2026 ASEE Annual Conference & Exposition

This Complete Research Paper compares spatial visualization outcomes across three instructional delivery modes: Traditional face-to-face, Virtual-Asynchronous, and Flipped classroom, in a first-year engineering spatial visualization course that implemented a widely used 10-module multimodal method (guided tutorials, sketching workbook, manipulatives, and interactive software). Spatial visualization is a well-established contributor to early success in engineering coursework and retention, yet many entering students begin with weaker preparation. Because these skills are teachable, first-year interventions provide a timely opportunity to strengthen performance and confidence and to support equitable participation and persistence in engineering pathways.

Using a quasi-experimental, cohort-based design, we collected data from three consecutive first-year cohorts taught by the same instructor across three 15-week fall semesters, each semester using one delivery format. The course integrated the same 10-module sequence in each format and embedded the Purdue Spatial Visualization Test: Rotations (PSVT:R) as a pre-test at the beginning of the semester and a post-test after module completion. We defined spatial visualization gain, Gain, as PSVT:Rpost − PSVT:Rpre and computed descriptive summaries of pre-test, post-test, and gain distributions for each delivery mode.

To examine outcomes for lower-prepared students and support equitable early engineering success, we conducted the primary gain comparison for the subgroup with PSVT:Rpre ≤ 18 (n = 58; Traditional, n = 26, Virtual-Asynchronous, n = 13, Flipped, n = 19). Mean gains were positive across all modes (Traditional 5.42; Virtual-Asynchronous 4.46; Flipped 4.42). We used Welch’s one-way ANOVA to compare mean gains across modes under unequal group sizes and potential variance heterogeneity and followed with Games–Howell pairwise comparisons; the omnibus test detected no statistically significant differences in gain by delivery mode (F(2, 32.62) = 0.51, p = .603), and no pairwise comparisons reached statistical significance after adjustment. A Kruskal–Wallis sensitivity analysis likewise did not detect differences in gain distributions across modes ( χ2(df = 2) = 1.48, p = .477).

We next evaluated whether spatial visualization improvement relates to course outcomes and whether that relationship varies by delivery mode. We modeled course grade points, GradePts, (0–4) as a function of Gain, PSVT:Rpre, delivery mode (Mode), and a Gain×Mode interaction. Gain positively predicted grade points (β = 0.057, p = .040), and the Gain×Mode interaction was not statistically significant, suggesting that the association between gain and course performance did not differ detectably across delivery modes.

Together, these results suggest that, within this instructional context, multiple delivery formats can support comparable PSVT:R gains for students entering with lower spatial visualization scores, and that students who improve more on PSVT:R tend to earn higher course grade points.

This study’s inferences are constrained by its non-randomized, single-instructor, single-site design, and cohort-level differences may confound delivery mode comparisons. Even so, the findings provide practical evidence for programs choosing among traditional, asynchronous, and flipped implementations of established spatial visualization curricula and underscore the value of replication across institutions and instructional settings.