The Case Against School Choice: Politics, Markets and Fools by Kevin B. Smith & Kenneth J. Meier

Author:Kevin B. Smith & Kenneth J. Meier [Smith, Kevin B. & Meier, Kenneth J.]
Language: eng
Format: epub
Tags: Political Science, Public Affairs & Administration, General
ISBN: 9781315286556
Google: jIkYDQAAQBAJ
Goodreads: 32205942
Publisher: Routledge
Published: 1995-03-01T00:00:00+00:00

Notes

1. Witte's argument deals with Catholic, as opposed to all private, schools. Since Catholic schools constitute the majority of private schools, this generalization is not seen as doing too much violence to Witte's more focused argument.

2. Based on the high school and beyond data, Witte reports that blacks and Hispanics made up 10.9 percent of private school enrollments and 20.8 percent of public school enrollments. Handicapped students made up 1.5 percent of private school enrollments and 4.9 percent of public school enrollments.

3. The methodology employed here is based on previous work (Meier and Smith 1994) used to examine the causal links among politics, bureaucracy, and minority employment.

4. Despite a large Hispanic population, Florida is not an especially Catholic state.

5. See the extensive literature on "white flight" referenced in Meier, Stewart, and England (1989).

6. It should be clear that the cohort's previous pass rate and the lagged version of the dependent variable are not the same thing. The previous pass rate represents the percentage of the cohort who passed the eighth-grade test. The lagged version of the dependent variable represents how many students passed the tenth-grade test at t— 1, t—2, etc.

7. The fully specified models are as follows.

Private school enrollment model:

PRI = b1LPRI + b2LPERF + b3PBSTUD + b4CATH% + b5INC

Public school performance model:

PERF = b1PASS + b2LPERF + b3LPR1 + b4WGIF + b5INC + b6FREELUN

Where:

PRI = private school enrollment measured as a percentage of total district enrollment attending private schools.

PERF = public school performance measured as a percentage of students passing standardized tenth-grade math or communications tests.

LPRI = lagged private school enrollment (i.e., PRI at t-1).

LPERF = lagged public school performance (i.e., PERF at t — 1).

PBSTUD = percentage of district enrollment that is black.

CATH% = percentage of district residents who are Catholic.

INC = mean district family income.

GIF = percentage of students in gifted classes.

PASS = cohorts pass rate on previous test.

FREELUN = percentage of students receiving free lunches.

8. Missing data prevented including more time points in the public school performance model.

9. Pooled designs often require substantial statistical manipulation to over- come the methodological challenges inherent in the technique. In this instance, however, the approach is fairly straightforward. An OLS design was considered appropriate because the models contain lagged dependent variables as predictors and because the design is unit (i.e., cross-sectional) rather than time dominant (see Stimson 1985). In some models dummy variables for years were included as controls for autocorrelation, and dummy variables for individual school districts were included where diagnostics indicated outliers. In general, when such corrective controls are made, diagnostics show neither autocorrelation nor hetero skedasticity to be at unacceptable levels in the final models (see Stimson 1985; Hsiao 1986, chap. 4). The single exception was the public school performance model using the math-based performance variable. Even after employing year dummies, diagnostics indicated autocorrelation in this model was high enough to make the estimates unreliable. Accordingly, to correct for autocorrelation in this model it was necessary to use a generalized least-squares technique, and GLS coefficients are reported.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.