Excuse me for being pedantic, but a CRC is capable of correcting a specific number of errors on its own. That's why its called a Cyclic REDUNDANCY Check - it actually carries limited corrective information. The number of errors it can accommodate is proportional to the length of the CRC (can't remember my college lectures that well - it's too long ago!). We are talking CHECKSUMs here - good/not good type checks. Might be worth looking at CRC theory for the smallish data tables under discussion. It has to be more economic than duplication or triplication of data...
Dave