A County Level Analysis of Educational Attainment in the United States by Social, Economic and Geographic Variables BY Brandon Hallstrand (University of Wisconsin – Stout) Kunjan Upadhyay (University of Wisconsin - Stout) 2010 Wisconsin Economics Association Annual Conference
Outline Introduction Prior Studies Model Data and Descriptive Statistics Regression Analysis Conclusion Future Work
Introduction Education is Important – Huge Disparities within the country. US is currently Ranked 16 th in Education amongst 26 other OECD Countries. – Organization for Economic Cooperation and Development (OECD) – Dropped from 1 st position in 1995
Figure 1: its “Percentage of Tertiary-Type A Graduates to the Population at the Typical Age of Graduation Measure for 2010,” (Organization for Economic Cooperation and Development, 2010).
Prior Studies Racial, gender cohort dropout rates in Chicago Public Schools (Allensworth & Easton 2001). High school Drop outs and graduation rates in central region (Randel, Moore & Blair 2008). Focus on Specific Regions, gender, race One Study Points Out Data Problems – Hidden Crisis in High School Dropout Rate (Sum et. al 2003).
Full Models
Reduced Models
Data and Descriptive Statistics 1990 VariableCount MeanStDevMinimum Maximum Dropout Rate Per capita personal income yr Lag Edu Spend per Child yr Lead Edu Spend per Child Averaged Edu Spending per child Males per 100 Females Percent White, Non Hispanic Percent Black Percent Hispanic Percent Asian or Pacific Percent Native American Percent Other Race
Data and Descriptive Statistics 2000 VariableCountMeanStDevMinimumMaximum Dropout Rate Per capita personal income yr Lag Edu Spend per Child yr Lead Edu Spend per Child Averaged Edu Spending per child Males per 100 Females Percent White, Non Hispanic Percent Black Percent Hispanic Percent Asian or Pacific Percent Native American Percent Other Race
Data and Descriptive Statistics Panel VariableCountMeanStDevMinimum Maximum Dropout Rate Per capita personal income yr Lag Edu Spend per Child yr Lead Edu Spend per Child Averaged Edu Spending per child Males per 100 Females Percent White, Non Hispanic Percent Black Percent Hispanic Percent Asian or Pacific Percent Native American Percent Other Race
Regression Analysis Used Minitab 16 Statistical Software Best Subsets Chose Models for Simplicity and Fit
NOTE: * : denote the variable is statistically significant at 1% ** : denote the variable is statistically significant at 5% *** denote the variable is statistically significant at 10% Regression Analysis Predictor Per capita personal income (-6.83)*(-6.31)*(-9.7)* 2yr Lag Edu Spend per Child (-5.93)*(-1.26)(-5.04)* 2yr Lead Edu Spend per Child (-0.16)(1.7)***(-0.55) Averaged Edu Spending per Child (-0.55)(-2.96)(-2.42)** Males per 100 Females (-1.55)(-4.61)*(4.34)* Percent White, Non Hispanic (1.76)***(-0.9)(1.69)** Percent Black (2.35)**(1.66)***(2.27)* Percent HispanicN/A N/A(-0.55)(-1.31) Percent Asian or Pacific Island (-0.11)(-0.17)(-1.04) Percent Native American or Alas (4.32)*(1.76)***(2.61)* Present Other Races (4.05)*(3.00)*(4.06)* Midwest (-5.06)*(-2.47)*(-5.16)* South (6.32)*(4.22)*(7.7)* West (-0.14)(-0.7)(-0.74) Year 1990=0, 2000 =1N/A N/A (-4.28)* R-sq23.80%22.50%23.30% R-sq(Adj23.50%22.10%23.10%
Regression Analysis (cont.) Predictor Combined Per capita personal income (-7.18)*(-7.23)*(-10.71)* 2yr Lag Edu Spend per Child (-7.45)*(-4.65)*(-8.16)* Males per 100 Females (-1.56)(4.33)*(4.21)* Percent White, Non Hispanic (2.01)**(2.89)*(3.26)* Percent Black (2.59)*(7.06)*(6.82)* Percent Native American or Alaskan (4.65)*(6.70)*(7.79)* Percent Other Race (4.45)*(7.00)*(8.00)* Midwest (-5.09)*(-1.96)**(-4.88)* South (6.31)*(5.08)*(8.21)* West (-0.19)(-0.55)(-0.67) Year (-7.27)* NOTE: * : denote the variable is statistically significant at 1% ** : denote the variable is statistically significant at 5% *** denote the variable is statistically significant at 10% R-sq23.77%22.07%23.20% R-sq(Adj23.53%21.82%23.07%
Conclusion Local Educational Spending and Per Capita Income have consistent inverse effects – Effective way of reducing High School Dropouts – increase in spending and income from 1990 to 2000 coincides with a substantial decrease in the dropout rates. Whites, blacks, Native Americans and others have positive coefficients – Relative to areas with high numbers of Hispanics and Asians; Areas with high numbers of whites, blacks, Native Americans and or others, have higher dropout rates. – This Differs from Model to model, area to area.
Future Work Better way to manage racial categories – 1990 Data Set Problem – Relative Population Size Vs. Exact Sampling Change in local spending & lagged spending Perhaps Panel Year Value takes away from Spending value
Questions & Comments
Thank You!!!