Brookings-Wharton Papers on Urban Affairs 2000 (2000) 194-200
[Further Evidence on the Effects of Catholic Secondary Schooling]
Eric A. Hanushek: Understanding the performance of alternative school organizations has become increasingly important. The necessity of school reform has been accepted by a wider and wider segment of the population. And, while the majority still emphasizes finding ways to improve the existing public schools, an increasing sentiment favors introducing alternatives and more parental choice--either charter schools or expanded private schools. Unfortunately, experience with such alternatives remains limited, at least in the United States, so predicting the outcomes of some of these reform ideas rests on inferences and extrapolations from "close substitutes." Jeffrey Grogger and Derek Neal turn to performance in Catholic schools to provide insights. While the approach is not new, their paper provides more evidence on the potential gains from introducing competitive schools.
Catholic schools offer an interesting but complex laboratory for considering competitive alternatives to public schools. Catholic schools, which educate about 5 percent of K-12 students, have evolved over a long time. Thus, in contrast to recent experiences with charter schools or with public or private voucher experiments, they provide observations of outcomes that reflect long-term development of alternatives. The size of the Catholic sector also means that these schools are found across the nation, offering a broad comparison group to the public schools. They are particularly important in urban settings, where Catholic schools offer viable alternatives to urban public schools--schools that remain the object of intense policy concern.
On the other hand, as many previous researchers have noted, inferences about the impacts of these schools are difficult, in large part because of the difficult selection issues. Students in Catholic schools are different from those in public schools. When faced with free public schools, they choose to pay extra to attend the Catholic-school alternative. This fact by itself suggests the [End Page 194] potential for problems in extrapolating the experiences of Catholic-school students to others who currently are not attending private alternatives.
The long history of work in this area, reviewed in an earlier paper by Neal, has concentrated on understanding the reasons for choosing Catholic schools and then using that information to modify the estimates of performance differences in the two sectors. 1 This approach is of course difficult to implement in a convincing manner. First, it is difficult to find factors that influence selection but do not influence student performance, making identification of parameters of the selection process difficult. Second, there is a strong suspicion that selection follows from unobserved and unmeasured factors, such as the importance parents attach to school quality. Selection on the basis of unobservable factors makes the task even more difficult.
Grogger and Neal, building particularly on Neal's previous work, extend this analysis to data from the National Education Longitudinal Study (NELS). The NELS data set has both advantages and disadvantages for this analysis. It is more recent than the data sets previously used, allowing insights into more current situations. It also draws a large sample with many private-school observations. But it employs a sampling design that complicates the analysis, because it involves choice-based sampling on important outcomes (dropout behavior).
The analysis has a series of innovative features. Specifically, looking at median scores to provide robust estimation of models in the face of missing test data is very interesting. Additionally, the investigation of the supply of Catholic schools is novel and provides entirely new information about the scope and changes of Catholic education. Finally, while contained in previous work by Neal, the attention to the appropriate public-school comparison group is noteworthy.
The results. The central result of their work is the finding that Catholic schools on average appear to provide more learning than the relevant public counterparts. The effectiveness of Catholic schools is concentrated in urban settings, where the public-school performance is of greatest concern and where school choices, at least by the poor, are probably most limited. The differential effectiveness of Catholic schools is much less apparent in suburban areas.
The estimates are...