Explaining the optimistic performance evaluation of newly proposed methods: A cross‐design validation experiment