Timeline for Chart indicates homoscedasticity but Breusch-Pagan test p<.001
Current License: CC BY-SA 4.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 11, 2020 at 14:32 | history | edited | CommunityBot |
Commonmark migration
|
|
| Nov 26, 2019 at 19:01 | comment | added | Nick Cox | I think this nails part of the question nicely, namely why conventional significance is easily attained. | |
| Nov 26, 2019 at 15:07 | history | edited | PsychometStats | CC BY-SA 4.0 |
added 417 characters in body
|
| Nov 26, 2019 at 14:58 | comment | added | PsychometStats | @SalMangiafico I agree exactly. I will further edit the answer | |
| Nov 26, 2019 at 14:55 | comment | added | Sal Mangiafico | I appreciate the edit in the answer, but you might further reconsider the word "spurious".... I guess there are two issues going on in the OP's results: 1) It looks like there is enough heteroscadasticity to be a consideration. 2) With 15,000 data points, the B-P test may be overly sensitive to small deviations from homoscedasticity. As mentioned in this answer, examining model assumptions from visual methods is probably a better approach than relying on statistical tests for e.g. normality and homoscedasticity. | |
| Nov 26, 2019 at 14:42 | history | edited | PsychometStats | CC BY-SA 4.0 |
added 41 characters in body
|
| Nov 26, 2019 at 14:41 | comment | added | PsychometStats | @SalMangiafico quite the opposite. The plot appears to be heteroskedastic | |
| Nov 26, 2019 at 14:37 | comment | added | Sal Mangiafico | Are you agreeing with the OP that the plot of residuals suggest homoscedasticity? I'm not familiar with the idea that studentized residuals vs. standardized predicted values should be triangular... | |
| Nov 26, 2019 at 10:01 | history | answered | PsychometStats | CC BY-SA 4.0 |