ABSTRACT
The regression output produced by PROC REG for the maths data presented in Table 8.2 is shown in Figure 8.6.
Model: MODEL1 Dependent Variable: SMATHS
Analysis of Variance Source DF Sum of Squares Mean Square F Value Prob>F Model 1 827.76424 827.76424 32.680 0.0004 Error 8 202.63576 25.32947 C Total 9 1030.40000
A refinement on the regression plot which shows the fitted regression line (see predicted vs. explanatory variable in Figure 8.1) can be produced by using PROC GPLOT. This produces a least squares fitted line and prints this on the plot. This output would be suitable for inclusion in a report or research paper. SAS code to produce the fitted regression line as well as plots of the 95 per cent confidence intervals for a mean response and a 95 per cent prediction interval (individual response), is shown in Figure 14, Appendix A3. The output for these three plots is shown in Figures 8.7a-c. The 95 per cent confidence interval for the mean standardized maths score for the mean teachers’ score is narrower than the comparable 95 per cent confidence interval for individual observations (compare Figures 8.7b and 8.7c).
