Timeline for why do we need to sometime jitter the data when using correlation (Spearman) in R
Current License: CC BY-SA 4.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 11, 2020 at 12:26 | answer | added | alan ocallaghan | timeline score: 2 | |
| Feb 11, 2020 at 10:13 | comment | added | Glen_b | Try calculating a p-value several times on the sane data using jittering each time. What do you notice? ... If you do it a large number of times, what do you notice about the average correlation after jittering? | |
| Feb 11, 2020 at 4:26 | comment | added | jbowman |
The warning is caused by the fact that you have ties in your data and Spearman is a correlation based on the ranks of the data; the ties imply there is not a unique ordering, and therefore ranking, of the data, among other things. cor.test.default does not calculate an exact p-value in this case.
|
|
| Feb 11, 2020 at 1:01 | comment | added | alan ocallaghan |
Who suggested to jitter your data? What justification did they give? I would just use exact = FALSE or instead use Kendall's tau (method = "kendall").
|
|
| Feb 11, 2020 at 0:44 | history | asked | user3639557 | CC BY-SA 4.0 |