What I would do in that situation is: 0. I would NEVER work on a data entry system and rely on data in one place or on one computer only.
1. Make sure to have back-up of all files including "corrupt ones".
2. Make a challenge test with the software. E.g. for entry and .not do as described by Jamie Hockin in a previous mail.
3. If you experience problems after 40 observations, then copy the .not file to some other location and rename, e.g. notes1-40.not notes41-79.not etc
4. For finding any backup copy or left over copy of a "pre-corrupt" file I would search my entire harddisk, including hidden files, for a text string that I knew was in the "pre-corrupt" file. E.g. parts of your observation 1 text. Several software tools will do this e.g. the crimson editor or total commander.
5. To recover the text from "pre-corrupt" or "corrupt" .not file I would use an editor such as "Crimson editor" or "pfe" or similar ascii oriented file. Then you would almost certainly towards the end of the file, find the cause of the problem - be that size of the .not file or some particular control sign.
regards Jens Lauritsen EpiData Association On 2010-08-16, Suzanna wrote:
We set up an EpiData database with a notes file<filename>.not. The notes file appears to have corrupted and has taken the qualitative data from over 40 surveys with it. I was wondering if there was any way I could recover this or if I'll need to re-enter all the data manually into a word document.
Do you know of some recovery technique for this kind of problem?