How can I get the log odds associated with the levels in strata()
within a clogit() model? I'm running R-1.9.0 on a Linux platform.
I am using clogit() to run a Rasch model in Item Response Theory in
psychometrics. Symbolically, the model is:
logit(p_{j,k}) = \log({\Pr(p_{j,k}) \over \Pr(1-p_{j,k})}) \theta_j -
\alpha_k,
That is, the log odds of answering an test item correctly is a function
of the item difficulty (\theta_j) and the ability of individual
students (\alpha_k). The following R syntax gives me the \theta
coefficients associated with items i5 to i17, but summary() does not
automatically summarize the log odds associated with the strata "id".
I suppose the log odds associated with strata(id) gives me \alpha_k.
But I can't seem to get them with summary(). Help is greatly
appreciated.
Yuelin Li.
---- my R syntax is -----------
> exam2.clog<- clogit(resp ~ i5+i6+i7+i8+i9+i10+i11+i12+i13+i14+
i15+i16+i17+ strata(id), data=exam1.1)
> summary(exam2.clog)
Call:
clogit(resp ~ i5 + i6 + i7 + i8 + i9 + i10 + i11 + i12 + i13 +
i14 + i15 + i16 + i17 + strata(id), data = exam1.1)
n= 476
coef exp(coef) se(coef) z p
i5 0.514 1.67 1.025 0.501 6.2e-01
i6 0.923 2.52 0.991 0.931 3.5e-01
i7 0.514 1.67 1.025 0.501 6.2e-01
i8 1.875 6.52 0.955 1.964 5.0e-02
[snip]
exp(coef) exp(-coef) lower .95 upper .95
i5 1.67 0.597954 0.224 12.5
i6 2.52 0.397477 0.361 17.6
Rsquare= 0.525 (max possible= 0.659 )
Likelihood ratio test= 355 on 13 df, p=0
Wald test = 108 on 13 df, p=0
Score (logrank) test = 277 on 13 df, p=0
