Milgram's obedience study is high in internal validity (the setup controlled variables well) but criticised for low ecological validity, as few real-world situations involve a lab authority figure ordering shocks.