JVER v25n4 - The Effects of School Size and Leadership on Participation in the School-to-Work Movement
The Effects of School Size and Leadership on Participation in the School-to-Work Movement
William J. Stull
Nicholas M. Sanders
Judith C. Stull
What determines the extent to which a high school offers its students support in making the transition from school to work? This study uses multiple regression analysis and data on 1,144 public comprehensive high schools from the NELS:88 administrator survey to identify the crucial factors that determine the breadth of a school's STW programming. Using school leadership indices, school and district size, student body characteristics, and various geographic indicators as independent variables, the estimated model accounts for over 50% of the variation in the number of STW activities offered. The study finds that schools with stronger leadership as measured by indices of innovation, school climate, and external cooperative relationships have significantly broader STW programs. It also finds that the scope of STW programming only increase slightly with increases in school or district enrollment, suggesting that small schools are not necessarily at a disadvantage in providing a range of STW activities.
The school-to-work (STW) movement can trace its origin, like many other recent education reform efforts, to the 1983 publication of A Nation at Risk ( National Commission on Educational Excellence ). The central premise of this report was that the nation's preeminence in science, technology, industry, commerce, and military preparedness was threatened by its mediocre schools. In the decade that followed, several other widely-circulated national reports took its general concerns as points of departure and used them to focus on the specific educational and career problems of young people who do not attain a bachelor's degree. These included The Forgotten Half: Pathways to Success for America's Youth and Young Families ( William T. Grant Foundation Commission, 1988 ), as well as America's Choice: High Skills or Low Wages ( Commission on the Skills of the American Workforce, 1990 ) and What Work Requires: A SCANS Report for America 2000 ( Secretary's Commission on Achieving Necessary Skills, 1991 ). All of these publications stressed the need to raise the job skills of noncollege youth so they can make a successful entry into and progress within what has come to be known as the sub-baccalaureate labor force (Grubb, 1996).
The cumulative effects of these reports, and of the problems to which they were responding, have been felt at every level of the American education system. Individual states and localities began developing their own STW programs in the late 1980s, and these were reinforced by a series of federal legislative efforts in the 1990s. The first of these were the 1990 Amendments to the Carl A. Perkins Vocational and Applied Technology Act of 1984. The Perkins amendments were a response to research showing that the majority of vocational education students do not benefit from the occupationally specific training they receive in high school ( Boesel, Hudson, & Masten, 1994 ; McCormick, Tuma, & Houser, 1995 ). The amendments required federally-supported vocational education programs to "integrate academic and vocational education" as a way of providing vocational education students with more general skills.
The second important piece of federal legislation was the 1994 Goals 2000: Educate America Act. Goals 2000 enacted the National Education Goals announced at the 1989 National Education Summit attended by the nation's governors and President George P. Bush. Two of the goals directly addressed the concerns of the STW movement. One (Student Achievement and Citizenship) specified that all American students will leave school prepared for "productive employment." Another (Adult Literacy and Lifelong Learning) specified that every adult American will be prepared to compete effectively in the global economy.
The culmination of the STW movement, however, was the passage of the School-to-Work Opportunities Act (STWOA) of 1994 ( Jennings, 1995 ). States received grants under this legislation once they established a statewide STW partnership that included representatives from the governor's office, state agencies, business firms, organized labor, nonprofit human service organizations, and the education sector. The statewide partnerships distributed funds to local partnerships that then worked to establish STW systems in individual schools and districts. The STWOA did not specify in detail what such local systems should look like except that: (1) they must consist of school-based learning, work-based learning, and connecting activities, and (2) they must be sufficiently comprehensive and structured so that students can choose among one or more sets of linked school- and work-based learning paths each leading to defined postsecondary education or employment opportunities. By 1999, all 50 states plus Puerto Rico and the District of Columbia had received STWOA development and implementation grants.
Purpose and Scope of the Study
The purpose of this study is to determine what factors explain high school involvement in the STW movement as measured by the number of such activities available to students. As discussed in more detail below, we hypothesize that school (and, by extension, district) size and several dimensions of school leadership are important determinants of the number of STW activities that a school offers. We also hypothesize that certain student body, community, state, and regional variables exert an influence on STW programming. These hypotheses are tested using a multiple regression analysis, the results of which are presented and evaluated in the final sections of the paper.
Secondary education in the U.S. is dominated by large comprehensive public high schools that draw students from relatively wide feeder areas and provide a variety of academic and nonacademic programs and services. In their classic study, Powell, Farrar, and Cohen ( 1985 ) describe these schools as "shopping malls" which attempt to serve the diverse educational needs of their student "customers, " subject to a set of budgetary and political constraints imposed by local school boards, state boards of education, and, to a much smaller extent, the federal government. Operating within these constraints, high schools allocate their resources -- primarily staff time and energy -- among a broad range of curricular and extracurricular activities.
Over the past fifteen years, STW programming has become a legitimate contender for these resources. Anecdotal evidence ( Olson, 1997 ) as well as the data used in this study suggest, however, that schools vary greatly in the extent to which they have chosen to participate in the STW movement. Some high schools became deeply involved in STW programming, others (the majority) made a partial commitment, and still others ignored the movement entirely. The general purpose of the research being reported here is to explain this variation.
