Comparison between methods of spatial analysis of field experiments

Abstract

A study on three methods of spatial analysis of field experirnents, was carried out with a very high number of varieties. The methods were: Papadakis method, moving average, and moving blocks. Comparison among the methods used data from a simulated uniform trial with known parameters; linear and cyclic trends were imposed to the data and a set of 600 different experimcnts was analysed by alI methods, including de usual analysis of variance. Through the average values of some relevant statistics the behavior of the methods was compared. The results indicated the Papadakis method as the most efftcient in the reduction of experimental error in the presence of cyclic and strong linear trends. In other situations the Papadakis method showed at least an efficiency similar to the other methods. In the presence of cycie and weak linear trends the moving average and moving blocks methods yielded very low estimates of the experimental error.