August 2009
Intermediate to advanced
648 pages
17h 35m
English
While conditional logistic regression often serves to counterbalance the small counts in a strata by conditioning away the strata effect, sometimes the data are so sparse that these methods also become inappropriate. The conditional exact inference described in Chapter 8 also applies to the stratified setting. Note that in the asymptotic logistic regression setting, the methodology for the unstratified and stratified analysis is different (the former is based on an unconditional likelihood and the latter is based on a conditional likelihood). In the exact setting, you use the same methodology (involving conditioning). The only difference is that, in the unstratified case, ...
Read now
Unlock full access