Right on both points, correlation and observational studies. They are as reliable as self-reports, but if that's all we have, it's a starting point and we use that to rework hypotheses.