Sample size to estimate the average of variables agronomic in cassava

Abstract

In agricultural experiments, measuring all plants that constitute an experimental unit is the most appropriate method for estimating averages. This study analyzed sample sizes (number of plants) for estimating averages of agronomic traits in cassava. The experiment for seedling production was conducted on four sowing dates: September 22, October 7, October 19, and November 25, 2017. The evaluated characters were plant height at planting, number of leaves at planting, number of leaves at 7 days after planting, number of leaves at transplanting, and plant height at transplanting. The seedlings were transplanted to the field on November 9, November 24, December 3, and December 26. Root characteristics and root yield were measured at harvest. The following statistics were measured: minimum, maximum, range, mean, median, variance, standard deviation, and coefficient of variation. The sample size was determined by resampling with replacement when the range of the 95% confidence interval (CI) was 10%, 20%, 30%, and 40% of the average estimate. Larger sample sizes were needed to estimate trait averages with a 95% CI range of 10% of the average estimate, i.e., when estimation accuracy was higher. In contrast, lower sample sizes were required to estimate averages of all characters in all planting times with a 95% CI range of 40%. For measuring seedling production and stem and root yield, 63 and 153 plants are sufficient to estimate averages when the range of the 95% confidence interval was 20%.