A number of recent articles report that analyses which account for spatial field trials in terms of correlations between pilot errors are more efficient than the classical randomized blocks analysis of variance. In most cases, these efficiency comparisons are in terms of model-based or "predicted" estimates of precision. However, the validity of estimates of precision has not been generally deomonstrated for these correlated errors (CE) analyses. We describe a simlulation study to assess validity (as well as efficiency) of several CE and alternative fixed effects spatial analyses. We focus on situations typical of large field trials with limited replication and realistic levels of both fixed and random components of spatial variation. Results show that when spatial autocorrelation is present, the CE analyses are robust with respect to validity, except when strong fixed trend is underfitted in the analysis. Also, when spatial autocorreltaion is present, efficiency of the CE analyses is substantially greater than that for the classical blocks analysis, fixed effects trends analysis, and row-column analysis, but is only slightly better than an uniterated Papadakis analysis. These results are illustrated using a corn yield dataset that is also used to demonstrate a graphical technique for assessing adequacy of a CE model.

Key Words
Autocorrelated errors; Papadakis method; Trend analysis.

Cavell Brownie is Professor, and Marcia L. Gumpertz is Associate Professor, Department of Statistics, North Carolina State University, Box 8203, Raleigh, NC 27695-8203.