Gerald W. Bracey

In the May column, we described some ways in which rich schools and poor schools differ in the literacy experiences they offer young children. In the Winter 2000 issue of the American Educational Research Journal, S. Jody Heymann and Alison Earle of the Harvard School of Public Health report on how poverty influences another aspect of school life, parental involvement.

Heymann and Earle point out that "parental involvement" comes in a number of forms. It can occur from "developing and using skills to support effective learning, engaging in home-to-school communication about student progress; volunteering at school; assisting their children with homework; becoming involved in school governance issues and decisions; and coordinating and integrating community services that will enhance the learning experience…Numerous studies have shown that regardless of how it is defined, parental involvement is important to children's success at school."

The researchers began by examining the working conditions of "substantially employed" mothers. "Substantially employed" is a term taken from the National Longitudinal Survey of Youth and means that the mother was not self-employed, had held at least one job within 30 days of the 1994 survey, and worked at least 20 hours a week (when the researchers used cutoffs of 30, 35 and 40 hours a week, the results remained unchanged). The NLSY did not collect employment data on fathers.

Twelve hundred and eighty of the mothers took the Behavior Problems Index (BPI). The BPI is a 28-item questionnaire about peer conflict, antisocial behavior, anxiety or depression, and dependency that a child might have exhibited in the 30 days before the mother took the questionnaire. The NLSY had administered the Peabody Individual Achievement Test (PIAT) in reading and math.

Of the 1280 mothers, 582 had a child in the top quartile of the BPI. Two hundred and eighty had a child in the bottom quartile of the PIAT reading test, and 334 had a child in the bottom quartile of the PIAT math test.

One analysis looked at the working conditions of mothers who had a child in the lowest reading quartile. When divided into families living in poverty and families above the poverty line, substantial differences appeared in the amount of time a parent would have to become involved with the school--or even with the child. Forty percent of the families not living in poverty had no paid sick leave, while, 67 percent of those below the poverty line had no such leave. Smaller differences appeared between the two groups for the presents of vacation leave or the flexibility to leave the job site.

Similar results were found looking at the working conditions of mothers who had children in the lowest quartile of the PIAT math score.

The largest differences and numbers, though, appeared for the poverty mothers whose kids acted out often enough to land them in the highest quartile of the BPI: Seventy-one percent of mothers in poverty had no paid sick leave (36% of mothers not in poverty), 46% had no paid vacation and 67% did not have the flexibility to leave the job site. The researchers offer a number of suggestions on how to ameliorate the problem. Non school-hour meetings are one possibility, but this imposes extra work on teachers, many of whom are mothers themselves. The Family Medical Leave Act could be expanded. It currently covers only 50% of families. The leave is for up to 24 hours, but since it is unpaid, this would be less of an aid to poor families. The authors seem most enthusiastic about two other possibilities:

Alternatively, paid family leave could be provided through an insurance program. National survey data have shown that the majority of Americans would be in favor of family leave insurance. An insurance program could be modeled after current public insurance programs such as workman's compensation or private insurance programs that are typically offered through workplaces such as disability insurance.

Tax incentives offer a different approach. Incentives could be provided to encourage companies to provide more paid leave and flexibility. These tax incentives could be modeled after the incentives that have been passed to encourage corporations to work for what economists classify as public goods, just as tax incentives for energy conservation do.

Don't look for compassionate conservatives to move on this any time soon.

SMALL CLASSES AND THE BLACK-WHITE ACHIEVEMENT GAP

This column has from time to time presented information from a variety of sources concerning the effects of small class size. Among the sources has been Princeton University economist, Alan Krueger. Krueger has recently teamed with Princeton colleague, Diane M. Whitmore to put a lot of their conclusions in one paper. They also add some research not previously seen elsewhere.

