Additional interview information on tablet's Completed Screen in 'census' mode

Feature Suggestions
#1

Dear team,

I’m planning a survey in (what used to be called) census mode, and have just started the testing of the questionnaire (not a field test yet) and entering some mock interviews. That gave me two ideas to suggest to you, but first I should say that the way I am planning to organise the survey and as we have done before, is for supervisors manually to check (so reopen) the interviews on the enumerators’ tablets every evening before synchronising (mainly for errors & missing answers). We have little access to internet in the field, so I do not plan to use the supervisor facility on the server (it is a fairly small team with just 2 supervisors and 10 enumerators). (Data quality of uploaded questonnaires is checked out of the field, and supervisors are sent reports with errors daily). For this set up, I thought it would be useful if

  1. it is possible to use a variable from the questionnaire - so an answer to one of the questions in the interview (e.g., name household head) - to identify assignments on the tablet’s completed screen. For example, next to the Survey Solutions identifyer (47-02-06-28 in the screenshot below).

Having an identifyer on that screen makes it easier to ‘find’ an interview (e.g., if an enumerator completed an interview but has a question about it they want to show the supervisor) and for the supervisors to see more easily which ones they have already checked.

Screenshot_20210423-111439
Screenshot_20210423-1114391024×768 63.6 KB

  1. I also wondered if the completed screen could show whether an interview has errors/missing values by showing a red/blue bar instead of the green one on the left of the individual interview. I mean the little green bar to the left of the interview ID and reopen button. In my tests, this little bar is green regardless of whether there are errors/missing answers.

I hope you find these suggestions useful. I am using Interviewer Version 21.01.8 on an Android Version 7.1.1.

Cheers,

Sebastiaan

Create variable in which single select question code is converted to the equivalent string value
#2

Hi Sebastiaan!

On 1): “Identifying questions” will be your saviour! These are questions with a particular scope. Please note, since Version 20.07 you will have the Special Section: Cover Page. So in order to have identifying info from your questionnaire displayed on an Interviewers dashboard, the respective questions need to be placed into that section*. Have a look at the respective documentation!

*If you don’t use a questionnaire designed in the Designer tool before July 20, 2020, see release notes of Version 20.10.

Hope that helps on that question?
Best
Peter

#3

@peter_brueck is right hinting at the cover page. It is exactly to place the identifying fields, such as province, district, city, street, etc.

But in the recent December release it has also been updated to do much more!
image
Now you can include a variable to the cover page, and it will be visible to the person viewing the dashboard! With that you can formulate all sort of messages and stats based on the interview content, such as:

noteSection1 + " " + noteSection2 + " " + noteSection3....

to display the notes left by interviewer at the end of each section, or

nmales.ToString() + " males, " + nfemales.ToString() + " females"

to display number of males and females in the enumerated household.

There is no system variable containing the number of errors in the interview, but if you count them yourself in a variable, you can display that on the dashboard as well.

For number of answered/unanswered questions you can refer to the section functions

Hope this helps.

Best, Sergiy Radyakin

1 Like
#4

Thanks Peter and Sergiy,

I had looked at all the release notes since my last survey to update myself on new possibilities, but this one didn’t fully register apparently (I had been wondering what to do with that cover screen that kept appearing in Designer :slight_smile:

I’ll check it out in more detail. Sounds very promising.

Cheers,

Sebastiaan

#5

Ha, the feature of variables on dashboard slipped my attention as well! That’s pretty nice to have.

Thanks @sergiy for highlighting.

#6

Dear Sergiy and Peter,

Some time passed since your response, but I only now had time to try this. In my existing questionnaire I had an introduction section with quite a few questions (intro text, consent agreement, enumerator and supervisor name, HH ID, village etc.) I also let enumerators re-enter the HH_ID and village name to prevent mistakes. I had hoped that I could select a few of these questions as identifying (HH ID,village name), and therefore appear on the virtual cover page and dashboard, but to leave the structure of the introduction section unchanged (i.e., not have an actual cover page section visible to enumerators in the questionnaire).

However, I found that even with a virtual cover page (rather than a ‘real’ cover) it takes the questions I indicated to be identifying and creates a separate ‘section’ (the four identifying questions I had selected were in the cover section and missing from the intro section).

