Free Trial

Safari Books Online is a digital library providing on-demand subscription access to thousands of learning resources.

Share this Page URL
Help

Chapter 38. The GLIMMIX Procedure > Examples: GLIMMIX Procedure - Pg. 2418

2418 ! Chapter 38: The GLIMMIX Procedure The (marginal) covariance matrix of the data is shown in Output 38.17.4 to confirm that it matches the A matrix given earlier. Output 38.17.4 R-Side Covariance Matrix Estimated V Matrix for Subject 1 Row 1 2 3 4 5 6 Col1 0.007462 -0.00522 0.01023 Col2 -0.00522 0.01820 -0.01059 Col3 0.01023 -0.01059 0.01600 0.01126 -0.00910 0.01579 -0.00910 0.03949 -0.02000 0.01579 -0.02000 0.02617 Col4 Col5 Col6 The LSMESTIMATE statement specifies three linear functions. These set equal the parameters from the groups. The step-down Bonferroni adjustment requests a multiplicity adjustment for the family of three tests. The FTEST option requests a joint test of the three estimable functions; it is a global test of parameter homogeneity across groups. Output 38.17.5 displays the result from the LSMESTIMATE statement. The joint test is highly significant (F D 30:52, p < 0:0001). From the p-values associated with the individual rows of the estimates, you can see that the lack of homogeneity is due to group differences for 1 , the log clearance. Output 38.17.5 Test of Parameter Homogeneity across Groups Least Squares Means Estimates Adjustment for Multiplicity: Holm Standard Error 0.1368 0.2402 0.2054 Effect Parameter Parameter Parameter Label beta1 eq. across groups beta2 eq. across groups beta3 eq. across groups Estimate -0.5560 0.04443 -0.1788 DF 92 92 92 t Value -4.06 0.18 -0.87 Pr > |t| 0.0001 0.8537 0.3862 Least Squares Means Estimates Adjustment for Multiplicity: Holm Effect Parameter Parameter Parameter Label beta1 eq. across groups beta2 eq. across groups beta3 eq. across groups Adj P 0.0003 0.8537 0.7725 Least Squares Means Ftest Num DF 3 Den DF 92 Label Homogeneity F Value 30.52 Pr > F <.0001