Parental Alienation and Abuse: What the Evidence Shows (Sharples, Harman & Lorandos 2025)

Love Over Exile is a plain-language research archive on parental alienation, written by Malcolm Smith — an alienated parent and author of the forthcoming book Love Over Exile — for non-specialist readers (other alienated parents, family members, therapists, lawyers) who want to understand the evidence base without a psychology degree or a journal subscription. This page is one entry in that archive.

What the Researchers Asked

When a court finds parental alienation, who is actually the abuser?

The question is the load-bearing one in the entire parental-alienation policy debate. If PA findings systematically attach to families where the alienating parent is the corroborated abuser, the construct is doing what its defenders claim — naming a behavioural pattern that runs alongside other forms of intimate-partner aggression and child psychological abuse. If PA findings systematically attach to families where the alienated parent is the corroborated abuser, the critic camp’s argument is correct — PA is a tactical label that displaces genuine abuse victims.

Joan Meier and colleagues (2020) made the strongest version of the critic-camp argument in a US Department of Justice-funded study of 4,338 custody cases. Mothers’ custody-loss rates roughly doubled when fathers raised PA cross-claims after the mothers raised abuse claims. The framing in advocacy work that followed was direct: PA is a weapon against survivors.

Jennifer Harman and Demosthenes Lorandos’s 2020 appellate-records study reached different conclusions on a different sample, and drew formal critiques from Mercer (2022) and Meier and colleagues (2022) on methodological grounds — sample selection, coding reliability, the lack of pre-registered hypotheses, gendered framing.

Sharples, Harman and Lorandos (2025), in Journal of Family Violence, is the methodological response to those critiques. The research question is now narrow and testable: in the subset of US appellate cases where parental alienation has actually been found by a court — not merely alleged — what is the substantiation pattern of abuse claims? Do they run from the alienating parent, from the alienated parent, or in both directions equally?

The pre-registered hypothesis was that perpetrators of parental alienating behaviours would have a history of co-occurring forms of abuse, consistent with the coercive-control model. The two-by-two coding frame — direction (alienating → alienated, or vice versa) by substantiation (upheld by a third party, or not) — is what makes the headline numbers interpretable.

What They Did — Methods in Plain English

The contribution begins with what the previous case-record studies in this area did not have.

Pre-registration of hypotheses, blind coders, and a tighter inclusion criterion are the three design improvements over the 2020 precursor study. Each one was demanded specifically by the 2022 Mercer and Meier critiques. Sharples 2025 implements all three.

Sample frame. The authors searched US appellate court opinions for cases in which parental alienation was found — meaning the trial court, a court-appointed professional, or both made an explicit finding that PA had occurred. Cases where PA was merely alleged but not found were excluded. This is the key tightening over the 2020 Harman-Lorandos study.

Attribute	Value
Source	Published US appellate court opinions
Sample size	492 cases where parental alienation was found by the court or a court-appointed professional
Cases with abuse allegations	288 of 492 — 58.54%
Total separate abuse claims coded	1,112
Inclusion criterion	Appellate opinion records PA as found / upheld
Coding scheme	2-by-2 frame: direction × substantiation
Pre-registration	Yes (per the abstract)
Coder blinding	Yes — coders were blind to the study hypotheses

The two-by-two frame is what generates the headline numbers. For each abuse claim, the coders recorded the direction it ran in (was it made by the alienating parent against the alienated parent, or vice versa, or by a third party such as Child Protective Services?) and whether the claim was substantiated (upheld by a court ruling, a CPS investigation, a criminal conviction) or unsubstantiated (rejected, dismissed, withdrawn, unsupported).

Substantiation is defined by third-party rulings, not by self-report. This is a much higher evidentiary standard than the survey-based prevalence research the field has relied on for decades.

Statistical analysis is logistic regression / odds-ratio comparison: the probability that an alienating parent has a substantiated abuse claim against them versus the probability that an alienated parent does, and the mirror comparison for unsubstantiated claims. The “81.62% greater probability” and “86.05% greater probability” numbers are reported as percentage differences in probability, not raw counts.

The full odds-ratio confidence intervals are behind the Springer paywall — flagged for re-verification once Malcolm has access. The directional finding is consistent across the open-access secondary sources that have summarised the paper, including the National Parents Organization, the Springer landing page, and the Varavei 2025 critical reply.

Sharples 2025 vs Meier 2020 — what the two studies actually answer

The most important methodological point about this paper is that it does not directly contradict Meier 2020 — even though both camps in the policy debate often treat the two studies as if they do. The two papers ask different questions on different samples.

