Introduction Tablet-based assessments offer benefits over scannable-paper assessments; however, there is little known about the impact to the variability of assessment scores.

Methods Two studies were conducted to evaluate changes in rating technology. Rating modality (paper vs tablets) was manipulated between candidates (Study 1) and within candidates (Study 2). Average scores were analyzed using repeated measures ANOVA, Cronbach’s alpha and generalizability theory. Post-hoc analyses included a Rasch analysis and McDonald’s omega.

Results Study 1 revealed a main effect of modality (F (1,152) = 25.06, p < 0.01). Average tablet-based scores were higher, (3.39/5, 95% CI = 3.28 to 3.51), compared with average scan-sheet scores (3.00/5, 95% CI = 2.90 to 3.11). Study 2 also revealed a main effect of modality (F (1, 88) = 15.64, p < 0.01), however, the difference was reduced to 2% with higher scan-sheet scores (3.36, 95% CI = 3.30 to 3.42) compared with tablet scores (3.27, 95% CI = 3.21 to 3.33). Internal consistency (alpha and omega) remained high (>0.8) and inter-station reliability remained constant (0.3). Rasch analyses showed no relationship between station difficulty and rating modality.

Discussion Analyses of average scores may be misleading without an understanding of internal consistency and overall reliability of scores. Although updating to tablet-based forms did not result in systematic variations in scores, routine analyses ensured accurate interpretation of the variability of assessment scores.

Conclusion This study demonstrates the importance of ongoing program evaluation and data analysis.