The honest version is more reassuring than the popular one.

On average, children whose parents divorce score slightly worse than children whose parents stay married — and the gap is small enough that the two groups overlap almost completely

Two meta-analyzes covering roughly 160 studies find real but small average differences on schoolwork, conduct, psychological adjustment, self-concept and social relationships. The median difference in the 1991 analysis was 0.14 of a standard deviation. That is a difference you can detect across thousands of children and cannot reliably see in any one child. Better-designed studies tended to find smaller differences, not larger ones.

Amato and Keith (1991) pooled 92 studies comparing children in divorced single-parent families with children in continuously intact families. Children of divorce scored lower across a range of outcomes, with a median effect size of 0.14 standard deviations. The authors reported two things that get dropped in retellings: methodologically stronger studies produced weaker effect sizes for some outcomes, and more recent studies produced weaker effect sizes than older ones.

Amato's 2001 update analyzed 67 studies published in the 1990s. Children with divorced parents again scored significantly lower on academic achievement, conduct, psychological adjustment, self-concept and social relations. The pattern across decades was curvilinear, not linear: the gap narrowed through the 1980s and widened again in the 1990s.

Lansford's 2009 review in Perspectives on Psychological Science notes that in the 1991 meta-analysis 70 percent of the 92 studies showed lower well-being for children of divorce, and that in the 2001 update 88 percent pointed the same way but only 42 percent of those comparisons reached statistical significance. A large majority of small, mostly non-significant differences is what a small real effect looks like.

A separate 2019 meta-analysis of long-term mental health outcomes (54 studies, 117 effect sizes, 506,299 participants) found elevated odds across every outcome examined — depression OR 1.29, anxiety 1.12, suicide attempt 1.35, suicidal ideation 1.48, psychological distress 1.48, alcohol 1.43, smoking 1.64, drug use 1.45 — and that these effect sizes decreased between 1990 and 2017. Odds ratios of this size mean a modest increase in risk applied across a whole population, not a likely outcome for a given child.

None of this literature is experimental. Nobody randomises families to divorce. Every figure here is an association, and associations of this size are exactly the ones most vulnerable to being explained by something other than the separation itself.

Most children recover to look like their peers on measures of disorder, while still carrying painful feelings about the divorce years later — both things are true at once

The resilience finding is real: the large majority of children of divorce do not end up in the clinical range on symptom measures. But a well-designed study that asked young adults about painful feelings, beliefs and memories rather than about symptoms found those were substantially more common in people whose parents divorced, in the same people who reported few psychological symptoms and said they accepted the divorce. Researchers call this the difference between disorder and distress.

Kelly and Emery's 2003 review in Family Relations is the standard citation for the resilience position. It reviews the longer-term adjustment literature from the twin perspectives of elevated risk and protective factors, and concludes that the majority of children demonstrate resilience. It also addresses head-on the controversy about adult children of divorce.

Laumann-Billings and Emery (2000) set out to test whether the research literature and the clinical literature were actually contradicting each other. In two studies — college students, then a community sample of young people from low-income divorced families — participants reported significantly more painful feelings, beliefs and memories about their parents' divorce than peers from married families, while simultaneously reporting that they accepted the divorce and having few psychological symptoms. Greater distress was associated with the child's living arrangements, how often they saw their father, interparental conflict, and psychological symptoms.

Two figures put a number on the tail. Chase-Lansdale, Cherlin and Kiernan (1995), following a British birth cohort to age 23, estimated that the share scoring in the clinical range on an emotional-disorder measure rose from 8 percent to 11 percent among those whose parents had divorced. Hetherington and Kelly (2002) reported a wider gap in their own sample: 25 percent of adults from divorced families with serious long-term problems against 10 percent from continuously married families. Both are reported in Lansford's 2009 review.

Amato's figure is the one most often misread. His estimate that around 10 percent is the gap between the two groups is a statement about how far the two distributions fail to overlap — roughly 90 percent of them do. It is not a prediction about any individual child, and it does not say what would have happened to a particular family had the marriage continued.

There is no reliable, replicated figure for how long acute distress lasts. The two-year figure that circulates widely traces back to Hetherington's longitudinal work on a specific sample and is a reasonable rule of thumb, not a finding that has been pooled across studies. Say it as a rule of thumb or not at all.

The children who do badly are largely identifiable in advance: those who were already struggling, those exposed to sustained conflict, those who lose a parent relationship, and those who go through repeated family upheaval

This is the most useful thing on this page. Poor outcomes after divorce cluster among children with pre-existing behavioral or emotional problems, children whose parents keep fighting through and after the separation, children who effectively lose a relationship with one parent, and children who experience several changes of family structure rather than one. Where studies control for how the child was doing before the split, the divorce effect shrinks sharply.

