By Heidi Dietzsch
“To get answers, ask questions; but to get good answers, ask good questions.”
The above quote from Ghanaian author Israelmore Ayivor is particularly relevant for research work. The questionnaire is one of the most important links in the chain of information that begins with respondents and ends with the reader of the research report. It is critical to design it well.
Designing a good questionnaire is an art and no research can achieve success without a concise, well-thought-out questionnaire. The questionnaire has a great impact on completion rates, data quality, the experience of the respondent and the eventual research results.
A good questionnaire should motivate and inspire respondents to provide complete, accurate information and should counteract the problem of respondent disengagement. By not asking the right questions, in the right order and in the right way, you might generate a pile of data garbage, rather than the useful feedback and insights that you were hoping for.
When designing a questionnaire, there are issues that could make or break a research study.
First, there needs to be a proper introduction that informs respondents exactly what the survey is about and what the results will be used for. Respondents need to be assured that their details will be kept private and anonymous and that it will not be used for any marketing purposes.<
To yield data that are robust and diverse, a questionnaire needs to include a variety of different question types. The most basic question type is the one with a simple yes or no answer. For instance: “Do you have a private banker?”
<spThe second type of question allows a respondent to choose from a multiple set of answers. For instance: “Which services does your private banker offer?” The choice of answers might be: 1) Advice on managing your investments; 2) Helping with your will and structuring your estate; 3) Advice on insurance cover; 4) Transactional banking and 5) Lending products.
Probably the most often used question type incorporates rating scales. Likert scales are the most popular scales used by researchers, named after Rensis Likert who developed them in 1932. Likert was an American social psychologist who was interested in measuring people’s opinions or attitudes on a variety of items.
Likert scales are considered one of the most reliable ways to measure opinions, perceptions and behaviours. They are typically five-, seven- or nine-point scales that offer a range of answer options, from one extreme attitude to another, with a moderate or neutral midpoint. For example: “How satisfied are you with the service you receive from your private banker?” The answer choices might be: 1) Very dissatisfied; 2) Somewhat dissatisfied; 3) Neither satisfied nor dissatisfied; 4) Somewhat satisfied; 5) Very satisfied.
The fourth type is the open-ended question that aims to garner qualitative data and is suited to adding some colour. These questions allow respondents to provide answers in their own words. Open-ended questions are exploratory in nature and can provide researchers with rich data that shed more light on respondents’ true feelings about a subject. An example is: “Can you please describe how your private banker can improve his / her service?”</span>
Although this cannot always be avoided, a questionnaire should ideally not be too long. Lengthy questionnaires are intimidating and might prevent prospective respondents from participating. They can lead to respondent fatigue with respondents either not completing the questionnaire or rushing through it, providing random or low quality answers.
Researchers should ensure that questionnaires are engaging with a sensible flow, incorporating questions that are succinct and clear. Unless you are interviewing a specific segment, such as stockbrokers for example, specialised jargon should be avoided. Also, steer clear of abbreviations and slang.
Questions should be worded in such a way that respondents are comfortable answering them honestly. It might happen that respondents provide the answer they feel is the most socially acceptable, or the one they feel the researcher is seeking. Confusing, double-barrelled questions should be avoided. For instance: “How satisfied are you with the professionalism and knowledge of your private banker?” The private banker may have been perceived as professional but not knowledgeable, or vice versa. Rather split such a question into two.
It is also important for researchers to make an effort to construct well-written, precise and clear questions. A poorly written questionnaire, that is cumbersome, difficult to understand and riddled with spelling and grammar errors might not only compromise data quality but can also seriously undermine the researcher’s credibility with respondents and clients.
It is also important that respondents should be respected. Researchers should remember that they are human and not just a source of data, and that their time is valuable.
Finally, before a questionnaire can go live, it needs to be thoroughly tested. The researcher needs to ensure that the questionnaire doesn’t exceed the intended length. No questions should be missing and questionnaire routing should work properly. Spelling and grammar errors are taboo. If it is an online questionnaire, it should be visually appealing with a good layout. Tests should include work by internal audiences, but also a testing phase of fieldwork when respondents’ use of the questionnaire can be carefully examined to insure it is working as intended.
Designing a good questionnaire may not be as simple as it seems. These days, with a variety of online questionnaire design apps available, anyone can create a questionnaire. However, businesses seeking market research data might be better off avoiding these and leave questionnaire design to the professionals. Intellidex has researchers and analysts with several decades of questionnaire design experience and our methodologies are informed by behavioural science. We believe that behavioural science has a major impact on market research and that it can substantially increase response rates and data quality.
