Failure Without an Exit Event Is Invisible
Marriage has a number attached to it that everyone can cite. Roughly some fraction of marriages end in divorce, and that number gets treated, correctly, as a real measure of how often the arrangement fails. Parenthood has no equivalent number, and this gets treated, incorrectly, as evidence that parenthood mostly succeeds. The absence of a comparable statistic is quietly read as an absence of the corresponding problem. It isn't. It is the absence of a mechanism that would have recorded the problem in the first place.
The reason is structural, not emotional. Divorce is a legal event. Something has to happen, filed, finalized, dated, for a marriage to register as failed in any dataset. A parent-child relationship has no equivalent threshold to cross. It can go distant, dutiful, conflictual, or psychologically empty without ever producing a single recordable event that would move it out of the "still connected" column. The relationship can be hollow and the data will still show continuation, because continuation, in this category, is measuring contact, not quality, and contact is a much lower bar than the thing people actually mean when they ask whether the relationship worked.
A category that has no way to record its own failure will always look more durable than a category that does, and that appearance of durability is a fact about the measurement, not a fact about the relationships.
Two different things being measured
This is a case of two categories being compared as though they were tracking the same variable when they are not. Marriage statistics track dissolution, whether the formal bond was terminated. Parenthood statistics, when they exist at all, mostly track contact, whether some relationship persists in any form. These are not the same axis. A marriage can be miserable and intact, in which case divorce statistics undercount its failure. A parent-child relationship can be distant and intact in exactly the same sense, but because there is no dissolution event to fail to record, there is no undercounting visible at all. The failure simply has nowhere to be filed [1].
Survey data on adult children's relationships with their parents makes this gap visible when it is measured directly rather than inferred from continuation. Only a majority, not the near-totality that "children rarely divorce their parents" would imply, describe the relationship as excellent or very good, and a meaningful share describe it as merely adequate or as fair to poor [2]. The relationships that fall short of excellent did not become failures in the way a divorce is a failure. They simply continued, unrecorded, indistinguishable in the top-line data from relationships that are genuinely close.
The self-report asymmetry is a symptom, not noise
A related pattern shows up when the same relationship is rated by both parties: parents consistently rate the relationship as stronger than their adult children rate it, often by a wide margin [3]. This is not simply two people with different opinions. It is close to a predictable consequence of the measurement problem itself. The party for whom the relationship has no formal exit mechanism, and no external, comparative benchmark forcing an accurate assessment, is the party more likely to round the relationship up to something closer to intact-by-default. The party with more freedom to disengage, and less institutional pressure to describe the bond as successful, gives the less flattering, and probably more accurate, report. The asymmetry is the fingerprint of an unrecordable relationship, not a coincidence of perspective.
Why this generalizes past parenthood
The mechanism here is not specific to families. Any relationship, or any system, that lacks a formal dissolution event will tend to look more durable in aggregate than one that has such an event, regardless of the underlying rate of actual failure, because failure in the first category has nowhere to register. Employment relationships without a clean termination process, long friendships that quietly hollow out without a defined ending, institutional roles people keep occupying long after they've stopped functioning in them, all share this signature. Continuation gets mistaken for success because continuation is the only thing being tracked, and the absence of a countervailing statistic gets misread as the absence of the problem the statistic would have measured.
What this reframes
The practical correction is to stop treating "still connected" as a proxy for "working," in any relationship category that lacks a formal way to record its own ending. Contact, presence, technical continuation, these are real facts, but they answer a much narrower question than the one usually being asked when someone points to a low dissolution rate as evidence of success. The right response to "there's no divorce-equivalent statistic for this" is not "so it must mostly be fine." It's "so we have no idea, from that number alone, how often it isn't," and the actual answer has to come from directly measuring the thing that matters, closeness, safety, being known, rather than the much easier and much less informative thing that happens to be recordable.
The point
Marriage has a widely cited failure rate because failure in marriage has a legal, recordable form. Parenthood has no equivalent statistic, and this absence is routinely misread as evidence that parent-child relationships mostly succeed, when it is really evidence that the category has no mechanism capable of registering its own failures. Continuation and quality are different variables, and a relationship can be hollow while still counting, permanently, as intact, simply because there was never an event available to move it out of that column.
The asymmetry in how parents and adult children rate the same relationship supports this: the party with no formal exit and no external benchmark tends to round the relationship up, while the party with more freedom to disengage gives a less flattering, more accurate account. This pattern generalizes beyond families, to any relationship or role that lacks a formal dissolution event, and the correction is the same in every case: stop treating technical continuation as a proxy for success, since the absence of a failure statistic is a fact about what got measured, not a fact about how often the thing actually failed.
Sources
- Cherlin, A. J. (1978). Remarriage as an incomplete institution. American Journal of Sociology, 84(3), 634-650. On how relationships lacking clear institutional norms or exit procedures produce ambiguous, hard-to-classify outcomes that resist the kind of clean failure statistics available for formally dissolvable arrangements like marriage.
- Pew Research Center. (2024). Young Adults' Relationship with Their Parents. On the distribution of relationship quality among adult children who remain in contact with a parent, showing a meaningful minority rating the relationship as merely adequate or as fair to poor even absent any formal estrangement.
- Pillemer, K. (2020). Fault Lines: Fractured Families and How to Mend Them. Avery. On the documented asymmetry between parents' and adult children's assessments of relationship quality, and on estrangement as a widespread but under-recorded phenomenon precisely because it lacks the formal, legal markers that make marital dissolution countable.