Nursing Students Learn Statistics With Their Bodies in Embodied Learning Trial

Nursing students who physically acted out abstract statistical concepts—forming human pyramids to represent Bloom’s taxonomy, lining up by height to embody measures of central tendency, and gesturing reliability and validity with their fingers—reported markedly more positive attitudes toward educational measurement and stronger self-efficacy than classmates taught the same material through traditional lectures. That is the central finding of a randomized controlled trial published in BMC Medical Education, which tested whether embodied learning, a pedagogy rooted in the idea that cognition is grounded in bodily experience, could help prospective nurses master one of the most abstract units in their curriculum. The study, led by Özgür Tutal at Hakkari University, enrolled all 31 senior undergraduate nursing students taking the Education and Instruction in Nursing course during the fall semester of the 2025–2026 academic year.

Because the nursing department had admitted its first cohort only in 2022, this was the first time the fourth-year course had been offered, and no previous cohort existed for comparison. Participants were individually randomized to a treatment group of 15 students receiving embodied learning instruction or a comparison group of 16 students receiving conventional lectures. Baseline tests confirmed the two groups were statistically equivalent in academic performance, attitudes, and self-efficacy before the intervention began, with pre-test comparisons yielding p-values between 0.446 and 0.874. The theoretical backdrop is a challenge to a long-standing assumption in education. Traditional lecture-based teaching, the paper argues, inherits from Cartesian dualism the notion that mind and body are separate, with cognition residing purely in the brain and the body serving as a passive instrument. Embodied Cognition Theory counters that thinking emerges from dynamic interactions among the brain, sensorimotor systems, and the environment, meaning that meaning is actively constructed through movement and physical experience rather than passive processing of symbols.

Embodied learning translates this theory into practice by treating gesture, spatial orientation, and sensory experience as essential components of learning, guided by three design principles: sensorimotor involvement, action-based compatibility with the concept being taught, and perceived immersion. The approach also draws on Cognitive Load Theory. Humans process biologically primary information—gestures, instinctive movements, innate behaviors—with relatively little cognitive effort, and integrating such primary channels into the learning of biologically secondary information, like abstract academic content, can reduce extraneous load on working memory. Gestures in particular can act as a secondary representational format, offloading verbal processing onto other cognitive systems. Notably, however, prior research is mixed: some studies of anatomy education found no advantage for embodied methods over traditional instruction, and one experiment showed that pointing and tracing gestures improved performance without reducing subjective cognitive load, cautioning that gestures which do not genuinely facilitate learning may simply add extra burden.

Over seven sessions, the treatment group covered exactly the same content as the lecture group—fundamental concepts, assessment types, measurement errors, test formats, Bloom’s taxonomy, central tendency, item analysis, reliability, and validity—but through structured physical activities. In a session on measurement errors, three groups staged short theatrical sequences: the constant-error group uniformly added five points to all scores, the systematic-error group adjusted scores by seating row, and the random-error group assigned arbitrary values with their fingers while peers guessed the error type. A dartboard drawn on the board let students model reliability and validity combinations by finger placement, with tight clusters at the bullseye representing high reliability and high validity. In the central tendency session, students wore cards displaying their height and arranged themselves into a living number line, allowing the instructor to ask the group to locate the median or identify where the mode would fall. The item analysis session had students line up in descending order of hypothetical total scores, physically demarcating the upper and lower 27 percent of the distribution on the classroom floor; for each test item, students who answered correctly stepped forward, and the discrimination index was computed in real time, with raised hands signaling strong discrimination above 0.30 and a discarding gesture marking weak items below 0.19.

The statistical picture that emerges from the trial is more nuanced than the headline affective gains might suggest. Both groups improved significantly on the academic performance test over the course of the intervention, and in a simple unadjusted comparison of post-test scores the embodied learning group came out ahead, with an effect size of g = 0.748 and a p-value of 0.041. Yet when the researchers applied their prespecified analysis plan—an analysis of covariance that adjusted for baseline scores, combined with a Bonferroni correction for the three outcomes tested—the academic performance difference fell just short of the corrected significance threshold of 0.017, landing at p = 0.047. The adjusted mean difference of 11.07 points, with a confidence interval stretching from 0.13 to 22.00 points, still favored the embodied learning group, but the authors are careful to label this result exploratory rather than definitive.

