The Analysis of Errors in Assaying Fecal Androgens in the Dairy Cow

Open Access
Publisher Website
Google Scholar
Abstract
The errors for the chick-feeding method of estimating the androgenic activity of cow feces were assessed using data from 9 assays. Methods of reducing the variability of such estimates were explored. Large "between chick" sampling variation and significant "between_pen"effects were found, thus making the "between-pen" mean square the appropriate error for testing treatment differences and indicating that several small pens per prepn. would give more precise results than 1 or 2 large pens of chicks. The use of body wt. in conjunction with comb wt. reduced the coefficient of variation (c. v.) from 40% to about 33% and the logarithmic coefficient of variation (lambda) from 0.48 to about 0.35 whether body wt. was introduced as a co-variable in an analysis of co-variance or directly to form a ratio of comb wt. to body wt. or body wt. squared. Methods of determining the no. of chicks to use per prepn. to obtain a desired sensitivity are presented, using the c.v.''s and lambdas found in these data.