|Title||Sample size determinations for studying selected cattle foraging behaviors|
|Publication Type||Journal Article|
|Year of Publication||1991|
|Authors||Havstad K, Olson-Rutz K.M.|
|Journal||Applied Animal Behavior Science|
Experimentation on livestock foraging behaviors under free-ranging conditions is frequently constrained by few experimental units (animals). Effective research designs require knowledge of expected sampling variability, but this information relative to range foraging behaviors is lacking. Sixteen 5-year-old range beef cows were used to estimate daily grazing time (DGT) and fecal output (percentage of body weight, FOBW) for 46 consecutive days during a winter grazing season. Estimates of components of variability associated with cows, days, and theoretical treatment effects were constructed from these data. Implied sample size requirements for a half-confidence-width =10% of the mean and a Type 1 significance level of 0.05 were 7 and 9 cows for DGT and FOBW, respectively, with day as a fixed effect. For day as a random effect, implied sample sizes were 5 days and 5 animals for DGT, and 5 days and 9 animals for FOBW. Variability among animals was greater when estimating FOBW than for DGT estimates. Detecting differences between two treatments sampled over 5 days with Type 1 and Type 2 probabilities of 0.05 and 0.10, respectively, implied sample sizes of five and > 20 cows for DGT and FOBW estimates, respectively. In general, these data indicated that <5 animals was insufficient and > 20 was excessive. These sample size estimates were obtained from a relatively uniform set of animals grazing a uniform quality and quantity of forage. Estimates of N would have been greater under more variable experimental conditions. Given constraints of statistics and the logistics of range nutrition experimentation, unless sample size requirements can be satisfied it may not be appropriate to examine certain questions with current experimental methodologies.