Measuring a milk response

lactation models add power

Abstract

Milk production is often the most important outcome variable in both field investigations and research trials involving dairy herds. Nutrition, health, and environment all have strong and rapid effects on milk production, but precise and unbiased quantification of such effects usually requires either matching of cases and controls py lactation stage and parity or use of a lactation model to minimize the confounding effects of lactation stage. Use of a model is often the only practical option, but little attention has been paid to measuring the precision and bias of lactation models, especially for individual lactations. An estimate of expected performance is essential in choosing among competing lactation models for maximum sensitivity and specificity in detecting changes from normal milk production, and in power calculations.