On her “Bridging Differences” blog, educator Deborah Meier began a discussion with Mike Petrilli of the Thomas B. Fordham Institute, on whether urging disadvantaged women to defer childbearing until they had sufficient income (whether from work or marriage) to adequately support their offspring would result in better outcomes for those children. This, in turn, led to an extended discussion (not on the blog, but widely circulated among some education policy experts and commentators by e-mail) about whether alleviating poverty would raise student achievement, whether alleviating poverty through tax reform or income redistribution might be effective for that purpose, whether poor children in the United States have worse outcomes than poor children in other countries, what the best way might be to calculate poverty levels across countries, and whether school reform in the absence of alleviating poverty can be significantly effective.

The shortcoming of this discussion is that because Americans are averse to acknowledging the concept of social class and hold to a widely shared myth of unrestricted mobility (that is less and less reflective of reality), we tend to use the term “poverty” as a proxy for lower social class status. This shortcut causes great mischief in educational policy. Lower class children are not only characterized by having families with low current money income; they also have a collection of interacting characteristics, each of which affects the ability to learn.

Years ago, the Heritage Foundation published a report called No Excuses, by Samuel Casey Carter. Among others, one school it found enrolled a majority of children who were eligible for subsidized lunches yet who still had high achievement. According to the report, this (along with other, equally flawed examples) proved that poverty is no bar to high achievement. The school in question was in Cambridge, Massachusetts, and it turned out that the students mostly had parents who were graduate students at Harvard or MIT, whose stipends were low enough that their children were eligible for the lunch program.

Of course, how much money a child’s parents earned last year (the qualifier for the lunch program) does not itself impede learning. But poverty is a good proxy, sometimes, for lower class status because it is so highly associated with other characteristics of that status.  Lower class families have lower parental literacy levels, poorer health, more racial isolation, less stable housing, more exposure to crime and other stresses, less access to quality early childhood experiences, less access to good after school programs (and less ability to afford these even if they did have access), earlier childbearing and more frequent unwed childbearing, less security that comes from stable employment, more exposure to environmental toxins (e.g., lead) that diminish cognitive ability, etc. Each of these predicts lower achievement for children, but none of these (including low income) itself causes low achievement, and lower social class families don’t necessarily have all of these characteristics, but they are likely to have many of them. Sociologists used to define social class solely by assigning a reputational status to a father’s occupation. That no longer is as useful, but sometimes it might lead to less mischief than “poverty.”

One problem that has puzzled observers of education for some time is the fact that after controlling for poverty, black achievement is still lower than white achievement, and some conclude from this that they have now disposed of background characteristics and can be certain that an important cause of the achievement gap must be poorly qualified and motivated teachers and/or low teacher expectations. Without denying that some teachers are poorly qualified and motivated, and do have low expectations, and that this should be a target of educational improvement efforts, this explanation as a major cause of the achievement gap is mostly false, because black children have many social class characteristics that are different from those of white children whose parents have the same current-year income, even when all of them—black and white—have excellent teachers. Some years ago, in the Jencks-Phillips book, The Black-White Test Score Gap, Meredith Phillips addressed this issue, and showed that adding very few other social class controls, besides family income, explained much of the black-white gap. Patrick Sharkey’s new book, Stuck in Place, shows that black children living in high poverty neighborhoods are more likely than white children to have parents who also lived in high poverty neighborhoods. Black poverty tends to be multi-generational, white poverty tends to be episodic. Others have shown that permanent (i.e., multi-year) income explains more than current year income.

In short, the poverty rate itself is not an adequate explanatory factor. Lower class status better explains what we puzzle about. Unfortunately, it is a lot harder to find simple data to compare children across countries by social class status than to pick an easy (but mischievous) one like “poverty rate,” no matter how it is adjusted. If I had to pick a single one, it would not be poverty but mother’s educational attainment, although this single measure would also be unsatisfactory. In the report that Martin Carnoy and I did earlier this year on international test scores, we compared the achievement of children across countries by the number of books they reported having at home (actually, the number of feet of shelf space devoted to books). This, too, was an inadequate measure, but we thought it the best one available cross-country because it was more academically relevant.

Addressing any one of the characteristics of lower class status can make some difference, but if addressed in isolation, will not make the difference to which we aspire. So while the start of this discussion had some merit—it would make some difference if lower-class women could be encouraged to delay maternity, or if they bore children only with a stable co-parent—telling women to marry if they live in communities where unemployment and underemployment of young men reaches 40 percent (as is true in some African American communities today), or where the failures of economic policy lead to few “marriageable males,” that advice, even if followed, won’t accomplish all that much. For example, if children are born into more stable marriages, or get better early childhood care and education, but still lose 5 to 10 points of I.Q. because of lead exposure (as the medical literature shows is the result), having a husband present or attending pre-school, in themselves, won’t substantially narrow the achievement gap.

As Martin Carnoy and I showed, there is a social class achievement gap in every country, and it is of roughly similar size, even in countries which purportedly have much superior school systems. This should suggest that school improvement alone will not narrow that gap in the absence of addressing its social and economic causes. Certainly we should improve schools, but if we want to raise the achievement of disadvantaged children substantially in our own country, we will have to improve the collection of interacting and mutually reinforcing characteristics of lower social class status. Addressing any one of them alone—whether it be income, or teen childbearing, or some other—will be a good thing to do, but won’t get us very far on the path we hope to take.