extracted from each court order.
The number of coding forms completed by RAs for each case depended on whether there were any allegations of abuse (e.g., domestic violence, child abuse)—one separate form was completed for each time an allegation of abuse was made in each case. For example, if a father called CPS and reported that the mother had neglected the children and hit one of them, one form was completed for this one allegation, even though it involved two types of alleged abuse. If, on another day, the father called the police to make the same allegation, this separate action was coded on another form because it occurred at another time. If the court decision noted that two people visited the police to report an incident, this was also coded as one allegation even though there were multiple accusers.
This study involved only objective recording of data reported in each case (no subjective ratings were made at this stage), so interrater reliability was not calculated. Rather, the second author reviewed the forms completed by both coders to identify discrepant fields. If there were discrepancies, the original court order was used to locate the correct information, and the data in the form’s fields were finalized. Two other RAs who were not involved with the initial coding independently entered the data from the forms into an Excel database using data entry numerical codes (see the OSF project page for these details) for analysis purposes. The two independent databases were then compared by the first author to identify inaccurate data entries, which were corrected by referring to the data reported in the final pdf forms for each case.
2.1. Coded variables
2.1.1. Gender of the alienating parental figure
This variable was coded as being the mother, father, both, and “other” (e.g., grandparent). When both parents were cited as being an alienating parent in the case (n = 11), it was because the court found both parents engaging in parental alienating behaviors, which is different than the outcome of parental alienation in the child (Harman et al., 2022). Therefore, “both” was recoded as being mother or father depending on the parent the child was described in the order as having aligned with. Dummy codes were created for this variable such that 1 = mother, 2 = father, and 3 = “other.” There were a few cases where different siblings were alienated from different parents, or it could not be determined from the details reported which parent the children had aligned with. These cases were coded as “other.”.
2.1.2. Physical custody of the children
Physical custody of the child(ren) was recorded based on what was described as the applicable court ordered parenting plan prior to and after the trial or hearing related to the case decision. We recorded whether the mother or father had sole or primary custody, the physical custody was joint or shared,8 it was split (with children in the sole custody of different parents), the children were in foster care or residing with another relative, or “other” if there was no information or the children had become adults.
Change in parenting time. Using the physical custody data, we created a new variable to reflect whether an alienated parent’s original court ordered parenting time prior to the hearing or trial was increased (coded 1), decreased (coded −1), or stayed the same (coded 0). To be coded as a change, the change had to be a 20% or more difference in days of physical custody each month. A change in parenting time of less than 20% was not considered to be significant, as it would just be a change of one or two days a month. A 20% change was also used by Harman & Lorandos (2021), so we used the same cut-off to have consistency in variable operationalization across the two replication studies.
Total loss of physical custody. Parents whose parental time was taken away, or whose parenting time was changed to only supervised or therapeutic visits several hours a week or less were coded as having lost custody as a result of the case decision. Loss of custody in a previous court decision was not coded as a loss of custody related to the current court decision. We created dummy codes for this variable (1 = loss of custody, 0 = no loss of custody) for both the alienating and the alienated parent.
2.1.3. Loss of the case
Loss of the case was coded based on the motion(s) or trial that was at issue in the case. The winner of the adjudication before the court was coded (father, mother, “other”), and sometimes there were multiple winners based on decisions made for multiple proceedings, or it was a case where there was not one winner (coded as “both”). The winner was recorded to indicate who lost the case to test our hypotheses.
2.1.4. Abuse claims against the alienated parent
Every instance that an allegation of abuse was made towards any individual mentioned in the court decision was coded on separate fillable pdf forms (see OSF project page for the forms). Thorough details about each allegation (e.g., dates, type of abuse, victim[s], evidence provided), investigation (e.g., parties who investigated), and outcome (including court outcomes, if applicable) were recorded. Our hypotheses pertained to allegations of abuse made about an alienated parent (not the alienating parent or others), so we restricted our statistical analyses to only those allegations. Forthcoming papers will be examining other allegations of abuse towards the alienating parent and other related parties to test other pre-registered hypotheses.
Types of abuse. The types of abuse allegations that were coded included domestic violence, child physical abuse, child neglect or maltreatment, child sexual abuse, child emotional or psychological abuse, or “other” (e.g., abusing a pet or assaulting a neighbor). Child maltreatment is often used as an umbrella term for other forms of child abuse, so we only used the term if that was what was stated in the order—most orders provided more specific details on the type of abuse mentioned (e.g., physical abuse). Parental alienating behaviors intended to make a child believe the alienated parent never loved them, abandoned them, is unsafe, and/or unfit (e.g., loyalty inducing behaviors, gatekeeping, derogation of the alienated parent) have been considered a form of child psychological abuse by a growing number of scholars (e.g., Harman et al., 2018, Kruk, 2018), but we did not code these behaviors as emotional or psychological abuse for this study because PA was an outcome determined to have happened to the children in the cases. Rather, the case had to mention some other form of psychological or emotional abuse not attributed to PA by the investigators, mental health service providers, or the court.
Third party investigations and outcomes. Some allegations of abuse were only made to the court (e.g., described in a pre-trial motion or through testimony) so these cases’ outcomes were only coded regarding their court involvement (see below). Other allegations were reported as having been investigated by one or multiple parties (e.g., police, special victims’ units, parenting assessment, family therapists, or Child Protection Services [CPS]). All investigative parties and the outcomes of their investigations were recorded using the exact terminology used in the decision. Table four describes the types of abuse found in the data set and Table five describes the agreement of MHPs and other professionals who came to determinations about allegations of abuse. Insofar as this analysis was focused on what was actually happening in the Family Court, any ultimate determination as to the reliability or veracity of an abuse complaint was beyond the scope of this analysis. Investigation outcomes were then coded as being “substantiated” if the abuse allegations reported were substantiated, validated, founded, or the individual was charged with a crime. The outcome was coded as “unsubstantiated” if the investigation findings were unsubstantiated, unsupported, unfounded, the case was dismissed or closed, or the person was not charged by the police. The investigation outcome was coded as “inconclusive” if the investigators stated there was not enough evidence, the evidence had been tainted, or was inconclusive. Outcomes were only coded as “false” if the investigator determined the allegation was false, not truthful, exaggerated, or the accuser/victim admitted to lying or fabricating the allegation. “Unknown” was entered as a code when it was unclear or unknown what the outcome of the investigation was, or the outcome was still pending. Conclusions from multiple investigative parties were recorded separately, and there were not any cases where the conclusions of different investigators were at odds with each other for each allegation—they were all in agreement. In other words, for example, in cases where CPS substantiated abuse and the police also investigated, both were in agreement.
