POWER AND SENSITIVITY OF SOME SIMPLE TESTS FOR DETECTION OF MAJOR GENES IN OUTBRED POPULATIONS

Three simple statistical tests (Bartlett test, log-ANOVA test, and the mixture test, a model that fits a mixture of normal distributions to data) for detecting major genes in hierarchical data structures in out bred livestock populations were investigated. The power of;he tests wa s evaluated when the major gene had 2-3 alleles and in the presence of a skewed distribution of polygenic phenotypes or a nongenetic mixture of normal distributions. The most powerful test, bur also the most se nsitive to skewness, was the mixture test. The Bartlett test was more Powerful than the log-ANOVA test, but was Sensitive to skewness of the data and to an underlying non-genetic mixture of distributions. All t ests were able to detect a dominant major gene with a difference of 1. 5 phenotypic standard deviations between homozygotes from a data set o f 5 000 records if the frequency of the dominant allele Tvas less than 0.9. The most suitable robust simple test for major gene detection un der a hierarchical data structure is the log-ANOVA test.