The aim of this research was to present ways to estimate the coefficient of soil heterogeneity from the intraclass correlation coefficient for different structure of experimental designs. The coefficient of soil heterogeneity estimates obtained by this method were compared with those obtained by the generalized least square method. Practically, there was no difference between the estimates obtained from both methods, and thus it is recommended to use the estimate obtained from the coefficient of intraclass correlation due to its simplicity and practicability. Several plot sizes were obtained from the coefficient of soil heterogeneity estimates, for different combinations of variation coefficient, replicate number, expected difference between treatments and significance level.
intraclass correlation coefficient; experimental precision; experimental design; variation coefficient