Court involvement and outcomes. If the allegation of abuse was made directly in court, or if the court became involved with determining the veracity of the allegation or guilt of the alienated parent after the investigation, this information was recorded. Details about court involvement were recorded by the RAs based on what was described in the court decision: whether the allegation was adjudicated in family and/or criminal court and what the final determination of the court was after the judge and/or jury reviewed all the presented evidence. Court outcomes were classified as “substantiated” if the person was reported to have been found guilty, pled guilty, or the court made a finding that the allegation was true or most likely true. The outcomes were classified as “unsubstantiated” if the person was determined to be not guilty or was acquitted, the case was dismissed, or the court determined the allegation was false or unsupported. Cases were classified as “unknown” if there were no details provided or the case was still pending. We also coded whether allegations of abuse that were later determined to be unsubstantiated or false by the court or another third party were labeled as being an alienating behavior (e.g., legal/administrative aggression). In other words, we recorded if the court order mentioned that the allegation itself was determined to have been made by the alienating parent in order to gain a custody advantage or harm the alienated parent.
Substantiation status of abuse allegations. Each allegation was then recoded into another variable as “unsubstantiated” if the family and/or criminal court outcome was unsubstantiated, even if an earlier third-party investigation found the allegation against the alienated parent to be substantiated. This new code was chosen to be the final determinant of an allegation’s substantiation because the outcome typically involved consideration of the preceding third-party investigations and/or consideration of all of the evidence from the parties and other sources presented at trial. If there was no court involvement, the investigation outcome was used to determine whether the allegation was substantiated or not. False and inconclusive allegations were reclassified as unsubstantiated. Cases were coded as 0 = unsubstantiated, 1 = substantiated, and 2 = unknown/pending. For any alienated parent who had one or more substantiated allegations of abuse, another dummy code was created to categorize them as an “abusive alienated parent.” Numbers of unsubstantiated allegations against each alienated parent were also calculated. This variable regarding the allegation’s final substantiation status was used in our analytic models.
3. Results
3.1. Sample characteristics
3.1.1. Case details
Nearly every case in the sample involved multiple motions or countersuits made by the parties, so it was not possible to discretely classify each case by particular types of court adjudications. Details on these case types are presented in Table 1. The cases varied considerably in time since separation/divorce (range of 70 to 7,704 days). Over a third (36.2%) of cases were published in French. The majority of cases were from Quebec (38.8%), Ontario (27.2%), and British Columbia (15.4%), with the remaining cases spread across the remaining provinces except for Nunavut and Yukon, which did not have any cases that met the inclusion criteria. These case details are presented in Table 2. Importantly, all but 29 cases involved a motion to modify of enforce parenting plans. These 29 cases involved financial matters that were unresolved from previous custody disputes, and seven of the decisions in those cases (24.1%) still resulted in a change in parenting time.
Table 1. Trial level motions.
| Motions | Number of cases |
|---|---|
| Modification of custody or parenting time | 302 |
| Financial matters (e.g., modification of child support, alimony, legal fees) | 97 |
| Divorce | 30 |
| Increase access, request for therapy/evaluation | 83 |
| Restriction of access, request for protection | 35 |
| Contempt/enforcement, appointment of a GAL | 50 |
| Relocation | 8 |
| Other (civil suit, appeal request, order reviews, transfer of jurisdiction) | 22 |
Note. The total number of motions is higher than the number of cases in the sample because many of the court actions involved multiple motions by both parties. Therefore, percentages of cases were not possible to calculate.
Table 2. General characteristics of Canadian Trial Level PA Cases (N = 500).
| Number (%) | |
|---|---|
| Cases published in French only | 181 (36.2%) |
| Province | |
| Alberta | 30 (6%) |
| British Columbia | 77 (15.4%) |
| Manitoba | 11 (2.2%) |
| New Brunswick | 12 (2.4%) |
| Newfoundland and Labrador | 10 (2.0%) |
| Northwest Territories | 3 (0.6%) |
| Nova Scotia | 11 (2.2%) |
| Nunavut | – |
| Ontario | 136 (27.2%) |
| Prince Edward Island | 2 (0.4%) |
| Quebec | 194 (38.8%) |
| Saskatchewan | 14 (2.8%) |
| Yukon | – |
| Number of childrena | |
| One | 176 (35.4%) |
| Two | 204 (41.0%) |
| Three | 88 (17.7%) |
| Four or More | 29 (5.8%) |
| Parental figure found to have alienated the children | |
| Mother | 322 (64.4%) |
| Father | 170 (34.0%) |
| Other (e.g., grandparent) | 8 (1.6%) |
| Cases where an allegation of abuse was made against the alienated parent | 238 (47.6%) |
| Cases involving alienated parent having a substantiated allegation of abuseb | 35 (7.0%) |
- a
While some cases involved step-children and other family members, we only coded the children directly or legally related to the parental figures for whom the case involved. Three cases had missing data on number of children, and so the % is based on an n = 497.
- b
The percentage reported in the table is different than the text, as this table was based on all cases (including parental figures such as grandparents), while the in-text percentage was based on the mother/father alienated parents only.
3.1.2. Third parties who found parental alienation
The court was identified as the sole determiner that PA occurred in 158 cases (31.6%)— there were no other professionals identified in these cases. Of the remaining cases, one or multiple parties found PA. Two hundred and thirty-eight cases involved a therapist, psychologist, or psychiatrist, 104 involved a custody assessor, social worker, or Guardian ad Litem (GAL), and 21 involved another expert (e.g., psychosocial) who found PA. There were 21 cases where at least one mental health provider and an assessor or social worker (not including the court) found PA had occurred in the family. In 53 cases, there was disagreement between the court and another third party that PA had occurred: The court disagreed with therapists, psychologists, or psychiatrists on 41 cases (17.2% of 238), with custody assessors, social workers, or GALs on 9 cases (8.6% of 104), and with other professionals on one case (of 21). The court also disagreed with the conclusions made by a therapist, psychologist, or psychiatrist and a custody assessor, social worker, or GAL on two cases (11.8% of 17). This finding indicates that courts are not blindly accepting the opinions of experts in their decisions.
3.1.3. Alienating parents and children
The majority of the 500 alienating parents in the sample were mothers (n = 322, 64.4%). Of the others, there were 170 (34.0%) fathers and eight parental figures (1.6%) who were not the biological parents (e.g., grandparents, foster parents). The average number of children in each case was 1.96 (SD = 0.93) with a range of 5. Most families in the 500 cases had two children (41.0%). Other details regarding parents and children represented in the sample are presented in Table 2.
3.1.4. Parenting plan status pre and post trial orders
The court ordered parenting plans (whether they were adhered to or not) before and after the court decisions in each case are presented in Table 3. The cases were at different stages of court involvement (interim and final adjudications, through later attempts to modify and enforce prior orders), so the custodial arrangements at the time of the court decisions we reviewed do not necessarily reflect custody change as an intervention to address PA at that time. Prior to the court decision, alienating mothers had sole or primary custody 51.5% (166 of 322 cases) of the time, while alienating fathers had sole or primary custody of the children 21.8% (37 of 170 cases) of the time. After the court decision, these percentages dropped slightly for mothers (to 45.0% mothers), but not substantially for fathers (20.0%). Interestingly, there were larger proportions of split custody arrangements (different children in the sole care of different parents) when fathers were the alienating parent (9.4% pre; 10.6% post of 170 cases) than when mothers were the alienating parent (2.5% pre; 5.3% post of 322 cases), although these differences were not statistically significant (ps > 0.05).
