Modern methods of field trials and data analysis
1994
Lauk, E. (Estonian Agricultural Univ., Tartu (Estonia). Inst. of Field Crop Husbandry)
The most common methods of field trials in Estonia do not assure using regression methods for analysis of results of field trials effectively. Regression analysis requires greater number of treatments and small differences between treatments. For that reason field experiments which have quantitative variants should be designed as unreplication experiments. At the same time the first level (treatment) should be repeated in two plots, repeating other treatments is not necessary but can be done. The results of trials should be corrected before regression analysis is done. The correction of data for the first and last of level takes place by two observed points. For other treatments results should be corrected by three observed points. The correction of results reduces error (standard error of regression, standard error of coefficients) considerably
显示更多 [+] 显示较少 [-]