School choice in Milwaukee - school-voucher plan

Public Interest, Fall, 1996 by Paul E. Peterson, Jay P. Greene, Chad Noyes

If Witte's handling of parental satisfaction is quite straightforward, the same cannot be said about his discussions of the ability of the program to retain its students. When assessing "benefits to students," he concludes that "attrition from the program is ... high, which means that few students are remaining in the schools over a period of years," a conclusion constantly trumpeted by teacher unions and other critics of the program. But, as McGroarty points out in his book, Witte's conclusion is contradicted by his own data.

Student retention within any one school is an extraordinary problem in Milwaukee, in good part because of the residential mobility of low-income, welfare-dependent populations. For central-city, female-headed households with children between the ages of six and 17 (which constitutes 77 percent of the choice-student population), the annual residential mobility rate is 30 percent for African Americans and 35 percent for Latinos. Not every change in residence dictates a change in school attendance. But, in Milwaukee's public elementary schools, nearly 20 percent leave even before the end of the school year in June. Come the following fall, 35 percent of the students are no longer in attendance at the same public elementary school that they were the year before.

The choice schools substantially reduced this migration from one school setting to another. All but 23 percent of the choice students returned to the same elementary school the following fall (as compared to 35 percent in Milwaukee's low-income public schools). Within the school year itself, the percentage leaving choice elementary schools was as little as 4 percent in 1993 and just 6 percent in 1994 (much lower than the near 20 percent changing from one low-income public elementary school to another).(1) It is thus wrong for Witte to say that these percentages are high. They instead constitute a substantial improvement on the turnover rates within the Milwaukee public schools.

Flawed evaluation methodology

Witte's claim that the choice program has "not yet led to more effective schools" relies upon three types of comparisons with Milwaukee public schools: (1) comparison of test scores of cohorts of choice students with those of both all and low-income public-school students; (2) comparison of the changes from year to year in the individual test scores of choice students with those of both all and low-income public-school students; and (3) analysis of the effect of choice by means of a multiple regression analysis of all changes in test scores. None provide useful information about choice-school effectiveness.

When comparing the test scores of cohorts of choice students with those of public-school students, Witte finds no stable, significant difference between them. But, even if he had, the finding would have been meaningless. Before entering the program, the soon-to-become choice students scored well below the average of a cross section of Milwaukee public-school (MPS) students. Only 23 percent of the first-year choice students scored above the national average on the reading test, and only 31 percent attained that score on the math test. For the so-called MPS control group, the scores were 35 percent and 43 percent, respectively. Since students had decidedly lower scores before entering choice schools, it is misleading simply to compare scores from before and after the students entered choice schools. Since even Witte admits that this is not a "way to accurately measure achievement gains and losses," one can only wonder why the comparison is reported.