Table 3. Physical custody of children before and after court decision (N = 492).
| Alienating parent | Custody Ordered | Before | After |
|---|---|---|---|
| Mother (n = 322) | Mother sole/primary | 166 (51.5%) | 145 (45.0%) |
| Father sole/primary | 73 (22.7%) | 107 (33.2%) | |
| Joint/shared | 60 (18.6%) | 42 (13.0%) | |
| Split custody | 8 (2.5%) | 17 (5.3%) | |
| Foster care/extended Family | 10 (3.1%) | 8 (2.4%) | |
| Other | 5 (1.6%) | 3 (0.9%) | |
| Father (n = 170) | Mother sole/primary | 69 (40.6%) | 76 (44.7%) |
| Father sole/primary | 37 (21.8%) | 34 (20.0%) | |
| Joint/shared | 37 (21.7%) | 32 (18.8%) | |
| Split custody | 16 (9.4%) | 18 (10.6%) | |
| Foster care/extended Family | 7 (4.1%) | 6 (3.5%) | |
| Other | 4 (2.4%) | 4 (2.4%) |
3.1.5. Allegations of abuse against the alienated parent
This exhaustive review of 500 cases documented that the courts took every allegation of abuse raised by any party seriously and integrated the concerns into their decisions. Across the 500 cases, it is noteworthy that over half (52.4%) did not involve any allegation of abuse against the alienated parent. Details about the allegations and third parties who investigated them are presented in Table 4. There were 768 total allegations made against 238 alienated parents, averaging 3.23 allegations, and ranging from 1 to 63 per person. Some allegations had multiple types of abuse reported (e.g., domestic violence and child physical abuse), so determining the exact percentages of types of alleged abuse was not possible. The most common allegations were domestic violence and child physical abuse (approximately 200 allegations each), followed by child maltreatment/neglect, child sexual abuse, and child emotional/psychological abuse. Similarly, there were sometimes multiple people making allegations, or the information was not reported in the court decision. Most of the allegations of abuse were made by alienating mothers (459 allegations), followed by children (152 allegations), and then alienating fathers (113 allegations). Fewer allegations were made by extended family members and third parties such as neighbors (58 allegations).
Table 4. Characteristics of abuse allegations made against alienated parents.
| Type of allegation | Na |
|---|---|
| Domestic Violence | 201 |
| Child Physical Abuse | 263 |
| Child Emotional/Psychological Abuseb | 55 |
| Child Neglect/Maltreatment | 90 |
| Child Sexual Abuse | 90 |
| Other abuse (e.g., harming a pet or co-worker, loitering, theft) | 61 |
| Unknown | 9 |
| Person(s) or parties alleging each abuse claim | |
| Alienating mother | 459 |
| Alienating father | 113 |
| Child(ren) | 152 |
| New partner/spouse, extended family member | 26 |
| Other (e.g., neighbor, childcare provider, pastor) | 32 |
| Unknown/anonymous | 9 |
| Allegation initially reported to: | |
| Child protection services | 267 |
| Police | 246 |
| Therapist or counselor | 42 |
| Doctor or medical provider | 31 |
| School teacher/educator | 14 |
| Court | 159 |
| Custody Evaluator/Assessor/Office of Legal Council | 20 |
| Others (e.g., friends, pastor, neighbor) | 15 |
| Unknown | 32 |
| Third parties who investigated each allegation of abuse (n = 654) | |
| Police | 211 (32.3%) |
| Child Protection/Youth Services | 227 (34.7%) |
| Custody Evaluator/Medical Provider/Office of Legal Council | 23 (3.5%) |
| Psychologist/therapist/pastor | 20 (3.6%) |
| Multiple parties (most involving the police and/or CPS) | 85 (13.0%) |
| Unknown/pending | 91 (13.9%) |
| How substantiation was ultimately determined (n = 768) | |
| Family court | 350 (45.6%) |
| Criminal court | 54 (7.0%) |
| Both courts | 31 (4.0%) |
| Investigation only: No reported court involvement | 310 (40.4%) |
| No details on investigation or court involvement | 22 (2.9%) |
| Number of allegations made immediately following a court action which negatively impacted the alienating parent | 268 (30.1%) |
Note: Only 238 cases in the entire sample (47.6%) involved any allegation of abuse made against the alienated parent. The numbers reported in the table are from 768 allegations across these 238 cases. Allegations were sometimes made against the alienating parent and other family members as well, but these were not central to our hypotheses, and will be explored in forthcoming papers.
- a
Percentages are only reported when there were not multiple values within categories; some categories had multiple entries, such as multiple forms of abuse made within one allegation.
- b
Although PA has been considered a form of psychological abuse, we did not code cases as emotional/psychological abuse if the allegation was specifically stated as being part of the PA.
Allegations of abuse against the alienated parent were most often reported to the police or CPS (around 250 each), but 159 allegations were made to the court only (e.g., via testimony). A significant proportion of allegations (40.4% of 768) were only investigated by third parties and were not decided in a criminal or family court trial. Investigations were conducted by third parties for 85.2% of 768 cases, and 654 of which were conducted by police (32.3%) or CPS (34.7%). Multiple parties were involved with investigating 85 of the allegations (13.0% of 768), most commonly by both the police and CPS. While some third parties are not technically investigators (e.g., pastors), they were listed in some cases as being the investigator of the allegation, and so they were recorded as such on the form.
3.1.6. Outcomes of investigations and court hearings regarding allegations of abuse
The substantiation status outcomes and level of disagreement in substantiation between third parties and courts are presented in Table 5. Four hundred and sixty two of the 654 allegations made against alienated parents that were investigated by third parties were determined to be unsubstantiated or unfounded (70.6%); 2.1% were determined to be false. It is possible that a many of the unsubstantiated allegations were actually false but were not labeled as such for a variety of reasons. For example, it is typical for the conclusion of a CPS investigation to be reported as “unverified,” even if there are indications it could be false. Other reasons may be the accuser not knowingly making a false claim, or investigator does not want to prejudice the accuser and disincentivize future genuine allegations. Only 10.9% of the 654 allegations were founded or substantiated, 1.6% were inconclusive, and the remainder (14.7%) did not have any details provided or were still pending an outcome at the time the court decision was written. Of the allegations against alienated parents that were brought before family court with a known outcome (350 allegations, 45.6% of the total allegations), only 36 (10.3% of 350) resulted in a substantiated finding of abuse. When the allegation was adjudicated in criminal court (54 allegations, or 7% of all allegations), 17 (31.5% of 54) of the allegations were founded or substantiated. There were also 31 allegations (4% of 768) that were decided in both courts, and seven of these were determined in both courts as being substantiated.
