What exactly does the “score” on a standardized test tell you? Higher is better, but is 100 twice as good as 50? 10 percent better? Three times better? To get around this conundrum, which economists call “ordinality,” Timothy Bond and Kevin Lang translate test scores into more meaningful measures, such as years of schooling. Then they use their translations to take a more careful look at the Black-White test score gap in K-7. (The authors also make important corrections for the unavoidable measurement error in test scores.) The results are more accurate and more useful than what we’ve known before, but they’re not going to make any of us happy.
- The race gap is large and roughly constant from early years through higher grades.
- If you measure the gap in years of education, then the race gap in math in early grades is larger than the gap in the actual years of education.
- If you measure the test gap in terms of future earnings, the gap is smaller than the gap in actual future earnings. This suggests–not surprisingly–that part of the racial earnings gap is due to post-school factors.
These results tell us that the race gap can largely be “explained” by differences that are observable in kindergarten. The authors carefully point out that this does not mean that what happens after kindergarten doesn’t matter. It may be that students with low scores in kindergarten attend, on average, not very good first grades. Later schooling interventions could change the initial race gap…but they don’t.