Pre-existing adjustment is the single most deflating variable in this literature. Lansford's review reports that children with pre-divorce behavior problems have worse long-term outcomes, and cites work by Sun and Li finding that academic achievement differences could be accounted for almost entirely by the child's achievement and the family's functioning before the divorce. Chase-Lansdale and colleagues found the opposite pattern for previously well-adjusted children: a steeper acute rise in problems at the point of separation, from a better starting place.

Sustained interparental conflict is the second. It is covered in its own topic below because it is also the main causal mechanism proposed for the average effects.

Repeated transitions matter separately from any one transition. Fomby and Cherlin (2007) used a nationally representative two-generation longitudinal survey to test whether the number of family-structure changes predicts child outcomes, or whether that association reflects the mother's characteristics before the child was born. For white children, the association with cognitive outcomes was largely explained by the mother's prior characteristics, but the association with behavioral outcomes appeared to be causal in part. They found no robust effect of number of transitions for black children. This is the honest version: instability probably matters for behavior, some of what looks like instability is really selection, and the pattern is not the same across groups.

Age at the time of the separation moderates but does not dominate. Lansford reports that children who were in kindergarten through fifth grade showed more internalizing and externalising problems than those in grades six to ten, while the older group showed larger academic declines. Malone and colleagues found that divorce during elementary school predicted a sustained rise in boys' externalising behavior, whereas divorce during middle school produced a temporary rise that then fell away.

Gender findings are genuinely mixed and should not be stated confidently in either direction. Amato and Keith's meta-analysis found smaller effect sizes for African American children than for European American children; later work has not produced a stable explanation for that.

Conflict between the parents, not the legal end of the marriage, is the best-supported mechanism — and it works by damaging parenting, not by upsetting the child directly

Amato and Keith's meta-analysis found the most consistent support for a family conflict explanation rather than parental absence or economic loss. A 2020 meta-analysis of 115 samples and nearly 25,000 divorced families then traced how conflict does its damage: it degrades parenting, and degraded parenting predicts children's internalizing and externalising problems. Reducing what a child witnesses and is drawn into is the highest-yield thing available to a separating parent.

Van Dijk, van der Valk, Deković and Branje (2020, Clinical Psychology Review) coded 2,257 correlations from 115 samples covering 24,854 divorced families and then tested mediation models. Three findings matter for a parent. First, the raw correlations between interparental conflict, parenting and child adjustment are mostly significant and in the expected direction, but small. Second, parental support, hostility, structuring, intrusiveness and role diffusion all functioned as mediating mechanisms linking conflict to children's problems — dyadic parent-child processes did not. Third, negative parenting behaviors appeared to matter more than positive ones, and parental role diffusion — pulling the child into an adult role, confidant or ally — was singled out as a particular risk in the context of post-divorce conflict.

The construct of the child 'feeling caught in the middle' comes from Buchanan, Maccoby and Dornbusch's 1991 study of adolescents in divorced homes in Child Development, which produced the Feeling Caught scale still used in this literature. It is the operational version of what parents intuitively call putting the child in the middle.

Cao, Fine and Zhou's 2022 review in Clinical Child and Family Psychology Review proposes a model in which pre- and post-divorce conflict shape children's adaptation trajectories through the child's own cognitive and emotional appraisals, coparenting quality and parent-child relationships. Their central argument is that the family environment during and after the divorce, rather than the legal event, drives adjustment. They also argue that treating conflict as a single undifferentiated thing is a mistake: frequency, intensity, and whether it is overt or covert affect children differently.

One caution about the direction of the arrow. Almost all of this is correlational, and children who are already struggling can increase the strain between their parents. The mediation models are the best available evidence, not proof of causation.

Children leaving a high-conflict marriage tend to do better after the divorce; children leaving a low-conflict marriage tend to do worse — and most divorces are of the second kind

Booth and Amato's national longitudinal study found that ending a high-conflict marriage was associated with better outcomes for offspring, while ending a low-conflict marriage was associated with worse ones — on relationship quality, social support and general psychological well-being. This is the strongest available answer to 'should we stay together for the children', and it is contested: it comes from one research group's analyzes of one survey, and it is an association between family types, not a prediction for a particular family.

Amato, Loomis and Booth (1995, Social Forces) reported that offspring well-being in early adulthood depended on the level of marital conflict that preceded the divorce. Booth and Amato's 2001 paper in the Journal of Marriage and Family extended it in two parts. The first found that dissolving low-conflict marriages appeared to have negative effects on offspring's later intimate relationships, social support from friends and relatives, and general psychological well-being, while dissolving high-conflict marriages appeared to have beneficial effects.

The second part of that paper is the part usually left out, and it is the more interesting one. Parents in low-conflict marriages who divorced differed from other parents beforehand: they were less integrated into their communities, had fewer impediments to divorcing, held more favorable attitudes toward divorce, were more predisposed to risky behavior, and were less likely themselves to have had divorced parents. So the low-conflict divorces that look harmful are also the divorces of a distinguishable kind of household.

