A dependency-light Python script (standard library only) that turns a ratings matrix into a defensible result. It computes (1) each clip's MOS with its 95% confidence interval both ways — the ITU-R BT.500-15 normal multiplier 1.96 and the ITU-T P.910 Student's-t multiplier correct for small panels; (2) the SOS-hypothesis parameter a (Hossfeld/Schatz/Egger 2011) as a sanity check on subject disagreement; (3) a significance test between two conditions, paired (same subjects) or two-sample, reporti
Download free PDF