Cambridge Colour Test: reproducibility in normal trichromats.

This study evaluated reproducibility of the Trivector subtest of the Cambridge Colour Test. Data for normal trichromats were obtained in Brazil ($ N = 111 $N=111) at T0, six months (T1), and 12 months later (T2), and in the United Kingdom (${ N}={79}$N=79), with the test directly followed by a retest. Coefficients of repeatability-Bland-Altman indices-for Protan, Deutan, and Tritan vectors were similar for both datasets. Intraclass correlation coefficients (ICCs)-measures of reliability-were low or moderate for these relatively homogeneous datasets; for a heterogeneous dataset, comprising color-normal and abnormal observers, ICCs were 0.80-0.98, indicating the high discriminative accuracy of the Trivector subtest.