Hetherington, Bridges and Insabella's 1998 American Psychologist review supplies the other half of the picture. As reported by Lansford, children in high-conflict intact marriages scored 0.32 standard deviations below children in low-conflict intact families on adjustment measures, and 0.12 standard deviations below children in divorced families. That is the empirical basis for the claim that a bad marriage can be worse for a child than a divorce.

Amato, Loomis and Booth estimated that high-conflict marriages accounted for something between 30 and 49 percent of divorces in their data, which means the majority of divorces are of the type where the research points the other way. This is uncomfortable and it is what the literature says.

None of this can tell an individual parent what to do. The comparison is between averages of two groups, and it says nothing about a marriage that is low in overt conflict but corrosive in ways a survey instrument does not capture.

Studies designed to separate genetic inheritance from the divorce itself reach different answers for different outcomes — and their results have not converged

Children-of-twins designs compare the children of identical twins who differ in whether they divorced, which strips out shared genetic and family background. Three such studies from the same research group reached partly conflicting conclusions: some outcomes look causally connected to the separation, others look like inherited liability that would have shown up anyway. A reader should know that the causal question is open, and that some of what is attributed to divorce is not caused by it.

D'Onofrio and colleagues (2005, Journal of Abnormal Psychology) compared offspring of adult twins in Australia and concluded that environmental influences associated with divorce accounted for higher rates of drug and alcohol, behavioral and internalizing problems — consistent with a causal connection.

The same group's 2006 paper in Developmental Psychology found results consistent with a causal role for divorce in earlier first sexual intercourse and emotional difficulties, plus greater probability of educational problems, depressed mood and suicidal ideation — but concluded that the increased risk for cohabitation and earlier drug initiation was explained by selection factors, including genetic confounds.

Their 2007 paper in the Journal of Child Psychology and Psychiatry reached a third pattern: the association with offspring substance use problems held up after controlling for genetic and environmental risk from the twin parent, consistent with causation, while the increased risk of internalizing problems in adult offspring appeared to be accounted for by shared genetic liability between parents and children — an artefact of selection.

The authors say plainly that the children-of-twins design controls only for one parent's genetic and environmental characteristics, and that additional genetically informed analyzes are needed. Lansford's review notes that the conclusions across these samples are discrepant and that the inconsistencies remain unresolved.

The practical upshot is not that divorce is harmless. It is that the honest sentence is 'divorce is associated with these outcomes, and for some of them the association may not be caused by the divorce', and that anyone stating a clean causal claim in either direction is going beyond the evidence.

A large share of what looks like the effect of divorce on children is the effect of a household's income falling — which is the one mechanism that can be directly offset

Amato and Keith's meta-analysis found some support for an economic-disadvantage explanation of children's outcomes, alongside stronger support for conflict. US Census data show the scale of the gap between household types: in 2024, 4.3 percent of married-couple families were below the poverty line, against 21.8 percent of families with a female householder and no spouse present. Economic consequences of divorce for mothers have fallen since the 1980s as women's earnings and child support receipt rose.

Amato and Keith tested three competing explanations for the outcome gap: parental absence, economic disadvantage and family conflict. Economic disadvantage received some support; family conflict received the most consistent support. Both mechanisms are in the data.

The Census figures are cross-sectional comparisons of family types, not measurements of what a divorce does to a particular household, and they include households that were never married. In 2024 the poverty rate was 8.1 percent for all families, 4.3 percent for married-couple families, 11.0 percent for families with a male householder and no spouse present, and 21.8 percent for families with a female householder and no spouse present. In 1993 the corresponding figure for female-householder families was 35.6 percent, so the gap has narrowed substantially over three decades while remaining large.

Tach and Eads (2015, Demography) used Survey of Income and Program Participation panels from 1984 to 2007 to track how the economic consequences of union dissolution changed over time. They found that the economic consequences of divorce declined since the 1980s, driven by growth in married women's earnings and their receipt of child support and income from personal networks, while the consequences of cohabitation dissolution worsened over the same period to closely resemble those of divorce.

The reason to put this in front of a parent is that it is the mechanism most amenable to direct action. Conflict requires two people to change behavior. Income can be affected by child support enforcement, benefits take-up, school meal and fee-waiver programs, and by not moving a child out of a school catchment if it can be avoided.

School-based groups for children of divorce have been running for forty years and have been tested in trials; the pooled evidence says they improve how children feel about the divorce and their self-esteem, and does not show effects on mental health

The Children of Divorce Intervention Program, developed in 1982 and still distributed by the Children's Institute in Rochester, is the most-tested school program, starting with a randomised trial in 1985. But a 2023 meta-analysis of 30 studies covering 4,344 children found no significant effects on children's mental health outcomes. It did find significant effects on children's adjustment to the divorce specifically and on self-esteem. That is a real but narrower benefit than the program literature usually claims.

