Friday, 12 March 2010

Just one indicator please!

So one day you decide to cut through all of the nonsense and insist that a single indicator is defined to represent how well our children are progressing. You have no time for half-measures, and your goal is a global one. All children, all societies, one indicator. As you wait impatiently in the canteen of the UN headquarters an important and flustered looking person (perhaps Ban Ki-moon) comes in and asks if it is you that has set this challenge. A positive reply, a concerned frown, and the gentle ushering of your good self to the podium leaves you in the position of making the first suggestion. What do you say?

“Income poverty, ah no wait, life satisfaction…errr, infant mortality… no, no, wait, low birth weights!” Everyone is thoroughly confused.

Child income poverty is a good choice. Everyone knows kids need money. It is easy enough to measure objectively, though from time to time families can mis- or under-report their income. When it comes to children is family income enough, or should we be talking about pocket money? Can we be sure household money is spent on the children? Income poverty is policy amenable, quick to adjust (if you’re willing to take taxes) and policy efforts are easy to evaluate and are likely show up before the end of a government term (so politicians might like this one!). But then how do we measure it: relatively or absolutely? Does a dollar mean the same to two different children on either side of the world, in developing or underdeveloped countries… certainly not. Would it make sense to compare globally… oh dear, probably not. Even before we begin to address the dreaded issue of equivalisation we have problems.

Life satisfaction says it all, right? How is your quality of life? First, everyone can answer that – and it can capture experiences of health, education, poverty and more. But ask it of children and you’re stuck. First of all there is general perception that children are easily swayed, their quality of life depends on whether they have had enough sweets that day (I am sorry to report I have had to challenge this perception in the past). ‘Soft measures’ abound in child well-being research – an easy in for the critic, and a risk that your hard work measuring this concept comes to nothing. Also, you can’t ask it of all children, how do you ask a two-year-old how they rate their life? What is the context in which any child of any age will respond to this question - compared to yesterday, last week, last year, in general, or compared to their peers or the people living on their street? Uh oh, more problems.

Infant mortality or low birth weights? Infant mortality may be considered an iceberg indicator measuring the experience of a few to represent the risk to many. But how do we identify those that were at risk, and what might be considered a natural level of infant mortality? Many low birth weight studies suggest low birth weight children’s long term earning and learning is impeded, but it is a measure defined by a threshold and the lack of a distribution limits our policy response as we don’t know who is at immediate risk or severe risk. Moreover infant mortality and low birth weight measures interact. The better we are at saving preterm births, the lower the infant mortality and the higher the low birth weight… oh dear, more problems.

By no means exclusive, these are some of the challenges we face when we try to define indicators for child well-being. I have not discussed culture, children’s rights, complementarity between indicators, prioritisation or weighting, the longevity of an indicator, or children’s participation even. Nonetheless the political pressure to identify catch-all indicators is real. Recently the European Commission has produced a report designed to identify a reduced set of indicators for monitoring child well-being to add to the so-called Laeken indicators set.

When few indicators are desired, selecting those indicators is a difficult job. We need to get it right first time. Any change to key indicators can mean we quickly lose our ability to create times series, and therefore restrict our ability to measure progress.

Take yourself back to the beginning, back to the podium… What do you say?


  1. This comment has been removed by the author.

  2. Nice post Dominic. A few years ago I was involved in some work in Australia looking at an indicator there that was hitting the headlines " Number of children in jobless families". The numbers were suprisingly high and there was a good deal of concern. But when we started to think about this, we asked "hang on .. what exactly is this indicator supposed to be telling us? B ecause at first glance it seems to be suggesting that these kids are in poverty (in which caase why don't we measure that directly) and also that they don't have any working adult role model in their lives (in which case why not measure that directly)." Because think about it... neither is true. There are kids in jobless families, who may live in a house where another adult is working... so they might well have a working role model. And there are families who might be jobless but also well off (kids living with relatively wealthy retired parents or grandparents etc). Then of course there are people every month - short term contractors say - who are unemployed only briefly (and at the time of the labour force survey) and who are not necessarily poor. Anyway, the long and the short of it was that once we disentangled the data and became clearer about the problem we were trying to measure, the story was rather different.

    So its not easy I agree. It is very important to measure a genuine problem (measure an outcome), rather than pick an indicator that seems to suggest a problem.


  3. Interesting! What about Infant Mortality Rate among the lowest socioeconomic quintile? Or, even better -probably - the ratio between first and fifth quintile Infant Mortality Rate? That would give us a lot of information about the quality of life and equity of that particular society and I think data is already available or, at least, rather easy to obtain.