The study is based primarily on data from a national sample of 1,144 comprehensive public high schools collected as part of the U.S. Department of Education's National Education Longitudinal Survey of 1988 (NELS:88). Included in the NELS:88 database are a set of school-level indicators of STW programming. Using these indicators in conjunction with various school, community, and state characteristics from NELS:88 and other sources, we seek to explain why some high schools became more involved in STW activities than others. The data are for 1992 -- after the appearance of the national reports on youth cited above but prior to the passage of the STWOA. The research thus analyzes the behavior of schools before the U.S. government intervened to provide broad-based financial incentives to establish STW programming. The STW activities shown in our data are therefore primarily the result of local or state initiatives rather than federal funding.
The shopping mall is a useful metaphor for public high schools in America. It does not, however, provide much insight into the process whereby resources are allocated to the different "stores" in the mall. Indeed, what it does imply is misleading. The metaphor suggests that resource allocation is determined by "market forces," which in a school setting are presumably best represented by student demand. In reality, under the "one best system" ( Tyack, 1974 ) public schools everywhere in the country are governed by democratically controlled bureaucratic structures whose actions are affected by a broad range of internal and external influences that only partially reflect the desires of students and their parents ( Chichura, 1989 ; Chubb & Moe,1988 ; Ferris, 1992; Hartman, 1988; Jones, 1985; Monk, 1981).
To capture this reality, we develop a conceptual model of STW programming at the high school level that is designed to capture as many of these influences as possible. In recent years, education researchers have focused considerable attention on two specific determinants of school resource allocation and student outcomes: school size and school leadership (see, for example, Hallinger & Heck, 1996 ; Lee & Smith, 1997 ). Special effort is therefore devoted in the study to determine the importance of these particular factors. In addition, we test for the influence of a wide variety of student body, community, state and regional variables on STW programming.
Despite an extensive review of the literature, we were unable to find any research similar to what is being reported on here. This turned out to be true both substantively and methodologically. Substantively, there is now a large literature on STW, but much of it is made up of case studies reporting on individual schools, districts, consortia, or partnerships that have created "successful" STW programs (for example, Olson, 1997 ). Two very useful reviews of the STW research literature have been prepared by the National Center for Research in Vocational Education ( Stern, Finkelstein, Stone, Latting, & Dornsife, 1995 ; Urquiola, Stern, Horn, Dornsife, Chi, Williams, Merritt, Hughes, & Bailey, 1997 ). Together, these reports provide a comprehensive overview of published research through 1997 on both STW implementation and on the effect of STW programming on student outcomes. Except for the Mathematica research discussed below, however, none of the studies cited by the NCRVE reports are based on quantitative data showing the extent of STW activity across a national sample of education entities.
The only research we found that quantifies the establishment of STW practices nationwide is the work carried out by Mathematica Policy Research, Inc. In the early 1990s, Mathematica completed several national studies of specific STW programs such as Tech Prep. More recently, it received a multi-year grant from the U.S. Department of Education to analyze the implementation of the STWOA. Thus far it has issued three STWOA reports ( Hershey, Hudis, Silverberg, & Haimson, 1997 ; Hershey, Silverberg, & Haimson, 1999 ; and Silverberg, Haimson, & Hershey, 1998 ) based on interview and survey data from STW partnerships and students across the country. This work is quite comprehensive, but it differs from the research being reported here in three fundamental ways. First, its central focus is on the implementation of the STWOA. Therefore, it is looking at a later stage of the STW movement when local programming was substantially driven by this legislation. Second, its primary units of analysis are the local STW partnership (typically made up of several school districts) and the individual student. Unlike the present study, it focuses little attention on decision making at the school level. Third, its analysis thus far is primarily descriptive. It does not use multivariate techniques to try to explain why schools, districts, or partnerships vary in their STW involvement.
More surprisingly, we did not find many methodological precedents for our work either. Looked at in the broadest context, the study seeks to explain one dimension of school resource allocation in terms of the internal and external influences on school decision makers. We expected to find empirical studies in the school change or implementation literature in which the dependent variables are school decisions and the independent variables are school characteristics. Such research appears to be nonexistent except for the work of David Monk. Using large data sets, Monk and his collaborators have sought to explain why high schools vary in the diversity of their curricula, with particular emphasis on the influence of school size ( Haller, Monk, Spotted-Bear, Griffith, & Moss, 1990 ; Monk, 1987 ; Monk & Haller, 1993 ). The research being reported here builds substantially on Monk's econometric approach and replicates some of his general findings concerning the effect of school size on the breadth of school offerings. To the best of our knowledge, however, his methodology has never been applied to STW programming.
The data used in this research are derived from multiple sources. The most important of these, as noted above, is the NELS:88 longitudinal study. The NELS:88 project began with a national sample of over 25,000 eighth grade students drawn from approximately 4,000 public and private schools in 1988 (base year). A subsample of these students (along with some additional students) were resurveyed in 1990, 1992, and 1994 (first, second, and third follow-ups). In the base year, first follow-up, and second follow-up, administrators (usually principals) of the schools attended by the students were surveyed as well. They were asked to provide information in seven areas: school characteristics, student characteristics, teaching staff characteristics, school admission policies and practices, grading and/or testing structure, school programs, and school climate. Questions pertaining to STW programs and services were included in several sections of the questionnaire.
