Models for estimating plot size in experiments

Abstract

The use of statistical models in the estimation of experimental plot size is a practice that contributes to the experimental planning by choosing a size that allows better efficiency in the comparison of treatments. This work aimed to present some statistical models, with the purpose of using them as alternatives in determining the optimum plot size in experiments. Some nonlinear models with a simple configuration similar to that proposed by the modified maximum curvature technique were proposed, which are derivable and have a curvature function. The curvature function was obtained for each model and plot size estimators were obtained through the critical point of the curvature function derivative. The models were shown to be feasible for estimating plot sizes with simpler estimators compared to those obtained by Meier and Lessman.As an illustration, data from two uniformity tests were used for the application of the proposal and comparison with the modified method of maximum curvature. Estimates of the optimum plot size varied according to model and method