There are various translations, including Spanish (40), while a Mandarin version showed an alpha reliability of 0.92 (41); a two-factor solution was obtained in a Cantonese version (42). A 10-item abbreviated Cantonese version had an alpha coefficient of 0.78, and a kappa agreement of 0.84 with the full-scale version in classifying cases (43). A Korean version has been validated (44). In a study of Japanese adolescents, response to the positive items in the CES-D differed markedly from that in U.S. samples, whereas responses to negatively worded items was similar; this resulted in poor psychometric properties for this Japanese version (45).