The research draws most heavily on data from the second follow-up school administrator survey. We chose the second follow-up results because they provided the most comprehensive information about STW programming at the school level. Approximately 1,400 high schools were included in the second follow-up. After eliminating private schools (both sectarian and nonsectarian) and specialty public schools (such as vocational and technical institutes), we were left with a sample of 1,144 comprehensive public high schools. Given the NELS:88 student and school selection process, this sample is reasonably representative of the national population of such schools in the early 1990s.
The dependent variable and most of the independent variables are drawn from the NELS:88 dataset. Additional independent variables come from the Common Core of Data (CCD). The CCD is the principal U.S. database for primary and secondary public schools and public school districts. It is published annually (since 1987), covers all public schools and districts in the nation, and provides data that are comparable across states ( Davis & Sonnenberg, 1993 ). We linked the CCD file for 1992 to the NELS:88 second follow-up administrator file so that variables from the former can be used in regressions across the NELS:88 schools. The most important of the CCD variables for our purposes are those that describe school revenues and expenditures broken down along a variety of dimensions. These represent significant additions to the data because NELS:88 contains no fiscal information. A third set of independent variables are derived from data on the timing of state participation in the STWOA. These variables are intended to provide a rough measure of overall state commitment to the STW movement in the period before the STWOA.
Our model takes the school as the decision-making unit whose behavior is to be analyzed. Therefore, its dependent variable and all the independent variables are measures of school characteristics. The specific variables used in the model are described in this section. More details are provided in a technical appendix available on request from the authors.
The term "school-to-work" as used both in the professional literature and the popular press has come to denote a wide and somewhat vaguely-defined set of school services and programs that have workforce preparation and/or placement as their objectives. At the deepest level any school activity that enhances the intellectual or emotional maturity of a student or adds to her skills or competencies could be labeled as STW because such gains can, and in many cases do, contribute to future success in the workplace. From this perspective, sports teams, history classes, foreign language clubs, and student government are all STW activities. Such a broad human capital definition may be useful for some purposes, but this study is based on a narrower conception that is closer to the general understanding of the term. Specifically, we define a STW activity as a school program or service that has as its overriding purpose the preparation of students for career choice, job placement, and/or labor market success.
To make this general definition operational, we divide STW activities into four general groups: job skill provision, work readiness training, job placement, and business involvement. This is only a rough classification because many activities can easily be placed in more than one group. Job skill activities directly provide general and specific skills to students that will be useful in the labor market. Traditional vocational and cooperative education programs, as well as more modern initiatives such as Tech Prep, are included here. Work readiness activities prepare students psychologically for entry into the full-time labor force. They include work readiness seminars, STW counseling, and ability/interest assessment. Placement activities focus on obtaining full time employment for students after they graduate from high school. Examples include job fairs, practice interviews, resume workshops, and placement courses. Finally, business involvement activities connect the school to specific local businesses. They include provision of student information to prospective employers, listing of jobs available in the local area, and business sponsorship of the entire school or specific programs.
In the NELS:88 data there is information on 24 school activities that could be reasonably classified into at least one of the four groups. These are shown in the top portion of Table 1. The dependent variable in the study -- the STW activities index -- is a simple count of the number of these activities present in a school. It is important to note that this index is a breadth measurement of STW programming and therefore does not take into account program depth (that is, the extent of student participation) or program quality. Descriptive statistics for this composite variable are presented at the bottom of the table. As these statistics indicate, there were substantial differences in the breadth of STW programming across the sample schools in 1992.
Table 1 STW Activities Index: Components and Characteristics
Job skill activities: Job placement activities:
Cooperative education Employment search help Other work experience programs Job fairs Tech-prep or other 2+2 program Provision of letters of recommendation Vocational education program Practice interviews Basic computer skills training Arrangement of interviews Computer programming training Job placement courses Credit for off-campus work experience Job placement counseling Job placement services in school/district
Business involvement activities: Work readiness activities:
School adoption by business Transition to employment counseling Incentive program sponsored by business Vocational interest and/or ability assessment Anti-drug program sponsored by business Provision of interest inventories Job lists provided by businesses Career readiness seminars Recommendations requested by businesses
Range: 0-24. Median: 14.00. Mean: 12.42. Standard deviation: 5.56.
Cronbach's alpha: 0.89. Note: All variables coded 1 for "yes" and 0 for "no".
Independent Variables As discussed in earlier sections, schools are subject to a variety of internal and external forces that influence their resource allocation decisions. We classify these influences into five groups: school and district size, school leadership, student body, community, and state/region. A general description of each group is provided below with some discussion of specific variables. More details are provided in the next section where the results are presented.
School and District Size In recent years a strong case has been made for smaller high schools on the basis that they provide more personal education for students. This is believed to be of particular importance for students who are educationally at risk. One of the principal arguments against smaller schools, however, is that they lack the economies of scale necessary to offer a broad range of alternative programs and activities. Following this line of reasoning, we hypothesize that school involvement in the STW movement depends positively on both school and district size measured in terms of the number of students. We include school district size because many STW programs are organized at the district level. However, as Monk and Haller ( 1993 ) found, size relationships are likely to be nonlinear in that they increase at a decreasing rate. This functional form is expected because adding (say) 100 students to large school will probably have a smaller impact on STW programming than adding 100 students to a small school.
