People and percentages: our approach to quantitative research
We are all individuals but data often shows us there can be unity in some of our behaviours and attitudes. Contrastingly, quant research can show how groups of people can be differentiated in their ideas or actions, apparently driven by variables such as age, life stage or socioeconomic status, for example. Herein lies a tension in research, perhaps laid most bare in quant, where people are represented by percentages; so how do we find the patterns in the data, the groupings of similarities and differences and why they exist - without reducing (and at worst, completely ignoring) the real people behind it all?
Here at Humankind Research, our quantitative approaches hang on some guiding principles:
Treat people like people, not numbers
We should be designing surveys for the people who are agreeing to complete them. To that end, we focus our efforts on being polite and clear; we’re non-judgemental, balanced and strive to make surveys as engaging and concise as possible. We’re always wary of triggering people by unnecessarily or carelessly diving into sensitive topics – and we avoid asking for personal information we don’t really need.
Let people use their own voice
Where possible, we champion hearing things in people’s own words. Whether this is doing qualitative research as well as quantitative, engaging a lived experience advisory group in our survey design, or perhaps more pragmatically making sure we have given people the opportunity to express their thoughts and feelings by including open questions - and particularly relish spending time reading each verbatim!
Question the groupings
Research is prone to many typical or standardised groupings, ranging from arbitrary age bands, location defined solely by regionality that hides deeper detail such as spectrum of urban/rural, through to frankly out of touch socio-economic classifications. It’s important that we keep pushing the industry to challenge out of date definitions. We ourselves can try out something new in our proprietary pieces, as well as staying abreast of the efforts of academic researchers and industry leading bodies to define and test improvements and alternatives.
Peel back the layers
From survey design to analysis, we are always thinking about ways we can ascertain where nuance comes into play, even if a core data point looks to tell one joined up story. Let’s take a completely made-up statistic as an example – if we find that 75% of people in the UK aged over 65 donated to a cancer charity in a given year, our next task is to extrapolate the detail behind this number: Why? How? Who? And within these, where are there similarities and differences, and why do they exist?
Don’t just assume cause and effect
A serious research pitfall is concluding ‘this is this because that is that’. We acknowledge that from the outset, the research is already bound by the objectives, what is included in the survey, and what exists in the ensuing data. Yes, it’s possible to establish mathematically a circumstance whereby one variable is predicting the outcome of another, but we must always remind ourselves that our sub-conscious is always at work, and as such we can’t necessarily accurately explain our own behaviours – let alone those of others. Instead, we need to look at our data through multiple lenses; we need to take into account other sources, contexts and hypotheses – and use all of this to tell a wider story.
Keeping humans at the heart of all our research approaches is a fundamental principle at Humankind Research. We acknowledge and mitigate as best as we can the pitfalls in any research methodology, but push forth knowing that research grounded in a genuine and respectful desire to learn about people and inform decision making is the best type of research.
Laura Moore
Head of Quantitative Research