The Children of Divorce Intervention Program (CODIP) runs 12 to 15 group sessions in school, with age-specific manuals and materials for kindergarten through eighth grade, led by trained group leaders. The original trial (Pedro-Carroll and Cowen, 1985, Journal of Consulting and Clinical Psychology) randomly assigned 75 fourth- to sixth-graders to a 12-week program or a delayed control; teachers rated improvements in shy-anxious behavior and learning problems, and children reported significantly less anxiety than controls. Alpert-Gillis, Pedro-Carroll and Cowen (1989) adapted it for urban second- and third-graders and found greater improvement in feelings about self and family and in coping. Several later evaluations used quasi-experimental rather than randomised designs.

The Children's Support Group is the other long-standing school program, a 14-week cognitive-behavioral group for ages 7 to 13 covering problem solving, anger control, impulse control, communication and relaxation. Stolberg and Garrison (1985) found improved self-concept at post-test and five-month follow-up against a no-treatment control; Stolberg and Mahler (1994), with 103 children aged 8 to 12, reported significant reductions in internalizing and externalising problems at post-test and one-year follow-up.

Herrero, Roca, Cormenzana and Martínez-Pampliega (2023, Family Process) is the most rigorous synthesis available. Thirty studies, 4,344 individuals, 258 effect sizes, random-effects analysis. Their headline conclusion is that these interventions affect specific variables rather than children's global adjustment: there were no significant effects on children's mental health outcomes, and none of these were qualified by any of the moderators examined. There were significant effects on children's divorce adjustment and on self-esteem. Of 20 moderators tested, only one — child age, affecting family functioning — had a significant effect, and the authors themselves say to treat that cautiously because it rested on few studies.

The gap between the program literature and the meta-analysis is worth naming. A 2021 review chapter in the Encyclopedia on Early Childhood Development describes child-focused programs as producing an array of improvements. Three of its four authors are from the research group that developed one of the parent programs reviewed. The meta-analysis, by an unaffiliated Spanish group, is more conservative, and it is the design better suited to the question.

The same review concedes something important about availability: despite the evidence base, none of these programs has been widely implemented, which it attributes to the cost of group programs.

The strongest long-term results in this entire field come from a program that never worked with the children at all — and its effects shrank sharply when it was scaled up

The New Beginnings Program teaches divorced parents specific parenting and discipline skills over ten group sessions. In a randomised trial of 240 families, benefits were visible at six years and still visible at fifteen, concentrated among the children who started at highest risk. Then a much larger effectiveness trial in real family courts found only small parent-reported effects at post-test and no significant main effects at ten months. Both halves are the finding.

New Beginnings was developed by Sharlene Wolchik at Arizona State University's REACH Institute. The format is ten weekly two-hour group sessions led by master's-level clinicians, plus two individual phone sessions, covering listening skills, anger management, effective consequences and positive family activities. Blueprints for Healthy Youth Development, which rates prevention programs against independent evidence standards, rates it Promising.

In the main trial, 240 divorced mother-child pairs were randomised to a mother-only intervention, a mother-and-child intervention, or a literature control. Effect sizes at post-test were small to moderate, from .22 to .57, on internalizing and externalising problems, with improvements in mother-child relationship quality, discipline and conflict. At six years, the combined intervention groups had better grade-point averages, fewer sexual partners, fewer mental health diagnoses and greater academic competence. The moderation result is the striking one: among children who started at highest risk, mental disorder diagnoses fell from 46.1 percent in the control group to 25.0 percent in the intervention group. At fifteen years there was roughly a 35 percent reduction in major depression diagnoses, fewer days in jail, and fewer mental health services used, with odds ratios of .26 to .34 for internalizing disorder.

The fifteen-year results also include null findings that are usually omitted: no significant main effects on externalising disorders, no significant main effects on marijuana or alcohol use frequency, and no effect on recent internalizing and externalising problems.

The effectiveness trial (Sandler et al., 2020) took the program into Arizona family courts with 830 parents, mothers and fathers, an ethnically diverse sample and an active control. Average attendance was 5.59 of 10 sessions. At post-test there were parent-reported improvements in relationship quality (d = 0.22), discipline, and children's internalizing, externalising and total problems. At the ten-month follow-up there were no significant main effects at p < .05. Blueprints did not certify this study, and notes that the only significant post-test effects were parent reports, from parents who were themselves involved in delivering the program.

There is a second null worth reporting: reviews of this literature note that combining parent-focused and child-focused programs did not produce additive effects. Running both did not beat running one.

