Count number of answered question in Roster

Hello all !

The title says it, I’m trying to make summaries for my interviewer where they can see the response rate for each person in a household, which invovles to count the number of answered and unanswered questions in a roster.

My current syntax is : EnabledAnsweredQuestionsCount(MyRoster[((int) @rowcode)].MySubSection)

I’m aware of this post : Get count of unanswered questions for each person in a roster

To which @sergiy 's answer was correct : One can only use those functions (like “EnabledAnsweredQuestionsCount”) on subsections (like in my code above), but apparently not directly on rosters.

The turnaround is to move all roster questions into a giant subsection, but that makes the navigation in the questionnaire lesse practical.

What is the best way to do this ? If these functions exist, shouldn’t they be applicable to rosters ? If not, why ?

Many thanks,
Noé

The subsection doesn’t have to be giant. One can almost certainly separate the questions by various logical topics/modules into reasonable subsections:

Hi Sergiy,

Thanks for your answer.
Indeed, one can always find logical sections, but what if one does not want to ?
What if there are only 5 questions in the roster ?

I understand that there is currently no way to do what I’m asking, and that’s ok, but maybe the development team could consider this as an intertesting feature to implement ?

Cheers
Noé

Then generate a variable (NumUnuansQuest) inside the roster (R) (and if there are no enabling conditions):

NumUnansQuest = 
  IsAnswered(q1)?1:0 + 
  IsAnswered(q2)?1:0 + 
  IsAnswered(q3)?1:0 +
  IsAnswered(q4)?1:0 + 
  IsAnswered(q5)?1:0

And then add up outside of the roster:

RostUnansQuest=R.Sum(x=>x.NumUnansQuest)

Hi Sergiy,

Thanks, I appreciate the the effort ! I thought about this solution as well, but it quickly becomes a war machine to code if there are enabling conditions (as you pointed).
It is a good solution for very simple surveys, but I still feel like there is a need to adapt “EnabledAnsweredQuestionsCount” to Rosters. Maybe in a future development.

Cheers,
Noé

Going back to the original problem that you described:

  • where did you get guidance on this approach?
  • Why is the standard count of not answered questions not sufficient?
  • What is this survey?
  • Where is the questionnaire/ interviewers’ instructions?

Best, Sergiy

Hello Sergiy, and happy new year :slight_smile:

I got guidance from … myself !
This is part of a census survey, so we will have plenty of unexperienced interviewers. We would have liked to allow them to check where they are at in the interview, and what’s left to do.
Indeed, the blue/green/red color code serves that purpose (partially at least).

I’m not sure what you mean by “the standard count of not answered questions” ?

Anyway, I let this idea aside, but I still think that “EnabledAnsweredQuestionsCount” could be applied to rosters !

Cheers,
Noé

See the last screenshot at this page:

After reading this I still have absolutely no idea as to:

  • what country?
  • what organization?
  • what census?
  • who are your partners?
  • what is the timeline?
  • what is the budget available?

This was duly noted in #2973.

1 Like

Oh ok I see ,thanks.
The original idea was to authorise interviewers to leave some questions unanswered (for a number of reasons). The obvious problem was to find a way for them to assess where (precisely) they left questions unanswered, so that they could know if it was by design or by accident.

To avoid this problem, we just added a “I don’t know” equivalent to all questions, and instruct interviewers to answer all the questions. After all, this fits better the philosophy of Survey Solutions, in my opinion.

After reading this I still have absolutely no idea as to:

  • what country?
  • what organization?
  • what census?
  • who are your partners?
  • what is the timeline?
  • what is the budget available?

It’s not like it’s classified information, but how is that relevant to the topic ?
In short, this is the census of New-Caledonia, our partners are the INSEE, but they don’t use SuSo, so we’re on our own for the development of the questionnaires (with some experience now though !).

Anyway Sergiy, thanks again for your help, it is very much appreciated.

Interesting. I didn’t know about the Census in New Caledonia to be done with Survey Solutions. The SPC’s census and survey calendar indicates 2024 PHC in New Caledonia, but doesn’t indicate that it is to be done electronically/on tablets:

is that information inaccurate?

Priorities are affected by the estimation of the developmental impact. A feature that is necessary to an unknown survey of an unknown country done by an unknown individual gets priority zero, and thus will likely stay in the pipeline for a long time, if even considered.

SPC would be your closest and most resourceful advisor. They have done plenty of surveys with Survey Solutions, as well as population censuses in the Pacific islands.

We used Survey Solutions for other big surveys (like the LFS), but for the census we will only do a test with Survey Solutions (that is, for a small portion of the territory).
However, the next one (in 5 years) may be done entirely with tablets and SuSo, if the test is succesful and if we are ready by then.

I understand, thanks.

Indeed, we are in contact with the SPC, and we will benefit from their experience with Survey Solutions ! But thanks !