What is the difference between unstratified and stratidied binary data? How does the logistic model look like for each of them?
Data stratification is dividing into mutually exclusive, collectively exhaustive subgroups. For binary data, the Wikipedia example covers why it can reduce error in estimating the outcome.
For the logistic model, you should apply conditional logistic regression for the stratified case.