Even after error correction has been applied, some incompleteness and some errors in data are likely to remain. This incompleteness and these errors must be managed during data analysis. Doing this correctly is a challenge. Re-cent work on managing and querying probabilistic and conflicting data suggests one way to make progress.