Timeline for 1-way ANOVA contradicts boxplot?
Current License: CC BY-SA 4.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 28, 2023 at 13:09 | comment | added | whuber♦ | You can't use SDs to assess the significance of differences, either, because they do not account for the sample sizes. Since a CI width typically is proportional (at least approximately) to the SD divided by the root group size, in the boxplots you have added to this post, where there is no overlap among any of the middles (the "boxes"), it's immediately clear that all pairwise differences are significant provided the group sizes are larger than 5 or so. | |
| Jun 28, 2023 at 1:49 | history | edited | FlyingDutch | CC BY-SA 4.0 |
added 1 character in body
|
| Jun 28, 2023 at 1:48 | comment | added | FlyingDutch | @whuber Sorry I meant standard deviations. | |
| Jun 27, 2023 at 20:31 | history | edited | kjetil b halvorsen♦ | CC BY-SA 4.0 |
deleted 10 characters in body; edited tags
|
| Jun 27, 2023 at 20:10 | answer | added | Ute | timeline score: 4 | |
| Jun 27, 2023 at 19:58 | comment | added | whuber♦ | There might be some misconception about boxplots implicit here. (1) Boxplots usually don't display CIs (and when they do it's an additional graphical feature shown as a "notch"). (2) Overlaps of CIs are not appropriate statistical tests unless properly adjusted. | |
| Jun 27, 2023 at 19:43 | history | asked | FlyingDutch | CC BY-SA 4.0 |