Comparing discovery learning with direct instruction replicating (and extending) the famous Klahr and Nigam result