OPINION3 December 2015

Do feelings betray thoughts?


Do people genuinely believe what they say at times of national emotional distress? Martin Boon looks at how implicit reaction testing can dig deeper than question and answer data into public attitudes.


Front page pictures of dead children washing up on Greek beaches during the summer provoked understandable outpourings of grief and dismay among a British population historically noted for its compassion toward fellow humans in distress.

Calls for a relaxation of immigration controls inevitably followed, with the Prime Minister outlining a policy of permitting up to 20,000 new refugees into Britain over the next five years.

The public, for their part, also appeared ready to intervene. The Guardian reported being “inundated” by people wanting to support refugees in need by offering shelter in their homes, and a survey from the Charities Aid Foundation revealed that 7% – or 1.8 million households – would be prepared to offer a room or space in their home to a refugee.

But is such an admirable response a true reflection of the public mood, or merely an overstatement of gut-wrenching, but momentary, sentiment inspired by a need to metaphorically bow in favour of perceived socially desirable outcomes?

The issue of immigration into Britain is, of course, provocative and troubling to many people, with real ambivalence on display about immigration/refugees. This prompted ICM Unlimited to identify the issue as an ideal opportunity to investigate whether people truly believe what they say at times of national emotional distress.

ICM undertook an Implicit Reaction Test (IRT) to tap into the sub-conscious, an online technique applied within our omnibus to see whether what people think is reinforced by what they feel? The quicker the response (in milliseconds), the stronger the view held.

We tested 10 statements reflecting public attitudes toward the refugee crisis, with notable inclusions being whether or not respondents would welcome refugees into their own town, and indeed offer a room in their own home. Conventional response confirmed that a sizable chunk of people ( 10%) would make their homes available while more, ( 36%) would welcome refugees to their town. In both cases, however, more people, in total, would not offer a room ( 75%), nor be particularly welcoming ( 43%).

But the IRT had much more to say. Not even half of those willing to offer a room demonstrated certainty about their own offer to do so (only 4% did, in conjunction with a moderate speed of response score). However, uncertainty was not evident in reverse. In other words, those unwilling or unable to offer a room to a refugee implicitly confirmed as much, while many of those demonstrating potential generosity revealed uncertainty about what they might do. In this case, what they said was not in complete harmony with what they felt.

So maybe this is an interesting teaser that IRT reveals more about ‘true’ public attitudes than question-and-answer data is capable of.

So what implications does this have for public opinion measurement? Well, use of the technique might offer some bulwark against narrative setting instant reaction polls that headline writers might love but the substance of which often fades as interest in the subject dwindles.

And we’re still interested in accuracy right? More than ever actually. As the British Polling Council (BPC) election failure inquiry ramps up to reporting stage, we’re examining IRT techniques to see if it helps explain and improve our inaccurate 2015 General Election work. Hard evidence on implicitly ‘Shy Tories’? Polling nirvana.

Martin Boon is director at ICM Unlimited