The Misuse of Statistical Correlations

In this Kappan article, Ben Levin (University of Toronto) says it’s true that poverty, parental education, family interactions, and ethnicity are strongly correlated with students falling behind in school, dropping out, and not going on to post-secondary education. But all too many educators and parents make the mistake of using these aggregate correlations to predict that a particular eight-year-old will not be successful in life. 

This is a serious error, says Levin: “Many studies have found that a large number of students defy negative expectations based on their backgrounds.” One Canadian study found that nearly 40 percent of young adolescents with very low reading skills were in post-secondary education six years later, and an American study of failing third graders found that more than 75 percent eventually graduated from high school, including 70 percent from low-SES families. “In every study of this kind,” he says, “a significant number of students who seem to have everything against them end up having good results.”

It’s not that poverty and other factors aren’t important, or that we shouldn’t be working hard to alleviate social inequalities, or that schools can by themselves overcome entering disadvantages, says Levin: “What is important to keep in  mind is that we just don’t know how people’s lives will turn out… So, while many people in jail come from high-poverty backgrounds, most people who grow up in poverty… don’t end up in jail. For educators, there is a very heartening message in this research. What it says, very clearly, is that our work matters.” Small actions by teachers – or just doing their jobs well – can make a tremendous difference to struggling children.

“All this means that schools and educators should be very cautious about thinking we can predict any student’s future,” Levin concludes, “and even more cautious if that belief leads us, or the student, to lower expectations.”

“The One-Legged High Jumper and the Perils of Prediction” by Ben Levin in Phi Delta Kappan, October 2012 (Vol. 94, #2, p. 74-75), www.kappanmagazine.org; Levin can be reached at ben.levin@utoronto.ca. 

From the Marshall Memo #458