I’ve written a few times on measures of wellbeing on Troppo. For instance here and here. (In fact, reviewing it, I can’t find both of my articles for New Matilda on the Australia Institute’s GPI, so here they both are (pdf).) As ever Troppo was hip before the world caught up, but catch up it has and now everyone and their dog is getting a wellbeing index.
The French had a few Nobel Laureates drop in, the Canadians are off and running and the UK Conservatives (a tribute to Tony Blair as the Rudd/Gillard Govt has been a tribute to John Howard) want their index of wellbeing. And the OECD has just launched a very natty website where you can run the numbers on all the OECD countries over eleven thematic areas.
When you get into trying to set these indices up various things jump out at you. The first is the mountain of literature on it all. Another is the many different methodologies you can employ – should you measure things ‘objectively’ or ‘subjectively’? Then there’s how they relate to each other – are the measures commensurate? And if you’ve got several measures of some aspect, how should they be aggregated and do they make sense together? It’s amazing how much the various indicators are just added up and given equal weight. Still it’s hard to think of any obviously superior approaches.
Anyway, the prompt for this post was reading the approach on the new OECD website to measuring health. The OECD have used the women’s clothing store Susan’s slogan “this goes with that”. They’ve slapped together two simple measures – and why not one objective one and one subjective one? The first is life-expectancy at birth and the second is self-reported health status. Sounds fair enough. The problem is that a predominant influence on self reported health status is age. So those countries with older populations report lower health. There are other things going on of course, but ageing is obviously a large part of the self-reported health index. Japan which has the highest life expectancy at birth – a whopping 82 – and which has major ageing issues has the second lowest reported health status. By contrast, Australia has high life expectancy but relatively low ageing, at least compared with some of our OECD peers. The ageing Mediterranean countries are likewise high on the life expectancy index and low on the self-reported health index. I would have liked to see the self reported health index corrected for age.
“Still it’s hard to think of any obviously superior approaches.”
How about: keep the measures separate, or, if lots of them are highly intercorrelated, then find the meaningful principle components and report those instead? This reminds of intelligence research (and numerous other things) — everyone agrees that you can be smart in all sorts of different ways, yet somehow everyone wants a single measure of things so you can calculate league tables and so on to feed to an ignorant public. I’ll be interested to see how this data gets (mis)interpreted.
The purpose of removing age as a confounding variable is what though? To identify the world’s worst nation of hypochondriac whingers?
Name and shame!
FDB, the purpose of having age adjustment would be to measure health outcomes on a comparable basis. I thought that was the point of the exercise.
I guess it is doubly hard because of the explosion in new illnesses we didn’t previous realised we were suffering. Very interesting story about this on the Health Report.
I suspect that the age correlation is easier to see than to measure. Are old Aussies bigger or smaller whingers than Yanks or whatever?
It’s heartening to see the strides some countries have made since 1960 in life expectancy at birth (eg, India, Indonesia, Brazil, China, Turkey, Mexico, Chile, and South Korea). Of course, things haven’t been so rosy in Russia since 1960…
You could look and compare self reported health status at 10 year cohorts. EG .jp 30 – 39 year olds with .au 30-39 year olds.
Self reported and mortality aren’t all that good measures.
People could have high rates of age related macular degeneration as one example and still feel good health but it will lead to eventual legal blindness – but not effect mortality rates.
Thx FXH – I can certainly see that self reported mortality might not generate much of a result!
FDB’s right. How you self-report your health will have a lot to do with the mores of the country you’re in. Some have an Erewhon-like attitude (bad health as a moral failing), others are indeed hypochondriac whingers. So cross-country comparisons are affected by cultural mores that, unlike age, you can’t easily control for.
FXH is also right – you might be able to use difference-in-differences to wash out the fixed effect. But I suspect this needs data that isn’t gettable, and in any case the mores may not be a fixed effect over time – social attitudes are surprisingly plastic (which is one reason national stereotypes are a treacherous guide to them).