Court-mandated parent education more broadly has been meta-analyzed. Fackrell, Hawkins and Kay (2011) found an overall effect of d = .39 across 19 studies with a no-treatment control, with specific outcomes ranging from d = .19 to .61 — and stated plainly that there were important methodological limitations in that body of research.

One online program built for children to work through themselves has a randomised trial behind it — and the same trial found it slightly worsened things for the lowest-risk children

Children of Divorce-Coping with Divorce (CoD-CoD) is a five-module interactive online coping-skills program for 11- to 16-year-olds. In a randomised trial of 147 young people it improved self-reported emotional problems (d = .37) and reduced the rate of clinically significant self-reported problems. It worked best for the highest-risk half of the sample. For the 10 percent with the lowest parent-reported baseline risk, parents reported the control condition did better.

Boring, Sandler, Tein, Horan and Vélez (2015, Journal of Consulting and Clinical Psychology) recruited children aged 11 to 16 from public court records after their families filed for divorce, blocked them by risk score, and randomly assigned them either to CoD-CoD or to an internet self-study control condition called Best of the Net.

Main effects: young people in CoD-CoD improved more on self-reported emotional problems (d = .37) and had a lower rate of clinically significant self-reported mental health problems (OR = .58, p = .04).

The interactions are the interesting part. The 55 percent of young people with the highest baseline problems improved more than controls on total self-reported mental health problems. The 30 percent with the lowest baseline coping efficacy improved more on coping efficacy. And for the 10 percent with the lowest parent-reported baseline risk, those in the control condition had lower problems than those in CoD-CoD.

That last result is the one that should change how a parent thinks about handing a child anything. A program that helps distressed children can make a settled child worse, presumably by making a problem salient that the child was not experiencing. It is the same logic that makes universal grief interventions a bad idea, and it argues against giving a coping program to a child who is coping.

No public enrollment route for CoD-CoD could be found. It is described here because it is the best-evidenced thing aimed at children directly, and because its risk-moderation finding generalises even if the program itself is not available to you.

Individual therapy is the right answer for a child with a problem, not for a child having a hard year — and the clinician you want is one who will work with the parents too

There is no trial evidence that individual therapy for an otherwise-healthy child improves adjustment to a parental separation, and the pooled evidence on divorce-specific programs found nothing on mental health outcomes. Therapy is indicated when there is a problem to treat. Professional-body guidance is clear about how to find and choose a clinician and about the difference between the disciplines.

The order of operations that the evidence supports is: reduce what the child is exposed to, protect both parent relationships, restore routine, wait, and refer if a problem appears or persists. Referring first is common and is not supported by the intervention literature, where the largest long-term effects came from working with the parent.

AACAP's guidance on where to find help (Facts for Families No. 25, October 2017) names the starting points: friends and family, a spiritual counselor, the child's school counselor, the child's pediatrician, employee assistance programs, local medical and psychiatric societies, county mental health departments, hospitals with psychiatric services, and professional organizations.

The same fact sheet distinguishes the disciplines, which matters when you are choosing. A child and adolescent psychiatrist is a physician with two additional years of advanced training beyond general psychiatry, who can evaluate, treat and prescribe. A psychologist holds a master's or doctoral degree and provides evaluation, treatment and psychological testing. A social worker typically holds a master's degree and provides psychotherapy. AACAP's stated criteria for choosing are the professional's training and experience, and finding a comfortable match between the child, the family and the clinician.

One thing to add to that list from the outcome literature rather than from guidance: the mechanisms that predict children's adjustment after separation run through parenting and through what the child is exposed to. A clinician who will see the child alone for a year and never speak to either parent is not working on the variables the evidence points at. Asking, at the first appointment, how the clinician involves parents is a reasonable question with an evidential basis.

This is not therapy advice and cannot substitute for an assessment. Where a child's presentation includes risk of harm to themselves, that is not a referral question; see the next topic and the crisis lines in the resources list.

The signals that distinguish a child who needs help from a child who needs time are about function and duration, not about sadness

Professional-body guidance converges on the same idea: it is not the presence of distress that indicates a problem but its effect on how the child functions and how long it goes on. The AAP's threshold is that the reactions become excessive; AACAP routes parents to a pediatrician and then to a child and adolescent psychiatrist for evaluation. Neither body publishes a checklist specific to divorce, and this guidance rests on clinical consensus rather than trial evidence.

The AAP's guidance on adjusting to divorce (last updated July 2026) describes typical reactions by developmental stage: preschoolers with sadness, fear of separation from a parent, sleep and eating disruption, regression to babyish behavior and self-blame; school-aged children who become moody, distracted at school, aggressive, or who wish their parents would reconcile while worrying they caused the split; adolescents who withdraw socially, take risks, take on adult responsibilities prematurely, or become depressed. Its stated threshold is: if any of these behaviors become excessive, talk with your child's doctor.

