Timestamps sequence

Survey Solutions user Domenico has sent us the following question:

…we asked the enumerators to record the time several times over the course of the enumeration. More specifically, the first one at the beginning of the enumeration and the last one at the end. However, when downloading the data, in several cases the timestamp of the completion of the survey is antecedent to the timestamp of the beginning of the enumeration, which should be impossible.

Suppose the interview consists of two timestamp-type questions A and B. A is answered, then B is answered. The order of timestamps appears to be correct. Now if the interviewer goes back to question A and taps it again, the new timestamp will be recorded. It will have a value later than that in question B, so the sequence will appear to be incorrect.

Domenico is making an assumption that the interviewer will always ask the questions in sequence and never go back and change earlier answers. This is not how most interviewers work in reality, and Survey Solutions will let the interviewer change the answer to any enabled [interviewer-scope] question.

It is not clear from the question why there are multiple timestamps in the interview. This could be a rudimentary approach to calculate the duration of the interview. Instead one finds more insights from paradata.

If the data must exhibit a certain pattern, such as A<B is desirable, then Domenico should add a validation to mark this as an explicit error during data entry.

Hope this helps.

Best, Sergiy

A variation of the same question has been sent to us from Survey Solutions user Eugenia:

When I download the paradata for the surveys, in many cases it happens that the timestamp for some questions “goes back in time”… basically a full day before the “InterviewCreated” event.

For tablet interviews the timestamp may “go back in time” in cases when the clock of the tablet is adjusted. This is because when working disconnected, there is no other source of time information other than the tablet clock setting.