The affective outcomes tell a stronger story. Attitudes toward educational measurement, measured with a 31-item validated inventory covering interest, affective, and course-related dimensions, showed a mean increase of 17.74 points in the embodied learning group, and this difference survived the multiplicity correction with a very large effect size of g = 1.945. Self-efficacy perceptions, assessed across four domains spanning the selection of methods and techniques, data analysis and interpretation, providing feedback, and reviewing the evaluation process, also improved significantly relative to the lecture group, rising from a mean of 66.27 to 81.80 within the treatment group, with an effect size of g = 1.091. The comparison group’s self-efficacy gain was marginal and statistically non-significant, moving from 67.06 to 71.63. A post hoc sensitivity analysis clarified why these findings carry more weight than the academic performance result: with 31 participants, the study was only adequately powered to detect large effects, roughly g of 0.91 or greater. The attitude and self-efficacy effects comfortably exceeded that bar and achieved power of 0.99 and 0.88 respectively, whereas the academic performance effect, at g = 0.748, had achieved power of just 0.62—meaning there was a 38 percent chance of missing a true effect of that size.

The measurement instruments themselves deserve scrutiny when weighing these results. The attitude inventory showed good internal consistency in this sample, with a Cronbach’s alpha of 0.87, and the self-efficacy scale performed even better at 0.94. The academic performance test, however, was a 16-item multiple-choice instrument adapted from national examination items, and its KR-20 reliability coefficient of 0.55 indicates limited internal consistency. The authors acknowledge this openly, noting that short tests administered to small samples routinely yield depressed reliability estimates—classical guidance suggests that KR-20 values above 0.50 become acceptable only with examinee samples of at least 30, and reliability estimators suffer additional imprecision under range restriction. Similarly, the mean item discrimination index of 0.30 should be read as an unstable estimate vulnerable to sampling error rather than a verdict on item quality, since classic psychometric research has shown that discrimination indices calculated from samples of 15, 30, or even 60 examinees carry little value, and recent work suggests that useful item-level information requires samples approaching 100. Correlational patterns in the combined sample add an intriguing wrinkle: academic performance correlated moderately and significantly with self-efficacy, at r = 0.362, while attitudes toward measurement showed no significant relationship with academic performance, a reminder that how students feel about a subject and how well they perform on it can move along partly independent tracks.

Several limitations frame the interpretation of these findings. The sample of 31 students came from a single institution, and small-sample studies are known to produce inflated or unstable effect size estimates, so the very large g values reported for the affective outcomes should be treated as preliminary. The confidence interval around the academic performance effect spanned 0.03 to 1.46, encompassing values close to zero and indicating considerable uncertainty about the true population effect. The study also examined a single course unit, leaving long-term retention and transfer of learning unassessed. The same instructor who designed the embodied intervention also delivered instruction to both groups and conducted the assessments, which—while controlling for teaching style—introduces a risk of instructor-expectancy bias, since the researcher was not blind to condition. Moreover, the embodied condition differed from the traditional condition not only in its use of physical movement but also in its novelty, level of social interaction, and active participation structure, so some of the observed effects, particularly on attitudes and self-efficacy, may have been influenced by these extraneous factors rather than by embodied cognition per se. Future studies, the authors suggest, should compare embodied learning to an alternative active-learning approach matched for participation level but lacking sensorimotor integration.

Even with these caveats, the study offers a concrete template for educators wrestling with content that students find intimidating. Measurement and evaluation coursework routinely produces resistance among nursing students, who must master reliability coefficients, item difficulty indices, and taxonomies of cognition despite the subject’s highly symbolic and statistical character. The finding that physically enacting these concepts improved attitudes and self-efficacy, with suggestive but unconfirmed gains in performance, aligns with the theoretical prediction that grounding abstract symbols in sensorimotor experience makes them more accessible, and may help bridge what the literature calls the conceptual chasm between bodily knowing in clinical practice and cognitive knowing in theory. Confirming these effects in larger, multi-institution trials with longer follow-up, and using qualitative methods to trace exactly how embodied activities reshape students’ understanding, remains the necessary next step before embodied learning can be recommended as standard practice in healthcare education.

Subject of Research: Embodied learning instruction for teaching measurement and evaluation to undergraduate nursing students

Article Title: Concretizing the abstract: embodied learning as a pedagogical strategy for teaching measurement and evaluation

Article References: Tutal, Ö. (2026). Concretizing the abstract: embodied learning as a pedagogical strategy for teaching measurement and evaluation. BMC Medical Education, 26(1), Article 1728. https://doi.org/10.1186/s12909-026-10321-8

Image Credits: AI Generated

DOI: 10.1186/s12909-026-10321-8

Keywords: embodied learning, nursing education, measurement and evaluation, randomized controlled trial, self-efficacy, attitudes, cognitive load theory, embodied cognition, psychometrics, pedagogy, health professions education, Concretizing