The title of this post actually says what I want to ask about.
To detect heteroksedasticity for an ordered logit regression using the polr function in the MASS package in R can the standard error from the regression summary then be used to check if there exist heteroskedasticity?
If not then how can there be tested for heteroskedasticity for an ordered logit regression?
Asked
Active
Viewed 51 times
0
rr19
- 65
-
1Isn't the SE a number? How would that indicate heteroscedasticity? In a logit regression "heteroscedasticity" wouldn't seem to be an applicable concept. Maybe you're trying to test for goodness of fit? – whuber Dec 07 '22 at 22:23
-
@whuber I should mentioned this also in the post: I just read that the variance for an ordered logit is assumed to be pi^2/3.. so I thought that if it could be possible to use the s.e. to determine the variance to see if they are almost equal to this value.. if they may vary then there may be heteroskedasticity.... maybe I misunderstooth everything.. I just got very confused.. maybe heteroskedasticity is not relevant for an ordered logit regression? :/ – rr19 Dec 07 '22 at 22:30
-
@whuber so you don't have an answer to my question? – rr19 Dec 29 '22 at 20:37
-
I can't determine what the question might be, sorry. – whuber Dec 29 '22 at 20:41