Preliminary and Incomplete Comments Welcome Early Test Scores , Socioeconomic Status and Future Outcomes

This paper uses data from the National Child Development Survey of an entire cohort of children born in Britain in one week in 1958. We show that test scores measured as early as age 7 have significant effects on future educational and labor market outcomes, even after a wide array of observable characteristics have been controlled for. The effects are similar or larger than those of other observed characteristics of the children. Perhaps unsurprisingly, it is easier to predict future test scores given past test scores, than to predict schooling attainments or labor market outcomes. A striking result is that among men, early test scores are better predictors of wages and employment at age 33 than at age 23. Among women, one observes the same pattern for wages, but it is generally more difficult to predict employment at age 33 than at age 23, presumably because many women take time out of labor market careers to care for children. We test several hypotheses about the interactions between socioeconomic status and high or low test scores at age 7. Compared to other children (and in terms of educational attainments), lowSES children reap both larger gains from having high age 7 test scores and smaller losses from having low age 7 test scores. The opposite is true among high-SES children who suffer large losses from low scores and small gains from high scores. The hypothesis of systematic test bias against low-SES students offers one way to interpret these results. Janet Currie Duncan Thomas UCLA and NBER RAND and UCLA Dept. of Economics, UCLA 1700 Main St. 405 Hilgard Ave. Santa Monica CA 90403 Los Angeles 90095-1477 dt@rand.org currie@simba.sscnet.ucla.edu