Timeline for How to measure confidence in non-binary (e.g. ordinal) choice tasks?
Current License: CC BY-SA 4.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 6, 2021 at 18:28 | history | edited | Arnon Weinberg♦ |
edited tags
|
|
| Mar 21, 2021 at 9:37 | vote | accept | Shelly C. | ||
| Mar 21, 2021 at 5:12 | answer | added | Arnon Weinberg♦ | timeline score: 0 | |
| Mar 21, 2021 at 1:12 | comment | added | Shelly C. | Thank you, Bryan. It’s really an insightful example. But I’m still wondering if people are good at probabilities (Bayesian), an ordinary single-step way for measuring confidence (as probability of choice among ABCD being correct) and a hierarchical way, e.g. first, prob. of AB being correct (e.g. 90% for sure), and then prob of A being correct between A and B (e.g. 50% for sure) make a big difference: i.e. the former may directly yield 45 %, and the latter does 90% x 50% = 45%. It would be appreciated if you or someone else could cite some theories or previous studies if any. Best, | |
| Mar 20, 2021 at 23:44 | comment | added | Bryan Krause♦ | Confidence gets complicated when you have more than two choices. If you have options ABCD you may be very confident it is either A or B but not actually certain that A is the correct answer. In this case you have a mixture of confidences at different levels of comparison. Two-alternative tasks are easier to analyze for many reasons and often multiple alternative tasks can be distilled to two choice tests anyways. | |
| Mar 20, 2021 at 11:52 | review | First posts | |||
| Mar 23, 2021 at 9:03 | |||||
| Mar 20, 2021 at 11:49 | history | asked | Shelly C. | CC BY-SA 4.0 |