We draw conclusions from sample groups without considering whether the sample captures what we should be looking at.