How can I create a tool to make survey interviewers' jobs easier?
April 23, 2009 5:13 PM   Subscribe

I am conducting a survey with zero previous experience that includes skips (e.g. 'if respondent answers X, skip to question 12') and open-ended questions. How can I best create a tool for the interviewers to record the answers?

I was imagining something simple that includes the prompts the interviewers are supposed to read aloud to respondents and a simple place to record answers to reduce interviewer-related error.

Ideally, they just press the button related to the choice they'd like to select and then it (whatever it was) would take them to the next question, omitting the interviewer actually having to skip questions as appropriate. Initially I was planning to use an excel spreadsheet but then I'd need to have a different place to record open-ended answers and my pre-testing shows that even trained interviewers miss the occasional skip prompt (e.g. read questions they weren't supposed to, or skip questions they were supposed to read).

Also, it should generate the results automatically in some format that can be easily compared.

Is there a simple and obvious way to do this? Thanks!
posted by stewiethegreat to Technology (6 answers total)
 
This ends up being a somewhat complicated question to address sometimes, but to ensure you pick the correct tool you probably need to answer the question of whether the group of questions being answered constitutes a tree, a directed acyclic graph (DAG), or just a general directed graph.

In my experience most series of questions like this end up being a DAG but you should pay careful attention to the differences between a DAG and a tree, as a DAG is usually more difficult to implement than a tree.
posted by XMLicious at 5:28 PM on April 23, 2009


This is called CAPI or CATI and there are several packages out there. The good ones you have to pay for. If you are savvy, you can design your survey in one of the computer-survey packages and have the interviewer just read it to the respondent. There are many good commercial ones; limesurvey and php esp are two decent open sources ones that come to mind. Neither is well documents, but both can build in skips based on previous answers. You can also have BOLD BIG skip instructions that are hard to miss.
posted by a robot made out of meat at 5:50 PM on April 23, 2009


er well, documented. Poking around I see that they're getting better in this regard. And It's very much possible to put two questions on a page for the interviewer: a multiple choice which will determine skip logic, and a text box which might capture something you're interested in. All survey packages will export data in a usable format, but if it's more than something trivial the analysis is of course for you. You also might be able to get away with something dead simple like survey monkey.
posted by a robot made out of meat at 6:01 PM on April 23, 2009


Just set it up with Survey Monkey, which manages this kind of branching logic very well.

At the end, you have to set it to allow respondents (your data collectors) to submit multiple times, but SM can totally do this - and at the end gives you a nice bunch of statistical analysis.
posted by DarlingBri at 6:45 PM on April 23, 2009 [1 favorite]


Seconding limesurvey.

I've used to set up very complicated branching-logic questionnaires using this, and you can set up a webserver on whatever machine you are using then have the interviewer record responses using limesurvey.

Plus, limesurvey has the advantage that you are not handing off the data to a third party, which can be necessary if the information is confidential.
posted by butwheresthesushi at 9:12 PM on April 23, 2009


Oh, if you use limesurvey, and you can output all your data as an excel, spss, R, or csv file when you are done.
posted by butwheresthesushi at 9:13 PM on April 23, 2009


« Older Help me find this hat?   |   If not the Treasury, then who? Newer »
This thread is closed to new comments.