Which classic social psychology findings survived being repeated?

It depends on the finding, and "survived" can mean different things. Start with the baseline that the replication-crisis lesson described: in the 2015 Reproducibility Project, social psychology fared worse than cognitive psychology, with 25 percent of its studies replicating against 50 percent. Finding by finding, the famous results of this path fall into four bins.

Held, with a catch. Milgram's obedience study was rerun in 2009 with ethical changes, stopping at 150 volts, and found obedience rates virtually identical to the original. The bystander experiments produced one of the strongest and most replicable effects in social psychology, yet real surveillance videos of public conflicts show that in over 90 percent of them at least one bystander stepped in. Asch-style conformity keeps appearing, but its level varies with culture and has declined over time in the United States.

Held. A 2009 brain-imaging study replicated the basic findings behind cognitive dissonance, and a 2020 study replicated Cialdini's door-in-the-face experiment.

Did not hold. Two direct replications in the early 2010s failed to reproduce Bargh's "elderly-walking" priming study, and in 2022 Daniel Kahneman called behavioural priming research "effectively dead". Large multi-lab replications in 2016 and after found no ego depletion. Power posing kept its effect on felt power but not on hormones or behaviour.

Disputed. The Stanford prison experiment's story about roles was challenged by the 2002 BBC Prison Study, though in 2018 Zimbardo, Reicher and Haslam jointly called both experiments valid. Groupthink has only partial support.