Dealing with a Sensitive Subject
According to the Guardian, “there’s nothing that makes Brits cringe more than discussing our salaries.”[1] So how can we overcome this when we’re asked by our clients to deliver income details as part of a research project? We might just have found the answer – or at least got closer to it.
As some of you will know, one of the trickiest classification questions – particularly in online surveys – asks for annual household income and there are numerous ways of obtaining the answer. Asked too early, you risk losing participants before you get to the main body of the questionnaire. And regardless of whether the question asks Which of these describes your income last year? or similar, standard practice involves giving participants the option to choose from 4-7 different income bands, a ‘don’t know’ and a ‘prefer not to say’ option, and leave it at that.
Depending on the audience, this question usually sees a larger proportion of ‘don’t knows’ than other classification questions. But perhaps not for much longer. An online survey we recently carried out amongst UK homeowners and private landlords included classification questions with a twist, which could provide an answer to this tricky subject.
So, what did this survey do that others don’t? There were a number of classification questions at the end of the survey – there was a safety net for all those participants not willing (or able) to share this kind of information, obtained in a first question with standard wording. Participants were asked to choose from a list of 6 income bands, a ‘don’t know’ and a ‘prefer not to say’ option. Of the 2,265 participants who completed the survey, 186 responded ‘don’t know’ and 12 participants ‘preferred not to say’.
These 198 participants were then shown a follow-up question, asking Could you please tell us which of the below income classes you believe your household falls into? and reassuring participants of confidentiality of their answers. 58% of participants seeing this second question selected one of the three “value” options given – below average, above average, around average, decreasing the proportion of participants who could not be classified from 8.7% to 5.1%, a great result for our client since this was one of the crucial markers to answer their research question.
Given that the household income question is often used in market research as a key instrument to determine socio-economic group, or as a measure for segmentation exercises and cluster analysis, we can see the potential that this line of questioning can have for all our clients. Adding a caveat, such as referencing ONS data, average UK household income, or even more precise, average household income in a distinct area or age group, could further help categorising participants. Giving them a reference point could encourage them to self-identify into one of the three target groups.
This is certainly something to build upon as we design questionnaires in the future.
Anna Maria Haque-Everding