Sunday, February 18, 2007

Why UNICEF are wowzers

Sorry for not posting earlier as Femme-de-Resistance had promised. The UNICEF report (pdf) helpfully presents its raw data in a format that I couldn't just paste into Excel, so I had to do a lot of typing to understand what was going on.

The first point to make is that trying to produce a "league table" by combining indicators as disparate as relative income poverty and smoking rates is an inherently absurd exercise. Precisely what percentage of average household income is it worth not to smoke? The report tries a sensible approach (normalising every indicator to have the same mean and standard deviation) but still has to decide which indicators are most important.

The report weasels this by saying
Equal weighting is the standard approach used in the absence of any compelling reason to apply different weightings and is not intended to imply that all elements used are considered of equal significance.
which is rubbish. If the overall indicator is a measure of how well a country treats its children, then the weighting of the various elements reflects how important a certain indicator is to measuring this. In any case, the report doesn't weight each indicator equally. It rather arbitrarily divides the indicators into six dimensions, and three components within each dimension. Each component then gets equal weighting. This allows the report to (among other wierdnesses) give a three times higher weight to relative poverty (which gets its own component) than absolute poverty (which is merely one of three indicators of "reported deprivation"), thereby "proving" that children in Greece are better off economically than children in the UK.

So lets look at the six dimensions.

We start, reasonably enough given that poverty prevents children from reaching their potential, by considering poverty. The "component" system exaggerates the weight of a relative poverty measure that the report admits is close to meaningless, because it uses 50% of national median household income as a poverty line. The standard reasons given for why relative poverty matters, given that Bill Gates does not in fact make me any poorer, are
  1. Not being able to afford the same brand of trainers as their peers means children get picked on.
  2. Seeing images of incredible affluence in the mass media makes children envious and therefore unhappy.
Number 1 suggests that local rather than national incomes should be used to define the poverty line, whereas number 2 implies that international standards should be used (which makes relative and absolute poverty look remarkably like the same thing). Both of these changes would benefit the UK's position - a lot of income inequality in the UK is due to a north-south divide, with incomes being more equal within regions.

Next up is "Health and Safety". Here the UK does badly mainly because of a low rate of childhood immunization due to the MMR scare. However, patriotic Britons need not worry - British children are actually healthier and safer than foreigners, with a death rate 41% below the OECD average. Based on weighting vaccinations equally with not dying, the UK places 12th out of 21.

Then on to "Education". English children do better than the OECD average in native-language literacy, maths and science. But the UK nevertheless places 17th out of 21, because the "component" trick is used to give participation in further education (with no quality measure included) twice the weight given to the quality of primary and secondary education. "Achievement" is only one component, whereas participation in FE counts under both "Beyond basics" (where it gets a whole component to itself) and "Transition to employment" (where it reduces the number of 15-19 year olds not in employment, education, or training). FE in the UK is a national scandal, but it isn't 50% more important than getting the basics right. The last indicator in this dimension shows another unhealthy pattern in the report - it treats a value judgement by the authors as an objective measure of well-being. Countries are marked down if not enough children aspire to highly skilled work. Given that not everyone will get a highly skilled-job, this raises the question of whether aspirations which may well be unrealistic are an unalloyed good thing.

The dimension of "Relationships" is where the report starts getting silly. One of the indicators (heavily weighted, getting a whole component to itself) is "percentage of 11,13 and 15 year olds finding their peers kind and helpful". This is a completely subjective measure, and belongs in the dimension on "subjective well-being". The other two indicators both reflect value judgements - that children are best brought up by two biological parents, and that families should eat together. While I agree with these judgements, plenty of people dont, or don't think that these things are important.

Next comes "Behaviours and risks", on which the UK comes bottom by a country mile. Equal weighting has some interesting consequences here - using cannabis even once is considered as bad for a child's well-being as getting addicted to tobacco, or getting seriously drunk twice. My favourite value judgement is taking off points for 15-year olds having sex. When I was 15, sex would probably have improved my well-being. I imagine that is true of most 15-year olds, just as it is true of almost all adults. Apart from laughing at this point, I can't argue with the overall conclusion - that British children are more likely than foreigners to do the kind of things their mothers tell them not to do.

Last is "Subjective well-being". Wowze wowze wowze. Seriously, there is a whole field of happiness economics which starts with the assumption that asking people "How happy are you?" generates useful information, and goes downhill from there. As the linked article points out, individual answers to this question tell you more about who has a sunny and optimistic disposition than who has a good quality of life. The wowzer claims that when you do large-sample comparisons between countries then disposition should cancel out, but this assumes that there are no differences between national cultures. Since the English are notoriously a nation of grumblers, it is unsurprising that we come out bottom. I don't see what this says about how well our children are treated. My favourite riposte to the happiness wowzers is figure 5 of this report which shows subjective well-being being completely flat throughout the economic collapse of the 70's, the social collapse of the 80's, the unemployment and reposessions of the early 90's, and the recovery from all the above under Major and Blair. Perhaps it just isn't a very useful indicator.

Just for fun, I tried reanalysing the data in a slightly less wowzerish way. I excluded all the purely subjective indicators, as well as the ones based on controversial value judgements. I also weighted all indicators equally, without dividing them between components. This left three themes: poverty, health (including behaviour indicators like smoking) and education. The UK comes out below average in all three areas (only just in the case of education), but is nowhere near the bottom. You might think that the report was rigged so that the UK did badly.

For the real cynics out there, Peter Preston points out that the report's author, being a Brit involved in campainging on a number of issues that the report deals with, has an incentive to make the UK look as bad as possible because it gives his campaigning more impetus. I'm not that cynical - I just think that the report reflects the subconcious biases of the kind of authoritarian lefties who love to work for the UN and think that individualism is an ilness.

Labels: , , ,


Post a Comment

Subscribe to Post Comments [Atom]

<< Home