We study unrepresentative observational data in survival analysis. The first paper focuses on proportional hazards regressionwhen observed subjects have different selection probabilities. Wedevelop methods which are applicable when the selectionprobabilities are unknown but estimated using...