Top tips when asking sensitive questions in surveys
Many surveys will include sensitive questions of some kind. Whether you are asking respondents for personal factual details – such as their age, income or ethnic origin – or, alternatively, asking questions on sensitive topics – such as medical issues, politics, religion, sexuality, and so on – then these questions need to be handled with care. If not, respondents will give you inaccurate data.
Online surveys are arguably a better tool to handle sensitive questions than face-to-face or phone interviews. Respondents are often more hesitant to answer truthfully where a human interviewer is present and will want to project a "good" image of themselves.
However, even with online surveys there are risks to asking sensitive questions. Many respondents are naturally private and may be suspicious of parting with personal details or anything which could embarrass them or reveal private information. This may result in them answering untruthfully, skipping questions or, worse still, abandoning the survey completely.
Here we outline our top tips when asking sensitive questions, to avoid some of the risks outlined above.
- Put sensitive questions towards the end of your survey: Imagine talking to someone you've just met. You wouldn't start by asking them personal or intrusive questions right away; instead you would build up a rapport with them first, and only when you had established some trust with them, would you start broaching sensitive topics. The same applies for surveys. Even though you don't know your respondents personally, you still need to script it so that you build up their trust – which you can do by asking neutral questions first and building up slowly to more sensitive topics.
- Know your audience: If your respondents are interested in or work in a specialist area that may be a sensitive one, they are more likely to be amenable to answering in-depth questions on that topic. For example, a survey sent out to doctors may be full of sensitive questions, which they would have no hesitation in answering and would expect. The same questions, however, would not work in a survey of the general public. Knowing your audience will give you an idea of how many sensitive questions you can safely ask without fear of dishonesty or high drop-out rates. Limit sensitive questions to only those that are absolutely necessary – do you really need to know their household income or religious beliefs, for example?
- Make it clear the survey is anonymous and results are confidential: You can state this clearly in the introductory text of your survey – for example, by saying that responses will only be used for analysis purposes and will be treated confidentially. Guaranteeing anonymity will increase your chances of getting honest replies. In particular, employees in a staff survey may not be truthful about how they feel about their organisation if they believe their answers can be linked back to them. If you do need to collect personal data, such as for a prize draw, state that these details will not be used for anything other than selecting the winner.
- Ask indirect questions: People tend to give what they perceive to be as socially acceptable answers instead of telling the truth, something known as "social desirability bias". To avoid this, consider asking your respondents to answer questions from another person’s perspective, who is similar to them. An example would be "What would your best friend do if they saw two people fighting in the street?"
- Try to normalise the sensitive issue: This requires phrasing the question to make it clear that you think it's completely normal, which it most likely is, but may need to be reiterated. An example would be "At our company we are proud of having a workforce with diverse religious beliefs. What is your religious affiliation?"
- Hide your sensitive issues in a multiple choice or matrix question: This involves having a list of items, some innocuous and some more sensitive ones. Having a mix of topics will make the sensitive issue seem less of a big deal. An example, in this case regarding tax evasion, would be:
Which of these have you done in the last year:
Overslept and been late for an important meeting
Locked yourself out of the house
Accused someone of tax evasion
Exceeded the speed limit while driving
Randomising the order of the answer options in this type of question is also recommended to eliminate any bias to certain options.
- Give them a way out of answering: Another option is to let your respondents choose if they want to answer the sensitive or personal questions by either not making them compulsory or by adding another answer option such as "Prefer not to say". This will at least avoid people being dishonest in their response if they really don't want to answer a particular question.
In summary, consider the above factors next time you are including sensitive questions in your survey as this can make all the difference between getting truthful responses and dishonest or incomplete responses.