We conclude (not exclusively) that demographics and income level play a vital role in determining composite scores. Our next task is to conduct a regression analysis to examine the relationship between the two variables. Using the Minitab program, we plugged in the variables and came up with the following analysis:
Regression Analysis: Spending per pupil ($) versus Composite Score
(chart)
The regression estimate equation given above is where Y equals the composite score and X equals the spending per pupil. The regression analysis tells us that the R- Square is at 11.6% meaning that the relationship between the two variables is not strong. Also, the F value is at 4.33 is just another indicator that the relationship is indeed very weak. The P-value at 0.003 tells us that there is no significance. Another interesting point with this regression analysis is its unusual observations (possible outliers?). A total of three unusual observations were listed indicating states that either spent much more or much less money than the average. We decided that it might prove worthwhile to conduct another regression analysis omitting these unusual observations and came up with the following data:
(chart)