Fig. 2

Cumulative sums of parameter metrics. a) Confidence intervals sorted in ascending order. Reference line is at 1.65, which is the threshold for intervals larger than expected by chance. Thus, most confidence intervals are within a reasonable range. b) Learning effects sorted in ascending order. Approximately 60% of learning effects were negative, indicating a lower parameter Z-score at the second assessment than the first. c) ICC-C values plotted in ascending order. Approximately 5% of each distribution were considered ‘good’ (> 0.75) and approximately 50% were > 0.50 (fair)