Dimension	Meier 2020	Sharples-Harman-Lorandos 2025
Question	When mothers raise abuse and fathers cross-claim PA, who wins custody?	In cases where PA has been found, who has the substantiated abuse findings?
Sample	4,338 US custody cases (2005–2014)	492 US appellate cases where PA was found
Inclusion criterion	Custody case with abuse allegations + PA cross-claim	Appellate opinion records a PA finding
Outcome variable	Custody loss / win, by gender + alienation cross-claim	Substantiation rate of abuse claims, by alienating-vs-alienated role
Headline finding	Mothers’ custody-loss rate roughly doubles when fathers cross-claim PA	Alienating parents have 81.62% greater probability of substantiated abuse against them; alienated parents have 86.05% greater probability of unsubstantiated abuse against them
Primary axis	Gendered (mothers vs fathers)	Role-based (alienating vs alienated)
Pre-registration	No	Yes
Coder blinding	Not reported	Yes
Funding	US DOJ / NIJ grant	Funding source not visible from open abstract

Both findings can be true at the same time. Meier’s worry — that PA cross-claims are deployed strategically against abuse-alleging mothers in the broad population of custody cases, with effects on custody outcomes — is about how courts use PA as a label.

Sharples’s finding — that within confirmed-PA cases the substantiated abuse runs toward the alienating parent — is about who the abusers actually are when PA is confirmed.

The intellectual fight between the two camps is partly about which sample frame is more valid, and partly about which population we should care about. Both are legitimate questions; they require different studies to answer.

The article body should not let readers conclude “Sharples disproves Meier.” It should let them conclude “the two studies together describe a more nuanced picture than either one alone.”

Direction × substantiation, drawn to scale

The figure below is the 2-by-2 finding put alongside a clean visualisation of the 1,112 abuse claims by their two coding axes.

What They Found — Six Key Findings

1. 81.62% greater probability of substantiated abuse against the alienating parent

The first headline finding is the substantiated-claims direction. Across the 1,112 abuse claims coded in the 492 appellate cases, parents found by a court to have alienated their children had an 81.62% greater probability of having a third-party-substantiated abuse claim against them, compared with the alienated parent.

Substantiation in this study is defined narrowly — by court ruling, CPS investigation, or criminal conviction. Self-report does not count. A claim that the alienating parent’s lawyer disputed but a CPS investigation upheld counts as substantiated; a claim that was dismissed at preliminary hearing does not.

The interpretation matters more than the precise number. Substantiated abuse is the kind of abuse the institutional system has already acknowledged. In the cases where PA was confirmed, the institutional system had also already acknowledged that abuse was running toward the alienating parent at a rate substantially higher than toward the alienated parent.

This is the empirical floor under the family-violence classification of Harman, Kruk and Hines (2018). PA and other forms of corroborated abuse co-occur — they are not competing categories.

2. 86.05% greater probability of unsubstantiated abuse against the alienated parent

The mirror finding is the unsubstantiated-claims direction. Alienated parents had an 86.05% greater probability of having an unsubstantiated abuse claim made against them, compared with alienating parents.

Unsubstantiated does not mean false in every individual instance. It means the institutional system — court, CPS, prosecuting authority — did not find the underlying claim supported by sufficient evidence to act on. Some unsubstantiated claims are genuinely unfounded; others fail substantiation for procedural reasons.

Read alongside the first finding, the pattern is consistent with the coercive-control model: the alienating parent uses unsupported abuse claims against the alienated parent as a tactic, while themselves having a higher probability of being a corroborated abuser.

This is what the Harman-Kruk-Hines 2018 family-violence classification predicts. False allegations against the targeted parent are not random failures of the family-court system. They are part of the documented pattern.

3. 1,112 separate abuse claims across 288 cases — the texture of the data

The headline percentages emerge from 1,112 separate abuse claims coded across 288 cases. The texture matters.

288 of 492 cases (58.54%) involved abuse allegations. That is a high base rate — it tells you that in the appellate-level slice of confirmed-PA cases, abuse is on the table in roughly three out of five cases. PA findings and abuse allegations co-occur often.

The typical case with allegations contained roughly four claims. Some involved one allegation; many involved multiple, often running in both directions and across multiple categories — physical, emotional, financial, child-related. The 2-by-2 frame (direction × substantiation) is what makes the headline numbers interpretable across this complexity.

This volume of claims also means the analytic conclusions are not driven by a handful of dramatic cases. The pattern is robust across the full sample.

4. Pre-registered, blind coders, public sample — design rigour

The design itself is part of the finding. Pre-registered hypotheses meant the analytic plan was committed before the data was coded. Blind coders meant the people reading the appellate opinions did not know what the study was looking for.

The sample is US appellate case opinions — public documents. The Westlaw search strategy is in principle re-runnable. Substantiation is defined by third-party rulings, not by self-report. This is a much higher replicability standard than the clinical case-study and survey-based research the PA literature has historically relied on.

