- validity might still be questioned as obedience didn't fall that much. The study was still clearly in a lab
- by taking ppts into 'real world' this should measure 'real' obedience, but the controls in the study (generator, verbal prods) are all likely to show ppts the task is far from real, which questions task validity
- 19 ppts obeyed in the office and 26 at Yale, which Milgram claimed as not that different. However, there's still less obedience, some people think its enough to claim using Yale meant the findings lacked validity