Table 5. Substantiation status and disagreement of abuse allegations made against alienated parents.
| Status/Disagreement | Percentage | |
|---|---|---|
| Third Party Substantiation (n = 654) | ||
| Unsubstantiated or unfounded | 70.6% | |
| False | 2.1% | |
| Founded or substantiated | 10.9% | |
| Inconclusive | 1.6% | |
| Pending or Unknown | 14.7% | |
| Civil court with known outcome (n = 350) | ||
| Substantiated | 10.3% | |
| Criminal court with known outcome (n = 54) | ||
| Substantiated | 31.5% | |
| Civil and criminal court with known outcome (n = 31) | ||
| Substantiated | 22.5% | |
| Disagreement about substantiation | ||
| Third party did not substantiate, court substantiated | 0.003% (1 of 290) | |
| Third party substantiated, civil court did not | 45.8% (11 of 24) | |
| Third party substantiated, criminal court did not | 37.5% (9 of 24) | |
| Third party substantiated, both courts did not | 16.7% (4 of 24) | |
There were 290 abuse allegations against alienated parents (of 768) where the allegation was investigated by a third party and had a court determination (criminal and/or family court). There was only one allegation where the investigators did not make a finding of abuse, but the family court did. There were 24 allegations involving a third-party investigator deciding that abuse was substantiated, but the criminal (11 allegations), family (9 allegations), or both courts (4 allegations) determined the allegation was not substantiated. All other allegations had consistencies in findings across investigators and the court(s). The data for abuse allegations made against alienated parents are available as a spreadsheet in the original Excel database (https://osf.io/3ngqm/?view_only=f3ebcbfc511548919f10536616b0803a).
4. Pre-registered hypothesis testing
Each of the hypotheses and the analytic plan are presented in Table 6. This plan was pre-registered on OSF before the decision was made to only include cases where PA was found to have happened. In other words, cases where PA was alleged, but not found to have occurred by the court or a third party (mental health professional, custody assessor) were not included in the sample. This change required an adjustment to the planned statistical models after they had been pre-registered, such that the only independent variable in a few of the models was gender of the alienating parent (a nominal dichotomous variable). Consequently, some of the binomial logistic regression models that were initially planned were not the most appropriate analytic model when gender was the only predictor, so we instead conducted Chi-square tests of independence for those analyses. The pre-registered multinomial regression models were still conducted as planned.
Table 6. Hypotheses and Analysis Plan.
| # | Hypothesis | Analysis | Dependent Variables | Support for the hypothesis |
|---|---|---|---|---|
| H1 | When a mother is found to be undermining the father’s paternal rights and alienating their child(ren), she is more likely to get a decrease in parenting time, lose custody of her children, and lose her case than a father. | Multinomial logistic regression Chi-square test of independencea Chi-square test of independence1 |
Decrease in parenting time | No |
| Total loss of custody | Yes | |||
| Alienating parent loses case | Yes | |||
| H1a | H1 results will be statistically significant even when the alienated parent is proven to be abusive. | Multinomial logistic regression Chi-square test of independence1 Chi-square test of independence1 All using only cases where there was an Abusive Alienating Parent (variable coded as 1) |
Decrease in parenting time | Unable to test |
| Total loss of custody | Unable to test | |||
| Alienating parent loses case | Unable to test | |||
| H2 | When mothers claim intrafamilial abuse in family court and the father is the alienated parent, her reports of abuse will be determined by the court to be unfounded more often than if the father claimed abuse and the mother was the alienated parent | Linear regression using only cases where there was an Abusive Alienating Parent (variable coded as = 1) | Number of unfounded claims of abuse | No |
| H3 | Mothers will have a decrease in parenting time or lose all custody more often than fathers when a GAL or custody evaluator is involved in the case. | Logistic regression model using only cases where there was a Third Party (variable coded as = 1) | Decrease in parenting time | Unable to test |
| Total loss of custody | No | |||
| H4 | When a mother claims that both child abuse and sexual abuse occurred and one or both were substantiated,b she is more likely to be penalized than fathers by getting a decrease in parenting time or lose all custody. | Logistic regression model using only cases where Sexual Abuse and Child Abuse Allegations are both made, and at least one is substantiated | Decrease in parenting time | Unable to test |
| Total loss of custody | Unable to test | |||
| H5 | The greater number of unsubstantiated allegations of abuse that a mother makes, the more likely it is for the father to have a decrease in parenting time or lose all custody. | Multinomial logistic regression model Binomial logistic regression |
Decrease in parenting time alienated parent | No |
| Total loss of custody alienated parent | No | |||
| H6 | Fathers are more likely to be an alienated parent than mothers. | Chi-square Goodness of Fitc | Proportion of cases | Yes |
- a
After the decision to sample only cases where parental alienation occurred, the only predictor in the model was gender which is an interval variable. Therefore, the analytic plan that was originally pre-registered (binomial logistic regression) was adjusted to a Chi-square test of independence.
- b
We originally used the term “corroborated” rather than substantiated, as that was the terminology used by Meier et al. (2019). It is not clear how the original authors defined corroborated, as this could mean there are multiple witnesses who believe it to be true, even if other investigators determine the allegation to be unsubstantiated or false. A more accurate term to use is “substantiated,” as this is an outcome of an investigation of the entire case and set of facts from all parties.
- c
The analysis plan for this hypotheses was initially a binomial logistic regression, but with the change in design to include only those cases where PA was found to have occurred, we could not test this hypothesis as planned. As the hypothesis is a comparison of frequency of the cases in two categories (gender), we changed the analytic plan to be a Chi-square Goodness of Fit test.
4.1. Hypothesis 1
Our first hypothesis examined whether a mother found to have alienated her child(ren) was more likely to get a decrease in parenting time, lose custody of her children, and lose her case than a father who was found to have alienated his children. A multinomial logistic regression model was conducted with gender of the alienating parent as the predictor (1 = mother; 2 = father) and change in parenting time as the dependent variable (-1 = loss of parenting time; 0 = no change; 1 = increase in parenting time). The model did not fit the data better with gender as the only predictor than a null model, χ2(2) = 1.47, p =.48. Gender of the alienating parent was not statistically related to whether the parent lost (B = 0.48, SE = 0.43, OR = 1.62, 95% CI [0.70, 3.72]) or gained more parenting time (B = -0.03, SE = 0.21, OR = 0.97, 95% CI [0.65, 1.45]) as compared to no change in parenting time, ps > 0.05. Similar proportions of alienated mothers and fathers received an increase in parenting time after the court order was made (103 of 322, 32.0% and 57 of 170, 33.5% respectively). Only a small proportion of alienated parents lost parenting time (32 of 492, 6.5%), and gender differences in this outcome were not found. We did not find support for H1 for this outcome.
One hundred and three alienating parents (20.9% of 492) lost custody of their children by court order after the trial, meaning that their parental time was revoked, or their parenting time was decreased to only supervised or therapeutic visits of several hours a week or less. The Chi-square result from the test of independence indicated that more alienating mothers lost custody after the court order than did alienating fathers, χ2(1) = 4.99, p =.03. Of alienating mothers, 77 of 322 (31.4%) lost custody, while of 26 of 170 (18.1%) alienating fathers did. Ninety-eight (20.0% of 492) alienating parents lost their case. As with the loss of custody outcome, the Chi-square was also statistically significant, χ2(1) = 14.18, p <.001. Alienating mothers were more likely to lose their case (80 of 322, 33.1%) than alienating fathers (18 of 170, 12.2%). Therefore, we found support for H1 for these latter two outcomes, however the effect sizes for both of these statistical models were quite small (φ = -0.10, and φ = -0.17 respectively). Gender of the alienating parent only accounted for approximately 10% of the variance in the scores so the remaining 90% of the differences must be attributed to variables other than gender, notwithstanding the small effect sizes. And across most tests, there was sufficient power to detect at least a medium effect.
