To demonstrate the effects of unbalanced data on analysis of variance, we now consider the true original data set of maize RILs, which includes about 4.5% missing data (MaizeRIL_miss.csv). We will show that the pattern of missing data has important effects on estimation; note that RIL- 5 has data from only two of four locations and RIL-51 has data only from three locations. When we fit the linear model to these data with SAS Proc GLM (“Example2B.sas”), we obtain the following results: