Sample Size For Estimating Components Of Variance In Maize Experiments In a Tropical Rainforest Ecology.

Sample size determination is the act of choosing the number of observations to include in a statistical sample of empirical studies in which the goal is to have sufficient statistical power for making inferences about a population from the sample. The objective of the study reported herein was to determine the optimum sample size (N) for estimating components of variance in maize (Zea mays L.) yield trials conducted in a tropical rainforest location in Sw Nigeria. In a field study involving 100 maize varieties (G), divided into sample sizes of 10, 20,........, 100 with three replications at two locations (L) in each of two years (Y), data obtained for grain yield were subjected to analysis of variance to obtain variances (𝜎²) and coefficients of variation (CV). Most variance components were relatively overestimated when N<50. Estimates of 𝜎_g², 𝜎_gy², 𝜎_gl², 𝜎_gyl² , and 𝜎_e² showed negative quadratic response as N increased (R²= 0.61 to 0.97), stabilizing at about N=60-70. Contrarily, 𝜎_yl² increased linearly as N increased (R²= 0.92) while CV had a positive curvilinear response to N, reaching a peak of about 46% at N=70 (R²= 0.94). It was concluded that relatively large sample sizes are needed for maize studies from which variance components are to be estimated.