Faculty Development Programs May Change Teaching Less Than Teachers Report, Major Review Finds

Universities around the world spend enormous sums on faculty development: workshops, year-long certificate programs, mentoring schemes, communities of practice, and, more recently, online and blended formats. The premise behind all of it is simple and rarely questioned. Train the teachers, and their teaching will change. A new scoping review published in Frontiers in Education puts that premise under the microscope for the first time outside health professions education, and what it finds is uncomfortable. The claim that development changes what teachers actually do in the classroom rests overwhelmingly on the teachers’ own testimony, and when it is tested against evidence from anyone other than the teacher, it holds far less often.

The review, conducted by Arianna Marci of the University of Genoa and Martina Ragusa and Flavio Manganello of Italy’s National Research Council, followed the PRISMA Extension for Scoping Reviews and searched Scopus and Web of Science for studies published between 2005 and 2025. From 491 deduplicated records, 69 sources met the inclusion criteria: empirical or theoretical work reporting a change in the teaching practices of university teachers following structured faculty development, in any discipline other than medical, dental, nursing, veterinary, or pharmacy education. The authors deliberately excluded health professions education because that field, with its accreditation mandates and routine clinical observation of teaching, has already been synthesized twice and operates under conditions most of higher education does not share. Setting it aside allowed the rest of the field to be described on its own terms for the first time since 2008.

The reviewers charted 498 reported effects across the included sources, organized using the four-level adaptation of Kirkpatrick’s model developed by Stes and colleagues for faculty development in higher education. Level 3, teacher behaviour, is the level most directly tied to what development programs promise: the transfer of learning into actual classroom practice. Of the 498 effects, 217 sat at this behavioural level. And here the review’s central finding emerges. Of those 217 behavioural effects, 150 rested on teacher self-report, either narrative accounts or responses to questionnaires, while only 66 rested on evidence external to the teacher, such as peer or expert observation, student perceptions, or analysis of course artefacts. One effect could not be classified at all.

The direction of effect depended dramatically on who supplied the evidence. Positive effects accounted for 65 percent of self-reported behavioural changes but only 42 percent of externally documented ones, with a quarter of the externally documented effects reporting no change whatsoever. The same asymmetry appeared for measurement design: 65 percent of effects measured without a pre-intervention baseline were positive, against 49 percent of those measured with one. When the two factors were combined, the most demanding configuration, external evidence measured against a baseline, was the only one in which no change was the modal outcome, reported for 14 of its 24 effects. In other words, the predominance of positive findings in this literature is partly an artefact of the methods that produced them.

The methodological weaknesses run deep. Of the 64 empirical studies appraised with the Mixed Methods Appraisal Tool, 29 used non-randomised or purely descriptive quantitative designs, and only a single study employed random assignment. Forty-seven of the 69 sources had no comparison group of any kind. Quantitative and mixed methods studies recurrently showed incomplete outcome data, unaccounted confounding, and undocumented intervention fidelity; in 13 of 23 appraised non-randomised studies, attendance or completion data were absent, leaving the exposure whose effect was being estimated entirely unverified. Only two of those 23 reported complete outcome data. The qualitative work, by contrast, largely met its methodological criteria, though its recurrent weakness was asserting changes in classroom practice on the basis of participants’ self-reports alone.

Perhaps the most striking finding concerns evaluation frameworks. Fifty-three of the 69 sources, more than three quarters, named no evaluation model at all. They described an intervention, chose instruments, and reported results without locating the exercise in any framework. The 16 that did name one did not converge: Kirkpatrick appeared in four, Rogers’ diffusion of innovations and the Community of Inquiry framework in two each, and eight further frameworks in a single source apiece. Notably, the Stes framework designed specifically for faculty development in higher education was used as an evaluation scheme by no source other than the Antwerp papers by its own authors. The field’s apparent theoretical pluralism, the authors observe, rests partly on a category difference that passes unremarked, since outcome taxonomies, theories of individual change, and intervention characterisations are not models of the same thing.

The instrumentation tells a similar story of fragmentation. Only 75 of the 217 behavioural effects rested on a published instrument. The Reformed Teaching Observation Protocol supported 30 of them, more than any other, followed by the Approaches to Teaching Inventory with 15. But 87 behavioural effects rested on instruments the study authors devised themselves, and 55 on no instrument at all. Two thirds of the evidence on what teachers do therefore rests on measures that exist only in the study that used them and cannot be compared across the corpus. The field’s instrumentation is bimodal: a small observational tradition anchored on RTOP and COPUS, and a large body of study-specific self-report measures.

Two sources in the corpus demonstrate the divergence between self-report and observation internally. Ross and colleagues found that the Approaches to Teaching Inventory and the Reformed Teaching Observation Protocol were uncorrelated in the same teachers, with no observed change in RTOP scores while both ATI scales rose significantly. Stes and colleagues found no change in teaching behaviour as perceived by students in the very programme for which an earlier paper had reported an effect on teaching approach. These are not marginal anomalies. They suggest that the field’s principal self-report instrument and its principal observation protocol may not be measuring the same object, and that the choice between them is not a matter of convenience.

The evidence base is also strikingly concentrated. Thirty-six of the 69 sources report on the United States, and 34 of the 39 sources that identify a disciplinary context are in engineering or other STEM fields, chiefly biology. The median study involved just 36 teachers, and 41 of the 69 sources were published between 2020 and 2025, making this a young and geographically narrow literature. What is known about faculty development’s impact on teaching practices, the authors conclude, is largely what is known about STEM teaching in North American universities. Whether the findings transfer to other disciplines and other systems remains an open question the corpus cannot answer.

The authors are careful about what their findings do and do not imply. None of this, they write, means that faculty development fails to change teaching practice. It means the literature, as presently constituted, is not in a position to establish that it does. The remedies they propose are modest and mostly consist of reporting what programs already know about themselves: naming the instruments used, recording attendance and completion, reporting attrition, and collecting a baseline of some kind, however crude. External observation of a sample of teaching, they argue, is not a refinement reserved for research but the only evidence that speaks directly to what teachers actually do, and the RTOP and COPUS tradition demonstrates that it can be organised at the scale of hundreds of sessions. For institutions that fund faculty development, the message is pointed: an evaluation specified as end-of-program participant feedback buys a result it cannot use, because the modal study of that design returns a positive finding almost irrespective of what occurred in the classroom.

Subject of Research: The impact of faculty development programs on teaching practices in higher education, assessed through a scoping review of evaluation methods and evidence quality.

Article Title: Impact of faculty development on teaching practices in higher education: a scoping review

Article References: Marci, A., Ragusa, M., & Manganello, F. (2026). Impact of faculty development on teaching practices in higher education: a scoping review. Frontiers in Education, 11, Article 1947549. https://doi.org/10.3389/feduc.2026.1947549

Image Credits: AI Generated

DOI: 10.3389/feduc.2026.1947549

Keywords: faculty development, higher education, teaching practices, scoping review, Kirkpatrick model, self-report bias, STEM education, classroom observation, RTOP, evaluation methods, teacher behaviour, research methodology