01-21-2015 06:01 PM
I need your input setting up a linear mixed model. I have corn yield data collected from several fields (over 2000) across three different regions and over multiple years (2004-2013). All the fields did not report data for each single year. I also have information about the previous crop. My interest is to find variability in corn yield in each region separately due to year effect, field effect, and variability coming from the previous crop. I would also be interested to test these effects. I am not sure if fields and years would be considered nested and how to setup the model. Kindly help on it.
As per my current understanding the following model will fit to the problem, however I am not sure. When I ran this model I found that the test of previous crop came significant even if the estimates were so close for previous crop corn and soybeans.
Proc glimmix data = Master namelen=200 plots=residualpanel (conditional marginal) noclprint;
class yeaR1 Legals PREV_CROP_Yr1;
model ACTUAL_YIELD = PREV_CROP_Yr1/ddfm=kr ;
NLOPTIONS TECH=NRRIDG GCONV=0;
Random yeaR1 Legals;