I think a good rule of thumb is to spend at least as much time thinking about your data as you did obtaining it. It's common for people to spend a lot less time thinking than doing which gives rise to all kinds of problems.