Talk:Statistical hypothesis test

Common test statistics

I corrected the erroneous last test, ("regression t-test") to a correct F-test. Harald Lang, 2015-11-29.

Discuss changing the lede here

Template:Reply to Discuss the lede changes here. You may well have a point, but you removed a reference when you trimmed it to one sentence. Maybe others can weigh in to a discussion and reach a consensus. Geoff | ^{Who, me?} 00:07, 5 March 2022 (UTC)

Here's why the lede should be shortened: (1) The function of a lede is to be a short summary, not to be a sample platter of verbose details on a topic. (2) There's no fundamental reason to be attached to the aforementioned reference. But even if an editor insists on including that reference, that can be done without re-adding a huge amount of text along with it. To say we should include a bunch of text just because we want to include the reference is backwards and nonsensical. We should include a reference when it's necessary to support the text—not include text just to have something to go with a reference! (3) The long lede is full of poor wording with unclear meaning that likely was made up by an editor and not directly derived from the cited reference anyway. For example, consider this phrase: "to determine a possible conclusion from two different, and likely conflicting, hypotheses." Why "likely?" The alternative is—by definition—nonoverlapping with the null (that's why it's "alternative"). And how does one determine a conclusion "from" hypotheses? Did the editor mean "regarding" the hypotheses? There are several other examples of peculiar, confusing wording (e.g., "difference between the sample and the null hypothesis"). Indeed, it seems highly unlikely that anyone looking to understand what statistical hypothesis testing is would read the lede as-is and come away feeling that they gained the understanding they were hoping for. (4) The long lede contains objectively incorrect information. For example, consider this phrase: "Type I error and Type II error which are controlled by the pre-specified significance level." Although the significance level affects the Type II error rate, it does not "control" Type II error. Unlike the Type I error rate, the Type II error rate is controlled by power (which depends on sample size), not by alpha alone. Or consider this statement: "every hypothesis test based on significance can be obtained via a confidence interval, and every confidence interval can be obtained via a hypothesis test based on significance." That is blatantly false; there are many significance test procedures (e.g., permutation-based methods and Simes-based multiple-testing procedures) that have no associated confidence interval procedure, and in any case one cannot infer a confidence interval from a p-value alone. Yet that blatantly false statement is the one that cites the reference that one editor is inexplicably attached to.

In short, an editor implied that we should keep an overly long lede, which is bloated, confusing, poorly written, and in some cases objectively wrong, simply because the last sentence—which happens to be incorrect—cites a reference. I think it's obvious that is not a good reason. 23.242.195.76 (talk) 10:27, 5 March 2022 (UTC)

Wiki Education assignment: Research Process and Methodology - FA23 - Sect 201 - Thu

— Assignment last updated by ISHIKAWAYI (talk) 16:49, 28 October 2023 (UTC)

Poorly organized

I suggest defining the null hypothesis before citing its historical importance.159.83.248.41 (talk) 22:46, 29 August 2024 (UTC)