Enhancing quality of dairy cattle research through adequate power analysis

Animal scientists use controlled experiments to test hypotheses and evaluate treatment effects within populations. Due to the impracticality of studying entire populations, representative samples are necessary. Assessing statistical power is a crucial step in designing robust experiments. However, many published articles in dairy science lack adequate power analysis, raising concerns about the applicability of findings. An analysis of trends from 1995 to 2023 based on publications in the "Production: Animal Nutrition" section of the Journal of Dairy Science reveals that only 4.73% of 4,376 articles reported power analysis. Additionally, 59% of papers that conducted power analysis did not report the tools used. With the large variability among the tools being used, and the predominant tool, SAS, being non-open access, there is a need to standardize power analysis and sample size calculation with an open-source tool. To address this gap, we developed the pwr4exp R package, an open-source tool that provides comprehensive model-oriented power analysis tailored for dairy science research. This review discusses key experimental designs for performing power analysis and sample size calculation using the pwr4exp R package. Developed based on linear mixed model inference, pwr4exp was validated by comparing with SAS procedures for completely randomized, randomized complete block, Latin Square, split-plot, or split-split-plot designs. It emphasizes the importance of model specification for downstream statistical analysis at the design and planning phase of an experiment and allows for precise power calculations for main effects, interactions, and contrasts in animal studies. Despite its advantages, pwr4exp in the current version has its limitations, such as the inability to analyze non-normal response variables or specify the correlation structures of repeated measures. This paper reviews the theoretical foundation behind model-oriented power analysis and provides practical guidelines for standardizing power analysis and sample size calculations in R, with examples of various experimental designs relevant in animal research.