Evidence is accumulating (though much of it is anecdotal) that a necessary condition for school change is effective leadership. We hypothesize that an important factor in the establishment of STW programming is leadership at the level of the individual school, primarily by the principal. Using NELS:88 data from the school administrator survey, we created indices for three different dimensions of school leadership that seem important for the establishment of STW programming: the ability to innovate, the ability to maintain a school climate conducive to learning, and the ability to establish cooperative relationships with external constituencies. Tables 2-4 present the components of each index and reports their descriptive statistics -- including Cronbach's alpha, a measure of internal reliability ( Bohrnstedt & Knoke, 1994 ; Helmstadter, 1964). In each case, Cronbach's alpha is greater than 0.75 indicating that the components of the index are substantially correlated with each other and thus likely to be measuring the same underlying construct. Each index is discussed in more detail below.
School innovation index A crucial task for any leader is to bring new ideas into the organization. Since a broad-based STW program agenda represents a fairly radical break with traditional thinking about how high school education should be organized, we hypothesize that innovative school leaders are more likely to embrace the STW philosophy than those content with the status quo. To measure "innovativeness" we created an innovation index as one of the independent variables ( Table 2 ). This variable is a simple count of the number of non-STW innovative practices (from a list of seven) that were established in the school in the three years prior to the NELS:88 second follow-up administrator survey. These practices include curriculum reform, interdisciplinary teaching, broad changes in instructional methods, and new procedures for student assessment. Approximately 20 % of the sample schools established no innovative practices (possibly overstated because of missing data) and almost 7% established all seven.
Table 2 Leadership Indices: School Innovation Index
School innovation index: Innovative practices established in past 3 years
New procedures for making school policy
Major new curriculum programs
Changes in student ability grouping policy & practice
School-wide changes in instructional methods
New roles for teaching and supervisory staff
Interdisciplinary teacher teams
New school-wide procedures for student assessment
Range: 0-7. Median: 3.00. Mean: 2.94. Standard deviation: 2.20
Cronbach's alpha: 0.78 Note: All variables coded 1 for "yes" and 0 for "no".
School climate index Perhaps the single most important responsibility of an academic leader is establishing a school climate that is conducive to learning. Studies of successful schools repeatedly emphasize the importance of high academic expectations; orderly classrooms; student effort; and mutual respect among teachers, students, and administrators. Our hypothesis is that school leaders who have established strong learning climates are more likely to embrace STW than those who have not because a commitment to the complexities of work-based learning requires that the fundamentals of high school education already be in place. To test this hypothesis, a climate index was created for each school in the sample based on a list of thirteen characteristics drawn from the NELS:88 data that are indicators of strong learning climates ( Table 3 ). The value of the index for each school is the number of reported characteristics. Slightly less than 15% of the sample schools have eleven or more of the characteristics and an almost equal number of schools have three or fewer.
Table 3 Leadership Indices: School Climate Index
School climate index: Characteristics of schools with strong learning climates
Teachers encourage academic achievement.
Teachers encourage students to enroll in academic classes.
All students are expected to do homework.
Students are encouraged to compete for grades.
Seniors must pass a test to receive a diploma.
Students place a high priority on learning.
Teachers do not find motivating students difficult.
Discipline is emphasized.
Classroom activities are highly structured.
Teacher morale is high.
Student morale is high.
Teachers have positive attitudes toward students.
There is no conflict between teachers and administrators.
Range: 0-13. Median: 7.00. Mean: 7.28. Standard deviation: 2.83.
Cronbach's alpha: 0.76. Note: All variables coded 1 for "yes" and 0 for "no".
School external cooperation index. Though less emphasized in the leadership literature than the previous two elements (see, however, Fullan, 1993 and 1997 ), a school leader's ability to establish cooperative relationships with external constituencies is of central importance to the establishment of a comprehensive STW program. The model STW system ( Stull and Sanders, 1999 ) requires a coherent program of school-based and work-based learning activities for a broad range of students. Establishing such a program necessarily entails an enormous amount of collaborative participation with outside groups -- particularly parents, employers, and district administrators. It seems reasonable, therefore, to conjecture that school leaders having good external relationships will be more successful in establishing STW programming than those that lack them. To measure the strength of such relationships we developed a cooperation index based on seven questions in the NELS:88 data set ( Table 4 ). The value of the index for a particular school is the number of affirmative responses to the questions. Approximately 50% of the sample schools had cooperative relationships with all seven external constituencies and 14% had them with none (again, a possible overstatement due to missing data).
Table 4 Leadership Indices: School External Cooperation Index
School external cooperation index: Cooperative relationships with external groups and individuals
Central office administrators
Teachers association or union
Range: 0-7. Median: 6.00. Mean: 5.37. Standard deviation: 2.39.
Cronbach's alpha: 0.91. Note: All variables coded 1 for "yes" and 0 for "no".
