So, Ipsos MORI have done some research for the Royal Statistical Society to put some figures on what we already know: that (in MORI’s words) “perceptions are not reality” (although, as Oliver Payne has pointed out “reality is perceived”).
In summary, we tend to over-estimate the incidence of bad stuff. Or of, err, difference. One example: on average, we (the British people) think that 15% of girls under 16 get pregnant each year. The true figure is 0.6%. Second example: on average, we think 30% of the population is black or Asian (compared with 11-13% in reality).
None of this is hard to understand, once you factor in insights from behavioural sciences. Availability bias looms large: it’s not news, or local gossip, when a girl reaches the age of 16 without being pregnant, but everyone will know about a teen pregnancy locally. And most people in the UK are white, with mostly white friends, family and acquaintances: why wouldn’t non-white people be more noticeable to them?
On top of this inherent bias, we have a hard-to-measure impact from public policy/discourse: politicians, media and opinion formers often talk up ‘problems’. It’s in their interests to talk up the profile of teenage pregnancy, perceived to cost the rest of us hard cash. And those who present diversity as a problem raise its profile, and use emotive language to embed it in our minds. “Swamping” is one of their favourites.
What can we learn from this? Bobby Duffy of Ipsos MORI says “We need to avoid dismissing public opinion: everyone has a vote, misperceptions have always been with us and they may reflect concerns – that is, people may over-estimate issues because they are worried about them, not the other way round.” Hard to argue with that. Peter Harrison of Brainjuicer stresses this: “People aren’t scared of immigrants because they believe there’s lots, rather they believe there’s lots because they’re scared”, he tweeted.
I would add two things.
First, framing is really important. This was survey research, in which questions can be framed in different ways. The pregnancy question went thus: In your opinion, what proportion of girls under the age of 16 years in Britain get pregnant each year? I’d suggest that you’d get a more accurate answer (ie closer to reality) with a wording like : Out of every 1,000 girls under the age of 16 years in Britain, how many do you think get pregnant each year? The likes of Ipsos MORI could experiment with split samples (that’s where you run a survey where some questions have two versions, each asked of half the overall sample of respondents) to find wordings that give a more accurate average answer. This of course, would tell us nothing additional about public opinion or perception as fact – but it would tell us how to frame a question to get an answer more grounded in reality. And, if we’re interested in good policy-making, we’d also learn from it to help us frame policy discussions.
Second, these results don’t mean people are stupid or ill-informed. I know that you know this already, or you wouldn’t be reading my blog. But the implication of reporting these findings as indicating wrongness is that it’ll get read as implying lack of knowledge or ability on the part of the great British public. The temptation then is to say that the appropriate response is ‘education’, or what Bobby Duffy calls “myth-busting”. The thing is, this response is stuck in the past. Now that we know that people aren’t rational economic beings, who marshall information to optimise their decisions, let’s stop acting as though that’s the case. Let’s not think that throwing information at people is the right response to this report.
The myth-busting I suggest we need is a growing understanding of human nature. As I say all the time, this is what we’re like! The reasons we nearly all mis-estimate what Ipsos MORI have measured are innate, and result from the way our species has evolved and succeeded. I hope there will come a time when this becomes a mainstream view. At the moment, I know it’s a left-field perspective, but it’s based on plenty of evidence.
To finish, I should ask you to do something. How about: the next time a friend or colleague says something along the lines of “we need to educate people” or “we need to get more information to people”, ask them to work with the grain of human nature instead. Use information that’s already available, but frame it differently. And measure the impact.
And next time someone is talking up the sort of risks and differences covered in the RSS report, don’t call them out on the inaccuracy of their data (you’ll just look like a statto); call them out on the impact they’re having on public opinion.