FREC 834: Money and Schools

Critics of public schools argue that simply “throwing more money at bloated, wasteful public education bureaucracies” does nothing to improve public education, and may actually reduce school quality.  Defenders of public schools argue that school systems are seriously underfunded and student performance will improve with more funding.  (Edison Schools, the first and largest for-profit school management company in the US, has never actually made a profit in 30 years!)

Use Google to locate College Board average SAT scores by state for some recent year (2004, 2005 or 2006). Cut and paste these into an Excel spreadsheet. (You may need to copy the PDF table into Notepad, and creating a comma-delimited datafile.)  Sum the verbal and math averages to obtain a combined SAT index of state performance. 

The US Census Bureau publishes annual state-level Public Elementary-Secondary Education Finance Data at http://www.census.gov/govs/www/school.html.  Download the state-level school finance spreadsheet for the same year or a prior year. The first four columns in Table 1 of the finance data contain total revenues by source (total, federal, state and local); copy and paste these four columns into your own worksheet.  Table 4 reports the property-tax component of local revenues; copy this column into your own worksheet. Replace the dashes in your worksheet with zeroes. Subtract local property tax revenues from all local revenues to obtain other (non-property-tax) local revenues. Table 18 contains state populations, public school student enrollments and total personal income.  Copy the total enrollment data into your worksheet. Calculate total K-12 public school funding per pupil, by state. 

Be sure to omit any blank rows and the District of Columbia row from your dataset. 

Graph state SAT performance (Y-axis) against total per-pupil funding (X-axis). Now regress combined SAT performance (dependent variable) against total per-pupil funding. What does this univariate regression model tell you?

If you include the percent test participation rate as a second independent variable in the regression, what does this model tell you? (Now you know why the College Board cautions against comparing states by SAT alone.)

The NAEP (National Assessment of Educational Progress) is a periodic series of standardized tests administered nationwide by the US Dept. of Education (DOE); unlike the SAT, these are specifically designed to assess the performances of states’ public education systems.  Download the state-level 4th and 8th grade reading and math scores for 2005 from DOE’s data explorer (http://nces.ed.gov/nationsreportcard/nde/) into Excel.  Sum the four test scores to create a combined NAEP index of state performance.  Omit Washington DC again.

Graph state NAEP performance (Y-axis) against total per-pupil funding (X-axis). Regress state NAEP performance (dependent variable) against total funding per pupil.  What does this model tell you? 

Now decompose total funding by source: divide federal, state, local property tax and other local revenues by total K-12 enrollment to obtain federal, state, property tax and other local funding per pupil. (Make sure the sum of these matches your total per-pupil funding.) Regress NAEP performance against these four per-pupil funding variables. What does this model tell you? Why do you think the different funding sources affect test scores differently? 

Write up a brief research report, including: an abstract, an introduction explaining the context of the research, a clear statement of the research problem, the data sources you used, the econometric specifications of your hypotheses, interpretation of your estimation results, your conclusions, and perhaps some additional research questions that arise from your findings.  Include graphs and data in an appendix.