borderline significanceP ≈0.07
(Minor exceptions occurred when comparing models 1 and 2, for instance when the LR test showed borderline significance ( P ≈0.07) for Sweden, but the AIC criterion showed evidence of observer heterogeneity).
(Minor exceptions occurred when comparing models 1 and 2, for instance when the LR test showed borderline significance ( P ≈0.07) for Sweden, but the AIC criterion showed evidence of observer heterogeneity).
Significant or borderline significant heterogeneity was found in all the countries except Belgium, but the absolute estimates of error rates were typically quite low.