Learning styles statistics reveal a clear gap between widespread belief and the evidence used to test whether instruction should be matched to a learner’s preferred style. Educator surveys, a large VARK questionnaire sample, higher-education reviews, and individual matching studies show why study strategies should be treated as flexible tools rather than fixed rules.
Contents
- Educator belief in learning-style matching
- VARK preference distributions
- What learners say about their results
- Learning-style models in higher education
- Measured effects of matching instruction
- Pupil and teacher agreement
Educator belief in learning-style matching
A systematic review covering educator samples collected from 2009 through early 2020 found that 89.1% of 15,045 educators agreed with the statement that people learn better when taught in their preferred learning style. The simple average across 34 samples was 86.2%. The weighted result gives a sense of the total sample, while the simple average treats each sample equally.
Agreement varied substantially between samples, ranging from 58% to 97.6%. In the same review, weighted agreement reached 95.4% among pre-service teachers and 87.8% among qualified teachers. Across three higher-education studies conducted from 2009 through 2019, the weighted agreement rate was lower, at 63.6%.
These results describe beliefs, not proof that matching instruction improves achievement. The review found no statistically significant relationship between study year and agreement: Spearman r = −0.290, p = 0.102. Teaching experience also had no statistically significant relationship with agreement, with Spearman r = −0.158, p = 0.642. The comparison between qualified and pre-service teachers was not significant either (Mann–Whitney U = 82, p = 0.529).
Evidence-focused training changed reported beliefs in four pre/post studies conducted from 2015 through 2020. Agreement that learning-style matching is effective fell from a weighted 78.4% before training to 37.1% afterward. The pooled change had a large Cohen’s d of 3.6, and the paired comparison was statistically significant (p = 0.012).
Practice did not differ significantly from belief in the available comparison. Across seven studies from 2009 through early 2020, 79.7% said they used or planned to use learning-style matching in teaching, compared with 89.1% who believed it was effective. The difference was not significant (U = 76.5, p = 0.146).
Source: How Common Is Belief in the Learning Styles Neuromyth, and Does It Matter? A Pragmatic Systematic Review, covering 2009–early 2020 and, for the training comparison, 2015–2020.
VARK preference distributions
VARK uses four preference categories: Visual, Aural, Read/write, and Kinesthetic. In the VARK research statistics reported for respondents completing questionnaires from September 2022 through August 2023, the sample contained 1,048,292 respondents. The distribution was more multimodal than single-mode:
| VARK profile | Share of respondents |
|---|---|
| Single preference | 34.0% |
| Bimodal preference | 20.1% |
| Trimodal preference | 15.0% |
| Four-part preference | 31.0% |
| Multimodal total | 66.0% |
Among single preferences, Kinesthetic was the largest category at 23.2%. Aural accounted for 5.7%, Read/write for 3.3%, and Visual for 1.9%. The multimodal categories included 20.1% bimodal profiles, 15.0% trimodal profiles, and 31.0% four-part profiles.
Several combinations were reported separately. The AK bimodal combination accounted for 11.9% of respondents. The ARK trimodal combination accounted for 4.5%, while the VAK trimodal combination accounted for 7.5%. The VARK Integrative four-part preference accounted for 25.4%.
Looking at whether a modality appeared anywhere in a profile gives a different picture from counting single preferences. Some Kinesthetic modality was included in 87.6% of profiles. Some Visual modality was included in 48.7%, and some Read/write modality was included in 45.0%. A learner can therefore have a modality in a multimodal profile without having it as a sole preference.
These percentages come from the VARK research report for September 2022–August 2023; they describe questionnaire respondents during that period and should not be treated as a universal population estimate.
Source: VARK® Research: What do we know about VARK?, September 2022–August 2023.
What learners say about their results
The same VARK sample included demographic information for some respondents. Among respondents who provided gender information, 60% were female, 38% were male, and 2% selected a non-binary option. These percentages refer to the respondents who supplied gender information, not necessarily all 1,048,292 questionnaire respondents.
The age distribution also leaned young. Respondents aged 25 or under made up 55% of the sample. Those aged 12–18 accounted for 22%, and those aged 19–25 accounted for 33%. The two younger age groups together match the reported under-25 share.
When asked whether their VARK result matched their own perception of how they learn, 72% said that it matched. This is a report of agreement between a questionnaire result and a respondent’s perception. It does not measure whether teaching in that preference produces better test performance, retention, or grades.