Krueger and Whitmore first review trends in black-white achievement as indicated by NAEP and the SAT, graphically showing that while the gap narrowed, it widened a bit in the 1990's and has been large all along. They then review the importance of educational inputs literature, once again pointing out that in his analyses, Eric Hanushek would first extract a number of estimates of resource impact from one study and then treat all the estimates from one study as if they had come from independent studies. The observe once again that when one gives equal weight to all studies rather than to all estimates from the studies, there is strong evidence relating school inputs to achievement.

In their review of the literature, they observe that smaller classes seem to have a greater impact on minority students than on whites. For instance, in the evaluations of California's statewide class size reduction mandate, schools with larger proportions of minority students had larger gains. The gains, while small, are impressive in view of the many factors that would mitigate against finding any gains at all: teachers leaving low-income districts for new openings in the suburbs, the scramble to find space in which to hold the smaller classes.

After a review of the basic findings of Project STAR, Krueger and Whitmore update some of their findings concerning SAT and ACT testtaking propensity and scores. Their original analyses found that black students who had had small classes in the early grades were 25% more likely to take the ACT or SAT. The data were limited, though, because they came only from the graduating class of 1998. Students who had been accelerated or who had been retained in a grade would not have shown up. When other members of what would have been a cohort in the early grades were found and added it, no statistically significant differences were found among those in small classes, those in regular classes, and those in regular classes with an aide. Some 46.4 percent of small-class students took a college entrance exam, compared to 44.7% of regular class students and 45.3% of regular-plus-aide students.

For black students, though, the relationship held. Some 41.3 percent of black students from small classes took an entrances exam, compared to 31.8 percent of regular class students and 35.7 percent of the regular-with-aide classes. "Thus, assigning all students to a small class is estimated to reduce the black-white gap in test-taking rate by an impressive 60 percent."

Test taking is one thing, test scores are something else. The scores for students in regular and small classes were quite similar. However, there might well be a "selection" bias operating on students from small classes. Because students from smaller classes are more likely to take the test, it is reasonable to assume that small classes encourage more marginal students to give it a shot. Thus, the fact that regular and small class students score the same is itself an indication of some positive impact of the small classes.

Krueger and Whitmore conduct two additional analyses to correct for the selection effect. As with test-taking percentage, the adjustments do not change the results for white students. For black students, though, one procedure produces an effect size of +0.15 standard deviation change in favor of small classes, the other yields an effect size of +0.20 standard deviation.

No group of students committed a lot of crimes. Some 1.6 percent of Project STAR students were convicted of a crime compared to 2.6 percent of all students. Since 88 percent of the convictees were male, further analyses were conducted only on males. Differences favored small classes, but none were statistically significant. Students from small classes did get slightly shorter sentences, but there is no information on how much time they actually served.

Finally, Krueger and Whitmore looked at teenage birth rates. This is an important outcome. Other research has indicated than teen mothers are more likely to have low birthweight children and some 80% of teen mothers ended up on welfare. White females from small classes were 33% less likely to have a child and black males from small classes less likely to father one. For black females and white males, there were no effects from small classes.

The researchers then move on to compare the impact of reduced class size with the impact of vouchers as presented in a paper analyzing voucher results in New York, Dayton, and Washington, DC. "We conclude from this comparisons that, when comparable samples are considered, black students who attended a small class for two years in the STAR experiment improved their test performance by around 50 percent more than the gain experienced by black students who attended a private school as a result of receiving a voucher in the New York, Dayton, and Washington voucher experiments."

Krueger and Whitmore then ask "Why?" "Why do black students appear to gain more from attending a smaller class than white students?" They speculate that "One possible explanation for these findings is that teachers have to move very slowly through the curriculum if they have weak students--e.g., because they are disrupted frequently or have to explain the material multiple times to the slower students--but if they have smaller classes they can effectively teach more material." Or, I would conjecture, teach the same material more effectively. "Regardless of the explanation, our findings suggest that class size reductions will have the biggest bang for the buck if they are targeted to schools with relatively many minority students. But if such targeting is politically infeasible, then reducing class size generally would still lead to a reduction in the black-white test score gap."