His own work held up very well to replication. It's when he is citing the work of other scholars (in particular, that of social psychologists) that doesn't hold up well to replication.