The statistical power analysis of tests used in animal breeding studies on lamb birth weight in Turkey

Abstract

Abstract. This study aims on to evaluate the statistical power of 40 published articles in Turkey. In order to maintain consistency, only the most common statistical test (t-test) was power analysed, alpha was held at a uniform value of 0.05 and Cohen’s (1977) definitions of small, medium and large effect sizes were adopted. The Sample Power software was used in all analysis. A total of 40 articles, 24 of them include the gender effect and 16 of them include birth type effect on the birth weight of lambs, were used. An average power estimate for small effect size was 0.062 and 0.05, for medium effect size was 0.534 and 0.347 and for large effect size was 0.598 and 0.854. The statistical power appears to be adequate only for large effect size criteria in these studies. If 0.80 was taken as an adequate power, the results indicate, for gender and birth type factors that none of the studies had adequate power to detect a small effect size, % 57.1 for gender factors and % 83.3 for birth type did not have adequate power to detect a medium effect size, and % 60.9 for gender factor and % 28.2 for birth type factor did not have adequate power to detect even a large effect size. The application of the power analysis is recommended.