Adult Sex Roles and Mental Illness

The paper makes two arguments: a descriptive one (women show higher rates across multiple data sources) and a causal one (sex roles produce the difference). The descriptive argument is careful — the authors computed age-standardized rates, deliberately overestimated male VA hospital admissions to bias against their hypothesis, and assembled data from six independent treatment channels.

The causal argument has fundamental problems. First, the definition of mental illness excludes the category where men predominate. The only community study that used a broader definition found no sex difference and was excluded on definitional grounds. The paper treats this as methodological precision, but the definition was chosen in a way that guarantees the finding. Second, the proposed causal mechanism — that role frustration produces stress, which produces mental illness — is entirely unmeasured. No cited study connects any of the five proposed role characteristics to mental illness outcomes. The theoretical section draws on sociological commentary (Friedan 1963, Parsons 1942, Komarovsky 1950) rather than empirical measurements. Third, the three aggregate patterns offered as causal evidence are each consistent with multiple explanations. The married/unmarried pattern could reflect selection into marriage. The temporal shift compares pre-war hospital data with post-war survey data — different methods, not just different eras. The community variation comes from a single unreplicated study.

The refutation of labeling theory also has a problem: the paper argues that because men displaying symptoms face harsher reactions, labeling theory predicts more men in treatment. But harsher stigma could also drive men away from voluntary treatment. The paper's data contain a gradient consistent with this — the female-to-male ratio is smallest in mental hospitals (where others initiate admission) and largest in general-practitioner settings (where individuals seek help voluntarily). This gradient is never discussed.

The paper's most consequential move is definitional. Restricting "mental illness" to neurosis and functional psychosis is clinically coherent — these conditions share symptoms and treatment responses that differ from personality disorders. But the restriction removes the male-predominant category and retains the female-predominant ones. The Midtown Manhattan study, which used a broader definition, found male personality disorders and female neuroses roughly balancing out. A reader who accepts a broader definition gets a different answer.

The central causal construct — "sex roles" — is too vague to test. The paper names five role characteristics (restriction to one role, low prestige, lack of structure, employment discrimination, diffuse expectations), but none is defined precisely enough to measure. The theory accommodates any pattern of results. If women have higher rates, their roles are worse. If men have higher rates in a depressed community, the local economy hurt men's roles. If a traditional village shows low female rates, women's roles there resemble a pre-industrial pattern. No observation could disconfirm the theory.

The framework also cannot distinguish between its preferred explanation and obvious alternatives: that women report distress more readily, that clinicians diagnose neurosis more readily in women, or that biological sex differences affect vulnerability to depression. All three produce the same observable pattern.

The descriptive analysis is sound — it assembles data carefully, accounts for estimation issues, and shows its work. The problems are in how it moves from description to explanation.

The definitional exclusion of personality disorders does most of the argumentative work. Including them eliminates the sex difference; excluding them produces it. The clinical rationale for exclusion (different symptoms, different treatment response) applies with equal force to the differences between schizophrenia and anxiety neurosis, which the paper groups together. The exclusion is presented as precision, but it is also convenient.

The role explanation is unfalsifiable. The paper interprets every data pattern as confirming its theory. The closing acknowledges the discussion is "somewhat speculative" and calls for more research, but the body argues for the role explanation with more confidence than this warrants, and the refutation of labeling theory is presented as definitive when it addresses only a simplified version.

The paper also conflates treatment rates with illness rates throughout. Every data source measures something different from true prevalence — community surveys measure willingness to report symptoms, hospital admissions measure referral patterns, GP data measure help-seeking behavior. The paper treats convergence across these sources as proof the finding is real, but all six share confounds (differential help-seeking, gendered expressiveness norms, diagnostic expectations) that convergence cannot eliminate. If a single upstream cause inflated female counts everywhere simultaneously, the convergence would look the same.

Biological explanations for sex differences in depression and anxiety — including hormonal influences being studied by the late 1960s — are never mentioned. For a paper claiming to identify the cause of a sex difference in mental illness, omitting the most obvious alternative causal category is a serious weakness.

The paper is more transparent than many about some choices. It states that "implicit in our analysis is the assumption that stress may lead to mental illness." It acknowledges in a footnote that the claim about housewife prestige has never been systematically evaluated. It discloses its estimation procedures for VA data and deliberately biases estimates against its hypothesis.

The problems are at the inferential joints. The paper presents its definitional restriction as methodological precision without acknowledging it favors the hypothesis. The marital-status evidence — central to the causal argument — comes entirely from Gove 1972, a companion paper by the first author that was forthcoming at the time, with no data tables presented. The paper does not distinguish between what it has demonstrated (women appear in treatment and surveys at higher rates under a specific definition) and what it has suggested (sex roles cause this).

Major alternative explanations are either briefly engaged or absent. Diagnostic bias — the possibility that 1960s clinicians diagnosed neurosis more readily in women presenting the same symptoms — is never mentioned. The possibility that men's distress presents as alcoholism, violence, or antisocial behavior (conditions excluded by the definition) is never raised. Biological explanations are not addressed.

The citation base for the descriptive finding is thorough. The paper assembles 17 community surveys and multiple national administrative datasets, presents specific percentages by sex, and computes standardized rates.

The citations supporting the causal argument are weaker. Several key sources for the sex-role theory are feminist advocacy works (Friedan 1963, Gavron 1966) rather than empirical studies measuring role characteristics. Friedan's The Feminine Mystique is social criticism, not research linking role frustration to mental illness. The marriage-concentration finding — the strongest indirect evidence for the role explanation — comes entirely from the first author's companion paper, creating a self-citation dependency with no independent replication. The historical shift evidence compares data from different measurement methods (pre-war hospital admissions vs. post-war community surveys) without acknowledging this as a confound.

Missing from the citation base: any biological research on sex differences in depression susceptibility, any study measuring reporting bias in psychiatric surveys, and any cross-cultural evidence from non-Western societies that could test the role explanation against different gender arrangements.

There are no financial conflicts — the paper was funded by internal university money. The relevant pressure is intellectual. Gove cites seven of his works, several unpublished at the time. The marital-status data, the labeling-theory critique, and the suicide analysis all route through the first author's research program. Key evidence is presented as "reviewed elsewhere" by the first author rather than shown. Every conclusion reinforces positions Gove had already staked out, and the unfalsifiable structure of the role explanation means the paper was unlikely to encounter friction with its framework. This is standard for building a research program, but it means the paper's most important supporting evidence was not independently verifiable when published.