Precision and Bias of Various Soybean Dry Matter Sampling Techniques1
- P. G. Hunt,
- K. P. Burnham and
- T. A. Matheny2
A balance between sample sizes sufficient to detect real differences and resources necessary to obtain and process samples is a continual problem for scientists. In the first year, this study assessed the precision and bias associated With three subplot sampling techniques used for estimating soybean [Glycine max (L.) Merr.] shoot dry matter production: (a) random 1 m of row, (b) random 0.3 m of row, and (c) four randomly selected plants. The experimental design was a split-split plot. Three cultivars (Braxton, Coker 338, and Davis) were the whole plots, and three sampling date plots were the subplots. Measurement of the total plant dry matter within the subplots as well as the dry weight obtained by three sampling methods constituted the four split-split plots. When dry matter data were averaged over time and cultivar, the coefficients of variation (CV) for the sampling techniques (a), (b), and (c) were 0.187, 0.417, and 0.386, respectively. The CV was 0.088 when the 20-m2 subplot (27 m of row) was measured. The 1-m technique did not significantly overestimate dry matter production relative to the entire subplot measure, but the 0.3-m and four-plant (4-P) techniques significantly overestimated dry matter production. Cultivar, date of sampling, and cultivar ✕ date-of-sampling interaction effects in the analysis of variance were significant at P ≤ 0.05 only when data from the entire subplot or 1-m sampling technique were used. A second study in 1985 also showed that the 0.3-m and 4-P sample techniques were poor for both precision and accuracy. This second study also compared the precision and bias of 1- and 2-m samplings as well as the precision and bias associated with different technicians. The 1- and 2-m techniques gave equally unbiased results, but the 1-m technique did not give equally good precision (CV). Comparison of techniques was not significantly affected by technicians. We concluded that adequate accuracy and precision were obtained by use of the 1-m or 2-m sampling technique, but not by the 0.3-m or 4-P techniaues.Please view the pdf by using the Full Text (PDF) link under 'View' to the left.
Copyright © . .