However, when Rank and Jacobson repeated Hofling’s procedure, more realistically, obedience dropped to 11%. Suggesting Milgram’s results do not reflect real situations of obedience. Pps must have guessed the learner wasn’t really receiving electric shocks - high obedience rates would be due to ‘demand characteristics’. However, physical displays of stress indicate pps did believe they were hurting someone. To check this Milgram interviewed pps afterwards - many reported they thought the shocks were real. This suggests Milgram’s study has high internal validity.