Take one group of poor African-Americans and another similar group as a control. Do a study on effects of crack, come up (almost) empty. Kudos for publishing negative results.
Then try to leverage your results by comparing them to a non-control group and drawing conclusions based on ideology. Null hypothesis fail.
If the author wanted to test hypotheses specifically on the effects of poverty, she would have needed to start with a group that was more well-off but otherwise similar. Not just ‘everyone else’.
