A sample text widget

Etiam pulvinar consectetur dolor sed malesuada. Ut convallis euismod dolor nec pretium. Nunc ut tristique massa.

Nam sodales mi vitae dolor ullamcorper et vulputate enim accumsan. Morbi orci magna, tincidunt vitae molestie nec, molestie at mi. Nulla nulla lorem, suscipit in posuere in, interdum non magna.

Bleg on happiness research

The preceding post is on the economics of happiness (EOH) literature. Econometrics is not a particular strength of mine but I do have a (possibly) naive query.

EOH identifies relationships between a self-described measure of happiness for individuals (in the studies I discuss it is often a ranking on a scale from 0-10) which is then regressed on lots of variables.  Among these are measures of  individual absolute income  and measures of income relative to those elsewhere in society.  The key finding as I understand it is that, in wealthy developed countries, it is the relative income variable that explains much of subjectively-assessed happiness (at least in time series and cross-country studies) rather than the absolute income level.  That suggests that “keeping up with the Jones’s” or “other regarding behaviour” has strong effects on happiness rather than an individuals’s purchasing power.

That seems a plausible and not-at-all foolish view to me.

My inquiry however is whether that finding isn’t likely to be simply driven by the way data is assembled here. The happiness variable used is not an absolute happiness measure (e.g. how many “utils” of happiness the agent gets) but a scaled variable that ranks their happiness from 0-10.  The determination of an individual’s self evaluation of their happiness (or quality-of-life) status on this scale will almost inevitably reflect the valuation of their happiness relative to others.  If I feel I am at the bottom of society in terms of happiness I might choose 0 and, if I am materialistic and earn Bill Gate’s annual income I might rank myself very high on the scale.  Won’t that sort of reasoning inevitably mean that in explaining happiness it is necessarily relative rather than absolute income measures that will always do most of the work?

When I think about this issue I can’t see an obvious way around it.  You can’t easily elicit absolute measures of happiness or well-being.  If you choose a ranking on a scale of 0-10 then it would seem inevitable that relative income rather than absolute income variables do the explaining in time series and cross-country studies. Indeed it is a bit of a puzzle that the same result does not arise in a some cross-section intra-national studies?

Or have I misunderstood the way the happiness measures are elicited?  Or do the probit/logit statistical procedures (which do seem appropriate for these data sets) effectively deal with this problem?

7 comments to Bleg on happiness research

  • wjr

    The answer to your last question re. logit/probit regression is an unequivocal NO. All this methodology does is to
    enable the response variable to have finite support – the interval [0,1] or scale 1 to 10, for example.

    I agree with you that there doesn’t seem a simple way around your objection unless one can use some objective
    surrogate for happiness – e.g. some health or stress measure. But even then there could be problems. Quite likely
    happiness, in the way ordinary people think of it, is relative. Maybe there is a state of nirvana for the truly
    enlightened but I haven’t found it yet.

  • conrad

    I think you need more info about the scale here — like what questions were used and the extent to which the underlying validity has been investigated.

    Whether happiness is going to be relative or not is going to be dependent on your measure for happiness. I don’t see why you couldn’t construct a scale that uses relative measures and one that uses absolute measures, although I think absolute measures are probably harder to construct (but probably not impossible). I imagine people have done this (perhaps taking out the cross-correlated bit to get a purer absolute measure).

    For example, if the scale is full of questions like:”When you look at your life, do you feel lucky compared to others” you are going to get relative answers, but if your scale is constructed with, say, questions like “I feel healthy at this stage of life”, it might not be relative to others.

    You can dissociate different aspects of things like happiness into different component. Off the top of my head, Yoshi Kashima over at Melbourne does this sort of thing with similar variables (e.g.,

  • hc

    Thanks Bill that is helpful. H.

  • hc

    BTW, It seems to me reasonable to suppose people can make absolute appraisals of their happiness and that the richer they are the happier they are: See:

    But what is difficult to understand are the reasons for this happiness. Is it because they can do more things (e.g. fly business class) or is it because they can then feel advantaged relative to their neighbour? I assume its a bit of both of these things and hard to disentangle.

    If it is the “other regarding” component which is dominant then people can improve their happiness by being less envious/covetous. Or simply by retracting their material aspirations. In short, they can become more happy by rearranging the way they think rather than by increasing their wealth.

    Of course sages like the Buddha and Christ said exactly this thousands of years ago.

  • conrad

    One of my comments seems to have been killed, but I think it depends on your measure of happiness. It’s most certainly multi-factorial, so if you’re just giving people one hammer question “how happy are you”, then you are confounding multiple things. Well being must be one of the most over-done topics in psychology, so I imagine the answers to your questions most probably exist.

  • Peter Whiteford


    I am an outsider in this without much knowledge (but just supervised a successful PHD in the area!). What is striking to me is the distribution of answers to questions of this sort – in Australia at least. Just about everbody in the Australian population (in HILDA at least)puts their satisfaction in life between 6 and 8 (on a 1 to 10 scale) – almost nobody answers these questions as if they were miserable and no one says they are ecstatically happy with their lives (rather similar to answers to questions about where you rank yourself in the income distribution with 60% of the Australian population saying they are in the middle 20% of the income distribution).

    So I think in Australia we have a sort of inbuilt bias against saying we are at the extremes of the distribution of happiness – perhaps on the one hand we think things could get worse (say no man is happy until they are dead, as the Ancient Greeks put it) and at the other end of the spectrum no one is willing to say they are actually miserable, because you would probably actually have to be seriously depressed to think this – and even then you might not be willing to accept this.

    On the other hand the distribution of responses in a number of Eastern European countries suggests that there are significant differences in the distribution across countries (they tend to say they are a lot more unhappy than us) – although whether this is objective circumstances or national temperament is hard to say.

  • conrad

    I annoyed someone at work about this. If you’re interested, one of the commonly used happiness measures is the Oxford Happiness Index, which you can find here . This is a uni-dimensional scale, and the questions should give you some idea of the types of things that might go into happiness. It would of course be interesting to look at scales with more than one dimension that encapsulate both self and comparative happiness.

Leave a Reply