Interpretation and design of nonregulatory on-farm feeding trials

Nonregulatory feeding trials are used to determine the frequency and magnitude of response to a nutritional treatment, the odds of success, and factors that influence the odds and to provide an estimate of potential economic impact. These types of trials are not appropriate for elucidating mechanisms or modes of action. Correct experimental design is critical. Important factors that impact the validity of these trials are animal randomization, potential confounding with time, identification of the proper experimental unit, and adequate replication. Animals should be assigned to treatment without a systematic influence of environmental or physiological factors. The experimental design must account for and remove these influences. The experimental unit is the smallest entity to which the treatment can be applied randomly. In commercial settings, this is often a pen of animals. In a pen feeding situation, the experimental unit is the pen, although measurements may be taken on an individual animal basis. Frequency of data collection generally does not influence the treatment effect; however, it does influence the variance associated with the observations and the power to detect differences in treatments. Statistical methods that account for repeated observations over time are required when analyzing data with multiple observations on the same animal. Improper accounting for environmental changes over time is the most common error. Other errors include improper assignment of animals to treatment, failure to replicate the treatment across multiple experimental units, lack of on-farm oversight, and poor calibration of test equipment.