Author(s): MAIA, A. de H. N.; DEL PONTE, E. M.; GODOY, C. V.

Summary: This work aimed to conduct an exploratory, prior to meta-analysis, assessment of the heterogeneity of the linear relationship between soybean yield (Y) and the severity (X) of a foliar fungal disease (soybean rust) from 144 fungicide trials conducted in five years and 45 locations in Brazil. We used graphical analysis to explore the within- and between-trial variability of model parameter estimates (effect sizes) and summary measures for model goodness-of-fit and influence analysis. In the estimation of the weighted mean regression coefficients, following the metaanalysis concept, we propose a method that takes into account the magnitude of influence measures as well as the uncertainties of parameters estimates, usually the inverse of the standard error of for the parameter estimate. High between-trial heterogeneity was observed for parameter estimates, standard errors (within trial variability), determination coefficient and influence measures. Graphical analysis suggested influence of timing of disease onset and control yield class on model slope, which will be quantified further using a meta-analytical approach.

Publication year: 2011

Types of publication: Paper in annals and proceedings

Unit: Embrapa Environment

Keywords: Análise estatística, Between-trials heterogeneity, Doença de planta, Ferrugem, Ferrugem asiática, Influence analysis, Linear model, Regressão linear, Soja