Background

Datasets are not editable so if a dataset has invalid information in it, the only way to address the issue is to import a valid dataset and recode it.

Solution

1. Correct the dataset content before importing into NVivo.
2. Import the corrected dataset into NVivo
3. Keep the coded dataset in the project file so it is available for the next step.
4. Autocode the new dataset using "Autocoding Based on Existing Coding Patterns" 
   1. This method may not result in an exact match to the coding in the original (invalidated) dataset but it will potentially reduce the amount of re-coding needed.
   2. The more coding completed in the original file, the better performance you will get with the autocoding feature.