Do Men Really Benefit More From Marriage Than Women?

The statistical principle at the letter's core — that finding a result in one group but not another doesn't prove the groups differ without a formal test of the difference — is correct and well-cited to Jaccard (1990).

The letter frames itself as correcting Ploubidis et al. (2015), but the claim it attacks — that marriage benefits men but not women — appears to come from a Globe and Mail newspaper column, not from the paper itself. Ploubidis et al.'s response states: "We simply did not find and certainly did not report that marriage does not benefit women." The letter never quotes the original paper making the strong claim it critiques.

The letter's prediction that a proper test would show equal marriage benefits is stated as personal belief, with no reanalysis, simulation, or data behind it. The gender convergence argument is asserted without citations and doesn't address whether the 1958 British birth cohort — whose formative years predated most of the convergence in question — would reflect current gender dynamics. The three supporting studies all measured mental health outcomes (depression, psychological well-being), while Ploubidis et al. measured biomarkers like metabolic syndrome, blood pressure, and inflammatory markers — different physiological systems.

The letter's central methodological distinction — interaction test versus eyeball comparison across stratified models — is precise and correctly defined.

The "gender role hypothesis" is selectively defined as resting on women's economic dependence, when the original literature proposed multiple mechanisms (social integration, health-behavior regulation, selection effects). This selective framing sets up the convergence argument: if the hypothesis rests on economic dependence, and dependence has declined, the hypothesis falls. But if it rests on mechanisms unaffected by labor-force participation, the argument doesn't follow. The letter never specifies how much convergence would be needed.

"Marriage benefits men and women equally" equivocates across outcome domains — the supporting studies measure depression and psychological well-being, while the study being criticized measures metabolic biomarkers. The letter treats these as interchangeable without justification. "Improper analysis" assumes Ploubidis et al. ran gender-stratified models specifically to compare coefficients between genders, when researchers commonly stratify for other reasons — different covariate structures, different baseline rates, different confounding patterns. The original authors' response suggests this assumption is wrong.

The letter opens with a newspaper column's claim, then treats the underlying study as the source of that claim — a conflation the original authors explicitly reject.

It assumes Ploubidis et al. stratified by gender to compare coefficients across gender, without considering that stratified models serve other legitimate purposes. It does not consider that biomarker outcomes may behave differently from mental health outcomes, or that the 1958 British birth cohort may not reflect current gender dynamics.

The letter critiques others for drawing conclusions without a formal test, then draws its own conclusion without one. The Ploubidis et al. response is cut off mid-sentence in the available text, but the visible portions are direct and unambiguous.

The letter's central premise — that Ploubidis et al. concluded men benefit more from marriage — is never verified against the original paper. The one direct quote from the study ("observed effects differed between men and women") is a weaker descriptive statement than "men benefit more," but the letter treats them as interchangeable. No limitations are acknowledged. The letter does not note that its supporting studies measure different outcomes in different populations. The gender convergence argument is asserted without citation. One of the three supporting studies is the author's own prior work, and this self-citation is not flagged. The letter does not distinguish between what it has demonstrated (the statistical principle is correct) and what it has merely suggested (that applying that principle here would change the conclusion).

Ploubidis et al. (2015) is misrepresented — the authors explicitly deny making the claim Strohschein attributes to them. Jaccard (1990) is used correctly for the statistical principle. The three studies supporting the prediction of equal marriage benefits (Strohschein 2005, Uecker 2012, Williams 2003) all measure mental health outcomes, not the biomarkers Ploubidis et al. studied. One is the author's own work. Two were over a decade old at the time of writing, despite being called "recent." No studies finding gender-differentiated marriage effects are cited. For a letter claiming broad support from the literature, three studies — one self-cited, two measuring different outcomes — is a thin base presented as consensus.

No financial or industry conflicts are present, and the marriage-and-health field has no known pattern of commercial distortion. The only structural pressure is modest academic self-interest: the author cites her own prior work as one-third of the evidence for her central claim, and her conclusion aligns with that earlier finding. This is normal in academic commentary, but the letter advocates for a position the author has previously published on.