How autonomy-supportive climates shape gender gaps in STEM success

Women studying science, technology, engineering, and mathematics appear to experience their academic lives in fundamentally different ways from their male classmates, and a new study suggests that the psychological engines driving their performance and satisfaction run on distinct fuel. The research, published in the journal BMC Psychology, examined how an autonomy-supportive educational climate, the use of metacognitive learning strategies, and academic self-efficacy relate to performance and satisfaction among STEM students—and it found that the pathways connecting these factors diverge sharply between men and women.

The study, conducted by Ricardo Navarro, Monica Takushi-Rodriguez, and Lucía Gurbillon of the Faculty of Psychology at the Pontificia Universidad Católica del Perú, set out to address a persistent problem in higher education: although STEM programs are central to scientific and technological advancement, the academic experience of women in these predominantly male environments is often not the same as that of their male peers. The researchers wanted to know whether the same educational factors that support men’s success in STEM do the same work for women, or whether the psychological machinery differs beneath the surface.

At the heart of the investigation lies academic self-efficacy, the belief a student holds in their own capability to organize and execute the actions required to succeed academically. Decades of research have tied self-efficacy to persistence, effort, and achievement, but the new study asks where this belief comes from in the specific context of STEM classrooms and what consequences it carries. The researchers focused on two upstream factors. The first is an autonomy-supportive climate—an educational environment in which instructors acknowledge students’ perspectives, offer meaningful choices, minimize controlling language, and nurture intrinsic motivation. This construct is grounded in Self-Determination Theory, the influential framework holding that autonomy, competence, and relatedness are basic psychological needs whose satisfaction fuels motivation and well-being. The second factor is the use of metacognitive strategies: the planning, monitoring, and self-regulation techniques that learners deploy to manage their own thinking—setting goals before studying, checking comprehension along the way, and adjusting tactics when material refuses to yield.

To trace the relationships among these constructs, the team employed structural equation modeling, a statistical framework that allows researchers to test whether a hypothesized network of cause-like associations fits observed data. Their general model posited that an autonomy-supportive climate and metacognitive strategies would each be associated with academic self-efficacy, which in turn would be associated with academic performance and academic satisfaction, with direct pathways from the climate to satisfaction and performance as well. The hypothesized general SEM model demonstrated good fit indices, meaning the observed pattern of relationships among variables was statistically consistent with the proposed structure. The researchers evaluated fit using established benchmarks, including the Comparative Fit Index, the Tucker-Lewis Index, the Root Mean Square Error of Approximation, and the Standardized Root Mean Squared Residual—metrics that together assess whether the model reproduces the covariance structure of the data adequately.

Critically, the team did not stop at a single pooled model. Because their central question concerned gender, they needed to verify that the same measurement instruments were functioning equivalently across men and women before making comparisons. The model met invariance criteria, the statistical prerequisite for meaningful multigroup comparison, after which the researchers conducted multigroup SEM analyses—essentially fitting the network of associations separately for male and female students and testing whether specific pathways held in each group.

The results were revealing. In both gender groups, the factors associated with academic self-efficacy were the same: an autonomy-supportive climate and the use of metacognitive strategies were each significantly linked to students’ belief in their own academic capabilities. In other words, whether students were men or women, environments that supported autonomy and study habits grounded in self-regulated learning were associated with stronger confidence. This consistency suggests that these two levers remain powerful entry points for supporting all STEM students, regardless of gender.

The downstream picture, however, fractured along gender lines. Academic satisfaction—the sense of contentment and fulfillment students derive from their studies—was significantly associated with academic self-efficacy among women but not among men. For male students, the association with an autonomy-supportive climate was significant instead, while for women it was not. This pattern implies that what makes a woman in a STEM program feel satisfied with her academic life may be tied closely to whether she believes she can succeed, whereas men’s satisfaction appears more directly sensitive to the motivational character of their learning environment.

Academic performance told a similar story of divergence. Among women, both academic self-efficacy and an autonomy-supportive climate were associated with academic performance; among men, neither association reached statistical significance. The finding is striking because it inverts a common assumption that self-efficacy functions as a universal predictor of grades. In this STEM sample, confidence in one’s capabilities and an educational climate that respects student autonomy tracked with measurable achievement only for women, while men’s performance appears to ride on other factors outside the modeled network.

The researchers caution, appropriately, that the study’s design captures associations rather than experimental causation, and the divergent significance patterns raise as many questions as they answer. Yet the implications are considerable. If the same interventions do not produce the same psychological consequences across genders, then STEM retention programs, teaching training initiatives, and academic support services built on one-size-fits-all models may be quietly failing a substantial share of their audience. A mentoring scheme that boosts men’s satisfaction by enriching their learning environment might do little for women whose satisfaction hinges on self-efficacy; conversely, programs that build women’s confidence through mastery experiences and strategy training may be precisely the lever that both grades and satisfaction require for female students.

The broader context sharpens the stakes. International assessments such as the Programme for International Student Assessment have long documented gender gaps in mathematics and science confidence, and bodies like the OECD and national ministries—including Peru’s Ministry of Education, whose institutional landscape frames this research—have prioritized widening STEM participation. Women remain underrepresented in many STEM careers, and attrition during undergraduate study is one point at which the pipeline narrows. Understanding which psychological factors predict performance and satisfaction differently for men and women offers a route toward targeted, evidence-based support rather than blunt institutional policy.

Methodologically, the study demonstrates the value of multigroup structural equation modeling for educational psychology. Had the researchers stopped with the pooled model, which showed good overall fit, they might have concluded that a single unified story explained all students’ experiences. Only by testing invariance and then partitioning the sample did the gender-specific pathways emerge—an analytic lesson with relevance well beyond this single dataset. The work also underscores the continuing utility of instruments derived from frameworks like the Motivated Strategies for Learning Questionnaire, a widely used measure of students’ learning strategies and self-efficacy, when combined with rigorous measurement-invariance testing.

The research was funded by Pontificia Universidad Católica del Perú under grant CAP PI-1139 and approved by the university’s ethics committee under approval number 085–2024-CEI-CCSSHHyAA/PUCP, with all participants providing informed consent. Published as an open-access article in BMC Psychology, the study was received in August 2025, accepted in August 2026, and published on 03 September 2026, with a citable version carrying a permanent DOI.

For educators, the takeaway is that autonomy-supportive teaching and explicit training in metacognitive strategies remain foundational—both were linked to self-efficacy in every group examined. But the study also warns that confidence, satisfaction, and performance are woven together differently for women and men in STEM fields. For women, whose self-efficacy connects to both grades and satisfaction, cultivating belief in capability may be the single most consequential intervention. For men, satisfaction appears tied to the climate itself. As universities worldwide scramble to close gender gaps in science and engineering, this research suggests that the path to equity runs not through uniform policies but through a finer-grained understanding of how the same classroom can be, in psychological terms, two very different places.