Combining diverse methods in a single study raises a problem: What should be done when the findings of one method of investigation conflict with those of another? We illustrate this problem using an example in which three study phases--quantitative, qualitative, and intervention--were applied. The findings coming from the quantitative phase did not fit those coming from the qualitative phase; there were discrepancies within the qualitative phase itself, and the findings coming from single-case evaluations of the intervention using standardized scales did not fit the findings derived from self-made scales. We explain these inconsistencies by way of the complementary approach: conflicting findings should be integrated, and consistency is restored by admitting complexity in the phenomenon under investigation. (Contains 3 tables and 4 figures.)