The shopping mall model suggests that programs and activities are established in high schools at least in part to satisfy the demands of students (and their parents). This suggests that otherwise similar schools with different student bodies might have different responses to the STW movement. To capture such effects, we include independent variables measuring the composition of the student body along the following dimensions: course of study, family income, educational aspirations, and race. We hypothesize that, other things equal, schools with higher proportions of low-income students, higher proportions of vocational students, and higher proportions of noncollege students have broader STW programs. We make these conjectures because the STW movement is strongly oriented toward improving the high school experience for students not destined to graduate from a four-year college or university (more so in its earlier years than later). We make no a priori judgments concerning the influence of racial composition since there appears to be no basis for doing so once other variables are controlled for.
Public schools are substantially creatures of the community in which they are located. Local taxpayers provide significant financial support, and local residents typically elect representatives to serve on the board that governs the school system. Given these important connections, it is natural to suppose that the community has an influence on school decision-making that is separate and distinct from that of students or school leaders. In principle, there are many community characteristics that might affect a school's involvement in STW programming, including the socioeconomic characteristics of its citizens (income, age, racial composition, etc.) and their willingness to spend tax dollars on public education. Because of data limitations, the focus in this study is on economic variables.
State and Region
Finally, it is reasonable to suppose that a school's involvement is STW activities might be shaped in part by political and economic forces that are external to its local community. Regional economic and political conditions might exert a significant influence. In addition, as noted above, there have been many STW initiatives at the state level, most of which have been implemented through the actions of individual districts and schools. Our expectation is that schools in states that made a substantial commitment to the STW movement are more likely to have significant STW programming than otherwise identical schools in states with lesser involvement.
Various versions of the foregoing model were estimated using different combinations of independent variables. The final results are shown in Table 5. The dependent variable, as noted above, is the number of STW activities offered by the school, and the independent variables are various internal and external factors that might reasonably be expected to influence STW programming. In the table, the latter are divided into the five groups identified in the previous section. To facilitate interpretation of results, both regular and standardized regression coefficients are presented. However, unless otherwise indicated, discussion in the text will always refer to the former.
Overall, the model fits the data reasonably well. The R2 is .55, which is excellent for a cross-section study with a large number of observations. In addition, most of the eighteen independent variables are statistically significant with the anticipated sign. The specific results shown in the table are discussed below. In each case of statistical significance, we attempt to provide some measure of substantive significance by computing the effect on the dependent variable of a "large" change in the independent variable. For some independent variables, a one standard deviation change is used. When this is not appropriate, different measures are employed. For example, the school and district size variables have very large standard deviations because their distributions are highly skewed toward the high end. A percentile-based measure of change is used for these two variables. Some research and policy implications of the findings are presented in the last section.
Table 5 The Effects of Student Body, School, Community, State, and Regional
Characteristics on the Number of STW Activities Provided by the School
(n = 1144 Comprehensive Public High Schools in 1992)
School and district size Log of district size 0.2226 0.0698 2.383 0.017 Log of school size 0.553 0.0698 2.600 0.009 Number of grade levels in the school -0.241 -0.05444 -2.533 0.011 School leadership Innovation index 0.451 0.1784 7.556 0.000 Climate index 0.236 0.1248 4.912 0.000 Cooperation index 1.214 0.5219 19.409 0.000 Student body Percent students non-Hispanic White -0.011 -0.0611 -2.028 0.043 Percent students eligible for free/reduced cost lunch 0.002 0.0059 0.225 0.432 Percent students enroll in 4-yr. college upon graduation -0.009 -0.0339 -1.449 0.148 Percent students in academic program -0.006 -0.0270 -1.119 0.263 Percent students in vocational program 0.017 0.0551 2.502 0.012 Community Median household income (in $1000) -0.030 -0.0512 -1.847 0.065 District expenditure per pupil (in $1000) 0.073 0.0224 0.786 0.432 Percent district revenues non-local -0.017 -0.0582 2.104 0.036 State and region State received early STWOA granta 0.651 0.0585 2.453 0.014 South regiona 0.593 0.0511 1.368 0.172 Midwest regiona 0.521 0.0402 1.302 0.192 West regiona 1.086 0.0821 2.354 0.019
Intercept -0.517 -0.266 0.790
R2 = .55 aNote: Variable coded as a dummy with 1 = yes and 0 = no.
School and District Size
Three size variables are used in the estimation shown in Table 5; all have coefficients that are statistically significant and have the expected signs. However, the magnitudes of the estimated coefficients are not large, so the variables are not of great substantive significance in explaining why the schools in the sample differ in STW involvement. As one would expect, there is a positive correlation between school size and district size (0.42), but it is not enough to affect the results (as evidenced by the fact that the coefficients of both school and district size are statistically significant.)
School size. Larger schools, not surprisingly, have more STW activities than smaller schools. The estimated relationship is logarithmic rather than linear, however, so the size effect tapers off as the number of students increases. This finding parallels Monk and Haller's ( 1993 ) results for curriculum diversity mentioned above. In addition, the strength of the effect, statistical significance notwithstanding, is not large within the range of typical high school sizes. Based on the value of the regular regression coefficient, an increase in school size from the 25th percentile of the sample (720 students) to the 50th percentile (1125 students) only brings about a 0.25 increase in the STW index -- less than 2 percent of the index value for the median school.
