Timeline for Is it possible that X and Y are uncorrelated but X can significantly predict Y? [duplicate]
Current License: CC BY-SA 4.0
19 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 19, 2022 at 17:31 | history | closed |
Sextus Empiricus Adrian Keister mkt |
Duplicate of Why zero correlation does not necessarily imply independence | |
| Jul 19, 2022 at 16:08 | answer | added | Dave | timeline score: 2 | |
| Jul 19, 2022 at 10:07 | comment | added | Dave | @EmmaN. When you run the “ON” regression, do you include predictor variables other than X? | |
| Jul 19, 2022 at 6:55 | comment | added | Glen_b | More details might help to unpick what happened | |
| Jul 19, 2022 at 2:00 | comment | added | Emma N. | @Glen_b I'm reworking on the code. I might have made a mistake in running the analysis if theoretically speaking, the hypothesized scenario would never happen. | |
| Jul 18, 2022 at 22:25 | comment | added | Robbie Goodwin | Can you give any examples either of general research or specific examples you've considered? | |
| Jul 18, 2022 at 0:55 | comment | added | Glen_b | @EmmaN. What you appear to describe seems unlikely -- both a test of Pearson correlation and a simple linear regression should have the exact same t-value and the exact same p-value; can you give some more details about what exactly you did? (I'm not specifically asking for code - though ultimately that may be where the issue is - but a clear explanation of exactly what you attempted, perhaps you could add a few sentences to your question? It may be that you didn't quite do what you thought you did.) .... how many observations were there? | |
| Jul 17, 2022 at 20:26 | answer | added | statmerkur | timeline score: 17 | |
| Jul 17, 2022 at 19:36 | review | Close votes | |||
| Jul 19, 2022 at 17:36 | |||||
| Jul 17, 2022 at 15:34 | history | edited | kjetil b halvorsen♦ | CC BY-SA 4.0 |
deleted 20 characters in body; edited tags
|
| Jul 17, 2022 at 3:06 | history | became hot network question | |||
| Jul 16, 2022 at 21:00 | history | tweeted | twitter.com/StackStats/status/1548412171431788547 | ||
| Jul 16, 2022 at 20:21 | comment | added | whuber♦ | Let $X$ be symmetrically distributed around $0$ and let $Y=X^2.$ The latter is perfectly predictable from $X$ but the correlation is zero. | |
| Jul 16, 2022 at 19:53 | answer | added | Tim | timeline score: 14 | |
| Jul 16, 2022 at 19:40 | comment | added | Emma N. | Yes. To be more specific, I used the WITH function in Mplus to run the correlation analysis between X and Y first, which shows no significant correlation. However, when I used ON function (Y as the outcome, X as the predictor), the p-value becomes significant. I didn't add other variables to it. | |
| Jul 16, 2022 at 19:14 | comment | added | Dave | Welcome to Cross Validated! What exactly do you mean? For instance, do you mean if $X$ and $Y$ can be uncorrelated yet $X$ can have a significant parameter in an OLS regression? | |
| Jul 16, 2022 at 19:13 | answer | added | Georg M. Goerg | timeline score: 23 | |
| S Jul 16, 2022 at 19:06 | review | First questions | |||
| Jul 16, 2022 at 19:27 | |||||
| S Jul 16, 2022 at 19:06 | history | asked | Emma N. | CC BY-SA 4.0 |