Timeline for ANOVA (assuming different distributions) vs Kruskal-Wallis and post hoc tests
Current License: CC BY-SA 4.0
18 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 1, 2022 at 18:00 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Mar 2, 2022 at 9:05 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Oct 31, 2021 at 22:03 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Sep 30, 2021 at 12:14 | history | edited | Nick Cox | CC BY-SA 4.0 |
deleted 23 characters in body; edited title
|
| Sep 30, 2021 at 12:03 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| May 27, 2021 at 22:07 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Apr 11, 2021 at 22:10 | comment | added | Alexis | The Kruskal-Wallis test only provides omnibus evidence against equality of means if (a) all groups share the same distribution shape, and (b) all groups share the same variance. The K-W hypothesis generalizes from the rank sum null thus: $\text{H}_{0}\text{: }P(X_i > X_j) = 0.5$ for all $i\ne j$, and where $i$ and $j$ index group. In words the null is the probability that a randomly selected value from group $i$ is greater than a randomly selected value from group $j$ is one half, for all pairs of different groups. | |
| Apr 11, 2021 at 21:58 | answer | added | Frank Harrell | timeline score: 1 | |
| Apr 11, 2021 at 18:06 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Dec 12, 2020 at 14:41 | comment | added | dbwilson | The one option you didn't list is a zero-inflated negative binomial model. Based both on your description and the frequency distribution, this would seem to be a sensible choice. | |
| Dec 12, 2020 at 14:01 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Aug 13, 2020 at 9:06 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Apr 10, 2020 at 16:01 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Nov 29, 2019 at 2:02 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Jul 18, 2019 at 15:03 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Mar 18, 2019 at 13:01 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Feb 2, 2016 at 14:04 | answer | added | Dolf Andringa | timeline score: -1 | |
| Jan 28, 2016 at 8:34 | history | asked | Dolf Andringa | CC BY-SA 3.0 |