District size. Holding school size constant, schools in larger districts have more STW activities than those in smaller ones with the estimated relationship again being logarithmic. As might be expected, the regression coefficient for district size (0.226) is smaller than the one for school size (0.553). School districts have a much broader range of sizes than individual schools, however, so percentile changes have similar effects in the two cases. Increasing district size from the 25th percentile (3500 students) to the 50th percentile (10,500) also increases the STW index by 0.25.
Number of Grade Levels Not all schools in the sample have four grade levels (i.e., freshman, sophomore, junior, senior); some have three and some have more than four. Holding the size of the student body constant, the number of grade levels in a school is inversely related to the number of its STW activities. This is expected because such activities are mainly provided for juniors and seniors. Again, however, the effect, though statistically significant, is small. Having three grade levels instead of four (school and district size constant) only increases the STW index by 0.21.
All three indices of school leadership are both statistically and substantively significant in the regression analysis. All are significant at the .001 level and their standardized regression coefficients are appreciably larger than those of the other variables. These results provide strong support for the position that leadership is an important determinant of STW programming.
Innovation Index The innovation index is a count, from 0 to 7, of the number of innovative practices reported by the school. Our a priori expectation is that schools that are more innovative in general would be more active participants in the STW movement as measured by this index. The results in Table 3 provide strong support for this hypothesis. A one standard deviation increase in the innovation variable (2.20) increases the STW index by 0.99 -- approximately 7 percent of the index value for the median school. Using the standardized regression coefficient, this translates into a 0.18 standard deviation increase in the STW index, the second largest in the study.
Climate Index Schools also differ in the extent to which they have a climate conducive to learning. Our hypothesis that schools with strong learning climates are more likely to be active providers of STW programming is supported by the results in Table 3. The estimated coefficient of the climate index is statistically significant, though smaller in size than those of the other leadership indices. The climate index takes on integer values between 0 and 13. A one standard deviation increase (2.82) raises the STW index by 0.66 -- more than 0.12 standard deviation (as shown by the standardized regression coefficient). This latter effect is the third largest in the study.
Cooperation Index Among the three leadership variables, the index measuring the strength of cooperative relationships with external constituencies is the most important in explaining STW programming. The index takes on integer values from 0 to 7. A one standard deviation increase (2.39) raises the STW index by 2.9 -- almost 21% of that index's value for the median school. Using the standardized regression coefficient, this translates into a 0.52 standard deviation increase in the STW index, by far the largest in the study. These results strongly support the hypothesis that having cooperative relationships with external groups greatly contributes to the establishment of a broad-based STW program.
Student Body Table 3 shows the estimated coefficients for five student body variables. Contrary to the spirit of the shopping mall model, the variables as a group are not important determinants of STW programming. Only two are statistically significant, and in both cases the estimated coefficients are relatively small. We tentatively conclude that the characteristics of a school's students, particularly their socioeconomic status (SES), do not have much influence on the extent of its STW programming.
Percentage Non-Hispanic White The coefficient of this variable has a negative sign and is statistically significant -- indicating that schools with higher percentages of minority students, other things equal, tend to have more STW programming. This result was neither anticipated nor unanticipated since there is no prior maintained hypothesis concerning minority involvement with STW after holding constant SES. As noted above, however, the strength of the effect is small. Decreasing the percentage of non-Hispanic Whites by one standard deviation (32.35), and hence increasing the percentage minority by the same amount, only leads to a 0.36 increase in the STW index.
Percentage Eligible for Free or Reduced Cost Lunch The percentage of students eligible for free or reduced cost lunch is widely recognized as a flawed measure of SES for high school students because many students from low-income families do not enroll due to the stigma attached to the program. The variable was used only because it is the only direct indicator of student income available for the sample schools. (In addition, it is perhaps reasonable to assume that underenrollment is proportionally the same in all schools and therefore that its presence will not affect the results.) We expected that the free lunch coefficient would be positive and significant because students from poor families have traditionally been the main participants in high school work readiness programs. However, in all estimations -- including those excluding the non-Hispanic White and four-year college variables -- the coefficient is statistically insignificant.
Percentage Attending Four-Year College Administrators in the sample schools were asked to report the percentage of recent graduates who attend a four year college or university soon after graduation. We anticipated that this variable would have a negative influence on STW programming for the same reason that we expected the percentage of students eligible for a free lunch to have a positive influence: high school curriculum programs that are not primarily academic have usually focused on noncollege students. However, like the free lunch variable, it is consistently insignificant regardless of model specification.
Curriculum All students in the sample schools were enrolled in an academic, vocational, or general curriculum. In the equation reported in Table 3, the percentages of students in academic and vocational tracks are included as independent variables. (The percentage in the general curriculum track was excluded to avoid perfect multi-collinearity.) The academic track variable is not statistically significant, indicating that increasing the proportion of academic track students at the expense of general track students has no effect on STW programming. The percentage vocational variable, however, is statistically significant and positive in sign. Schools with relatively high proportions of vocational students have more STW programming than those with relatively high proportions of students in either of the other two curricula. This is the anticipated result but, somewhat surprisingly, the effect is not particularly large. A one standard deviation increase in the percentage of vocational students (17.86) increases the STW index only by 0.30.
