Hi,
See the extract below from an exact conditional logistic regression
analysis using SAS PROC LOGISTIC.
Can anyone explain to me the differences in the p-values between the
panel "Conditional Exact Tests for 'Exact1'"
and the paner "Exact Parameter Estimates for 'Exact1'"? I.e., why is the
p-value for smok equal to 0.3400 in the former
panel and 0.4000 in the latter panel, and the p-value for ui equal to
0.5385 in the former panel and 0.7692 in the latter panel?
Thanks,
David
The SAS
System 8
The LOGISTIC Procedure
Exact Conditional Analysis
Conditional Exact Tests for 'Exact1'
--- p-Value ---
Effect Test Statistic Exact Mid
Joint Score 4.0003 0.2618 0.2617
Probability 0.000206 0.3046 0.3045
lwt Score 1.7380 0.1984 0.1964
Probability 0.00391 0.2622 0.2602
smok Score 1.9128 0.3400 0.2500
Probability 0.1800 0.3400 0.2500
ui Score 0.9186 0.5385 0.3654
Probability 0.3462 0.5385 0.3654
Exact Parameter Estimates for 'Exact1'
95% Confidence
Parameter Estimate Limits p-Value
lwt -0.0332 -0.0966 0.0137 0.1988
smok 1.5962 -1.0762 5.8661 0.4000
ui 1.3241 -2.0857 5.8646 0.7692
Exact Odds Ratios for 'Exact1'
95% Confidence
Parameter Estimate Limits p-Value
lwt 0.967 0.908 1.014 0.1988
smok 4.934 0.341 352.884 0.4000
ui 3.759 0.124 352.330 0.7692
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|