Why the ‘don’t know’ answer compromise your survey data
We think there’s a common and underestimated error that can compromise your survey data. When filling a questionnaire, it’s not rare to find “don’t know” as a possible answer option.
In this article, we analyze the “don’t know” issue with pros and cons.
Let’s start from the conclusion: why giving the respondent the possibility to answer “don’t know” is a bad choice for the survey data?
Because you’re giving the interviewee the possibility to cut across the questionnaire without focusing on the question nor the answer.
If so, why the “don’t know” is still so frequently used in questionnaires?
A lot of companies think it can help the respondent during the compilation and limit inaccuracy.
The general idea is that it’s better not to answer rather than giving a random and inaccurate response.
This strategy makes things more difficult.
Let’s split our survey respondents into two categories:
- those who have a strong opinion on any issue and are determined to express it;
- and those who are often uncertain of their opinions or avoid a definitive answer because of a competency limit;
The first category of respondents would decisively answer our survey questions regardless of the “don’t know” option that they wouldn’t consider anyway. The second category is the one that could opt more for the no-answer.
At first glance, adding the “don’t know” seems like a good choice: after all we want our respondent to be sincere.
When we include the “don’t know” option, we presume that people who opt for this answer don’t have an opinion on the subject.
The truth is that there are also other reasons why the “don’t know” can be picked. Some examples:
- they don’t understand the question;
- they avoid answering to a “complex” question;
- they’re not motivated or not suitable to answer.
Survey data and Satisficing theory
In 1991 Krosnick outlined the Survey Satisficing theory: answering a survey involves a cognitive effort not everybody is willing or able to make. Avoiding this effort with a “don’t know” is sometimes the easiest strategy.
Survey Satisficing theory is more and more popular in market research. As a consequence, the “don’t know” option is now frequently omitted from questionnaires.
There’s strong evidence of the unreliability of the “don’t know” answer and how compromising it is for your survey data.
Dismissing this option for good will put respondents into a different position may be a more demanding one but for sure more useful to the research. You will minimize the risk of getting an unmotivated and sloppy answer.
Our suggestion is to avoid the “don’t know” and to focus on better answers to cover all possible hypotheses and provide to your respondents a detailed selection of options.
If these reasons are not enough for you, just remember: getting a better answer is more than getting no answer at all!