Eliminating bias in market research
This year’s International Women’s Day has got us thinking about the impact of sex and gender on research and strategy. It’s easy to forget how recent much of the progress on gender equality has been. When we look beyond the obvious, there are many ‘averages’ and ‘norms’ that by default are still based on men.
Recently, I’ve been reading Caroline Criado Perez’s ‘Invisible Women’ which has prompted me to consider this bias further. As a new member of Incite’s Health team, I have been fascinated by the impact of this often unseen bias on the treatments and care that we accept as ‘standard’.
Perez explains how centuries of analytical bias towards men has shaped the field. Up until the 1990s it was common for clinical studies to overlook female-specific biology, anatomy and pathology, with many under-representing or simply excluding women from their samples. Upon reflection, some of the impact of this is clear; think of all those anatomical posters and models you saw at school or on the walls of doctors’ surgeries – they are often by default male.
More poignantly, Perez presents heart attacks as a compelling example of the fundamental consequences of this that remain today. Research found that in the UK women are 50% more likely to be misdiagnosed following a heart attack, rising to 60% in some types. She explains that not only can women suffer from ‘atypical’ symptoms with no chest pain, but the common diagnostic threshold (biomarker) used to test blood may be too high for women. Even the traditional angiogram scan used for detection looks for obstructed arteries, which are more common in men and often absent in women.
Well, good job we’d never make these sorts of mistakes in research, planning and strategy, right?
Here are three ways that these types of biases can sneak in, despite our best intentions, and three ways to avoid them:
1 / The flaw of averages
Research going back decades has demonstrated – in often startling ways – that designing for the ‘average’ person, even on a balanced and representative sample, can lead to deeply flawed and incomplete answers. If you want to know more about the maths behind this, you can deep dive on Anscombe’s Quartet, datasets that look like Dinosaurs and the problems experienced by the US airforce fitting cockpits to a non-existent ‘average pilot’.
Curiosity! Get to know your data before settling on a single approach or metric. In particular, visualising data should be a central part of any first investigation of it – not just a way of preparing it for presentation. Our eyes are much quicker at identifying patterns and problems than our rational cognition.
2 / Setting the wrong scope
A focus on in-going hypotheses and unquestioned assumptions can stop research from exploring wider, often important, issues at play. In Perez’s example above, the faulty assumption was that male and female physiology is essentially the same.
Challenge the question before setting out an approach: why are we asking this and what are we looking to do rather than focusing on what we currently think. In our research and consulting we use critical thinking tools like issue trees to make sure that happens.
3 / Relying on just one perspective
Using a single research method or approach can narrow findings and prevent you from uncovering the whole story. Equally, a lack of team diversity in analysis sessions can cause key dimensions and considerations to be overlooked.
Creative research methodologies and approaches that leave space for spontaneous discussion are crucial to capture previously unconsidered thoughts. An analysis team with diversity across different ages, genders, races and backgrounds is key to challenging thinking and combating unconscious biases.