SAS OUTPUT 1 Dependent Variable: HDL1 year 1 HDL cholesterol (mg/dl) The GLM Procedure Least Squares Means Source DF Sum of Squares Mean Square F Value Pr > F Adjustment for Multiple Comparisons: Tukey-Kramer Model A D E G <.0001 physact1 hdl1 lsmean number error f corrected total c in r-square coeff var root mse mean least squares means for effect pr> [t) for HO: LSMean(i)=LSMean(j) Standard Dependent Variable: HDL1 Parameter Estimate Error t Value Pr > It 3 Intercept 49.74806801 8 0.54289412 91.63 1 0.0225 <.0001 physact1 b sas output dependent variable: hdl1 year hdl cholesterol source df sum of squares mean square f value pr> F Standard Model 5568.5601 1113.7120 5.84 <.0001 parameter estimate error t value pr> It Error 2602 496078.2984 190.6527 Intercept 50.09829868 8 0.60033496 83.45 <.0001 corrected total physact1 r-square coeff var root mse hdl1 mean b smoking source df type i ss square f value pr> F 0.07 0.9412 physact1 2 4891.807273 2445.903636 physact1"SMOKING 3 1 0.14499250 B 1.96398691 12.83 <.0001 physact1 b smoking source df type iii ss mean square f value pr> F physact1"SMOKING 1 1 0.00000000 B physact1 2 2340.410047 1170.205024 6.14 0.0022 physact1"SMOKING 1 0 0.00000000 B SMOKING 409.443473 409.443473 2.15 0.1429 physact1" SMOKING 2 246.430781 123.215391 0.65 0.5241