I am wondering why NaN errors would be appearing in 2nd level analysis when
there were no errors during 1st level analysis and the 1st level results
look as expected. This is, for now, a small group of subjects performed a
number of different tasks. For each task, we keep on receiving the same
errors. We have tried different combinations of subjects to try to determine
whether there is a corrupt data set, but this is not remedying the problem.