AACAP's fact sheet describes younger children becoming more aggressive and uncooperative or withdrawing, older children feeling deep sadness and loss with schoolwork suffering and behavior problems common, and difficulties with relationships and self-esteem persisting into adolescence and adulthood. It says a family doctor or pediatrician can refer for evaluation and treatment.

Neither body ties a specific duration to the threshold. The practical translation of 'excessive' that follows from the literature rather than from guidance is: the behavior is new or much amplified, it is present across settings rather than only at handovers, it is interfering with school or friendships, and it has not eased over a period of months rather than weeks.

Two things are different in kind and are not on this scale. Any statement or sign suggesting the child may harm themselves, and any indication of harm to the child in either household, are immediate matters. In the United States, 988 reaches the Suicide and Crisis Lifeline by call or text.

Be aware of a second-order signal that is easy to miss. The distress literature says a child can be functioning perfectly and still be carrying a great deal. A child who says nothing at all about the divorce is not necessarily a child who is fine, and is also not necessarily a child who is unwell.

When researchers ask children directly, they describe a crisis they actively coped with, and they say adults underestimated them

The qualitative research on children's own accounts is smaller and mostly British, but it is consistent. Children describe parental divorce as a crisis and describe themselves as competent participants who mobilised internal and external resources — not as passive casualties. The researchers' conclusion is that cultural assumptions about children get in the way of adults hearing them. Separately, children in a 238-child community study named friends, not parents, as their key confidants.

Butler, Scanlan, Robinson, Douglas and Murch (2002, Children & Society) reported on a study of children's experience of divorce. Their findings were that children experience parental divorce as a crisis in their lives, that they are able to mobilise internal and external resources to reach a new equilibrium, and that in doing so they demonstrate how far they are active and competent participants in the process of family dissolution rather than objects of it. They explicitly discuss the cultural presumptions that militate against hearing what children have to say about their experiences.

Dunn, Davies, O'Connor and Sturgess (2001, Journal of Family Psychology) studied 238 children from a community sample across stepfamilies, single-parent families and intact families, focusing on confiding and communication — including children's recall of communication about family transitions. Friends emerged as key confidants, and children's family lives, including whether they were drawn into parental conflict, were linked to their friendships.

Cashmore's 2011 paper in Children and Youth Services Review reviews theoretical approaches to children's participation in family law decision-making, and is the standard reference for the distinction between a child having a voice and a child being asked to choose. That distinction is the one children in the qualitative work reliably draw themselves: they want to be informed and consulted, not made the decider.

A caution about how to use this. These are qualitative and mostly UK studies with modest samples. They are good evidence of what children say and weak evidence of what would happen if you acted differently. There is no trial showing that consulting a child more produces better adjustment.

One practical implication follows directly from the Dunn finding rather than from any program: a child's friendships are load-bearing during a separation. Protecting a child's access to their friends — not changing schools if avoidable, keeping sleepovers possible across both households, not scheduling the parenting time so that weekend social life becomes impossible — is supporting the confiding relationship the children themselves used.

Both parents keep full rights to a child's school records after a divorce unless a court order says otherwise — and the school needs the practical facts, not the story

Under FERPA, US schools must generally give both custodial and non-custodial parents the same rights to their children's education records, unless the school has been given evidence of a court order, state law or other legally binding document that says otherwise. Schools can also help in specific ways: watching for changes, running short-term groups, and adjusting the practicalities of who is contacted and who collects. What they cannot do is provide therapy.

The US Department of Education's Student Privacy Policy Office states: unless a school is provided with evidence that there is a court order, state law, or other legally binding document relating to such matters as divorce, separation or custody that specifically provides to the contrary, FERPA gives custodial and non-custodial parents alike certain rights with respect to their children's education records. The authority is 34 CFR § 99.4. A school may ask for legal certification denoting parenthood, such as a birth certificate or court order, from the parent requesting access.

Two things follow. If you are the parent the child does not live with, you are entitled to the records unless an order says otherwise, and you should not be told to ask the other parent for them. If there is an order restricting a parent's access, the school cannot act on it unless someone gives them a copy.

What is worth telling the school is operational: that the parents have separated, which addresses the child now lives at and on which days, who should be contacted for what, who is authorized to collect, whether report cards and portal access need to go to two households, and whether both parents should be invited to parent conferences. Schools handle this routinely; they mishandle it when the information arrives verbally from one parent months late.

What the school can offer is bounded by ASCA's own standards. School counselors provide small-group counseling that is planned, goal-focused, evidence-based and short-term, and refer out when a student needs long-term counseling or family therapy. Best practice, per ASCA, includes informing parents of a student's participation in a small group and its purpose, and obtaining informed consent covering the group's purpose, the limits to confidentiality, and that participation is voluntary.

