Arnold Kling  

Happiness Research

Liberal Fascism Watch... Useful Research...

Helen Johns and Paul Ormerod write,

One could conclude from the lack of correlation over time between aggregate happiness and almost any other socioeconomic variable of interest one of two things. Either that attempting to improve the human lot through economic or social policy is futile, or that happiness data over time is an extremely insensitive measure of welfare. The evidence points to the latter.

...There is a body of research on happiness which is, from a scientific perspective, much more securely grounded. This is based upon the analysis, not of aggregate happiness data over time, but of so-called panel, or longitudinal, data, which tracks specific individuals over time. It shows that stable family life, being married, good health, having religious faith, feelings of living in a cohesive community where people can be trusted, and good governance contribute to happiness. Chronic pain, divorce and bereavement detract from happiness.

I agree with the first paragraph, but I have my doubts about the second. I still think that the question "How happy are you?" is going to deliver unreliable answers.

I cannot measure my happiness in absolute terms, so I have to answer in relative terms. I have to think, "Compared to Joe, I'm happy, but compared to Susan, I'm not." Or I have to think, "Compared to when I fell and broke my wrist, I'm happy, but compared to when I fell in love I'm not."

And am I thinking of my momentary happiness, my happiness over the last hour, or my happiness over the past year? If it is momentary happiness, what if I have spent the last minute reflecting on bad childhood memories? Should I say that I am unhappy? Maybe 10 minutes from now I will be reflecting on fond memories. Should I then say that I am happy?

Suppose we get past that. Suppose that research shows in some reliable way that most people are happy doing X. Is it not possible for people to have different tastes? If research shows that people who eat tuna fish are happy, does that mean I should eat tuna fish?

I hate tuna fish.

But I like Bruce Charlton, who gave me the pointer.

Comments and Sharing

COMMENTS (8 to date)
conchis writes:

"I have to answer in relative terms. I have to think, 'Compared to Joe, I'm happy, but compared to Susan, I'm not.' Or I have to think, 'Compared to when I fell and broke my wrist, I'm happy, but compared to when I fell in love I'm not.'"

Well that's interpersonal and intrapersonal-intertemporal comparability right there. So what's the issue? As Ferrer-i-Carbonell and Frijters have shown, you don't even need interpersonal comparability for much of the panel analysis to work.

The only way an issue could arise out of this is if you think people only pay attention to how happy they think they are relative to others, and that this benchmark shifts over time. That's a potential problem, but I find it less plausible a priori than the idea that we use internal reference points ("How happy am I relative to yesterday"). It's also a problem that should lead to particular patterns in the data that don't really seem to be borne out. (E.g. it shouldn't be possible for the distributions of happiness within populations to change much over time, but they sometimes do.)

P.S. For the avoidance of doubt, I'm not suggesting there are no issues with happiness data. They're noisy, subject to survey artifacts, and potentially biased in predictable ways in specific situations. But (at least in the context of panel data - I pretty much agree that the macro level stuff is mostly rubbish) you can work with those.

P.P.S. I find it interesting that that Johns and Ormerod ignore an entire body of panel data evidence that supports the idea that relative income matters, and appear to write the idea off on the basis of evidence that they themselves accept is inferior. Something's fishy there.

Patrick writes:

"I cannot measure my happiness in absolute terms, so I have to answer in relative terms. I have to think, "Compared to Joe, I'm happy, but compared to Susan, I'm not." Or I have to think, "Compared to when I fell and broke my wrist, I'm happy, but compared to when I fell in love I'm not."

- Perhaps others are more capable of doing this than you are? It seems odd that one of your main objections is so specific to your own personal way of looking at happiness.

dearieme writes:

I'm baffled that anyone could for a moment entertain the idea that happiness is one-dimensional.

Barkley Rosser writes:

On the relative income/happiness issue, I quote Robert Frank, who likes to tell this one in talks, although it is far older than he is:

"He is happy who makes more money than his wife's sister's husband."

Three points on data. One is that I think the panel data is the only sort that is remotely reliable. One is following specific individuals over time. Thus, maybe they do not know if they are happy or not, but at least you are capturing their own variations over time. After all some people are just generally happier than others, and also the cross-cultural comparisons are seriously bogus, there are major cultural differences in what people in different countries say about how happy they are or not (No, I shall eschew the tempatation to descend into ethnic jokes here).

On this other matter, the research now, or at least the better research, tries to distinguish moment-to-monent "happiness" (how do you feel right now?) versus presumably longer term "satisfacion." These do tend to be correlated, and strongly, but not perfectly, and some of the differences are relevant to the income issue. In short, moment-to-moment happiness seems especially disconnected from income or status. It has much more to do with who one is with and what one is doing. Overall satisfaction does seem to be influenced by one's relative economic status and position, as well as some of these other things like health and marital status and so on.

Chuck writes:

Happiness is simple to experience, difficult to understand. A blog about it that I enjoy is

www dot happiness-project dot com

Nathan Smith writes:

If you want some basis for a happiness scale that's not just comparative, how about suicide=zero? If you want to commit suicide, you have no happiness. Then, among those who don't want to commit suicide, let them compare themselves to each other.

Derek Scruggs writes:

Maybe it's better to think in terms of risk factors. Not all smokers will get cancer or heart disease, but the risk is much higher. So as a general rule it's a good idea to avoid cigarettes, even though you don't know for sure it will make a difference in your particular case. The outliers who don't get cancer don't therefore negate the research.

Similarly, there are various behaviors, foods, experiences etc that are more likely to lead to happiness, so why not give them a try? The good news is that you'll probably know pretty quickly whether it's working, certainly much faster than knowing whether smoking will give you cancer.

Sadly, I don't know what to do if it turns out that smoking makes you happy. :)

Tiger writes:

Actually, tuna fish will make you happier despite your distaste:

Comments for this entry have been closed
Return to top