The open-source computerized Trail Making Test in PEBL: Preserved task structure, practice effects, and limited agreement with paper-based testing

This pilot study examined preliminary evidence regarding the measurement properties of the computerized Trail Making Test implemented in PEBL. Thirty-seven Brazilian adults completed the paper TMT and four administrations of three PEBL conditions: number sequencing, alternating number and letter sequencing, and letter sequencing. We evaluated short-term practice effects, task hierarchy, relative switching cost, and correspondence between formats using nonparametric tests, correlations, intraclass correlations, and Bland-Altman analyses. Completion times decreased significantly across administrations for all conditions, with the largest improvements occurring after the first administration. The expected hierarchy of difficulty remained stable. Relative switching cost decreased gradually, with a significant difference only between the first and fourth administrations after correction for multiple comparisons. PEBL and paper completion times were positively associated for TMT-A and TMT-B. However, intraclass correlations indicated limited consistency, while Bland-Altman analyses showed systematic bias and wide limits of agreement, with longer completion times on paper. Age was associated with performance only during the first administration, and no mood associations remained significant after correction. These findings indicate that the PEBL implementation preserves expected task ordering and short-term performance patterns, but provide limited evidence of cross-format correspondence; scores from the two formats should not be considered interchangeable.