The most valuable thing a school can actually do is observational. They see a large sample of children of this age daily, in a setting you do not, and they can tell you whether the change you are worried about is showing up there too.

Books for children about divorce are a custom with almost no outcome evidence, but the good ones do a specific job: give a child accurate words and normalize having two homes

No trial evidence supports handing a child a book as an intervention, and the CoD-CoD result is a caution against giving structured material to a child who is not distressed. What picture books and workbooks do well is more modest: they supply vocabulary, show the child that other children live in two homes, and give an adult and a child a shared object to talk around. Every book below was checked for author and publisher.

Ages roughly 2 to 6. Two Homes by Claire Masurel, illustrated by Kady MacDonald Denton (Candlewick Press, 2001) is the standard first book: no conflict, no explanation, just a child who has two of everything and is loved in both places. Living with Mom and Living with Dad by Melanie Walsh (Candlewick, 2012; published in the UK by Walker as Living With Mum and Living With Dad) does the same job with lift-the-flap pages. Mama and Daddy Bear's Divorce by Cornelia Maude Spelman, illustrated by Kathy Parkinson (Albert Whitman, 1998) is for a child who needs reassurance about being loved and about it not being their doing. Todd Parr's The Family Book (Little, Brown, 2003) is not about divorce at all; it is about families being different shapes, which is sometimes the more useful book.

Ages roughly 4 to 8. It's Not Your Fault, KoKo Bear by Vicki Lansky, illustrated by Jane Prince (The Book Peddlers, 1997) is written so that each spread carries a note to the parent as well as text for the child. Was It the Chocolate Pudding? A Story for Little Kids About Divorce by Sandra Levins (Magination Press, the children's imprint of the American Psychological Association, 2005) is built specifically around a young child's belief that they caused it. I Don't Want to Talk About It by Jeanie Franz Ransom, illustrated by Kathryn Kunz Finney (Magination Press, 2000) is for the child who will not discuss it. My Family's Changing by Pat Thomas (Barron's, 1999; later Hachette Children's Group) is part of a first-questions series and asks questions rather than telling.

Ages roughly 5 to 9. Dinosaurs Divorce: A Guide for Changing Families by Laurene Krasny Brown and Marc Brown (Little, Brown, 1986) is the most widely used book in this category and is also forty years old; its cartoon-panel format works, its assumptions about custody arrangements are dated. Standing on My Own Two Feet by Tamara Schmitz (Price Stern Sloan, 2008) and When My Parents Forgot How to Be Friends by Jennifer Moore-Mallinos (Barron's, 2005) both center a child narrating their own situation. Divorce Is the Worst by Anastasia Higginbotham (Feminist Press at the City University of New York, 2015; later Dottir Press) is angrier and less soothing than the rest, which some children want.

Ages roughly 8 to 12. What in the World Do You Do When Your Parents Divorce? A Survival Guide for Kids by Roberta Beyer and Kent Winchester (2001) is written to the child in question-and-answer form; the edition we could verify is a Tandem Library library binding, so check which imprint you are buying. The Divorce Workbook for Children by Lisa M. Schab (Instant Help, 2004; New Harbinger, 2008) is an activity workbook rather than a story. Mom's House, Dad's House for Kids by Isolina Ricci (Fireside/Simon & Schuster, 2006) is a practical handbook for a child living between two households. For fiction, Weekends with Max and His Dad by Linda Urban (Houghton Mifflin Harcourt, 2016) is quiet and unsensational; The Suitcase Kid by Jacqueline Wilson (Doubleday, 1992) is the British classic on the subject and is told entirely from the child's point of view.

Teenagers. The Divorce Workbook for Teens by Lisa M. Schab (Instant Help, 2008; New Harbinger, 2008) is the standard structured option and is the closest thing in print to the skills that CoD-CoD teaches. Now What Do I Do? A Guide to Help Teenagers with Their Parents' Separation or Divorce by Lynn Cassella-Kapusinski (ACTA Publications, 2006) is a workbook from a Catholic publisher, which some families will want and others will not. How It Feels When Parents Divorce by Jill Krementz (Alfred A. Knopf, 1984) is nineteen children and teenagers, aged 8 to 16, each telling their own story with a photograph; it is dated in its details and remains the book teenagers most often say made them feel less strange.

For the parent rather than the child. The Truth About Children and Divorce by Robert E. Emery (Viking, 2004; Plume, 2006) and Two Homes, One Childhood by the same author (Avery, 2016) are written by a researcher who publishes in this literature. Putting Children First by JoAnne Pedro-Carroll (Avery, 2010) is by the developer of CODIP. For Better or For Worse: Divorce Reconsidered by E. Mavis Hetherington and John Kelly (W. W. Norton, 2002) is the popular account of the longitudinal study behind several of the figures on this page — and should be read knowing that its 25-percent estimate is at the high end of what other samples found.

