Milwaukee Choice Studies: Compare Apples to Apples
To the Editor:
Your article on the Milwaukee school-choice evaluation ("New Studies on Private Choice Contradict Each Other," Sept. 4, 1996) accurately reports that our study of the Milwaukee choice program found that choice students outperformed a comparable control group of Milwaukee Public Schools students on standardized tests by a considerable amount after three and four years of experience in the choice schools. These findings differ from an earlier evaluation conducted by a team led by John F. Witte. The article on these studies, however, was missing some important information.
Our results differ because Mr. Witte compares choice students to a far more advantaged cross-section of Milwaukee Public Schools students, while we compare choice students to those who applied for the program but were randomly rejected for lack of space. We believe that our comparison is superior because it compares apples to apples. Just as in a medical experiment, students were randomly assigned to treatment (a choice school) or control (the Milwaukee Public Schools). The two groups were almost identical except that some won the lottery to go to a choice school.
John Witte, on the other hand, compares apples to oranges. The average Milwaukee Public Schools student had nearly twice as much household income, was twice as likely to live with married parents, and began with significantly higher test scores. Mr. Witte says that he made statistical controls for those differences, but these controls are not magic. They cannot make groups that are extremely different comparable, because they cannot adjust for the numerous differences that are not measured. The more inadequate the controls, the worse the more disadvantaged choice students will appear to do.
But Mr. Witte objects that our results are invalid because too many cases are missing. This is a curious argument for him to make, given that his evaluation is missing more than 90 percent of the cases in its control group, considerably more than is missing from our analyses. The reason for the difference is that an experiment with random assignment does not require information on each student's background because the two groups are made similar by lottery. Mr. Witte's approach, with controls for family background, requires a considerable amount of family-background information and therefore suffers from the extremely low response rate to Mr. Witte's survey of households.
To ensure that the comparability of our treatment and control groups is not undermined by missing data, we examined the characteristics of these two groups over the four years of the experiment. In every year, the choice and control students continue to have nearly identical family backgrounds and initial test scores. We also checked to see if students who continued in the choice experiment differed from those who left. They did not differ. The students who continued in the choice schools did not have higher test scores than those who left the study each year. Students who continue in the choice program three or four years make gains in test scores not because they are the best students who remain, but because they are learning more. The absence of students from the study may result from the high mobility of low socioeconomic-status households, the irregular testing schedule in the Milwaukee Public Schools, or various other factors. The data suggest that missing students did not have different academic abilities.
We agree that further experiments are necessary to have greater confidence in the positive effects of school-choice programs. But the results from Milwaukee are promising enough that additional choice programs should be adopted to obtain more evidence.
Paul E. Peterson
Jay P. Greene
University of Houston
'Paraeducators' Hold Key to Class-Size Reductions
To the Editor:
We couldn't agree more with David Haselkorn on the need to more effectively utilize instructional assistants in public education ("Breaking the Class Ceiling," Commentary, Aug. 7, 1996 ).
We've come a long way from the days when classroom aides were primarily housekeepers and clerical workers. In today's classrooms, labs, and career centers, these "paraeducators" are technicians and specialists who support and enhance the work of teachers in all components of the instructional process. It is time to view them as a strategic resource in the education of children.
As Mr. Haselkorn points out, paraeducators are an obvious, but often overlooked, source for meeting the teacher shortage. Districts can gain a competitive edge in attracting and retaining the most qualified paraprofessionals by designing a career-ladder program that coordinates an education plan with the participant's current job and that provides incentives for the participant to continue along the path to teaching.
But beyond the need to meet the teacher shortage, the California School Employees Association shares Mr. Haselkorn's vision of paraeducators as full partners in education. Many of the most difficult problems facing administrators--reducing class size, increasing the diversity of the teaching force, and improving teacher preparation--could be solved through more strategic use of paraprofessionals.
These problems are particularly vexing here in California, where class sizes are among the largest, the student body is among the most diverse, and state education funding is among the worst. Here, cutting class size has seemed like an elusive, prohibitively expensive dream. But delivering "smaller" classes is achievable--if we can get beyond the traditional view of class size. Instead of focusing on staffing more schools with more teachers, increased use of paraprofessionals is a cost-effective alternative. Best of all, by coordinating the career-ladder program with class-size-reduction efforts, paraeducators could be used to reduce class size while they actively pursue their teaching certificates.
The need for these types of alternatives is evident. For example, many districts won't be able to participate in the new class-size-reduction program enacted here last month because of either lack of classroom space or lack of teachers. The CSEA is currently pursuing legislation to add a paraeducator option to the program, but until policymakers realize the practical implications of the class-size dilemma, we will continue to pursue the "grow your own" approach Mr. Haselkorn mentions. We're working with local districts to implement paraeducator career-ladder and class-size-reduction programs in the hopes that more empirical evidence will convince policymakers of the merits of these programs.
California School Employees Association
San Jose, Calif.
Vol. 16, Issue 02