T HE SAT S CORES TELL A STORY OF THE U.S. D EMOGRAPHIC ECON 240A
T EAM M EMBERS Tore Stautland Bjøndal Chungkai Gao Eric Howard Dan Helling Chien-Ju Lin Matt Mullens
C HOICE OF S TUDY : SAT S CORES Everyone can relate to the SATs No ambiguity in the numbers Source: Moore, The Basic Practice of Statistics, 2nd Edition
W HY ARE SAT S CORES INTERESTING ? Can check dependencies on other variables Standardized test, highly valued as a college admission criteria Locate differences throughout the U.S.
A NGLE OF A TTACK Scatterplots Bar-Charts Regressions Wald Test Test for equality of means
W HAT ARE WE LOOKING AT ? How does teacher salary affect the SAT Scores of a particular state or region? How does region play in to SAT Scores? How does population without a high school education affect the SAT Scores? Does percentage of population taking the test matter?
D EPENDENT V ARIABLE : SAT S CORES Independents: Teacher average salary Region ( 9 different ) Percentage of population taking SAT Percentage of population with no high school diploma Regions: ENC (East North Central) - Illinois, Indiana, Michigan, Ohio, Wisconsin ESC (East South Central) – Alabama, Kentucky, Mississippi, Tennessee MA (Mid-Atlantic) – New Jersey, New York, Pennsylvania MTN (Mountain) – Arizona, Colorado, Idaho, Montana Nevada, New Mexico, Utah, Wyoming NE (New England) – Connecticut, Maine, Mssachusetts, New Hampshire, Rhode Island, Vermont PAC (Pacific) – Alaska, California, Hawaii, Oregon, Washington SA (South Atlantic) – Delaware, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia, West Virginia, District of Columbia WNC (West North Central) – Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, and South Dakota WSC (West South Central) – Arkansas, Louisiana, Oklahoma, and Texas
E XPLORATORY D ATA A NALYSIS
D ESCRIPTIVE S TATISTICS
D ESCRIPTIVE S TATISTICS #2
D ESCRIPTIVE S TATISTICS #3
S TATISTICAL A NALYSIS
S TATISTICAL A NAYLSIS Dependent Variable: SAT Method: Least Squares Date: 12/03/08 Time: 13:07 Sample: 1 51 Included observations: 51 VariableCoefficientStd. Errort-StatisticProb. TEACHER_PAY C R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regression Akaike info criterion Sum squared resid Schwarz criterion Log likelihood F-statistic Durbin-Watson stat Prob(F-statistic) Table 2. Regress SAT on TEACHER_PAY
S TATISTICAL A NAYLSIS Dependent Variable: SAT Method: Least Squares Date: 12/03/08 Time: 13:10 Sample: 1 51 Included observations: 51 VariableCoefficientStd. Errort-StatisticProb. REGION_WSC REGION_WNC REGION_SA REGION_NE REGION_MTN REGION_MA REGION_ESC REGION_ENS C R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regression Akaike info criterion Sum squared resid Schwarz criterion Log likelihood F-statistic Durbin-Watson stat Prob(F-statistic) Table 3. Regress SAT on all the Regions
S TATISTICAL A NAYLSIS Wald Test: Equation: Untitled Null Hypothesis:C(3)=C(10) C(4)=C(10) C(5)=C(10) C(6)=C(10) C(7)=C(10) C(8)=C(10) C(9)=C(10) F-statistic Probability Chi-square Probability Table 4. Wald Test for all the Regions
S TATISTICAL A NAYLSIS Dependent Variable: SAT Method: Least Squares Date: 12/03/08 Time: 13:19 Sample: 1 51 Included observations: 51 VariableCoefficientStd. Errort-StatisticProb. PERCENT_NO_HS PERCENT_TAKING POPULATION-2.76E TEACHER_PAY REGION_WSC REGION_WNC REGION_SA REGION_NE REGION_MTN REGION_MA REGION_ESC REGION_ENS C R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regression Akaike info criterion Sum squared resid Schwarz criterion Log likelihood F-statistic Durbin-Watson stat Prob(F-statistic) Table 5. Regress SAT on all the independent variables
C ONCLUSIONS As education level of state increases, the average SAT Scores decreases As average teacher pay increases test scores tend to decrease Differences in regional test scores are significant Test scores decreased as the percentage of the population taking the test increased