Tuesday, January 15, 2013

Family background matters, so measure it better.

A few months ago, the Census Bureau released data based on a relatively new, more sophisticated measure of poverty.   The old measure had been in place since the 1960’s and did not account for the realities of today’s living expenses.  The new measure considers housing, medical, and child care costs and does a much better job adjusting for support received through federal assistance programs.   In areas with high costs of living like California and Hawaii, the new measure classified substantially more residents as poor, while the reverse was often true in areas with lower costs of living. 

This new measure is by no means perfect, and it certainly does not do anything directly to help poor families in the U.S.  But this measure may allow policy analysts to better assess the needs of American families and the relative effectiveness of safety net programs.  The change is solely on paper, but it is an important change nonetheless.

The K-12 education sector is long overdue for improvements in how it routinely measures the social background of children.  More often than not, a student’s participation in the free and reduced school lunch program and his or her LEP status are the only available indicators of family background.  Although additional indicators are sometimes collected for research or special programs and assessments, free/reduced lunch and LEP tend to be the only measures that are available for all schools in regular enrollment data.

Free/reduced lunch is a lousy indicator of socioeconomic status for a couple of reasons.  First, it classifies all students into just one of three categories (free lunch, reduced lunch, no lunch support), losing valuable detail in the process.  With this approach, a family of four making $28,000 per year will be indistinguishable from a family of four making $14,000 per year, as both would be classified as free lunch.  Second, free/reduced lunch is based on income primarily, and income by itself is not a very good indicator of social class.  In Class and Schools, Richard Rothstein points out how the use of income via free/reduced lunch as the primary measure of socioeconomic status can lead to the misrepresentation of some schools’ populations.   One school that was nationally recognized as being both high poverty and high performing was actually a public school where many Harvard and MIT graduate students sent their children.  True, graduate students don’t make much money, but few sociologists would regard this group as a high needs population. 

Social scientists have used hundreds if not thousands of different indicators to measure class and socioeconomic status, and the measures will often vary depending on available data.   However, a handful of variables emerge more often than the rest, due to both availability and their quality as predictors of outcomes in the social sector.  If I had to pick a single variable to add alongside family income, parental educational attainment would be a good choice.  A common way of representing SES in richer datasets is to combine information on income, parent education, and occupational status or occupational prestige (e.g.).  While converting occupational status into a number can be tricky, it’s a bit more straightforward for parental education levels. In many cases, measures of parent education are even reduced to maternal educational attainment due to the prevalence of single-parent households.  Thinking back to the Boston public school that enrolls the children of Harvard and MIT Ph.D. students, it is easy to see how a combination of income and parental education levels would give you a much more accurate sense of the average socioeconomic status of some families. 

I am not a lawyer, and I don’t know what legal justification the feds or states would need to collect additional personal information from parents; but, from a researcher’s perspective, the case is easy to make.  The link between social background and academic achievement is well established, but the debate over the extent to which these links should influence educational policy continues.  Achievement gaps between racial and socioeconomic groups remain large, and school segregation along these lines may be getting worse.  Meanwhile, wage, wealth, and income inequality in the U.S. continues to worsen, as it has been doing since the mid 1970’s.  In this context, there is substantial need for better measures of students’ social background, particularly given the shortcomings of current measures. 

Moreover, the weaknesses of free/reduced lunch as a socioeconomic indicator are not just an inconvenience for researchers these days.  For better or worse, many states and districts are now using statistical models to influence the retention, tenure, and promotion decisions of teachers.  Better background variables on students may help improve these models.  With all of these factors in mind, one could make the case that looking beyond free/reduced lunch is not only in the best interest of federal and state departments of education but also that it is their responsibility to do so. 

No comments: