Top NAEP Cities have lowest poverty
I just can’t get off the topic of poverty and school performance. In the Hechinger Report, Jill Barshay has looked at the recent NAEP scores of 21 urban cities who participated in the NAEP Trial Urban District Assessments in her December 18, 2013 Education By The Numbers blog. She created an enlightening table of results that rank the cities by percent poverty and shows their grade 4 NAEP math scores. Interestingly, she also shows the math ranking. Check out the table in the blog above.
Math scores decline as poverty increases
These cities, all urban districts with populations over 250,000, range in poverty from 56 percent (Charlotte, N.C.) to Cleveland, with 100 percent poverty. A careful look at math rankings compared to the percent of poverty shows that, with but a few exceptions, as poverty increases, the comparative math rankings decrease. While there are a few exceptions (Boston, for example, ranks 15th in poverty but 5th in math—far above the usual pattern, while Washington, D.C. ranks 8th in poverty but 14th in math performance.
Correlation of math scores and poverty
Ms. Barshay suggests the correlation “is not as tight as you might guess,” but unfortunately does not provide any explanation of what analysis she did. Did she compare the 1-21 ranking of poverty to the math ranking, the percent poverty to the math ranking, the math scale score to the % poverty, or the math score to the poverty ranking? What was the correlation coefficient she came up with? Each of these correlations would produce a different number.
Based on my own decision to use the underlying raw data rather than the ranks (which have less detail than the raw data) I ran a correlation between the math scores and the percent poverty in Excel using the CORREL function, which yielded a correlation of -0.74564, which rounds nicely to -.75. In the social sciences, of which education is one, this is considered a very strong correlation.
Correlations are not indicators of causation. To give a simple example, think about correlating age with income. The older you are, up to retirement age, the higher your income. A graph of the relationship between income and age is quite flat for children—income doesn’t start rising with age until about 18-21 years of age, but then it rises steadily until retirement age, when it drops suddenly and levels off. The correlation between age and income for folks from 21-65 is very high, but it’s not causative. Simply put, as employees have more experience and better jobs, on average they earn more money. Likewise, you can’t say that poverty causes the low math scores, but the association between the two variable is very clear and very strong.
Role of poverty too often ignored by education reformers
Current discourse on the role of poverty in school results suffers significantly from inattention based on some distorted political correctness. ‘All children can learn’ is a phrase that has been pushed into meaning that every child, no matter how disadvantaged, can do just fine. NCLB started the chaos with the impossible goal of 100% proficiency for all. With rare, and I mean rare, exceptions, the average children of upper class Scarsdale, NY will always outperform the average children of New York City. And by substantial margins. No amount of platitudes from critics of public education will change this level of inequality. Inequality remains the biggest driver of low performance in America. I’ll offer more support for this in future blogs.