Timeline for Why is a "Correction" Required in Multiple Hypothesis Testing?
Current License: CC BY-SA 4.0
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 20, 2022 at 13:33 | comment | added | usul | Have you seen this xkcd comic? "Significant" xkcd.com/882 | |
| Jul 20, 2022 at 13:22 | history | became hot network question | |||
| Jul 20, 2022 at 12:00 | history | tweeted | twitter.com/StackStats/status/1549725833815887875 | ||
| Jul 20, 2022 at 9:47 | answer | added | Björn | timeline score: 8 | |
| Jul 20, 2022 at 8:03 | answer | added | Łukasz Deryło | timeline score: 5 | |
| Jul 20, 2022 at 6:28 | answer | added | Stefan | timeline score: 8 | |
| Jul 20, 2022 at 6:11 | comment | added | AdamO | Multiple regression is not well defined, but (IMO) is usually taken to mean one regression model with multiple adjustment variables. However, analyses somewhat like what you depict above are routinely done: usually you see some of X4 or X5 added or dropped or combined or... anyway, the goal might be modeling the x1 association with y. By fitting many models, you increase the risk that one is spuriously identified as significant, and you report the overall x1 and y association as positive. In that case the familywise error rate needs to be controlled to preserve the true alpha level. | |
| Jul 20, 2022 at 5:57 | comment | added | mkt | Thanks, that helps to understand this. Your question seems answerable but in the example you use, there are likely better ways to model the data (you could collapse all of these into one regression, for example). If the example is only intended to illustrate your point about multiple comparisons, that can probably be ignored. | |
| Jul 20, 2022 at 5:45 | comment | added | stats_noob | @ mkt: for your second comment - yes, that is correct! | |
| Jul 20, 2022 at 5:43 | history | edited | stats_noob | CC BY-SA 4.0 |
added 156 characters in body
|
| Jul 20, 2022 at 5:42 | comment | added | mkt | Specifically, does each id correspond to a different individual, and does each y value in the same row represent a different condition/outcome for the same individual? | |
| Jul 20, 2022 at 5:40 | comment | added | mkt | What precisely do each of the x and y values represent, in your data? | |
| Jul 20, 2022 at 5:38 | history | edited | stats_noob | CC BY-SA 4.0 |
added 227 characters in body
|
| Jul 20, 2022 at 5:13 | history | asked | stats_noob | CC BY-SA 4.0 |