4.2. Hypothesis 1a
One argument that has been raised by critics of PA scholarship is that mothers who are found to have alienated their children are just trying to “protect” them from abusive fathers (see Harman et al., 2018 for a discussion). We attempted to test this “protective parent” hypothesis by testing whether the first hypothesis would be statistically significant even when the alienated parent is proven to be abusive, however there were only 35 cases (7.1% of 492 cases) in which the alienated parent had any allegation of abuse that was founded or substantiated against them. We were therefore unable to test our pre-registered hypothesis due to this small number of cases.9 It is important to note that 25 “abusive” alienated parents were mothers (71.4% of 35), and 10 were fathers (28.6% of 35), so the presumption that “abusive” alienated parents are mostly fathers is not reflected in these data. Among alienated fathers (n = 10) who also had a finding of abuse against them, only one of the alienating mothers lost custody to them. In contrast, among the 25 alienated mothers who also had a finding of abuse against them, eight of the alienating fathers lost custody to them.
4.3. Hypothesis 2
The second hypothesis examined whether claims of abuse made by an alienating parent are more likely to be determined to be unfounded if the alienating parent is a mother than a father. In other words, are courts discrediting claims of abuse made by mothers or fathers more when it was determined they were an alienating parent? Harman & Lorandos (2021) restricted their test of this hypothesis to cases where there were findings of abuse against the alienated parent because Meier et al. (2019) reported doing so in their study. To replicate the test of this hypothesis, we conducted a linear regression analysis on the cases where the alienated parent had a substantiated finding of abuse. The outcome variable was the number of unsubstantiated claims of abuse, which could indicate that third parties had discredited other allegations of abuse made against the alienated parent by the alienating parent. Perhaps because the sample size of alienated parents with findings of abuse was very small (n = 35), the model fit for the analysis was not very good, R2 = 0.07, F(1,33) = 2.40, p =.13. Gender of the alienating parent was not a significant predictor in the model, t = 1.55, p =.13, 95% CI (-0.46, 3.38).
As a post-hoc test of this second hypothesis, we re-ran the linear regression analysis including all cases in the dataset (rather than just cases with a finding of abuse against the alienated parent) with gender of the alienating parent, a dummy coded variable for whether the alienated parent was abusive (1) or not (0), and an interaction term of the two variables predicting the number of unsubstantiated allegations in the model. The model fit was also not very good, R2 = 0.01, F(3,205) = 0.95, p >.05. There were not statistically significant main or interaction effects in the model. Both of these pre-registered and post hoc analyses indicate that we did not find support for our second hypothesis. Alienating mothers’ claims of abuse against known “abusive” alienated fathers were not being discredited more often than they were for alienating fathers, and there were very few cases where this was raised as an issue in court at all.
4.4. Hypothesis 3
The third hypothesis examined the role of court-appointed third parties in the cases. Specifically, we tested whether the involvement of a GAL or a custody assessor would affect the outcomes differentially for alienating mothers and fathers. The hypothesis prediction is that alienating mothers would be more likely to have a decrease in parenting time or lose all custody than would alienating fathers when such professionals were involved. We restricted the sample to only cases with data where such professionals determined PA had occurred (n = 102).
A multinomial logistic regression model was again conducted with gender of the alienating parent as the predictor and change in parenting time was the dependent variable. The model fit indicated that the model did not fit the data better with gender as the only predictor than a null model, χ2(2) = 1.82, p =.40. Gender of the alienating parent was not statistically related to whether the parent lost or gained more parenting time as compared to no change in parenting time, ps > 0.05. Therefore, the involvement of a custody assessor or GAL did not differentially affect this custody outcome for alienating mothers or fathers. The Chi-square result also indicated that alienating mothers were not more likely than chance to lose custody than alienating fathers when the third party who found PA was a custody assessor or GAL, χ2(1) = 3.06, p >.05. Therefore, we did not find support for the third hypothesis.
4.5. Hypothesis 4
Our fourth hypothesis examined whether an alienating mother who claimed that both child abuse and sexual abuse occurred, and one or both claims were substantiated, was more likely to be penalized by getting a decrease in parenting time or losing all custody than fathers making the same claims. This specific hypothesis was created based on reported findings in Meier et al. (2019), and there were not enough cases that met these criteria in the Harman & Lorandos (2021) database to test it. Therefore, we attempted to test the hypothesis again using the Canadian trial level cases. Among all of the allegations of abuse made against the alienated parents, there was not one case where there was a sexual abuse and child abuse allegation and one or both were substantiated. Indeed, there was only one case where a sexual abuse allegation was substantiated by anyone, and in that case, it was determined by a pastor (not a police officer or trained evaluator) to have occurred based only on an interview they had with the children.
4.6. Hypothesis 5
The fifth hypothesis was that the greater number of unsubstantiated allegations of abuse an alienating mother makes, the more likely it is for the alienated father to have a decrease in parenting time or lose custody of the children. The final substantiation status of each allegation of abuse was examined when a mother or father was a party who made an allegation against the other parent. We did not include allegations exclusively made by other parties (e.g., grandparents). There were 459 allegations of abuse made by alienating mothers against alienated fathers, 80.0% of which were unsubstantiated and only 6.8% were substantiated (13.3% were unknown). Although alienating fathers made fewer allegations of abuse against alienated mothers (n = 112), the percentage of substantiated allegations was similar (7.1%) to those of alienating mothers. Eighty-one (72.3%) of these allegations were not substantiated (20.1% were unknown or pending an outcome).
A multinomial logistic regression was conducted with gender of the alienating parent, number of unsubstantiated allegations of abuse against the alienated parent, and an interaction term using both variables serving as predictors in the model. As with H1, the model fit indicated that the model does not fit the data better with these predictors than a null model, p >.05. Gender of the alienating parent was also not related to whether the parent lost or gained more parenting time as compared to no change in parenting time, ps > 0.05, and neither was the number of unsubstantiated allegations, ps > 0.05. To test whether an alienated father would be more likely to lose custody than an alienated mother the more unsubstantiated allegations are made against them, a binomial logistic regression analysis was conducted with gender of the alienating parent, number of unsubstantiated allegations of abuse, and an interaction term of the two variables as predictors in the model. The model fit indicated that the model does not fit the data better with the three predictors than a null model, χ2(3) = 3.17, p >.05. The gender of the alienating parent, the number of unsubstantiated allegations of abuse, and an interaction of the variables were not predictors of whether an alienated parent lost custody of their children, ps > 0.05.
