Race in the OECD test of adults

This new OECD test, which is sort of PISA for adults, comes with endless documentation of the results, including reports by country. The U.S. report says:

Black and Hispanic adults are substantially over-represented in the low-skilled adult population. Data on race/ethnicity were collected as part of the US national data collection in the survey (comparable data from other countries are not available). While one in ten (10%) white adults scores below Level 2 in literacy, more than one in three (35%) black adults do so, and nearly one in two (43%) Hispanics do. Similar patterns are observed in numeracy: 59% of black and 56% of Hispanic adults score below Level 2, compared to 19% of white adults.  

You expect poor literacy from immigrant groups, but the poor-to-mediocre numeracy of Hispanics in the U.S. came as something of a surprise to me when I was looking at the data in 1991. Hispanics are not like Asians in being relatively better at math than verbal.

Given that blacks and Hispanics are three to four times more likely to have poor skills than whites, these groups are substantially over-represented in the low-skilled population. Half of those with the lowest level of literacy (below Level 1) are Hispanic, and a further one in five are black. Among those with Level 1 literacy skills, about one in four are black and a little over one in four are Hispanic. Among those scoring at the lowest level of numeracy (below Level 1), 31% are black and 37% are Hispanic, while 22% of adults scoring at Level 1 are black and Hispanic

So, there are three Hispanic innumerates for every two white innumerates … in absolute terms. Among adults with bad numeracy skills, there are 50% more Hispanics than whites. That`s not saying that Hispanics are 1.5 times more likely per capita than whites. Instead, there are 1.5 times as many millions of innumerate Hispanics in the U.S. as innumerate whites in the U.S.

That may explain something about why the subprime bubble was so disproportionately concentrated among Hispanics.

By the way, I have no idea how valid the results for this test are. There seems to be a paradox that we should bear in mind when evaluating test results:

  • The more everybody tried their hardest when taking the test, the more likely it is that some people cheated.
  • The less anybody cheated, the more likely it is that some people just blew off trying on the test.