Talk:Test–retest

Quantify?

What is the best way to quantify test-retest reliability? correlation coefficient? something else? —Preceding unsigned comment added by 128.112.151.175 (talk) 18:54, 18 February 2009 (UTC)

Merge

Test-retest and repeatability seem to essentially cover the same topic.Mikael Häggström (talk) 06:31, 8 December 2009 (UTC)

I believe it is important to note that "repeatability" is a defined term. It belongs neither to Test-retest or to Scientific Method (where it is definitionally related but not shown now), but to both. How do we handle providing definitions in Wikipedia? --Liftoph (talk) 16:01, 18 January 2010 (UTC)

On second thought, I think they should remain separate. Test-retest can also be interpreted as a repeated test on an individual as part of disease monitoring. Mikael Häggström (talk) 15:37, 10 October 2011 (UTC)

I merged the text in this article to Test-retest reliability instead. However, I didn't move the following text, because I think it needs a reference first: In some cases it is important to pick the right interval between tests. For example, the Center for Epidemiological Studies Depression Scale (CES-D) asks the participant about symptoms of depression over the past two weeks. If you administered the test on two occasions four months apart, you might expect only moderate test-retest reliability; depression symptoms can easily change over that time. But if you administered it on two consecutive days, test-retest reliability should be high, both because depression symptoms usually do not change that quickly, and because the participant was asked to consider nearly the same two-week period on both occasions. Mikael Häggström (talk) 16:38, 10 October 2011 (UTC)