4.7. Hypothesis 6
Our last hypothesis examined whether fathers are more likely to be an alienated parent than are mothers. We tested the proportional differences using a Chi-square Goodness of Fit test and found the proportions to be significantly statistically different from each other, χ2(1) = 45.30, p <.001, Therefore, we found support for our last hypothesis in that among the PA cases where the mother or father was the alienating parent (n = 492), a significantly larger proportion was alienated fathers (65.30%) than alienated mothers (34.70%). Lorandos, 2020a, Lorandos, 2020b analyzed thirty-four years of parental alienation cases in the US courts and found that about seventy-five percent of the identified alienating parents were female and twenty-five percent were male. The data developed in this study documented that a significantly larger proportion of alienated parents were fathers than mothers. An explanation of the socio-cultural bases for this significant discrepancy is beyond the scope of this analysis.
5. Discussion
The purpose of this study was to test the findings of Harman & Lorandos (2021) using trial level court decisions. To do so, we turned to publicly available English and French-language decisions from Canadian trial courts. Canadian jurisprudence has consistently recognized the concept of PA as a manifestation of maladaptive parenting practices after separation, and has evolved to the point where Court can take “judicial notice” of what is meant by the term “parental alienation” and deal with the competing narratives even without expert opinion evidence (D. v. T., 2021, A.M. v. C.H., 2019). This recognition of PA, and the ability of judges to access other provincial appeal and trial level opinions to inform their decisions, make Canada a different litigation context for PA cases than the U.S. Within this context, we found mixed support for our pre-registered hypotheses.
Our first hypothesis was that alienating mothers would be more likely to have a decrease in parenting time, lose custody, and lose their case than alienating fathers. Consistent with the findings reported by Harman & Lorandos (2021) we did not find gender differences among alienating parents regarding decreases in parenting time (Meier et al., 2019 did not test this outcome). Alienating mothers and fathers were just as likely to lose or gain more parenting time compared to having no change in custody after the court hearing. We did, however, find that alienating mothers were more likely to lose custody and their case than alienating fathers. This was a statistical difference, but not likely a clinical difference due to the effect size. There were possibly other factors that contributed to this finding that have nothing to do with gender or could be correlated with gender but analysis of such variables was beyond the scope of this study. These findings lend support for these two outcomes for the first hypothesis; however, the effect sizes were small. Bala et al. (2010) and Harman & Lorandos (2021) did not find that alienating mothers were statistically more likely to lose custody of their children than alienating fathers, so we are unsure why we found differences in the current study. A greater proportion of alienating mothers than alienating fathers had sole or primary custody of the children before and after the court orders we reviewed, so this proportionality may contribute to our findings. The analytic models for the other hypotheses indicated that gender alone was not a good predictor, the effect sizes were small, and there could be other constructs that could better explain loss of custody than gender, such as time since separation or severity and length of time that the alienating parent had engaged in parental alienating behaviors.
Harman & Lorandos (2021) also reported that alienating mothers were more likely to lose their appeal than alienating fathers, as did Meier et al. (2019). It remains difficult to interpret this effect, as the analysis only compared cases where the parents won or lost--there were many cases (n = 99) where both parents won different motions or other adjudications. In addition, there was a wide variety of issues in the adjudications that were being decided that could explain more of the effect than gender alone. Future research could focus on particular types of adjudication or relief sought (e.g., modification of custody) to further explore this effect.
We were unable to test the outcomes planned for hypothesis 1a, which was that the findings from hypothesis 1 would remain statistically significant when the alienated parent was found to have been abusive. We found that allegations of abuse made against alienated parents were considered very seriously and factored into the court decisions. We found only 25 alienated mothers and 10 alienated fathers who also had findings of abuse against them (35 total of 500 cases)—in other words, the base rate for these types of cases was very low (7.0%). Harman & Lorandos (2021) also found a low base rate of 7.9% using 967 U.S. appellate cases, so the two combined studies raise questions about how Meier et al. (2019) selected their cases, and how they obtained a large enough sample to statistically test this hypothesis. Media coverage of Meier et al.’s (2019) study (e.g., Schmidt, 2019) have reported that mothers are afraid to raise abuse allegations because they are allegedly losing custody of their children to “abusive fathers,” particularly when the fathers claimed to have been alienated. When an audience is not provided with the background base rate information and is presented only with descriptive statistics derived by unknown means, this can lead to a serious error in judgement known as the base-rate fallacy (Kahneman & Tversky, 1973). When base-rates are neglected or ignored, then the perceiver overestimates the occurrence of the specific information, and then harmful policies and laws may be created because there is an inaccurate perception about the prevalence of the problem.
We also found that only six alienated fathers and six alienated mothers with a finding of abuse had sole or primary custody of one or more of their children. Rather than assume that the courts must be handing children over to ‘abusive’ parents in these small numbers of cases, the investigation outcomes and details provided in the cases reveal another story: the finding of abuse was generally in the past and/or an isolated or transitory issue and the parent was never or is no longer a danger to the children, the abuse was not child-related, and/or the behaviors of the alienating parent were considered to be far more abusive and/or influential on the children than those of the ‘abusive’ alienated parent. For example, a case in Ontario (M.M.B. (V.) v. C.M.V., 2017) involved an alienated mother who was given temporary sole custody of her children, despite having substantiated findings of at times inappropriate reactions to the children’s behavior and inflexible attitudes towards various situations that negatively impacted the children. The court determined that these incidents did not indicate the children were in danger in the mother’s care at the time of the decision. In another case from Ontario, the alienating mother had a long history of physical aggression (e.g., throwing a butcher knife at the father) yet the alienated father was found guilty of one incident of domestic abuse ten years prior which also involved physical abuse by the mother. The judge in this case thoroughly reviewed the evidence of past charges and police involvement with the alienated father and determined there was no history of violence by him, and that the calls made by the mother to the police were inappropriate (Children’s Aid Society of Waterloo [Regional Municipality] v. L. (K.A.), 2010). Harman & Lorandos (2021) also reported that in 16 appellate cases where the alienated parent with a finding of abuse was given custody of the children, the parents were not determined to currently be a risk to their children for similar reasons (see Harman & Lorandos, 2021, pp. 21-22).
We failed to find support for our second hypothesis, which tested whether claims of abuse made by an alienating parent (in cases where the alienated parent had a finding of abuse) are more likely to be determined by the court to be unsubstantiated if the alienating parent is a mother than a father. Given that there were only a small number of cases where findings of abuse were made against alienated parents (n = 35), the pre-registered analytic plan for this hypothesis test involving the use of a linear regression model did not fit the data well, and gender was not a significant predictor of unsubstantiated allegations. Our post-hoc analysis using all cases where allegations of abuse had been made (<half the sample), regardless of whether the alienated parent had a finding of abuse, also failed to support the second hypothesis. While Meier et al. (2019) reported that mothers' allegations of abuse are discredited more often than fathers' allegations of abuse, neither Harman and Lorandos (2021) nor this data-set could replicate or substantiate that claim.
