```Date: Sat, 28 Jun 2008 12:13:15 -0400 Reply-To: "Ingrid K. Friberg" Sender: "SAS(r) