It could be correct, but that would depend on how much data you have. Typically I would create an index for the columns I was using in a GROUP BY, but in your case the optimizer may have decided that after using the field_two index that there wouldn’t be enough data returned to justify using the other index for the GROUP BY.