Does a job that suits you make you better at it, or just happier?
My argument is that the conditions around a person shape their work more than anyone accounts for, and that which conditions matter differs from person to person. Before making that case in public I went looking for whoever had already tested it. They exist, the literature is large, and it does not show what I would like it to show.
Evidence It is called person-environment fit. The main analysis covers 172 studies and 836 effect sizes.
Fit and job satisfaction, .56. Fit and commitment, .51. Fit and intending to quit, -.35. Those hold across roughly 40,000 people per result, and they survive being measured at different times.
Fit and actual job performance, .07.
That last number gets worse when you look at who did the rating. Where one person supplied both the fit rating and the performance rating it reaches .22. Where they did not, it is -.02.
Evidence The field says the rest itself. Its own 2023 review notes the research is mostly cross-sectional and between-person, that only a handful of studies used within-person designs, and lists clarifying causal direction as work still to do. So this literature cannot tell you whether changing a condition causes anything. It was not built to.
I first wrote that nobody had tested it experimentally. That was wrong, and worse, I had not looked. Experiments exist.
Evidence The largest one I have found is STAR, a randomized trial that gave employees more control over when and where they worked. On performance it found almost nothing. The one significant result was about an hour off the hours people expected to work each week.
Which points the same way as everything else here. Change the conditions and what moves is how work feels and how much of it people expect to do. What they produce stays where it was.
So the honest version is narrower than I would like. Matching conditions to a person is well evidenced for how they feel about their work. It is not evidenced for what they get done. Anyone selling you the second thing is citing the first.
One more number worth carrying. Fit and intending to quit is -.35. Fit and people actually leaving is -.14. Intent is not behaviour, and the gap between them is where a lot of confident retention claims live.
Where this stops. Kristof-Brown, Zimmerman and Johnson 2005 in Personnel Psychology, full text free at psychologie.uni-mannheim.de. The 2023 restatement by Kristof-Brown, Schneider and Su is free at the USC Center for Effective Organizations and revises none of the figures.
The intervention evidence. STAR is at PMC6719311. A meta-analysis of job crafting interventions exists and reportedly finds gains in engagement and contextual performance rather than task performance, which would fit the pattern. I could not open it, so I am naming it rather than citing it.
Still unchecked. Chronotype-based scheduling trials. Until I read those, my silence on them is unfinished work and not a verdict.