07-30-2014 05:48 PM
It looks like you may be running a Chi-square test of some flavor. Basic requirement is no empty cells. You'll also get a warning if fewer than 5 responses in a cell.
Since BMI, if body mass index, usually is more of a continuous variable unless you have a very large sample you likely have a number of cells with 0 counts for one or more values of your pm10q variable. You might reduce the issue by using a custom format to associate bmi with specific ranges to reduce the cell count.
What are actually attempting to accomplish? You may need another procedure/model than freq.