Hi,

I usually turn on timings in surveys, so I can check for speedsters and also because I need to know if the response was created in an exact time period (e.g. in August 2021).
I recently discovered an inconsistency I can't explain.
I substract the start date from the submit date and that way I should be able to get the time it took to complete the survey.
But LimeSurvey also provides timing data for every individual question (I display one question per page) and also for the whole survey. If I add the times of the questions it equals the time for the whole survey so that seems okay. However it does not match the time between the starting date and submit date, and I'm not sure how can that be. I might understand if the difference between beginning and finish would be more than the sum of individual questions (e.g the timer does not consider loading time or something like that) but in some cases the survey time (the sum of time spent on individual questions) is larger than the time difference between starting date and submit date and I don't understand how is that possible.

E.g.
Startdate: 2021-08-23 10:04:25
Datestamp/Submitdate (they are identical): 2021-08-23 10:07:01
So there is 156 seconds between the start and finish.
Yet, according to LS, the survey time is 168,46 seconds which does not seem possible.


I'd really appreciate if anyone has an explanation for this. I'm trying to figure out whether it's a problem/bug or I am the one misunderstanding something about the timing data, and I was not able to find an existing topic about this.
I'm using version 3.22.11+200330 if that counts.

Thanks in advance.

This survey have token + answer persistence + allow edit ?

Hi Denis,

I am using tokens, and if participants leave the survey they can resume later from the same point. They can navigate backwards if needed, but once submitted they can't fill in the survey with the same link again or edit their answers. Only I seem to be able to do that later as an admin (i have never tried that).

So the settings are:
Responses to the survey are NOT anonymised.
Enable token-based response persistance: on
Allow multiple responses or update with one token: off
Allow public registration: off

Thanks

OK Thank you for feedback.

I'm like you : i think it can not be possible … have 1/2 seconds, why not but not mote than 1 seconds clearly ¿

LimeSurvey version ? If last 3LTS or 5 : please report the issue.

And the thing is, it is not even consistent.
I mean in most cases: full survey time > submit date - start date
but there are cases when it is the opposite: submit date - start date > full survey time.
I have yet to find a case when its exactly the same. (but I did not check all of them of course)

Im using Version 3.22.11+200330

Hi,
isn't it an academic question?
Yes, the difference between "startdate" and "submitdate" is different from "interviewtime".

You have to ask the developers how both are implemented (or search the source code by yourself)

So, what?
If you need the correct interview time you use this value.
As you see, "interviewtime" and "sum of group times" are identical.

And "startdate" and "submitdate" you use in the answer table only for a quick rough overview.
And here the precision is in my opinion sufficient.
In a test of 700 responses I found a maximum difference in positive direction of 5.2% in negative direction of 4.3% 
with most differences in negative direction (653 vs. 47)

Here a screenshot
 

Joffm

 

Joffm wrote:

Hi,
isn't it an academic question?
Yes, the difference between "startdate" and "submitdate" is different from "interviewtime".


 

I don't understand the reason of this …