The design specifically addresses the methodological criticisms raised against the 2020 Harman-Lorandos precursor study by Mercer (2022) and Meier and colleagues (2022). Whether it succeeds in answering those critiques is itself contested — Varavei (2025) has published a reply arguing the correlation may be a coding artefact.

The point for a non-specialist reader is that this study was designed to be challenged, and the design makes the challenges productive rather than dismissive.

5. Findings are reported by role, not by gender

The single most deliberate framing choice in the paper is the gender-agnostic primary outcome. The 81.62% and 86.05% figures are reported by role — alienating versus alienated — not by sex.

The authors emphasise that the pattern holds regardless of the alienating parent’s gender. This is a direct response to the heavily gendered Meier 2020 framing, which structured its analysis around mothers versus fathers as the principal axis.

The role-based framing is more honest to the data. Alienation is a parental behaviour pattern, not a gendered one. Mothers and fathers both alienate; mothers and fathers are both targeted; the role someone plays in the family system is what predicts the substantiated-abuse pattern, not their sex.

This framing also defuses the most divisive part of the policy conversation. A research finding that says “the alienating parent’s role correlates with substantiated abuse” is harder to weaponise into a culture-war argument than one that says “fathers” or “mothers” are the problem.

6. What the paper does not claim — the honest framing

The strength of Sharples 2025 is partly what it explicitly does not claim.

It does not claim that PA is never used to cover up real abuse. It claims that in the specific subset of US appellate cases where PA was actually found by a court, the substantiated abuse findings ran toward the alienating parent. Those are different statements.

It does not claim that Meier 2020 is wrong. It addresses a different question on a different sample. The 4,338-case Meier study tells us about custody outcomes for abuse-alleging mothers in the broader population of custody disputes; the 492-case Sharples study tells us about substantiation patterns within confirmed-PA cases. Both findings can be true; they describe different populations.

It does not claim that every PA finding in every appellate case is correctly reasoned. The inclusion criterion treats PA findings as the input variable; the paper does not separately validate them.

It does not claim the family-court system is functioning well. A 58.54% rate of abuse allegations across PA-confirmed cases suggests significant institutional dysfunction, regardless of which direction the corroboration runs.

What the paper does claim — and supports — is that the broad-brush version of the cover-for-abuse argument does not survive contact with this data. That is a meaningful, defensible empirical contribution. It is also one piece of evidence in a larger debate, not a definitive ruling.

Why This Matters

The strongest argument against parental alienation as a recognised legal construct is that it is weaponised against genuine abuse victims. That argument has shaped policy, judicial training, advocacy framing, and the public conversation about post-separation family violence for nearly two decades.

If it is correct, every PA finding is a potential injustice. If it is correct in the broad-brush version critics deploy it in, the entire family-violence classification line of research — Harman 2018, Bernet 2022, Harman 2019 prevalence — risks being read as a sophisticated apparatus for shielding abusers.

Sharples 2025 is the first pre-registered, blinded-coder study to directly test that argument on appellate-record data. The finding cuts the other way.

In the cases where PA is found, the corroborated abuse runs from the alienating parent. The unsubstantiated allegations run toward the alienated parent. Neither pattern is what the broad-brush cover-for-abuse argument predicts.

The implications play out in three overlapping directions.

For the family-court system, Sharples 2025 is the empirical floor under the argument that PA findings should not be dismissed as a tactic. A judge or evaluator confronted with a PA pattern in a case with abuse allegations now has data to ground the question of which direction substantiation typically runs. The 81.62% / 86.05% figures are not a verdict on any individual case — but they are a meaningful prior.

For the policy conversation, the paper does not settle the Meier question. Meier’s claim — that PA cross-claims are deployed strategically against abuse-alleging mothers, with effects on custody outcomes — remains live. Both papers describe real patterns in different populations. A serious policy framework needs to engage both.

For alienated parents, the 86.05% figure is the answer to “is what is happening to me a recognised pattern?” Yes. Unsubstantiated abuse allegations against targeted parents are not random — they are part of the coercive-control pattern that Harman 2018 classifies as family violence.

The fact that the institutional system often does not read the pattern correctly is a separate problem from whether the pattern exists. The pattern exists, and now it has a defensible empirical citation.

What This Means for You

The practical weight of a direction-of-abuse paper depends on what you do with it.

If you are a targeted parent who has been falsely accused. Sharples 2025 gives you a research-anchored answer to the question “is what is happening to me a recognised pattern?” The 86.05% figure says yes — false allegations against targeted parents are part of the documented pattern. Document everything; pursue substantiation through the legal channels available to you; and use Sharples 2025, alongside Harman 2018 and the Five-Factor Model, as the citation backing for what you are experiencing.

