Bonus content: Making a New Interaction

Chapter 4 of my book, Questionnaires, demonstrates the hazards of being too clever with interaction devices. If you decide to go ahead anyway, here are some things you should think about.

If you must do a fancy interaction, then test the heck out of it 

If you’re going ahead with a new interaction device for your survey, it’s not enough to make it look good. It’s also got to be: 

  • easy to use for the people who answer, especially for anyone who isn’t that confident on the web, 
  • easy to program for the questionnaire builder, 
  • robust across different types of questionnaire,  and you must be sure it
  • works on the rich variety of different screen sizes and input modes. 

Users may want to talk about the question rather than the interaction device

It’s possible to achieve all the above, but not easy. For example, I worked with Kantar Operations, part of the huge Kantar group of world-wide market research organisations. They create and deploy around 300,000 different questionnaires every year so making those questionnaires easy for their millions of respondents is crucial.  

Kantar Millward Brown, another business in the Kantar group, wanted to develop a new response format. One of their many areas of expertise is in helping clients to develop advertising that supports a brand, and one of their questionnaires at the time included a brand adjective task – a question that asked people to say whether descriptive adjectives applied to an advertisement for the brand. 

 image missing here

0.20 “Unstyled” response format with adjectives 

 image missing here

0.21 “Styled” response format with scroll bar for the same adjectives 

Previous research showed that people who answered liked the ‘styled’ response formats, with the larger radio buttons and bigger text. But this might mean that a long list of adjectives required a scroll bar. Would this create difficulty? 

Kantar Millward Brown agreed to a usability test based on one of their ad testing questionnaires with the ‘thankyoucows.com’ ad for a premium UK brand of yoghurt, Muller Corner. You can read more about the test in this blog post: Using EEG in a usability test of a survey

 image missing here

0.22 Still from the “thankyoucows.com” ad for Muller Corner yoghurt 

We were testing a variety of topics alongside the specific response format. It was a lovely ad and I enjoyed watching it – a good thing as I calculated that I watched it at least 100 times over the course of the test and analysis. The good news was: participants barely noticed the scroll bar and the response format was fine.  

The slightly more tricky news was: a response format without a scroll bar was better, but limited the questionnaire designer to a maximum of 8 adjectives – which takes us back to ‘the reason you are doing it’ tentacle of the Survey Octopus. Can the brand owners be persuaded to limit themselves to getting feedback on 8 adjectives? 

And the even more tricky news for us, as designers: ideally, we would test the new interaction without any specific words, but that wouldn’t make any sense to the people who answer. So we have to put a specific question into it. What then happens is that the people who answer (quite rightly) much prefer to focus on the specific question than on the finer details of interaction. 

 image missing here 

0.23 When testing a new interaction device, we were interested in interaction and this experience. The people who answered were in the topic of the questionnaire and life in general (Coombe, Jarrett et al. 2011)