An Investigation of the Validity and Usefulness of Trend Analysis for Field Plot Data

Abstract

Abstract Blocking is sometimes ineffective in accounting for a field's systematic, spatial variation. In such cases, trend analysis has been proposed as an alternative method. This procedure fits a polynomial regression function (surface model) to account for some of the systematic variability. Computer simulated data were used to test the statistical validity of trend analysis. Data simulation facilitated this evaluation, because the experimental conditions of treatment variation, systematic variation, and random variation could be computer generated within prescribed limits. The validity of the significance levels in trend analysis was evaluated. Also, trend analysis was compared to the randomized complete‐block analysis with regard to power of test and estimation of treatment effect. These studies show that with proper restrictions on choosing the surface model, trend analysis has true significance levels close to assumed levels. In addition, trend analysis was often superior to the randomized block analysis in power, and offered less biased estimates of treatment effects, especially when the pattern of systemic variation was complex and curvilinear.