The Money Myth: School Resources, Outcomes, and Equity

Chapter 2: Multiple Resources, Multiple Outcomes:Testing the Improved School Finance with the National Educational Longitudinal Survey of the Class of 1988

Chapter 2 Multiple Resources,Multiple Outcomes: Testing the Improved School Finance with the National Educational Longitudinal Survey of the Class of 1988 THE CHALLENGE for improving schools, based on the approach presented in chapter 1, is to identify which school resources—now broadly defined as simple, compound, complex, and abstract—are effective. In this chapter, I discuss the rich data from the National Educational Longitudinal Survey of the Class of 1988 (NELS88) that enable me to do that. I present basic conclusions from the data and describe the powerful effects of school resources —most of them compound, complex, or abstract—on a variety of educational outcomes. I also clarify how existing analyses often understate the potential of educational resources. Some resources affect all outcomes; some affect learning, as measured by test scores, but not progress through high school; and some affect progress but not test scores. So confining educational outcomes to test scores, as do virtually all efforts to estimate educational production functions like equation 1.1, misses the powerful effects of some resources. To improve schooling outcomes, then, we must not only identify effective resources of all types but also determine ways to enhance those resources . In chapter 3, I examine which resources require additional funding and which must be developed in other, more complex ways—for example, by combining teacher participation with strong instructional leadership from principals and assistant principals. Although many researchers and many educators have tried to understand what school resources might be effective, the research underlying this chapter differs from prior work in at least four ways. First, it rests on a more complex model of the schooling process, illustrated by figure 1.2, which suggests what relationships are important. Second, rather than confining the analysis to the simple school resources that result from spending, it also includes compound,complex,and abstract resources.Third,NELS88 is a considerably richer data set than prior studies have used, and it enables me to measure a large number of different school and nonschool resources rather than a restricted set.And finally, rather than focus exclusively on math and English test scores, NELS88 makes it possible to construct many different outcome measures—twenty-nine in all,though I present here the results for only twelve. The next section describes the NELS88 data, the variables corresponding to the model of schooling in figure 1.2, and the estimation issues. In the following section, I analyze the effects of different school resources before going on to examine different types of outcomes, contrasting a common view that resources influence all outcomes in roughly the same way with a differentiated view that different resources affect different outcomes. I conclude with a summary of the implications for reform and policy, suggesting that many current policies are detrimental. This chapter answers questions with long histories to them—especially questions of whether and how schools make a difference. In chapter 3 it will be possible to examine the role of money in enhancing (or failing to enhance ) these effective resources and to determine when funding makes a difference and when it does not. THE NELS88 DATA ANDTHE CREATION OFVARIABLES Because NELS88 data go well beyond simple resources and include information on many school resources, as well as information on many measures of family background and student connectedness to schooling, they are admirably suited to the estimation of expanded production functions. The study initially drew a random sample of schools with eighth grades and randomly sampled eighth-graders within those schools in 1988.The students were questioned again in the tenth and twelfth grades (the first and second follow-ups), two years after they had completed high school (the third follow-up), and six years after that (the fourth follow-up, in 2000). Parents received questionnaires in the base year and again during the second follow-up, and teachers and administrators associated with grades 8, 10, and 12 were also questioned, with much more detailed information collected in grades 10 and 12. In addition, high school transcripts were collected during the senior year. I confine the analysis presented here to information collected as of the twelfth grade, except that I 54 The Money Myth obtained data on students’ initial enrollment in a two- or four-year colleges from the third follow-up. To be sure, the NELS88 data are now old in the sense that these students graduated sixteen years ago, in 1992. Much has...