I don’t want to place the whole intro section on the cover page, as that would be an overload both in the dashboard and when creating assignments (I run a ‘census-mode’ survey). The other option I could think of, i.e., splitting my original intro section in two also isn’t ideal as I don’t want to start with the cover questions (intro text and consent come first in my design) and given I have duplicate HH ID and village questions, there would still be quite a few questions on the cover page (with the duplicates not having an identifying function).

I wondered if there is any way in which to use the cover page as I hoped, that is to have an ‘invisible’ cover section that is made up of questions from other sections without changing those other sections.

I appreciate any advice you could give.

Best wishes,

Sebastiaan

#7

The problem turned out to be bigger than I thought. The questionnaire should not include any of the following questions, whether used as identifying or not:

- HH ID;
- interviewer name;
- interviewer code;
- supervisor name;
- supervisor code;

If you have these questions in your questionnaire remove them.

If your thesis supervisor, manager, coordinator, president or anyone else insists that this is a critical information, which must be recorded for quality control, audit, of whatever other reasons, you politely agree, but remind them that although you see the date and time of the withdrawal on every ATM receipt printed, no ATM in the world will require you to enter the current date and time before making a withdrawal. (I hope the example helps).

The rest depends on what type of survey it is and how the respondent gets located.
You have only indicated that you are doing a census. Census of what, where, how?

If you are asking for consent, you have to identify who gave you consent and who refused. At least for a simple reason of not approaching them for the second time. If you have another way of identifying them that prevents this double counting, then describe the parameters of that process, because that will form the essence of the cover page.

The person opening the door at 123 Main St, Pleasantville, may have refused to respond to your survey. He or she perhaps didn’t even agree to tell his/her name. But you record the fact that the refusal has been received and you identify the door, which you should not knock again. The identification of the door makes the mandatory part of the cover page. The rest (like the ever so popular name of the household head or visit date) can be constructed with variables pulling the information from the other parts of the questionnaire.

#8

Hi Sergiy,

Thanks for the reply (and questions). Hihi, I would probably be the supervisor who would say to add this information :slight_smile:

I think your point would be that e.g., a tablet is assigned to a specific enumerator,and enumerators to specific supervisors, so no need to enter these names (or codes). I find that in real life, tablets get swapped by mistake and enumerators also get swapped between supervisors depending on local or personal factors. I like flexibility to deal with such mistakes and unforeseen local factors to be built into the questionnaire (in other words to have as blank a questionnaire as possible).

I mainly use the tablet as a paper questionnaire with the other survey procedures largely outside the Survey Solutions structure. In part because there is usually very little internet access (rural Africa mostly), so supervisors cannot access the server (in these areas, there are no ATMs :slight_smile:). In part because there are always unexpected things, and I want the ground team to have the flexibility to deal with that without internet access or to have to report back. And in part, it is probably because this is the way I learned to do surveys and I am a little old fashioned.

I’m not doing a census, but surveying in the old ‘census mode’ of Survey Solutions. The sampling is often done only shortly in advance of the interviews (sample distribution over different villages is done in advance, but village sample frames and samples are drawn as the team arrives in a village or the day before). I have no identifying info about households prior to the interviews (or sampling). The selected households in a village are written down on paper and given numbers (IDs), maps are drawn and the IDs divided over teams and then enumerators. The supervisors don’t have time/internet access to go back to the software to enter these numbers and use them to create assignments.

Each enumerator has one assignment and can create as many interviews as they need. We need these IDs (together with the name of the household head) as part of the interview to compare to the sample in the evening data checks (e.g., to identify a household for repeat visits, to see all households have been interviewed).

Through the ID we know who gave consent and who refused. In case of a refusal, only the intro section of the questionnaire is filled out. The ID is linked to the sample list for the village (it is the front door).

I hope this gives you a better idea of my survey. Looking at the last paragraph of your reply, could I then perhaps achieve my objective by only using variables on the cover page, e.g., can I add a variable which outcome is simply the answer entered under the HH ID question, etc.?

Thanks again for your time and insight!

Sebastiaan

#9

Just tried that (using only variables on the cover page) and that seems to create what I wanted.

Cheers again,

Sebastiaan

1 Like