Alan Sloane writes:

The OECD put out a report drawing on their PISA and TALIS data:
http://oecdeducationtoday.blogspot.ie/2014/07/poverty-and-perception-of-poverty-how.html
I notice that it’s already attracted a NY Times op-ed by David Leonhart:

There are a number of things I find strange in its analysis and interpretation but, for starters, there’s the horizontal axis in the chart that’s reproduced in both the original and the NYT piece. As best I can tell the data is actually drawn from Table 2.4A here:
http://www.keepeek.com/Digital-Asset-Management/oecd/education/talis-2013-results_9789264196261-en#page43
So what’s actually being measured for each country is “the percentage of teachers working in schools whose principals estimated that 30% or more of their pupils came from socioeconomically disadvantaged homes”. Then what’s initially interesting in the discussion is how the measures on the vertical axis (a supposedly “objective” measure of disadvantage used in the PISA survey) differ from those on the horizontal, i.e. looking at points that lie significantly above or below the diagonal. So Brazil and the US are obvious outliers, although Singapore, Serbia and Croatia are by a proportional measure also fairly notable. So what caught me first is that this measure is obviously affected by the distribution of disadvantage across schools, e.g. if disadvantage (PISA-measure) is concentrated spatially then you can get a high score on the horizontal axis without having a correspondingly high score on the vertical axis. A highly skewed distribution of school size will also affect things (as I guess will a skewed distribution of teachers, but presumably that’s highly correlated with school size).

The discussion on the third dimension, shown in the bubbles, also seems to me to be dubious, but that’s more complicated.

I don’t really have anything to say on this except that I agree these numbers are hard to interpret.