your argument relies on the assumption that data gathered from surveys is accurate and so good that you can make appropriate insights.

For most purposes, I'd say this is good enough, but this is a whole other can of worms.