equiBSPD package: a computational edge for efficient equivalent estimation balanced split-plot designs

Agricultural, post-harvest, processing, engineering, and industrial experiments often involve factors that are hard to change, either due to high cost or the time required for adjustment. Such factors make complete randomization impractical. A cost-effective alternative in these situations is the split-plot design, where hard-to-change factors are assigned to main plots (whole plots), while easy-to-change factors are randomized within subplots. Traditionally, the analysis of split-plot designs requires generalized least squares (GLS) estimation. However, for certain designs, the ordinary least squares method produces estimates identical to GLS. These are known as equivalent-estimation split-plot designs. Their advantage lies in the fact that effect estimates are independent of variance component estimates, simplifying both computation and interpretation. When the number of subplots within each whole plot is equal, such designs are referred to as equivalent-estimation balanced split-plot designs. The present work introduces an online solution in the form of the R package “equiBSPD”, which enables researchers to generate optimal equivalent-estimation balanced split-plot designs tailored to diverse experimental setups. In addition, the package integrates tools to examine the relationship between equivalent estimation and statistical optimality such as trend robustness, making it a practical, user-friendly platform for designing efficient experiments across disciplines.