Our third hypothesis was also not supported, which was that involvement of a GAL or a custody assessor would affect the outcomes differentially for alienating mothers and fathers. Our findings are consistent with Harman & Lorandos (2021) and contrary to Meier et al., (2019), who stated that court professionals had gender biases and need to be trained on how PA is being used to deny claims of child physical and sexual abuse (pp. 26–27). Kayden’s Law incorporates Meier et al.’s (2019) recommendations to only allow training to such professionals on a limited set of family violence topics (PA is not included), and specifies that these trainings can only be taught by survivors of domestic violence or child physical or sexual abuse, or by professionals with substantial experience in assisting such survivors (Violence Against Women Act, 2022, Section 1504) rather than including other qualified professionals (e.g., scientists, law enforcement personnel). These training restrictions do not reflect recent empirically based recommendations by the Association of Family and Conciliation Courts and the National Council of Juvenile and Family Court Judges (Association of Family and Conciliation Courts National Council of Juvenile and Family Court Judges, 2022) that family law practitioners should receive training on factors such as parental alienating behaviors and child alignment issues. Given that two replication studies have not found support for Meier et al.’s (2019) argument that her data indicates there is “widespread gender bias in courts’ handling of..abuse claims” (p. 26), we are concerned that some court professionals in states that enact Kayden’s Law may receive mandatory training that presents a gender-biased perspective on family violence and lacks a broader scientific foundation.
Our fourth hypothesis was intended to examine whether an alienating mother who claimed both child abuse and sexual abuse occurred and one or both were substantiated, was more likely to be penalized than an alienating father by getting a decrease in parenting time or losing all custody. We were unable to locate even one case in our sample where both allegations occurred and one or both were substantiated. Some critics of PA scholarship have falsely equated sexual abuse allegations in the context of PA cases (e.g., Death et al., 2019, Milchman, 2017), and claimed that PA is just a legal defense used by abusive fathers in cases where sexual abuse is alleged by mothers (e.g., Meier et al., 2019). Other critics have argued that courts are discrediting sexual abuse allegations made by parents if they determine that the allegation was unsubstantiated or deliberately misleading to the court (e.g., Webb et al., 2021). Our data indicate that allegations of sexual abuse were taken very seriously and were closely investigated by many parties and/or the courts. Of the 90 sexual abuse allegations (15.7% were made towards mothers), 16 were tried in criminal and/or family court only. The majority of sexual abuse allegations (77.8%) were investigated and tried in court, and of these, twenty were investigated by the police, 26 were investigated by CPS, three by psychologists, and one by an investigator/detective appointed by the Office of the Children’s Lawyer in a province of Canada. Twenty of these allegations were investigated by two or more of these parties. Of the remaining cases that were only investigated and not tried in court, six were investigated by two or more parties. The data developed in this study, when compared with the Harman and Lorandos (2021) data set, indicate that the response of the courts and law enforcement to allegations of child sexual abuse are very similar.
Harman & Lorandos (2021) only identified 3 (out of 967) appellate cases that involved allegations of child and sexual abuse where one or both were substantiated, and not one of the three parents received custody of the children. The authors called into question the sample size used by Meier et al. (2019) who reported effects for their test of this hypothesis. Meier et al. (2019) wrote that two of the authors in their study “developed analyses for the statistical consultant to complete, reviewed the output, and, through numerous iterations, refined, corrected, and amplified on the particular analyses” (p. 8, emphasis added), which indicates the authors may have used a questionable research practice (p hacking) to create statistically significant models. The fact that we were unable to find even one case in this sample where a parent had both an allegation of sexual and child abuse and one or both were substantiated provides some support for this suspicion, but without more specific details about the Meier et al. (2019) sample and statistical models, we are unable to form any conclusions.
Our fifth hypothesis tested whether the greater the number of unsubstantiated allegations of abuse that an alienating mother makes, the more likely it is for the alienated father to have a decrease in parenting time or lose all custody of the children. Harman & Lorandos (2021) found that fathers were more likely than mothers to have a decrease, rather than increase, in their custodial time with their child(ren), and the more unsubstantiated claims of abuse that were made against a parent, the more likely they were to get a decrease, rather than increase, in their parenting time. We did not find gender differences on these outcomes in the current study. Neither gender of the alienating parent nor the number of unsubstantiated allegations of abuse was associated with a decrease of parenting time or loss of child custody. The appellate cases in the Harman and Lorandos (2021) data set were generally limited to high-level summaries of the trial court methodologies and findings, due to the deference given to the Trial-level factual findings and the scope limitations of appellate review. However, the current data set of trial court opinions did not reveal substantial differences in the treatment of substantiated and unsubstantiated allegations.
Our last hypothesis was not tested by either Meier et al. (2019) or Harman & Lorandos (2021), but we were able to test it due to the sequential sampling approach that was used in this study. A Chi-square testing the proportions of alienating parents by gender found support for our hypothesis that mothers were significantly more likely to be the alienating parent than fathers. This finding is consistent with Lorandos (2020b), who found that approximately 75% of alienating parents in 1,181 U.S. appellate cases were mothers. Population based national samples have not found gender differences in who the alienated parent is (e.g., Harman et al., 2019), and so it remains uncertain why gender differences are found when cases reach the court for intervention. Harman & Lorandos (2021) argued that fathers may be greater represented in appellate cases due to the financial cost involved, or they were more often subjected to biases than mothers, thus warranting an appeal. In trial level cases, it is possible that alienated mothers are more likely to settle out of court or just give up than fathers, or that many alienated mothers did not appear in our case search because they do not identify as being alienated, but rather as victims of domestic violence. More research will need to explore these possible explanations.
5.1. Limitations
Our results indicate that decisions about child custody are not strongly related to the gender of a parent in cases where there has been a finding of PA, or in cases where there have also been allegations of abuse. There could be many other factors that contribute to these decisions that were not tested or coded in our study, such as the frequency and duration of child exposure to parental alienating behaviors or the involvement of extended family members and other individuals in the abusive family dynamic. Future research should explore these other contributors and cease relying only on gender of the parent as the primary or most important factor.
Our sample included the most recent court order for families where PA was found to have occurred, so there was great variability across cases regarding when this determination was made. We do not believe this heterogeneity in the sample is a weakness: rather, it affords the ability to generalize to a wide variety of PA cases adjudicated in family courts in Canada. Had we used a narrower inclusion criteria set (e.g., cases where PA was first identified, or undefined “paradigmatic PA” cases selected by Meier et al., 2019), we would not be able to generalize to the greater body of PA cases in Canada. We also opted to use this sampling strategy because substantial modifications of parenting time or transfer of custody are more often employed in moderate to severe cases of PA after less invasive approaches have been attempted and failed (e.g., see Warshak, 2020 for a review). For example, if we had only included cases where PA was first identified, it is unlikely that the court would have ordered the more intensive parenting time interventions that were tested in this study. Although most Canadian family court decisions are publicly available, we still cannot fully generalize our findings to milder cases of PA that are not involved with litigation or to cases that were sealed and inaccessible (e.g., some child protection cases), and so it would be beneficial to explore how custody decisions are made in these other types of cases.