The recognition is not the end of the legal fight. It is the start of a defensible record.

If you are a lawyer or therapist reading this. The paper is the most rigorous case-record evidence currently available on the direction-of-abuse question in PA-confirmed cases. Pre-registered, blinded-coder, public-sample appellate research is unusual in this field. The 81.62% / 86.05% figures are appropriate for citation in expert witness reports, custody evaluation summaries, and judicial training materials — with the caveat that they describe the appellate-confirmed-PA subset, not the broader custody-dispute population.

The most important pairing is with Meier 2020. A clinically responsible expert engages both, not one. The two papers describe complementary patterns at different levels of the family-court funnel.

If you are a policy professional or family-court reformer. The headline implication is that the broad-brush “PA is cover for abusers” framing is empirically weak in the population the paper measures. Whatever institutional dysfunction the family-court system has — and a 58.54% rate of abuse allegations across PA-confirmed appellate cases is a strong signal of dysfunction — it is not principally a dysfunction in which PA findings systematically attach to families where the alienated parent is the corroborated abuser. The dysfunction runs the other way.

A reform agenda built on the assumption that PA findings are typically weaponised against survivors will misallocate institutional energy. The data suggests the substantive problem is the volume of unsubstantiated allegations against targeted parents, not the volume of PA findings against survivor-mothers.

If you are an advocate who has cited Meier 2020. Sharples 2025 does not refute Meier’s specific finding about custody outcomes. Both papers describe real patterns; a serious advocacy position now engages both.

The version of the argument that says “PA is always a tactic against survivors” is no longer empirically defensible at the population level. The version that says “PA cross-claims have specific impacts on custody outcomes for abuse-alleging mothers in the broader population of US custody disputes” remains defensible — and is what Meier actually argues.

The research literature is more nuanced than the policy debate. The advocacy work that runs on the more nuanced version will be more durable.

Limitations — What the Study Doesn’t Tell Us

Four honest limitations, named plainly.

First, appellate cases are roughly 5% of trial-court rulings. They are systematically the high-conflict, well-resourced end of custody disputes — not a random sample of the millions of US separations that never reach appellate review.

A pattern in 492 appellate cases does not necessarily generalise to the full population of custody disputes. This is the core Mercer / Meier critique of the entire Harman-Lorandos appellate-records research program. Sharples 2025 inherits the limitation; it does not solve it. The natural next study is replication on a non-appellate sample.

Second, the inclusion criterion conditions on PA being found. Critics including Mercer (2022) argue that PA findings in trial courts are themselves unreliable, so a sample selected on that variable inherits the unreliability. Defenders respond that this is exactly why blinding the coders matters — the coders are not independently validating the PA findings; they are taking the appellate court’s finding as given and asking what abuse pattern co-occurs with it.

Both sides have a point. The paper’s claim is appropriate for the sample frame it uses; whether the frame is the right one is a separate question.

Third, the paper does not directly test Meier’s claim. Meier 2020 asks about custody outcomes for abuse-alleging mothers in the whole population of custody cases. Sharples’s design cannot test that claim because it conditions on PA being found, not on PA being merely alleged. The paper is best read as orthogonal to Meier’s claim, not refutational of it.

The article body has tried to be clean about this. The LOE site should not let downstream summarisation overstate the contradiction. Both papers describe real patterns; they describe different populations.

Fourth, several methodological details are behind the Springer paywall. Items flagged for verification when full-text access is available include:

• Type-of-abuse stratification (was the substantiated abuse mostly domestic violence, child physical, child sexual, or emotional abuse?)
• Inter-rater reliability statistics
• Effect-size confidence intervals
• Year range of cases analysed
• Funding source and conflict-of-interest disclosure

The directional finding (81.62% / 86.05%) is robust across the open-access secondary sources that have summarised the paper, including the National Parents Organization, the Springer landing page, and the Varavei 2025 critic-camp reply that quotes the paper directly. Confidence in the headline finding is appropriate; confidence in any specific sub-analysis requires the full paper.

Varavei (2025) has published a critic-camp reply arguing the correlation may be a coding artefact — that cases where one party is already painted as the “bad parent” by the trial court will accumulate both PA and abuse findings against them. The reply has not yet been engaged by the paper’s authors in a published rejoinder. A responsible reader of Sharples 2025 holds Varavei’s critique alongside the headline finding, not against it.

None of these limitations is a reason to discard the paper. They are reasons to cite it precisely — Sharples, Harman & Lorandos (2025) for the direction-of-abuse pattern in the appellate-confirmed-PA subset; Meier and colleagues (2020) for custody outcomes in the broader population — and to engage the still-open methodological questions honestly. The aim of a good article about a paper is to leave the reader better able to cite it well. That is the aim of this one.