Your instrument or you have a good secondary source that you want to run a couple of different hypothesis tests against, what should you do to guard against making a Type I error?