Dispositional Optimism

What it is

Four different things get called optimism in popular writing and they are not interchangeable. Absolute optimism is the belief that you will beat the odds. Comparative optimism, Neil Weinstein's construct, is the belief that you will fare better than your peers. Explanatory style, Martin Seligman's, is a backward-looking habit of accounting for events that have already happened, measured with a different questionnaire. Dispositional optimism is forward-looking, general and unattached to any specific outcome. Paul-Choudhury's gloss is the cleanest short statement available: a general, all-purpose expectation that things will work out well for us that does not apply to any particular situation.

Two properties make it worth treating as a disposition rather than a mood. The first is stability. In the veterans' cohort in Lewina Lee's 2019 paper, the five-year correlation of the optimism score was 0.87. The second is heritability: the same paper records the construct as roughly 25 per cent heritable while also being shaped by social structural factors and learnable. A measure that holds its position over five years and has a genetic component is doing something more than recording how a respondent felt on the morning of the survey.

The headline evidence is a mortality association, and it is stronger than sceptics usually allow. Lee and colleagues followed 69,744 women in the Nurses' Health Study for ten years and 1,429 men in the Veterans Affairs Normative Aging Study for thirty. Adjusted for demographics, baseline health conditions and depression, the most optimistic quarter of the women lived 14.9 per cent longer than the least optimistic quarter, and the most optimistic fifth of the men 10.9 per cent longer. Hayami Koga and colleagues replicated it in 159,255 Women's Health Initiative participants followed for up to 26 years, at 5.4 per cent. Alan Rozanski's 2019 meta-analysis pooled 15 studies and 229,391 participants and reported a relative risk of 0.86 for all-cause mortality. The nurses' models adjusted for the participant's own education, her husband's education and her father's occupation when she was sixteen, which is a better socio-economic adjustment than most psychology papers attempt.

Three things then complicate every one of those figures.

The first is adjustment. Add smoking, alcohol, physical activity, diet quality, body mass index and screening attendance to Lee's model and the women's figure falls from 14.9 per cent to 8.7. The authors write that health behaviours may be confounders in this association and may also lie on the pathway from optimism to longevity, which means both figures are defensible and observational data cannot decide between them. Koga's team answered the question with a quantity instead, running a formal mediation decomposition with the lifestyle score measured three years after the optimism assessment, and put the mediated share at 24 per cent.

The second is confounding, and it is visible in the cohorts' own baseline tables. In Koga's data, comparing the least optimistic quarter with the most optimistic: 22 per cent were depressed against 4 per cent, 22 per cent had not finished secondary education against 9 per cent, and 23 per cent had a family income under twenty thousand dollars against 10 per cent. Optimism in these samples is substantially a reading of circumstances.

The third is the asymmetry, and it is the most important finding on this page. When the questionnaire is split into its positively and negatively framed halves, almost all the predictive work sits at the pessimistic end. Scheier's own consortium, pooling 61 samples and 221,133 people, found the presence of optimism predicting physical health at a correlation of 0.011, against 0.029 for the absence of pessimism, with the difference significant and the optimism subscale non-significant once two studies were dropped. Koga's subscales give 5.2 per cent shorter life span at the pessimistic pole against 2.5 per cent longer at the optimistic one. In genetics, Robyn Wootton's bidirectional Mendelian randomisation on subjective wellbeing found no clear causal effect on cardiometabolic health in either direction, and the one positive genetic result rests on a composite its own authors say is driven mainly by depressive symptoms.

In effect

The nurses, the veterans and the three models

In Lee and colleagues' paper the same comparison appears three times with three answers: 18.6 per cent longer life span for the women on demographics alone, 14.9 once baseline conditions and depression are added, 8.7 once behaviours are added. The paper supplies its own best calibration too. In the same model, never having been diagnosed with type 2 diabetes was associated with a 17.0 per cent longer life span and never having had a heart attack with 18.0 per cent. Optimism's 14.9 sits just below both, which is a real effect and not a small one, and the behaviour-adjusted 8.7 is a different claim about it.

A questionnaire that partly measures your circumstances

The everyday reading of these results is that cheerful people take better care of themselves. Koga's baseline table suggests something less flattering to the construct. Across quartiles of the same six-item questionnaire there is a fivefold difference in the prevalence of depression, a two-and-a-half-fold difference in not finishing secondary school and a fourfold difference in low family income. Paul-Choudhury himself makes the argument at page 185 of The Bright Side, about privately educated politicians, that for people who inherited fortunes things often do simply work out as far as they are concerned, and never carries it back 150 pages to his own health chapter.

Wellbeing provision sold on the longevity figure

The commercial use of this construct is in corporate wellbeing and insurance-adjacent marketing, where the 11 to 15 per cent longer life span is the number that travels. Two published facts bound it. Both circulated figures, the percentage range and the "50 to 70 per cent greater odds of reaching 85", come from the model before health-behaviour adjustment; the behaviour-adjusted versions are 8.7 and 9.8 per cent with odds ratios of 1.2 and 1.6 whose lower bounds touch 1.0. And Rozanski and colleagues say in print that nobody yet knows whether optimism enhanced through an intervention carries the health benefits of optimism that occurs naturally. Anybody selling the trait is selling a cohort association, in the model of their choice, for a trait the programme has not been shown to move durably.

What it does not say

It does not say that optimism causes longer life. No design capable of establishing that has been run on this construct. There is no Mendelian randomisation study of dispositional optimism, and no co-twin or sibling fixed-effects study of dispositional optimism and mortality. Both absences were searched for in standard and extended mode before being stated here.

It does not say the finding generalises widely. Both Lee's and Koga's cohorts were whiter, better educated and longer lived than their national populations, and both say so in their own limitations. Fifty-three per cent of Koga's eligible women reached 90. In the consortium reanalysis, over 90 per cent of participants were white and women and over 90 per cent of the studies were American. Two of four racial and ethnic subgroups in Koga's paper have intervals crossing zero.

It does not say that the behavioural mechanism is established. The behaviours that supposedly carry the effect correlate with optimism at 0.07 for physical activity, 0.12 for diet and 0.07 for smoking in Julia Boehm's meta-analyses, mostly cross-sectional and graded relatively low quality by her own team. One behaviour runs backwards: the least optimistic women in the Women's Health Initiative received about 5 per cent more frequent screening mammograms, against the researchers' own stated hypothesis.

And it does not say anything at all about whether a person's attitude affects their recovery from illness. It does not. Where attitude has been tested directly against cancer survival, the positive pole has not predicted it, and the trials of group psychotherapy for survival have not replicated. The one asymmetric result, helplessness and hopelessness predicting worse ten-year disease-free survival in Maggie Watson's cohort, came with her own recommendation that clinicians screen for it in order to target psychological care. That is a reason to notice despair and offer support. It is not a reason to tell anybody to cheer up, and persistent hopelessness is a matter for a clinician rather than for a questionnaire.

Who developed it

Michael Scheier and Charles Carver, in 1985, along with the instrument that measures it. Scheier is also first author of the 2020 reanalysis that found the optimistic half of his own scale to be the weaker predictor of physical health.

---