In the first week of a semester, a student writes down that she wants to be less anxious. Not as a plan, exactly. More as a wish made in September and rarely audited in December.
Whether that note predicts anything is a harder question than it sounds, and for a while the research on it pointed both ways.
Why their earlier results had wobbled
Two of these four authors had already published three intensive longitudinal studies reporting that stated wishes about a trait went with later movement in it. The trouble was that the effects came out somewhat inconsistently across the published studies, which left open how large they were and which traits they touched. The authors’ own summary of the field was that the picture remained somewhat unclear, and their diagnosis was mundane: individual studies were underpowered for effects this size, so what looked like disagreement was probably sampling error.
Nathan Hudson, R. Chris Fraley, William Chopik and Daniel Briley took an unusual route around that. Instead of running one more study, they merged every dataset their labs held on the question as of May 2019 into a single participant-level analysis: twelve samples, 2,238 people recruited from psychology courses for course credit, none of them excluded for any reason. They count 1,794 of the 2,238 people, four fifths, as unpublished. The paper also lists 2018 and 2019 papers among already-published samples, so that fraction is their accounting, not a claim that none of those records had appeared anywhere. The paper ran in Social Psychological and Personality Science in 2020.
The design was similar across samples rather than identical. Each study had its own research questions and three different trait questionnaires were used between them, inside a window of fifteen to sixteen weeks. In the first week, students rated how much they wanted to change on each trait, on a scale running from much less than I currently am through I do not wish to change this trait to much more. Then they rated the traits themselves, reading after reading. They were asked for one a week, though the site accepted them as often as every five days and set no deadline at the other end; on average people filed just under eleven across the term.
Because everything their own labs held went in, the result is not the residue of those labs’ file drawers. They mark the limit of that themselves: they had no way to be sure they had reached every lab working on the question.
The gap between wanting it a lot and wanting it the usual amount
Extraversion carries the clearest worked example. A student one standard deviation above average in wanting to become more outgoing was predicted to gain 0.217 standard deviations of it across the semester. A student at the average was predicted to gain 0.075. A student one standard deviation below average was predicted to lose 0.068.
That middle figure changes what the finding is about. Zero on the wanting scale meant I do not wish to change this trait, and the average student sat at 0.67, which is a wish to increase; even the student a full standard deviation below that sat at 0.22, still wishing upward. So the contrast is not between people who wanted to change and people who did not. It is between wanting it a lot and wanting it the usual amount, and the distance between those two ran 0.142 standard deviations across sixteen weeks, with a confidence interval from 0.122 to 0.162. Worth naming here: the paper describes that comparison group two ways, as peers with average change goals in its Table 2 note and discussion, and as peers “who did not wish to change” in a figure caption, its introduction and its results prose. The estimates were built from the first, so this piece follows the first.
For emotional stability the same figure was larger, at 0.181. For wanting to be more agreeable or more open it ran only about 0.05, or 0.053 and 0.039 taken separately, and there the authors add that the agreeableness result may depend on which questionnaire was used, since in the newer of the instruments alone it came out comparable in size to the other traits. (The paper’s abstract, and one passage of its discussion, round the extraversion and stability figures together to a single “approximately 0.16”; elsewhere the discussion reports stability at 0.18, in step with Table 2, and this piece uses the separate figures throughout.)
To make those numbers legible the authors reach for a benchmark: meta-analyses they cite put ordinary maturational change for people aged eighteen to twenty-two at roughly 0.16 standard deviations per trait over a median span of two years. That is a different 0.16 from the one above, and against it they say the monthly effects seen here — for extraversion, conscientiousness and emotional stability, not for the other two — may indicate that wanting predicts nontrivial variation in how people develop.
Then the finding starts qualifying itself. An analysis the reviewers asked for, which the authors mark exploratory throughout, let the relationship bend rather than run straight: a wish recorded in week one appeared to predict growth more strongly early and less strongly later. That pattern reached significance for extraversion, conscientiousness and emotional stability. For agreeableness and openness it did not, the estimates sitting at 0.002 and 0.000, close enough to nothing that no direction can be claimed in either. Their reading is a restriction rather than a reassurance: a wish recorded once may forecast only a stretch of several months, and may say nothing about how a trait is moving a year out. The wishes are not stable objects either, with a twelve-week test-retest correlation around 0.50 in the authors’ earlier work.
Wishes also mostly predicted movement in their own trait rather than the others, but about forty percent of the cross-trait effects were statistically significant in what the authors call seemingly random directions. Wanting to be more outgoing, for one, went with a predicted decrease in agreeableness.
And the sharpest counter-evidence comes from inside the same paper. Robinson, Noftle, Guo, Asadi and Zhang, writing in the Journal of Research in Personality in 2015, followed 170 graduating seniors, measured their traits twice a year apart, and found wishes unrelated to trait change — this by Hudson and colleagues’ account of that study, whose full text was not available to us. They set out at least five differences between the two designs, among them single-item against multi-item measures, two readings against eleven, a year against four months, and a settled semester against the disruption of graduating. They say it remains unclear why the findings differ, and float methodology as one candidate and short-lived or cyclical change as another.
A semester of psychology-course volunteers
The sample shapes everything above it. Seventy-one percent were women, the average age was a little over twenty, and although three universities contributed, 74 percent of the participants came from one of them. Ninety-four percent filed a second reading; 37 percent were still filing at the fifteenth.
Traits and wishes alike were self-reported, by the same person, into the same website, so what moved was self-description. The paper does not ask whether anybody noticed the change in themselves, and we cannot answer for it. We are writers reading a statistics table, not clinicians, and a table like this one describes a sample’s drift, not anyone’s particular month.
The design is correlational, and the authors say it cannot strongly speak to causal processes, which leaves the appealing story open and unproven. Wanting to be calmer may have led students toward the ordinary things that make a person calmer; or one underlying shift may have produced the wish and the drift alike; or rating your own steadiness over and over may alter how you rate it. Only the first is the paper’s own suggestion. The other two are ours, and none of the three was tested here.
That narrows the result without emptying it. Across a couple of thousand young students whose average wish on every trait pointed toward change, wanting it more ran in the same direction as slightly larger movement in their own ratings, and not because one lab published only its successes. It does not show that the wish caused the movement, that it survives past a semester, or that it behaves the same way at fifty. Anyone whose wish to be calmer is really a wish to stop being frightened all the time should take that to a therapist.
Which returns us to the note from the first week of term. Among these students, whose average wish on every trait pointed toward change, the size of the wanting sat with the size of the drift, and for three of the five traits it sat closest in the months nearest the writing. The paper never asks whether anybody could tell, by the end of term, that anything had moved.
