F: the variance–covariance matrix gives variances
and covariances for regression coefficients, not
variables.
T
Because matching has been used, the method of
estimation should beconditionalML estimation.
The variables AGE and SOCIOECONOMIC STATUS
do not appear in the printout because these variables
have been matched on, and the corresponding
parameters are nuisance parameters that are not
estimated using a conditional ML program.
The OR is computed as e to the power 0.39447, which
equals 1.48. This is the odds ratio for the effect of pill
use adjusted for the four other variables in the model.
This odds ratio says that pill users are 1.48 times as
likely as nonusers to get cervical cancer after adjusting
for the four other variables.
The OR given by e to0.24411, which is 0.783, is the
odds ratio for the effect of vitamin C use adjusted for
the effects of the other four variables in the model.
This odds ratio says that vitamin C is some‐what
protective for developing cervical cancer. In
particular, since 1/0.78 equals 1.28, this OR says that
vitamin Cnonusersare 1.28 times more likely to
develop cervical cancer thanusers, adjusted for the
other variables.
Alternative null hypotheses:
The OR for the effect of VITC adjusted for the other
four variables equals 1.
The coefficient of the VITC variable in the fitted
logistic model equals 0.
The 95% CI for the effect of VITC adjusted for the
other four variables is given by the limits 0.5924 and
1.0359.
TheZstatistic is given byZ¼0.24411/
0.14254¼1.71
The value of MAX LOGLIKELIHOOD is the
logarithm of the maximized likelihood obtained for
the fitted logistic model. This value is used as part
of a likelihood ratio test statistic involving this
model.
