The gap,
and the wrong confound.

The claim traces to Diane Halpern and Stanley Coren's 1988 Nature letter: surveying the reported handedness of recently deceased people, they found left-handers had died on average about nine years younger than right-handers. The finding made headlines for years. The standard rebuttal, developed over the following decade, is that it is a birth-cohort artifact, not biology: left-handedness was actively suppressed earlier in the 20th century (children trained to write right-handed regardless of natural preference), so older decedents who were still recorded as left-handed were a smaller, different group than younger ones — a confound with birth year, not a shortened lifespan. This run tests that resolution directly on a different population: 517 people in Wikidata carrying a handedness statement (P552) alongside a birth date and death date, pulled live via SPARQL, no survey, no synthetic rows.

The naive gap is real by this desk's bar. Left-handed people in the fit (133 of them) died at a mean age of 67.7; right-handed people (384) died at 72.5 — a gap of -4.74 years, 95% CI [-8.39, -1.08], p=0.011, excludes zero. A Welch's t-test agrees (t=-2.51, p=0.013), and a 4,000-draw bootstrap puts the gap at -8.47 to -1.03 years, negative in 99.5% of resamples.

The cohort-artifact explanation predicts that controlling for birth year should shrink this gap toward zero. It doesn't. Birth year does move with handedness in the predicted direction — left-handed people in this sample have an earlier mean birth year (1890) than right-handed people (1920), slope -0.00042 per year, CI [-0.00074, -0.00009], excludes zero — but adding birth year as a control to the age-at-death regression leaves the left-handed coefficient essentially where it started: -5.01 years, CI [-8.68, -1.33], p=0.008. Birth year's own coefficient in that same regression contains zero (-0.009 yr/yr, CI [-0.023, +0.005]) — in this particular population, the textbook resolution doesn't do the work it's supposed to.

A different confound turns out to matter more: who gets a handedness statement on Wikidata in the first place. Pulling occupations alongside handedness shows 224 of 517 people are racket-sport athletes (tennis, badminton) — for them, "handedness" on Wikidata usually just records which hand they played with, a routinely documented sporting fact rather than a reported trait from a mortality survey. Splitting the sample in two reverses the obvious guess: among racket-sport athletes, the gap contains zero and even leans positive (+3.11 years, CI [-4.20, +10.43], p=0.403). Among everyone else — writers, politicians, actors, musicians — the gap is larger than the naive full-sample estimate: -6.89 years, 95% CI [-11.22, -2.56], p=0.0019, and a bootstrap on that subgroup agrees almost exactly (CI [-11.47, -2.71], negative in 99.98% of resamples). The athlete-selection story, which looked like the obvious way to explain away a Wikidata artifact, runs backward.

An era split is more equivocal. Restricted to people born before 1950 (n=424), the gap is -2.79 years, CI [-6.36, +0.77] — contains zero, though barely. Born 1950 onward (n=93), the point estimate flips positive, +3.77 years, CI [-2.67, +10.20] — also contains zero, on a much smaller slice. Neither era alone clears this desk's bar in either direction; read together with the athlete split, the honest summary is that the full-sample gap is concentrated among older, non-athlete entries rather than spread evenly across the whole dataset.

Named, for scale. The left-handed entry with the earliest birth year in the fit is Aristotle (384–322 BC, recorded left-handed on Wikidata, off the chart's axis but included in every fit above), who lived to 62. The longest-lived left-hander in the sample is Paola Veneroni, an Italian actress, dead at 99; George H. W. Bush, also left-handed, died at 94. The longest-lived person in the whole sample is Morrie Markoff, right-handed, dead at 110. Long life on either hand is possible; the fitted gap is about averages, not a prediction for any one person.

Read plainly. In this Wikidata-notable-people population, left-handed people have died measurably younger on average, the gap survives a birth-year control the standard explanation says should remove it, and splitting off the sub-population most likely to carry a "playing hand" rather than "dominant hand" record makes the remaining gap larger, not smaller. That is not a confirmation that left-handedness shortens life — Wikidata's own notable people are not a population sample, and occupation, era of fame, and cause of death are all uncontrolled here — but it is a specific, stated failure of the textbook debunking to reproduce on this dataset, reported as found rather than discarded for not matching the literature.