Skip to main content
6 events
when toggle format what by license comment
Nov 24, 2017 at 22:44 comment added whuber +1. This is a good account. Out of an abundance of caution--never to be deplored!--you have enunciated overly restrictive assumptions, but they are accurately and well stated. And your remarks about subtleties and misconceptions are to the point: the choice of tests is never as simple as suggested by the preceding comment by@octern, because the nature of the data is one of the least important factors in the decision. The purpose of testing is the primary desideratum; next is a choice of statistical model for the experiment or data; and everything follows from those.
Jul 23, 2017 at 6:48 review Suggested edits
Jul 23, 2017 at 8:29
Aug 1, 2016 at 23:49 review Suggested edits
Aug 2, 2016 at 6:43
Nov 14, 2013 at 22:57 comment added octern This is correct, though I think the core of it is actually simple. Use chi-square if your predictor and your outcome are both categorical variables (e.g., purple vs. white). Use a t-test if your predictor is categorical and your outcome is continuous (e.g., height, weight, etc). Use correlation or regression if both the predictor and the outcome are continuous.
Nov 14, 2013 at 18:40 vote accept biogirl
Nov 14, 2013 at 18:32 history answered A. Kennard CC BY-SA 3.0