Although we coded each allegation of abuse thoroughly, some cases were still pending or lacked details about what the final substantiation status determination was for the investigation or court outcomes. Our trial level data contained considerably more detail about these allegations of abuse than the appellate cases used in Harman & Lorandos (2021) and Meier et al. (2019), and the sample is more generalizable because it reflects a wider range of cases than appellate cases. Similar to Harman & Lorandos (2021), our study closely examined outcomes of investigations across multiple parties. We did not presume, like others have, that all self-reported allegations are necessarily true (e.g., Ogolsky et al., 2022, Meier et al., 2019, Webb et al., 2021). In fact, we found that around 75% of 768 allegations of abuse that were investigated or tried in court were found by multiple parties to be unsubstantiated or false, a finding that is consistent with statistics reported by Moloney (2008) using national family court data from Australia. Only about 10% of the 768 allegations against alienated parents in our study were determined to be true/substantiated. Of note, this means that 90% of abuse allegations in cases of parental alienation were determined not to be true or otherwise unsubstantiated. The implications of such egregious allegations on the freedom of those wrongfully accused should not be underestimated. To date, law enforcement and CPS workers are unaware of this widespread violation of justice. Assuming that all allegations are true, particularly when many of them in our study (30.1% of 768, see Table 4) were made directly after a court decision or action (e.g., filing of a motion to modify custody), overlooks legal and administrative aggression tactics (Hines et al., 2015) used by some alienating parents to gain and maintain power over their children and the other parent (Harman et al., 2018, Harman and Matthewson, 2020).
Due to the considerable detail provided in the trial level cases (some were over 100 pages long) and over 1/3 of the cases were published in French, it took on average 1.63 h for each RA to complete each case (with a range of 15 min to 9 h). The decision was made after the first 10 cases were coded to stop data collection at 500 cases rather than the 1,000 cases that were initially planned because there was little missing data. Although our power analysis did not indicate that this change would affect our analytic plan, we still only had 35 cases where an alienated parent also had a finding of abuse, resulting in low power to detect gender differences for some of our statistical models. This small number of cases indicates that it was uncommon to find alienated parents who also had substantiated findings of abuse. We doubt that a larger sample would have remedied this issue. This finding is of great concern because recent legislation (e.g., Kayden’s Law) is based on the assumption that many mothers are losing custody of children to abusive fathers claiming to have been alienated from their children. Our data, as well as results from the U.S. (Harman & Lorandos, 2021), do not lend support for that assumption.
5.2. Direct versus conceptual replications
Harman & Lorandos (2021) intended to conduct a direct replication of the Meier et al. (2019) study, but the original authors did not share information about their methods or the cases included in their study after being asked directly for them (see https://osf.io/j9bh5/ for emails related to this inquiry). Despite claims to the contrary (Meier et al., 2022), Harman & Lorandos (2021) also attempted to conduct a case search using the search string that was later posted on-line by Meier et al., eleven months after the 2019 paper was published (see Harman & Lorandos, 2021, p. 191). Unfortunately, the results of this search could not be compared to Meier et al.’s (2019) search because to date their results have not been made available for review. It remains unclear why the full statistical models and fit statistics that produced the conclusions reported by Meier et al., (2019) have not been published or made publicly available. Such information is essential for scholars to fully understand the analyses that were conducted. Due to this lack of transparency, Harman & Lorandos (2021) instead conducted a conceptual replication study using open science practices that tested and failed to find support for Meier et al.’s (2019) conclusions. The current authors have read opinions written by critics of PA scholarship (e.g., Watson, 2021) and have attended professional presentations (e.g., Deutsch et al., 2021) where the Harman & Lorandos (2021) study was described as failing to rebut Meier et al.’s (2019) original findings, and who have stated that comparing the two studies is like comparing “apples and oranges.”.
We recognize that a failure to conceptually replicate a study does not necessarily invalidate the original study’s findings. Direct replications by scientists across multiple studies are viewed by some scholars as being the only way to differentiate true effects from sampling or unsystematic errors (e.g., Simons, 2014), but this does not mean that conceptual replications lack value. Conceptual replication is useful to extend psychological theory (Derksen & Morawski, 2022) because it involves operationalizing variables with different measures and/or manipulations (Stroebe & Strack, 2014), and tests hypotheses with different samples and in different contexts. The more that studies are conceptually replicated, the more confidence can be placed in their results and their contribution to scientific knowledge.
This second conceptual replication of Meier et al. (2019) again failed to find support for the majority of their conclusions. Using different sampling criteria and trial level cases (rather than appellate records) from Canada, our findings were consistent with those reported by Harman & Lorandos (2021).
Meier et al. (2022) have recently reported that they reanalyzed the Harman & Lorandos (2021) data and claim to have found support for their previous (2019) conclusions. The authors did not publish their commentary in the scientific journal where the original Harman & Lorandos (2021) paper was published. Rather, Meier et al. (2022) published their rejoinder in a professional journal that has been very critical of PA scholarship and was found by one scholar to have published numerous articles containing misinformation about PA (Bernet, 2021). It is outside the scope of this paper to address the inaccuracies and errors we find with Meier et al.’s (2022) arguments and reanalysis, and several of the arguments are verifiably untrue (e.g., claiming Harman & Lorandos (2021) did not attempt to use the search string; see Harman & Lorandos, 2021, p. 191). Unfortunately, there still remains a lack of transparency regarding how Meier et al. (2019) came to their original conclusions, and so it has not been possible to evaluate the scientific merit of their original work.
6. Conclusion
Kayden’s Law and similar legislation are based on spurious assumptions that allegations of abuse made by “protective mothers” are more likely than not to be accurate. These assumptions alter the burden of proof in family law. Such legislation also purports to remove the one most effective response to severe PA dynamics – the temporary placement of the children with the alienated parent to protect the child from the psychological abuse of the alienating parent and reset the family system. With laws such as this one, the accused are deemed guilty until they prove their innocence. This inversion of justice is troubling because the law is based largely on unverifiable and unreplicated research findings regarding an outcome of low prevalence and omits mention of other scientific research that has reported very different conclusions regarding abuse allegations and child custody. Greater efforts are needed to ensure that more evidence-based, scientifically peer-reviewed research is used to inform law and public policy that affects the lives of millions of families.
Declaration of Competing Interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: The authors have potential competing interests to disclose. Jennifer Harman has served as a general subject matter expert witness on an ad hoc basis, and she receives royalties from a co-authored book on parental alienation (Harman & Biringen, 2016). Christine Giancarlo receives royalties from a book she authored (Giancarlo, 2018). Demosthenes Lorandos has served as an expert witness on an ad hoc basis and receives royalties from several handbooks he has co-edited (e.g., Lorandos & Bernet, 2020). Brian Ludmer is a practicing family law attorney and co-author of a book for which he receives royalties (2014).
Data availability
All data is publicly available on Open Science Framework (https://osf.io/3ngqm/?view_only=f3ebcbfc511548919f10536616b0803a)
References
Association of Family and Conciliation Courts & National Council of Juvenile and Family Court Judges (2022). Joint Statement on Parent-Child Contact Problems. https://www.afccnet.org/Resource-Center/Center-for-Excellence-in-Family-Court-Practice/afcc-and-ncjfcj-joint-statement-on-parent-child-contact-problems.
A.M. v. C.H., (2019). ONCA 764. https://canlii.ca/t/j2nf4.
N. Bala, S. Hunt, C. McCarney
Parental alienation: Canadian court cases 1989–2008
Family Court Review, 48 (1) (2010), pp. 164-179, 10.1111/j.1744-1617.2009.01296.x