Table 3 presents estimated coefficients for three community variables, all of which represent economic characteristics of the school district in which the school was located. Experiments with other community socioeconomic variables were always unsuccessful. For example, some earlier versions of the estimated model included dummy variables for urban and suburban location (with rural location as the benchmark category) but these never proved to be statistically significant. Two of the three included variables have estimated coefficients that are statistically significant but their values are not large. Overall, the results suggest that community SES is not a major determinant of STW programming, a conclusion that parallels a similar finding for student body SES.
Median Income Median family income in the school district has a negative effect on STW programming that is small and of marginal statistical significance (.05 < p-value < .10). We interpret the result to indicate that more affluent communities are only slightly less likely to have STW activities than less affluent ones, other things equal.
Expenditure per Pupil We anticipated that district expenditure per pupil would have a positive effect because high expenditure districts would have more resources for implementing STW activities than low expenditure districts. Contrary to this expectation, however, the results in Table 3 show that the variable is statistically insignificant.
Percentage District Revenues Non-Local
All school districts receive some revenue from state and federal sources. Since federal funds for broad-based STW programs were limited in 1992, we hypothesized that schools with high percentages of non-local funding would have less STW programming than schools with low percentages (holding constant overall expenditure per pupil). This conjecture is supported by the results in the table, though again the effect is not a strong one. Increasing a school's non-local percentage by one standard deviation (18.82) only decreases its STW index by 0.32.
State and Region
As noted earlier, it seems reasonable to suppose that schools might differ in STW involvement because of differing economic or political conditions in their states or regions. At the state level, some states were much more involved in the STW movement than others in 1992. At the regional level, some areas in the country were harder hit by the recession of the early 1990s than others. We attempted to control for these influences primarily by using dummy variables of various kinds (including, at one point, separate dummies for each state). In general, we found state and region effects to be of marginal statistical and substantive significance. The results reported in Table 5 are representative.
State Received Early STWOA Grant This is a dummy variable taking on the value one if the state in which the school was located received its first STWOA grant in the first two years of the program (1994 or 1995) and zero otherwise. We hypothesized that schools in such states would be more likely to have had extensive STW programs in 1992, other things equal, than schools in other states because the state climate was likely to have been more supportive. Our results support this hypothesis, but the effect, though statistically significant, is not large. Being in an early recipient state only increased the STW index by 0.65, other things equal.
Regions These are dummy variables for three of the four census regions. The estimated coefficients show that schools in the South, Midwest, and West regions had slightly more STW programming than schools in the Northeast region, but the differences are small and only one is statistically significant.
Summary and Conclusions
Stepping back from the detail in the preceding section, the principal findings of the study are as follows:
- Differences in STW programming across schools are not random events. A substantial portion of the observed variation can be explained by school characteristics.
- The scope of STW programming does not increase very much with school or district size, even after controlling for the number of grade levels.
- Schools with strong leadership in the areas of innovation, school climate, and cooperation with external constituencies have broader STW programs than schools weak in these leadership dimensions.
- A variety of student body, community, state, and regional variables have statistically significant influences on STW programming but the effect sizes tend to be fairly small.
- Student body and community income variables do not seem to have much influence on STW programming.
These results have important implications for both future research and policy. We conclude the paper with a discussion of some of them.
This appears to be the first multivariate study of the determinants of STW programming at the school level using a reasonably representative national sample of public schools. It should not be the last. The work reported here is based on data from 1992, prior to the passage of the STWOA. Research on STW implementation using more recent data is obviously called for. Such research will be particularly crucial for determining the long-run impact of the STW movement whose future at this moment is quite uncertain. Also, there is a great need for the development and estimation of depth models - i.e., models that seek to explain the intensity of student involvement in STW at the school level rather than just the presence or absence of program elements in the school. The breadth model presented in this paper does not distinguish between programs that are "a mile wide and an inch deep" and those that both offer a broad range of activities and serve large numbers of students. Finally, in the estimation of both breadth and depth models there is a need for further experimentation with independent variables, particularly student body and community characteristics. The limitations of our database in these areas provide obvious opportunities for future research.
The policy implications of our work are twofold. First, the results on school size provide indirect support for the movement to make the public high schools more intimate learning environments by making them smaller. The coefficients in Table 5 indicate that size reduction at either the school or district level will not reduce the scope of STW offerings very much. Second, the statistical and substantive significance of the leadership coefficients in the table suggest strongly that developing strong academic leaders, either through pre-service or in-service training, is vital to the successful implementation of STW. The results are particularly suggestive with respect to the ability to establish cooperative relationships with external constituencies. This ability has not always been emphasized in the research on or the teaching of educational leadership. Perhaps this underemphasis is a mistake, not only from the perspective of STW programming but other large-scale education reforms as well.
This research is being supported in part by the Office of Educational Research and Improvement of the U.S. Department of Education through a contract to the Mid-Atlantic Laboratory for Student Success at Temple University, and in part by the Center for Research in Human Development and Education at Temple University. The opinions expressed do not necessarily reflect the positions of the supporting agencies, and no official endorsement should be inferred. We wish to thank Elizabeth Davis, Jessica Gudmundsen, and Christin Scholfield for outstanding research assistance and Margaret Wang for valuable comments and on-going support.
Boesel , D., Hudson, L., and Masten C. (1994). Final report of the National Assessment of Vocational Education, Volume II. Washington, DC: U.S. Department of Education.
Bohrnstedt , G., & Knoke, D. (1994). Statistics for social data analysis. Itasca, IL: F. E. Peacock.
