The difficult question of happiness
A new paper suggests that respondents to surveys on well-being are affected by the way they are asked
How happy are you with your life? Yes, it’s back to “happynomics”, because a new paper suggests that one of the oldest findings in the field may be an artefact of the way this question is often asked.
First, some background. Happiness – “subjective well-being” – is often measured by asking survey respondents how satisfied they are with their lives, taken as a whole. Sometimes this is a multiple choice question: “very happy, fairly happy, or not very happy?” The most influential paper in happiness economics, Richard Easterlin’s 1974 “Does Economic Growth Improve the Human Lot?”, extensively discusses evidence from surveys of this kind, in which very few American respondents said that they were “not very happy”. Almost half said they were “very happy”, the most blissful response available.
People, then, are generally happy with life – too happy, perhaps, for a three-point scale. Easterlin is often quoted as having proved that economic growth does not lead to happiness, but perhaps we should say he pointed out that economic growth might not eliminate misery.
Another fascinating titbit from the old Easterlin paper, based on a 1965 survey: 53 per cent of the British respondents were “very happy” but only 20 per cent of West Germans and 12 per cent of the French. Income per head was similar at the time, as it is today. What could explain such an extraordinarily large difference? Language? Culture? Something else?
More modern research often uses a 10- or 11-point scale of happiness, often a “Likert scale”, which offers a statement such as “I am satisfied with my life as a whole these days”, and asks respondents to say how strongly they agree or disagree. Clearly this allows for shades of grey, although problems remain: Bill Gates can earn 100,000 times more than you do, but even if that makes him 100,000 times happier, there’s no way for him to express that joy on a 0-10 scale.
And there’s another problem with Likert scales: maybe something about an integer scale itself distorts the survey results? There’s an alternative: the VAS, or visual analogue scale, which replaces the 0-10 options with a single line. The respondents can mark where they would place themselves on the line between an extremely satisfactory life and an extremely unsatisfactory one.
Raphael Studer and Rainer Winkelmann, economists at the University of Zurich, used a randomised trial to tease out differences between the two scales. Some respondents were asked to report their happiness on a Likert scale and then, a month later, on a visual scale. Others were given the visual scale first and the Likert scale a month afterwards.
There were broadly similar responses to a Likert and a visual scale. But there were intriguing differences. One was that Likert respondents tended to cluster around seven and eight, being positive but avoiding extremes. The visual scale attracted a greater variety of responses; the usual findings were borne out (the middle-aged are less happy; the married are happier) but the size of each effect appeared bigger on the visual scale.
And there was one really odd result. We have long believed that women are happier than men, on the basis of decades of answers to Likert-type questions. The recent foray into happiness research by the UK’s Office for National Statistics used a 0-10 scale and replicated exactly this finding. But in Studer and Winkelmann’s randomised trial, this tendency disappeared. Women and men, who had answered differently on the Likert scale, pointed to the same point on the visual scale.
The authors conclude that the long-standing finding that women are happier than men is simply an artefact of the way the question is asked. How fortunate that nobody had been recommending gender reassignment surgery to all men as the secret of happiness.
Also published at ft.com.