The strongest signal was confidence
Teachers in France and Australia reported greater teaching self-efficacy, or confidence in their ability to do the job, when the professional learning available to them exceeded the training they said they needed. The relationship was nonlinear, meaning the gap did not translate into a simple, steady change in confidence. The pattern appeared in nationally representative TALIS 2024 data covering 2,877 secondary teachers in France and 3,035 in Australia. The analysis accounted for 8% of the adjusted differences in self-efficacy in France and 12% in Australia.
The analysis compared teachers’ reported needs and supplies across 15 professional-skill areas. When the two measures diverged, the shape of the association differed by country: in France, the change in self-efficacy accelerated as misfit grew; in Australia, it leveled off at the highest levels of misfit.
To capture relationships that might bend or change direction, the researchers used response-surface analysis, a statistical method that maps an outcome across combinations of two measures. They compared 37 polynomial model families, including linear, quadratic and cubic forms, and weighed simplicity, explained variance, fit statistics and Wald tests when choosing among them.
Teaching’s social purpose stood out
The fit between professional-learning needs and supplies had little relationship with how much personal or social value teachers attached to teaching. Those models explained no more than 2% of the adjusted variation in either value measure. Reported needs were not significant predictors, while supply effects were small.
A clearer pattern emerged when researchers looked at self-efficacy together with well-being. In the models using personal utility, meaning the perceived personal benefits of teaching, the combination explained 5% to 14% of the variation in teaching enjoyment and job satisfaction, but less than 2% of the variation in intention to stay. High self-efficacy was associated with moderate well-being when personal utility was low. High personal utility, by itself, did not offset low self-efficacy.
Social utility, the sense that teaching benefits other people or society, showed a stronger association. The self-efficacy and social-utility models explained 30% to 50% more variation than the personal-utility models across the outcomes. The common pattern showed the highest well-being when confidence and social value were both high and balanced, with lower well-being when either was low. This interdependence pattern affected job satisfaction or intention to stay among 32% to 47% of French teachers, and enjoyment among 42% of Australian teachers.
Self-efficacy was part of the statistical pathway
The analysis also tested whether the needs-supplies pattern was statistically associated with well-being through self-efficacy, a procedure often called mediation. The needs-supplies models alone accounted for 2.2% to 4.5% of well-being variation, apart from 0.5% for Australian intention to stay. When the indirect association through self-efficacy was included, the models accounted for 7% to 13% of enjoyment and job-satisfaction variation and 3% of intention-to-stay variation. Direct effects often no longer met the study’s 5% significance threshold.
That mediation result needs caution. The procedure was a conservative approximation: it assumed a linear, non-interacting relationship for self-efficacy even though the main fitted patterns were nonlinear and interactive. It therefore describes a statistical pathway in the data, not a demonstrated sequence in which professional learning changes confidence and confidence changes well-being.
The evidence has a clear boundary
The study drew on a two-stage stratified probability sample. Schools were selected in proportion to size, and teachers were randomly sampled within schools. Models used robust full-information maximum likelihood, a way of using incomplete records, which accounted for 97% to 99% of French teachers and 99% of Australian teachers. Weighted checks that dropped incomplete cases produced substantively similar results.
But all measures came from teacher self-reports collected at one point in time. The design cannot establish whether professional learning led to higher self-efficacy or well-being, whether the relationship ran in the other direction, or whether other factors influenced both. Intention to stay was not objective retention or turnover, and the results may not apply beyond secondary teachers in France and Australia.
The findings describe associations in these two national teacher samples. They do not show that professional learning causes higher self-efficacy or well-being, affects actual teacher retention or turnover, or generalizes to primary teachers or settings beyond France and Australia.
Paper data and sources
Original title: Professional learning supports teacher well-being through needs-supplies fit and control-value processes.
Authors: Fernando Núñez-Regueiro, Helen M G Watt, Paul W Richardson et al.
Journal/Repository: Scientific reports
Status: Peer-reviewed
First online: 2026-08-20
DOI: 10.1038/s41598-026-57946-9
Original paper · Full text