Table 3.

Measurement equivalence between gamified and paper-based medication literacy assessments using a randomized crossover design (phase 2, n=85, grades 5 and 6, Nanjing, China, 2025). Order effect was tested using a 2 × 2 mixed-design ANOVA (sequence × modality interaction).

Domain	Order effect, F test (df); P value	ICCa (95% CI)	Score differenceb, P value; Cohen d (95% CI)	Bland-Altman bias (95% LoAc; 95% CI)	Decision consistency, κ (agreement %)
Knowledge	0.38 (1, 66); .54	0.94 (0.90 to 0.96)	.26; 0.047 (–0.04 to 0.13)	0.039 (–0.554 to 0.633; 0.19 to 0.27)	0.87 (93.4)
Attitude	0.01 (1, 66); .94	0.99 (0.98 to 0.99)	>.99; 0.000 (–0.04 to 0.04)	0.000 (–0.179 to 0.179; 0.06 to 0.06)	0.92 (96.1)
Perceived behavioral control	0.01 (1, 66); .93	0.98 (0.97 to 0.99)	.41; –0.017 (–0.06 to 0.02)	–0.010 (–0.213 to 0.193; 0.10 to 0.08)	0.92 (96.1)
Behavioral intention	0.00 (1, 66); .97	0.99 (0.98 to 0.99)	>.99; 0.000 (–0.03 to 0.03)	0.000 (–0.160 to 0.160; 0.06 to 0.06)	0.97 (98.7)

^aICC: intraclass correlation coefficient; ICC(3,1) was calculated using a 2‑way mixed‑effects model with absolute agreement.

^bScore differences were tested using a paired t test; Cohen d indicates the effect size.

^cLoA: limits of agreement.