Adolescents in new stepfamilies and adult children of a late divorce are the two groups that get least support, and the research on them is thinner than on younger children

The intervention literature is built almost entirely around primary-school-aged children. Reviewers of that literature say so directly: preschoolers and older adolescents are under-represented in the trials. Meanwhile a new stepfamily is another family transition, and family instability research suggests transitions matter for behavior on their own account. Adult children of a divorce that happens in their parents' later life are almost absent from the outcome literature.

The under-served-age problem is stated plainly by the 2021 Encyclopedia on Early Childhood Development review: most trials in this field involved school-aged children, with preschoolers and older adolescents under-represented, and very few of the samples were ethnically diverse. That is a candid admission from researchers describing their own field.

The age moderation that does exist points at teenagers having a different problem rather than a smaller one. Lansford reports that children who experienced the divorce in grades six to ten showed larger academic declines than younger children, even though younger children showed more internalizing and externalising problems. A teenager's response may show up in transcripts rather than in behavior at home.

For a young person entering a stepfamily, the relevant finding from the instability literature is that a repartnering is another family-structure transition, and Fomby and Cherlin's analysis suggests that the association between number of transitions and behavioral outcomes may be causal in part, even after controlling for the mother's pre-existing characteristics. That is an argument for pacing a new relationship's entry into the household rather than for avoiding one.

Dunn and colleagues' study of 238 children is one of the few that looked at children's own accounts across stepfamilies, single-parent families and intact families. Its finding about confiding — friends as key confidants, and communication about stepfamily issues as a distinct thing children do or do not do — is directly relevant to a teenager in a new household who has stopped talking to adults.

Adult children of a mid- or late-life divorce are a real and documented group but the literature is small and mostly qualitative, in journals such as the Journal of Divorce & Remarriage. There is nothing here of the quality of the meta-analyzes cited elsewhere on this page. Say that rather than filling the gap with plausible-sounding advice. What can be said with confidence is that the mechanisms established for younger children — being placed in the middle, being used as a confidant, being asked to carry messages — do not stop applying because the child is thirty, and role diffusion is the specific parenting variable the conflict meta-analysis singled out.

Several things separating parents do with good intentions have either no supporting evidence or evidence pointing the other way

Confiding in the child. Handing structured help to a child who is coping fine. Assuming a cooperative divorce is protective across the board. Buying both a parent program and a child program. Treating the number of hours with each parent as the thing to optimize. Each of these is either unsupported or has a specific study pointing against it.

Making the child a confidant. Role diffusion — pulling the child into an adult role — was the parenting variable that van Dijk and colleagues' meta-analysis of 115 samples singled out as a particular risk in the context of post-divorce interparental conflict. The construct of a child feeling caught between parents has been measured since Buchanan, Maccoby and Dornbusch's 1991 study. This is one of the few places where the evidence supports a flat instruction: do not do it.

Giving a settled child a coping program. In the CoD-CoD randomised trial, for the 10 percent of young people with the lowest parent-reported baseline risk, the control condition produced lower problems than the intervention. A program that helps a distressed child can make a settled one worse.

Assuming a good divorce protects the child across the board. Amato, Kane and James (2011) cluster-analyzed post-divorce parenting in 944 families into cooperative coparenting, parallel parenting and single parenting. Children in the cooperative coparenting cluster did have the fewest behavior problems and the closest ties to their fathers. But they did not score significantly better than other children on ten additional outcomes. The authors' own conclusion was that this provides only modest support for the good divorce hypothesis. Cooperative coparenting is worth doing; it is not a shield.

Doing both a parent program and a child program. Reviews of the trials note that combining a parent-focused program with a child-focused one did not produce additive effects — neither New Beginnings plus a child coping program, nor the Children's Support Group plus a mother-focused component or workshop, beat the single program.

Treating divorce education as automatically effective because a court ordered it. Fackrell, Hawkins and Kay's meta-analysis found an overall d = .39 across 19 controlled studies but stated that there were important methodological limitations in the body of research. The best-known court program in this space, Children in the Middle, was evaluated by Arbuthnot and Gordon in the 1990s using comparison groups drawn from parents who divorced the year before the classes began, and in the relitigation follow-up the comparison group was 23 parents against 89 who attended. Those are not randomised trials, and O'Hara, Hollis and Sandler's 2024 review in Family Court Review notes ongoing debate about whether court-ordered parent education is justified at all.

Two more that follow from the mechanism literature rather than from a single study. Reporting the details of the legal process to a child is exposure, not transparency. And a chore chart of parenting hours optimizes a variable that is not the mechanism: what predicts children's adjustment in the mediation models is parenting quality, hostility, structure and role diffusion, not the count of nights.