Chichura , E. (1989). The role of the board of education in the process of resource allocation for public schools . Paper presented at the annual meeting of the American Educational Research Association, San Francisco, CA.
Commission on the Skills of the American Workforce (1990). America's choice: High skills or low wages. Washington, DC: National Center on Education and the Economy.
Davis , C., & Sonnenberg, W. (1993). Programs and plans for the national center for education statistics. Washington, DC: U.S. Department of Education.
Fullan , M. (1993). Change forces: Probing the depths of educational reform. New York, NY: Falmer Press.
Fullan , M. (1997). What's worth fighting for in the principalship? New York, NY: Teachers College Press.
Grubb , W. N. (1996). Working in the middle: Strengthening education and training for the mid-skilled labor force. San Francisco, CA: Jossey-Bass Publishers.
Haller , E., Monk, D., Spotted-Bear, A., Griffith, J., & Moss, P. (1990). School size and program comprehensiveness: Evidence from High School and Beyond. Educational Evaluation and Policy Analysis , 12 (2), 109-120.
Hallinger , P., & Heck, R. (1996). Reassessing the principal's role in school effectiveness: A review of the empirical research, 1980-1995. Educational Administration Quarterly , 32 (1), 5-44.
Hartman , W. (1988). Understanding resource allocation in high schools . Technical Report No. 143, Center for Educational Policy and Management, University of Oregon, Eugene, OR.
Helmstadter , G. (1964). Principles of psychological measurement . New York, NY: Appleton-Century-Crofts.
Hershey , A., Hudis, P., Silverberg, M., & Haimson, J. (1997). Partners in progress: Early steps in creating school-to-work systems . Princeton, NJ: Mathematica Policy Research, Inc.
Hershey , A., Silverberg, M., & Haimson, J. (1999). Expanding options for students: Report to Congress on the National Evaluation of School-to-Work Implementation . Princeton, NJ: Mathematica Policy Research, Inc.
Jennings , J. (Ed.). (1995). National issues in education: Goals 2000 and school-to-work. Bloomington, IN: Phi Delta Kappa International; Washington, DC: The Institute for Educational Leadership.
Jones , T. (1985). Introduction to school finance techniques and social policy. New York, NY: Macmillan.
McCormick , A., Tuma, J., & Houser, J. (1995). Vocational course taking and achievement: An analysis of high school transcripts and 1990 NAEP assessment scores . Washington, DC: National Center for Education Statistics, U.S. Department of Education.
National Commission on Excellence in Education (1983). A nation at risk: The imperative for educational reform . Washington, DC: U.S. Government Printing Office.
Olson , L. (1997). The school-to-work revolution: How employers and educators are joining forces to prepare tomorrow's skilled workforce . Reading, MA: Addison Wesley.
Powell , A., Farrar, E., & Cohen, D. (1985). The shopping mall high school. Boston, MA: Houghton Mifflin.
Secretary's Commission on Achieving Necessary Skills (1991). What work requires: A SCANS report for America 2000 . Washington, DC: U.S. Department of Labor.
Silverberg , M., Haimson, J., & Hershey, A. (1998). Building blocks for a future school-to-work system: Early national implementation results . Princeton, NJ: Mathematica Policy Research, Inc.
Stern , D., Finkelstein, N., Stone, J., Latting, J., & Dornsife, C. (1995). School to work: Research on programs in the United States. Washington, DC: The Falmer Press.
Stull , W., & Sanders, N. (1999). School-to-work: Where are we, what do we know, where are we going? Paper presented at the annual meeting of the American Education Research Association, Montreal, Canada.
Tyack , D. (1974). The one best system. Cambridge, MA: Harvard University Press.
Urquiola , M., Stern, D., Horn, I., Dornsife, C., Chi, B., Williams, L., Merritt, D., Hughes, K., & Bailey, T. (1997). School to work, college, and career: A review of policy, practice, and results 1993-1997. Berkeley, CA: National Center for Research in Vocational Education.
William T. Grant Foundation Commission on Work, Family, and Citizenship. (1988). The forgotten half: Pathways to success for America's youth and young families. Washington, DC: William T. Grant Foundation.
WILLIAM STULL is Professor of Economics, Chair of the Economics Department, and Senior Research Associate in the Center for Research on Human Development and Education, Temple University, Ritter Annex - 9th Floor, 1301 Cecil B. Moore Avenue, Philadelphia, PA 19122. [E-Mail: firstname.lastname@example.org ]. His research interests are urban economics and the economics of education.
NICHOLAS SANDERS is Research Analyst in the Mid-Atlantic Regional Educational Laboratory for Student Success at Temple University, Ritter Annex - 9th Floor, 1301 Cecil B. Moore Avenue, Philadelphia, PA 19122. [E-Mail: email@example.com ]. His research interests are school reform, economics of education, and educational measurement.
JUDITH STULL is Associate Professor of Sociology, LaSalle University, and Senior Research Associate in the Center for Research on Human Development and Education, Temple University, Ritter Annex - 9th Floor, 1301 Cecil B. Moore Avenue, Philadelphia, PA 19122. [E-Mail: firstname.lastname@example.org ]. Her research interests are the sociology of education, school reform, and educational technology.