download bmml source – Wireframes created with Balsamiq Mockups

I'm redesigning a form that is used by office workers who interview respondents in person. In some cases, there will be two respondents.

The current design is (2). My suggested approach is (1).

a. The interviewer usually asks each question to each respondent in sequence
b. This is (unavoidably) a long form, with many questions
c. Important: each question may lead to a subsequent question -- depending on the respondent's answer. Follow-on questions are hidden by default, but displayed progressively. Thus, respondent 1's first answer may lead two "new" follow-on questions, while no follow-on questions may be displayed to respondent 2.
d. I will conduct usability testing but for now I'm looking for guidance

  • How often are there two respondents at once compared to just one? Is two always the maximum? The second option looks like it would be easier for the interviewer to manage, but, if you're doing any sort of statistical analysis with the results, I'd be concerned about one person's answers changing based on what the other person's extra questions were.
    – Karen
    Commented May 1, 2012 at 12:32
  • I don't have stats, but the two respondents scenario is significant and not an edge case. Two is always the maximum. There are no statistical analysis of the results -- the system makes a decision for each respondent and gives an answer at the end. One respondent's answers are unlikely to change based on the other's questions/answers. Commented May 1, 2012 at 12:45

1 Answer 1


One problem with option 1 is that with a long list of similar looking response types, it might be possible to confuse the respondent's answers between questions, and it won't be so easy to compare if you have onward answers from one respondent only. The upside of option 1 is that for long forms there is less side to side movement of the mouse/arm/hand [depending on device] which may be significant for repeated use

Here's another possible approach (number 3) into the mix:

For questions asked of both participants, span across both columns, but above the answers.

Where an answer leads to one participant having further questions then include that interaction in the appropriate column as in the mockup below

This gives a good side by side comparison. If you feel it's useful to be able to scan down a long form and see where responses differed, then perhaps consider some additional automated indicator of whether the answers match each other, for example in the mockup below, using a tick or cross in above the line that divides each respondent's answer.

Although, you might want to use icons that don't make it look like it's a right or wrong answer!!


download bmml source – Wireframes created with Balsamiq Mockups

  • 3
    +1 OP's option 2 was on the right way with the column approach, but was way too wide. This elegantly solves the width problem, is very clear and allows for a lot of permutations. Me like! Commented May 1, 2012 at 13:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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