I consider a proportional odds (ordinal regression) model, as described in Agresti (2002) and estimated by R base package function MASS. The software reports t-values but no p-values in summary. I wonder how I should determine the degrees of freedom for the t-test. In simple linear regression $df=n-2$ where $n$ the sample size. Is this also the case for ordinal linear regression?
Asked
Active
Viewed 405 times
1
tomka
- 5,874
- 3
- 30
- 71
-
Why not extract the residual df from the object? And I think polr is in MASS not base. – mdewey Aug 25 '16 at 14:39
-
Are you sure residual df is the correct df for the t-test? – tomka Aug 25 '16 at 15:13