For study planning, the distribution supports a practical interpretation: many learners report more than one preferred mode, and many profiles include Kinesthetic, Visual, or Read/write elements. A student may reasonably try reading, listening, visual organization, and hands-on practice, then keep the methods that help with the particular subject and task.
Source: VARK® Research: What do we know about VARK?, September 2022–August 2023.
Learning-style models in higher education
A review of 109 higher-education learning-style papers published from 2013 through July 2015 found that 103 papers, or 94%, began with positive intent toward learning styles. Ninety-seven papers, or 89%, reported a positive learning-style outcome. These figures describe the direction and outcomes reported in the reviewed literature, not a pooled estimate of learning gains.
Only one of the 109 papers tested the matching hypothesis as defined by the review. None contradicted the earlier critical conclusions of Pashler and Coffield. This distinction matters: a paper can report that learners have different preferences, or that a learning-style intervention was viewed positively, without testing whether matching instruction to a category improves learning.
The models used in the literature were not evenly distributed. Kolb classifications appeared in 34% of the 109 papers, while VAK-type classifications appeared in 33%. Felder–Silverman or Felder–Soloman classifications appeared in 12%.
Health-profession students accounted for 53 of 57 PubMed-indexed higher-education studies, or 93%. That concentration means the review’s research pattern is especially connected with health-profession education. It also limits how directly those papers can represent every university subject or every learner.
Source: The Learning Styles Myth is Thriving in Higher Education, review period 2013–July 2015.
Measured effects of matching instruction
Individual studies summarized in a meta-analysis reported different effects across auditory, visual, and verbal comparisons. Hedges’ g expresses the standardized difference reported for the matched-versus-unmatched comparison; the direction and size should be read with the participant count and study context.
| Study and subgroup | Participants | Hedges’ g |
|---|---|---|
| Aslaksen and Lorås (2019), auditory | 9 | −1.25 |
| Aslaksen and Lorås (2019), visual | 13 | 0.85 |
| Cuevas and Dawson (2018), auditory | 118 | −2.87 |
| Cuevas and Dawson (2018), visual | 65 | 2.13 |
| Ge (2021), auditory | 76 | −0.74 |
| Ge (2021), visual | 64 | 1.20 |
| Hazra et al. (2013), verbal engineering comprehension | 15 | −0.41 |
| Hazra et al. (2013), visual engineering comprehension | 124 | −0.10 |
| Kam et al. (2020), auditory | 29 | 0.76 |
The Aslaksen and Lorås (2019) auditory subgroup had 9 participants and a Hedges’ g of −1.25, while its visual subgroup had 13 participants and a g of 0.85. In Cuevas and Dawson (2018), the auditory subgroup had 118 participants with a g of −2.87, and the visual subgroup had 65 participants with a g of 2.13.
In Ge (2021), the auditory subgroup included 76 participants and had a g of −0.74; the visual subgroup included 64 participants and had a g of 1.20. Hazra et al. (2013) reported a verbal engineering-comprehension comparison with 15 participants and a g of −0.41, plus a visual comparison with 124 participants and a g of −0.10. Kam et al. (2020) reported an auditory subgroup of 29 participants and a g of 0.76.
The spread of directions and subgroup sizes is important context for interpreting these figures. The listed studies do not provide a single uniform effect, and several subgroup counts are small.
Source: Is it really a neuromyth? A meta-analysis of the learning styles matching hypothesis, including studies from 2013, 2018, 2019, 2020, and 2021.
Pupil and teacher agreement
A 2018 Athens primary-school study included 199 fifth- and sixth-grade pupils: 105 girls and 94 boys. It also included 19 teachers whose mean teaching experience was 20.05 years. All 19 participating teachers reported that tailored teaching enhanced pupils’ intake of new information.
The study also compared pupils’ self-identification with teachers’ labels. Among pupils who self-identified as Visual, 11 were also labeled Visual by teachers, while 9 were labeled Auditory and 17 Kinaesthetic. Among pupils who self-identified as Auditory, 15 were also labeled Auditory, while 7 were labeled Visual and 17 Kinaesthetic.
Among pupils self-identifying as Kinaesthetic, 64 were also labeled Kinaesthetic by teachers, while 32 were labeled Visual and 27 Auditory. The counts show that teacher judgments and pupil self-assessments did not always point to the same category in this study.
The study’s finding about tailored teaching is a report of what participating teachers believed. The pupil-label counts concern agreement between two classification approaches; they do not establish that category matching caused improved learning outcomes.
Source: The Learning Styles Educational Neuromyth: Lack of Agreement Between Teachers’ Judgments, Self-Assessment, and Students’ Intelligence, 2018 study.