Effect of missing values in multi‐environmental trials on variance component estimates

A common task in the analysis of multi-environmental trials (MET) by linear mixed models (LMM) is estimation variance components (VCs). Most often, MET data are imbalanced (e.g., due to selection). The imbalance mechanism can be missing completely at random (MCAR), (MAR), or not random. If missing-data pattern was caused selection, it usually MAR. In this case, likelihood-based methods preferred for as they account a MAR pattern. Likelihood-based used estimate VCs LMM have property that all VC estimates constrained non-negative; thus, estimators generally biased. Therefore, there two potential causes bias analysis: being MCAR, and estimators. current study tries dissect quantify both possible sources bias. simulation with typical cultivar evaluation conducted which size were varied. results showed simulated MET, similar under MCAR Thus, solely estimation. Bias increases when increasing ratio genotype error small. patterns. may concluded selection does increase