Abstract

A study was conducted to determine the size and shape of the experimental unit for rice yield trials in the canton of Santa Cruz, Guanacaste, Costa Rica. A uniformity trial was planted and the data obtained were used to apply multiple regression method. In this method, the residual variability, measured as a coefficient of variation, corresponding to the different sizes and forms of experimental unit that can be simulated, was modeled according to the length and width by a multiple regression. Then the partial derivatives for each regressor were estimated and equaled to -1; the result of this system of equations finds the combination of width and length of the experimental unit that minimizes the coefficient of variation. For this work the resulting combination was 8 meters long by 4 meters wide for an experimental unit of 32 m2 .

Key words: plot size; plot shape; corn; multiple regression; yield