You seem to have good control of data entry. I think the safest way is to use Analysis to check for records that do not match.

However, I wonder if you could populate your follow up file with JUST the fields used to create the ID field and the ID field, leaving all other fields blank. Everyone doing data entry would have this file.

Then use AUTOSEARCH to find the matching record. If no match is found, data entry proceeds as for a new individual. This would be especially useful if the forms used indicated when an individual was new, so that the person doing data entry would know when to expect no match.

I have never done double entry with AUTOSEARCH.

You will have only a bit of work to do to eliminate all of the duplicate records. This should be easy because all of the other fields will be missing. (MERGE might still work as long as you keep the non-blank fields).

Jamie

Emilie wrote:

I have a question about linking the follow-up round of a survey to the baseline which was data entered using EpiData. The follow-up is a new database that will be linked to the baseline during analysis. During data entry, I want to make sure that the IDs entered in the follow-up match those from the baseline in the household roster of the survey. No member that was in the household in the baseline should be missing, however new members can be added to the household. I have villageid, householdid, and personid, each of which is entered separately. I can then create a unique id which combines these to ID the unique individual.

The baseline data was entered using double data entry where the data was compared dynamically (Menu Option: "Double Entry verification") and the follow-up will be done in the same manner. The survey is large and the data are entered on several different computers that are not networked.

The only way I can think of to link the follow-up to the baseline in order to compare the IDs would be to do a mini double data entry after the first data entry of the household roster where I check only the IDs for the data and see what is in the follow-up that isn't in the baseline, but this seems like a poor solution. I would prefer to do something dynamic, which 1) doesn't allow the data entry clerk to move on to another household until they have entered all of the IDs from the baseline and 2) flashes a warning if there is an ID entered that does not match with the baseline. I don't think that EpiData has these capabilities.