1

I have several topics in a questionnaire for user profiles that they should be able to answer without a lot of hassle. They all contain the same components, but I'm not sure whether I should split them into many questions with few (possibly binary) answers or few questions (ideally just one) with many answers. How do I best determine that prior to user tests?

Components:

  • Have you tried X at least once? [Yes/No]
    • If not: Do you want to try X in the future? [Yes/No]
    • If yes:
      • Have you done X regularly in the past? [Yes/No]
      • Did you enjoy X? [Yes/No]

Alternative:

  • How do you feel about X?
    • I've never tried it and don't want to.
    • I've never tried it but hope to do one day.
    • I've tried it once and didn't like it.
    • I've tried it once and liked it, would consider to do it again.
    • I've done it regularly but quit.
    • I'm doing it regularly.
2

Ultimately, I think this is a perfect candidate for some user or A/B testing, but I'd also consider a third option, depending on your audience and application's tone: natural language sentence construction.

This would allow you to disguise your Boolean questions as a natural language response that might seem more friendly and less like a survey. With each selection, you can update the remainder of your sentence to reflect the branches of questions your users need to complete.

mockup

download bmml source – Wireframes created with Balsamiq Mockups

  • I like this approach and have recommended it in the past, but it can cause a translation/localisation nightmare. – Crissov Mar 7 '18 at 8:49
  • @Crissov good point—might not be the best option if you’ve got localization in mind. – maxathousand Mar 7 '18 at 13:59
1

I'd say "it depends":

  • on how much time you want the user to spend on your questionnaire

    Asking one closed question gets you the answer very quickly, which is in most cases the first thing that comes to the person's mind while several answers possible forces the user to read all possibilities first, then think on what suits best and maybe they can select some other answer to do better in your poll.

  • whether there is one unconditional answer to the question that is picked as "the most applicable for me"

    Think about a question "How often do you change your underwear" and multiple answers like

    • "Never"
    • *"Once a month"
    • "Once a week"
    • etc.

    in case of multiple answers and a series of questions "Do you change your underwear?", "Do you change it more often than once a month?", and so on just to get the same information.

  • whether what you are asking about can be described by a closed question without any truth hidden behind

    Imagine the question "Do you enjoy having sex with animals" with yes/no answers while the most appropriate would be "I haven't tried".

  • whether you can make your questionnaire interactive

    Should you be asking about food preference, you could start with "Are you a vegan/vegetarian?" and in case of "yes" simply drop all meat-related questions (they would be tagged as "not applicable") instead of forcing the user to go through all "Do you like burgers/stakes?" question with "no/I haven't tried" answers.

Bottom line - for the example you provided I'd surely go for the first solution as the second one forces the user to go through all the answers, deeply analyse them and perform "if this than that" analysis to actually select the answer that is the most applicable for them.

  • Just for the record, I know several vegans and vegetarians who could answer "Do you like steaks?" from experience and would even say "yes". – Crissov Mar 7 '18 at 8:59
1

Your first option seems to be best if you want to be hassle-free:

  • It's conditional, meaning you end up with very few questions which are very easy to answer (low cognitive load). Users are more likely to answer at least the first question since it's so easy to do.
  • You could make some of the questions slightly playful by not only using yes/no answers. For instance could there be a scale ? You might want to look at a tool like Typeform for inspiration.

Your second option seems like a lot of hassle because the answers are long and numerous. Looks to me like the cognitive load would be rather high.

I'm not sure a natural language form would be appropriate here. They are "fun" but they do require a bit of focus and commitment. Natural form languages are nice for doing a simulation, getting a quote etc. In your case it might be too much to ask of a user if they're not actively trying to achieve something.

Is there any chance you could do a bit of user testing? Even if it's just guerrilla tests?

  • I'm now leaning to combine some but not all questions, e.g. "How often have you done X?" with a scale or option list ranging from never over once to several times and often. I think this still requires low cognitive load compared to the composite answers. – Crissov Mar 7 '18 at 9:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.