Timeline for Are two asymptotic values enough to fail the test of normality?
Current License: CC BY-SA 4.0
12 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 15 at 5:52 | history | edited | Glen_b | CC BY-SA 4.0 |
added 372 characters in body
|
| Jun 4, 2016 at 2:46 | comment | added | Carl | Sometimes outliers are produced because the methodology for processing the data is flawed yielding results that are sporadically nonphysical. In those cases one has not only outliers but a physical reason to eliminate them. It is then tempting to use non-parametric methods, as mentioned above, instead of parametric ones, but non-parametric regression may confuse the reader more than trimming the data, and most people have not heard of non-parametric regression like Theil regression. | |
| Nov 27, 2015 at 3:15 | vote | accept | Antoni Parellada | ||
| Nov 26, 2015 at 17:58 | comment | added | Glen_b | @Nick Certainly. I just wanted to dispel the perception that there was any sort of statistical dogma about removal of outliers, as much as there seems to be a fairly widespread perception that there is. [Some statisticians will say "remove outliers" in some circumstances, but that's not at all the same thing] | |
| Nov 26, 2015 at 15:30 | comment | added | Nick Cox | You don't quite so often see the statisticians who respond saying "get rid of outliers" Agreed, and you'll be happy to affirm that non-statisticians can often give the same advice as statisticians. | |
| Nov 26, 2015 at 12:32 | comment | added | Antoni Parellada | As for the "heavy-tailed" labels (in my post; I don't know if when you are referring to a journal you suspect it's the material from the linked post that comes from a scientific journal) are simply a sort of reading of the shape of the plot. Frankly I was looking at one of your many outstanding posts: here | |
| Nov 26, 2015 at 12:27 | comment | added | Antoni Parellada | re: math definition I was going by the Wikipedia entry: The distribution of a random variable X with distribution function F is said to have a heavy right tail if: $\lim_{x \to \infty} e^{\lambda x}\Pr[X>x] = \infty$ for all $\lambda>0$. which I couldn't track down or explain to myself, but it looks like the result of some sort of transform? | |
| Nov 26, 2015 at 10:35 | history | edited | Glen_b | CC BY-SA 3.0 |
added 31 characters in body
|
| Nov 26, 2015 at 10:30 | history | edited | Glen_b | CC BY-SA 3.0 |
added 31 characters in body
|
| Nov 26, 2015 at 10:23 | history | edited | Glen_b | CC BY-SA 3.0 |
added 249 characters in body
|
| Nov 26, 2015 at 10:18 | history | edited | Glen_b | CC BY-SA 3.0 |
added 249 characters in body
|
| Nov 26, 2015 at 10:10 | history | answered | Glen_b | CC BY-SA 3.0 |