• Predicting college completion

    I’m a pediatrician.  I can’t help myself. Here’s one more chart from The State of Working America:

    This is the percent of children who finish college both by eighth grade test scores and income. Let’s acknowledge that students with high scores (yellow) are in general more likely to complete college than students with low scores (red).

    But note how much income dominates that determination. A student in the top 25% of income who has low scores is more likely to complete college than a student in the bottom 25% of income who achieves high scores. Moreover, if you’re in the bottom 25% of income, you’re not likely to complete college at all, regardless of scores. The opposite is true for those in the top quarter of incomes.

    Think about that as we fixate obsessively on test scores as a way to measure and predict future academic performance.

    Economic Policy Institute. 2011. Incomes matter more than test scores for college completion. The State of Working America. Washington, D.C.: Economic Policy Institute. Feb. 14, 2011. <www.stateofworkingamerica.org/jobs/figure224>.

    Comments closed
    • Interesting, I wonder what the cause is. Is it because they need to work to help the family?

      My first year in college I stayed and ate at my parents house and did the same my first summer, other than that I got no help at all from my parents. They did not contribute any money at all to me while I was in college. Also the less money one’s parents make the more financial aid one is eligible for so I do not see the expense of tuition or room and board as the biggest reason.

      It could be that students think that college will cost more than it does. It could be that they are more accepting of jobs that do not require a college degree.

      I wonder if you looked at those who get a sure income degree like engineering, accounting or nursing if you see the same relationship. That would show if wealthy children are just getting degrees in communications and art history just because they can more easily afford it.

    • Having seen this is my extended family, who are for the most part “low income,” my experience is that these parents do not expect their children to attend college just as they did not.

    • Most parents today push for their kids to attend college but many of them cannot afford the cost and put the burden on the child, and the child must apply for student loans and then they are in debt before they even reach the workforce. Some graduates do land good jobs but the majority of the graduates end up working the same jobs as their parents worked. A lot of low income families see this and do not encourage their kids to attend school.

    • The chart is based on Table 21 (page 50) in this U.S. Department of Education report. The test is described on page 21 of this related report. I think the following additional details are useful:

      Although the 8th grade students were given tests in reading, math, science, and social studies, the chart is based only on the math test. The math test contained 40 questions to be completed in 30 minutes.
      The “income” categories used in the chart are actually socioeconomic status (SES) categories based on a composite of parental education and occupations, and family income.
      The “high” and “low” categories, both for test scores and for SES, are quartiles, and the “middle” category combines the two middle quartiles.
      The test score quartiles are based on the entire 8th grade sample without regard to SAS.

      The association between SAS and college completion is striking but perhaps not surprising. A rich, dumb kid has a slightly better chance than a poor, smart kid of completing college. And a rich, average test score kid has a better chance than a middle-SES, smart kid of completing college. It appears our society is squandering a lot of brain power. This argues for financial support at least for low and middle income, high test score students.

      To me, the association between test scores and college completion is also impressive and much more surprising. Keep in mind that the test was a 30-minute math test administered in the 8th grade. Nevertheless, even among high-SES students where, presumably, finances are not an insurmountable obstacle to college completion,  high test score students were 2.45 times as likely as low test score students to complete college, and were 1.45 times as likely as middle test score students. Among low-SES students where finances are expected to be an enormous impediment to college completion, high test score students were 9.93 times as likely as low test score students to complete college. I am so surprised I must emphasize it again: this was a 30-minute math test administered in the 8th grade. Think about that as we fixate obsessively on test scores as a way to measure and predict future academic performance. This argues for identifying, developing, and financially supporting high potential kids from an early age.