19 Aug Question Design Principles for engaging, responsive surveys
Respondent focussed questionnaire design is essential to achieve high quality responses at at time when survey takers have so much choice to occupy their time. Our members on GooPoll are responsive to surveys, and are incentivised as well. But that doesn’t mean to say surveys don’t need to be designed with care to maximise their attention.
Here are a few principles we use when thinking about questionnaire design. We could list hundreds of points and write a lot about this subject, but here are 10 of the most common and important principles common to many surveys.
Keep it short
Questionnaire should never be longer than they need to. If you can obtain the key information from 5 questions rather than 20, that will save time, money and likely get a better quality response. Online surveys are cost effective, so it is practical to ask several surveys over a period of time rather than running one off exercises that results in overlong, bloated surveys.
In practice, try to plan your questionnaire design remembering to focus on MCQs and minimising NTHs.
MCQs (Most Crucial Questions)
Any survey will have a small number of crucial information objectives, and the survey will centre on them.
NTHs (Nice to Have Questions)
Sometimes we fall into the trap of asking questions which might be interesting to know, but are not essential. They can be dropped if we need more time to ask the MCQs. Often we can glean supporting data to get the answers for NTHs, either by checking if we have asked these questions before in other surveys, through publicly available data or some other data inference from other sources.
Make questions easy to recall / answer
The more complex questions are to read, understand, work out or recall, the more the effort is needed to answer them. Difficult, complex questions result in detachment from the respondent and ultimately poorer responses.
Make it human
Words matter. Clumsy, wordy surveys that use acronyms, expressions and sentence construction that feels more like a legal document than a simple question are all too common. We should be striving to ask questions in a human way, that seem natural, clear and easy to understand.
If surveys are designed with a good flow, progress naturally and make it easy to answer, we get better responses. If we are looking to get short answers and verbatims, it might be worth considering using a conversational survey style. This type of survey looks more like a chat bot conversation than a conventional survey. Even if we are using a more conventional survey style, incorporating instructions and feedback along the path of the survey will help to make the survey feel more like a conversation.
Survey tools are getting better at making the survey experience more visual and engaging. Pictures, icons and animated transitions all help to engage the respondent, and ultimately get a better response. Design led surveys do take more time to programme, and there may be insufficient time and budget to enhance the survey as much as we would like. But every survey can utilise good design, and on occasions a design rich survey is essential.
Most survey tools are accessible on mobile phones, but some tools are better than others for working well on mobiles. Given that often more than 60% of surveys are taken on mobiles or tablets, it is essential that surveys are easy to do on mobiles. We take a mobile first approach to survey design.
Easy to read question formats
Think carefully about which question formats to use. The question type used needs to be appropriate for the data we want to collect, but must also conform with our wider aims of designing easy to understand questions that attract the best responses. Too often hastily designed surveys consist of endless matrix/grid questions. Long girds need a lot of effort to read and can often result in respondents rushing through the answers. A well presented, small matrix question can work and they have the advantage of being quick to set up in a survey tool. But they need to be used sparingly.
Think about data outputs from question formats
Sometimes we see clients asking for a particular type of question format without knowing quite what to do with the format of the results. A common example is the ranking question. They are a useful question type but do not necessarily produce a simple result that people can easily report. If you are looking for a simple % of people choosing an attribute, a simple single/multiple choice question may be the best option.