Among the respondents who did have a single highest-scoring dimension, the Activist category dominated. Activists — learners who prefer direct involvement, new experiences and learning by doing — accounted for 28.7 percent of the full sample, or 429 individuals. Theorist came next at 12.0 percent, Reflector at 9.5 percent, and Pragmatist at just 5.3 percent. But the tie category dwarfed every single style, and its internal structure was far from uniform. The most common configuration was a four-way tie across all dimensions, accounting for 32.4 percent of tied cases and 14.4 percent of the entire sample. Two-way ties made up 39.2 percent of tied cases, three-way ties 28.4 percent, and four-way ties 32.4 percent.

The underlying score distributions help explain why ties were so frequent. Reflector scores had the highest mean at 17.59 with a standard deviation of 2.36, followed by Pragmatist at 16.87, Theorist at 16.81, and Activist at 13.70 with a much larger standard deviation of 4.71. Activist scores ranged from 3 to 20, while the other dimensions spanned narrower ranges. Crucially, no respondent scored at the floor of zero on any dimension, but ceiling effects were pronounced: 26.9 percent of respondents hit the maximum score of 20 on Reflector, 24.1 percent on Theorist, 22.9 percent on Pragmatist, and 17.1 percent on Activist.

The four dimensions were also far from independent. Pearson correlations between them ranged from r = .474 to r = .692, with the strongest association between Theorist and Pragmatist scores and the second strongest between Activist and Pragmatist. These moderate-to-strong positive correlations mean that a respondent who endorsed one learning preference strongly typically endorsed the others as well. The authors argue that this overlap, together with the tie frequency, undermines the practice of reading a maximum score as evidence of a mutually exclusive learner type. The internal consistency of the subscales added a further note of caution: Cronbach’s alpha was .662 for Activist, .758 for Reflector, .732 for Theorist, and .753 for Pragmatist, with the Activist subscale comparatively weak.

The researchers then tested whether these classifications were associated with three participant characteristics: gender, teaching-level specialisation, and programme family. All three chi-square tests of independence were statistically significant, but the effect sizes were small. Gender was associated with classification, χ²(4, N = 1,496) = 17.91, p = .001, with Cramér’s V = .109. Teaching level produced χ²(4, N = 1,496) = 41.76, p < .001, Cramér’s V = .167. Programme family showed the strongest association, χ²(12, N = 1,496) = 178.94, p < .001, Cramér’s V = .200 — still a modest magnitude by conventional standards.

The descriptive patterns behind those tests were nonetheless revealing. Tied classifications were the largest category for both female respondents (45.1 percent) and male respondents (43.8 percent), with Activist the most common single style in both groups. Primary trainees recorded ties at 51.3 percent, compared with 37.2 percent among junior high school trainees, who in turn showed higher proportions of Activist (31.3 percent) and Pragmatist (8.0 percent) classifications. Programme family produced the sharpest contrasts: ties reached 62.7 percent among Arts/PE students and 51.4 percent among Social/RME students, while Activist was the largest single category among Languages students (39.7 percent) and STEM/Tech students (43.7 percent), where ties fell to just 17.7 percent.

A multinomial logistic regression with Activist as the reference outcome sharpened the picture further. Gender was not a significant predictor of any classification; for example, males were not significantly more or less likely than females to be classified as Reflector rather than Activist (relative risk ratio = 0.44, 95% CI [0.18, 1.09], p = .075). Teaching level mattered only for the Pragmatist contrast: Primary trainees were significantly less likely than JHS trainees to be classified as Pragmatist rather than Activist (RRR = 0.14, 95% CI [0.03, 0.68], p = .015). Programme family predicted several outcomes. Compared with Arts/PE students, those in Languages, Social/RME and STEM/Tech were significantly less likely to be classified as Pragmatist rather than Activist, and Languages and STEM/Tech students were also significantly less likely to fall into the Tie category rather than Activist. The authors stress that several confidence intervals were wide and the estimates should be interpreted cautiously.

A sensitivity analysis re-estimated the model after excluding all 666 tied cases, leaving 830 respondents. The overall model remained significant, likelihood-ratio χ²(15) = 106.18, p < .001, with a McFadden pseudo-R² of .053 and a Nagelkerke pseudo-R² of .125. The same core pattern held: gender remained non-significant, the Primary-versus-JHS Pragmatist effect persisted (RRR = 0.15, 95% CI [0.03, 0.73], p = .019), and the programme-family differences for the Pragmatist category remained significant for Languages, Social/RME and STEM/Tech. This indicates the main regression findings were not driven solely by the large number of tied cases, although the wide intervals and modest pseudo-R² values underline how much variation remains unexplained.

One structural feature of the data deserves particular attention. College membership was strongly associated with gender, χ²(7, N = 1,496) = 802.66, p < .001, Cramér’s V = .732, because some participating colleges admit only women and others only men. The authors flag this as potential structural confounding and treat gender effects in the regression accordingly. By contrast, college membership showed no meaningful association with teaching level (Cramér’s V = .005) or programme family (Cramér’s V = .008), so those comparisons were not similarly entangled with institutional identity.

The study’s theoretical framing is deliberately cautious. The Honey and Mumford model, derived from Kolb’s experiential learning theory, describes four orientations — Activist, Reflector, Theorist and Pragmatist — but the authors use the LSQ as a measure of reported preferences rather than fixed learner identities, retaining tied maxima as a separate category instead of forcing respondents into a single label. They note that prior psychometric work on the LSQ found only limited support for four clearly distinct underlying constructs, and that major reviews have found insufficient evidence for matching instruction to diagnosed learning styles. The practical implication they draw is not that each student should be taught in a prescribed way, but that teacher education programmes should offer varied pedagogical experiences — active engagement, reflection, theoretical understanding and practical application — while treating learning-style scores as descriptive, context-dependent indicators rather than diagnostic labels. As a cross-sectional survey of Level 100 students in one region of Ghana, the findings describe score patterns at a single point in time and cannot establish causal effects of programme membership on learning preferences.

Subject of Research: Learning-style score patterns and tied classifications among pre-service teachers in Ghanaian Colleges of Education

Article Title: Patterns of Honey–Mumford learning-style scores and classifications among pre-service teachers in Ghana

Article References: Ntumi, I., Atakorah, P. O., Asumang, P., Cobbinah, E. A., & Aboagye, B. (2026). Patterns of Honey–Mumford learning-style scores and classifications among pre-service teachers in Ghana. Frontiers in Education, 11, Article 1907726. https://doi.org/10.3389/feduc.2026.1907726

Image Credits: AI Generated

DOI: 10.3389/feduc.2026.1907726

Keywords: learning styles, Honey and Mumford, pre-service teachers, Ghana, teacher education, Learning Styles Questionnaire, experiential learning, cross-sectional survey, multin logistic regression, chi-square tests, psychometrics, Patterns