Monday, June 17, 2013

Poverty matters more in the US

I'm on a bit of a wonkish kick, and I'm loving the PISA 2009 post-game analysis put together by the OECD about the relationship between poverty and the PISA data. Here's a table that I found illuminating from that report.
[Click to biggify.]

To give this some context, the question that everybody in the US wants to know is: when you control for poverty, does the US still ends up performing middling on these international exams?

There are a variety of ways of interpreting that question, but here's one: there's a spread of scores in the US, and there's a spread of socio-economic status of students. How much of that spread of scores could you have predicted if all you knew was the socio-economic spread? If you can predict 100% of the spread of scores based only on your knowledge of the socio-economic spread, then what matters is the poverty. If you can only explain 20% of the US's variance in scores based on the socio-economic spread, then there's more to the spread than the poverty.

* There are a lot of interesting things you have to do to tackle that question. You need to devise a way to measure poverty (or socio-economic status). You need to decide how you want to "control" for poverty. I'm eliding the decisions made in this particular report for now.

Anyway, the OECD did this analysis and here's what I find cool. On one end of their ranking is the US. In the US, if you just knew the socio-economic spread, you could've predicted about 80% of the spread in US reading scores.

So poverty matters, right? Ah, but check out the top of that table:

In Hong Kong the poverty spread only explain about 20% of the spread in scores? What gives with that?

Poverty sucks, and poverty makes learning harder. But, internationally, poverty isn't a singular, over-riding factor in determining the outcome of a child's learning. But in the US, poverty largely does play that role.

All of which makes for a confusing national debate. If you look only at the US, then it looks like poverty is close to fate. But if you look internationally, it seems as if poverty is only one major, important factor out of several in determining what sort of education a young person can expect.

So why is it that poverty matters more in the US? I don't know. I assume it's because in the US there are all sorts of other negative factors that happen to be coupled with poverty.

But don't let anyone ever tell you that this stuff is simple. It's complicated.


In the comments, Sue asks, "Maybe the schools aren't segregated as much by income in Hong Kong."

The OECD has a figure for that:

The US actually has a higher percentage of students in mixed income schools (as defined by the OECD, of course) than Hong Kong.


  1. Maybe the schools aren't segregated as much by income in Hong Kong.

    1. Hey Sue,

      I updated the post above to include a table from that same OECD report that states that the US has a higher percentage of mixed income schools than Hong Kong does.

  2. Dennis Condron wrote an article about this that people might want to check out at It should be open access.

  3. It makes a big difference whether you are comparing schools or individual students. If I am reading the report correctly, a large percentage of the variation between SCHOOLS in the US is explained by some sort of poverty index, but relatively little of the variation between students is explained by this index. See page 53 for a scatter plot of individual students (very weak correlation).

    This makes sense. If you randomly select a high income and a low income student, it would not be terribly surprising if the low income student was the higher performer. On the other hand, if you took two large groups of randomly selected high income & low income students, it would be surprising if the average for the low income group was higher. Who knows whether the poverty causes the scores, the scores cause the poverty, or there is some other factor responsible for the association.

  4. Isn't the explanation very simple: The PISA scores get worse the more children are left behind. This may or may not be connected to poverty.

    You can organize your school system in a very strict way, so that every child gets the same education until late in its teens, or you may fail to do so.

    Moreover, the attitude of the society to education can be such that even poor parents value schooling, or it can be oriented towards a caste system where people "know their place".

  5. Ignore poverty for the moment and just look at high and low performers. The US bottom 10% on PISA score better than the OECD average for reading. The top 10% do not. In math, the bottom 10% is also more competitive internationally than the top 10%, though both are well below the OECD average. Of course this means that the gap between the bottom 10% and top 10% in the US is lower than the OECD average. So, if every country could exclude their low performers, the US would actually compare less favorably to the OECD.

    Maybe low performers are more likely to be in poverty in the US (measures of poverty seem a little iffy). But our low performers don’t look much different from those in other countries. As you say, it is not a simple issue to think about.