Başer, EyüpYiğit, Soner2025-01-272025-01-2720211301-08912651-4494https://doi.org/10.33724/zm.948879https://search.trdizin.gov.tr/tr/yayin/detay/434648https://hdl.handle.net/20.500.12428/16566In this study, Wilks’ ? (W), Hotelling-Lawley Trace (H) andPillai’s Trace (P) tests which are used in testing of statisticallysignificance for canonical correlation coefficients werecompared in terms of actual type I error rate. As a result of10000 simulation experiments conducted, when samples weretaken from multivariate distributions which are normal anddeviate slightly or moderately from normality, the W test wasconservative in terms of protecting actual type I error rate inall cases. However, when there is excessively deviate fromnormality, actual type I error rates for the W test exceededthe upper limit of Bradley’s criterion (4.50-5.50%) almost inall cases. On the other hand, the H test and P test generallyobtained actual type I error rates which were outside Bradleylimits.eninfo:eu-repo/semantics/openAccessİstatistik ve OlasılıkArticle03729210010.33724/zm.948879434648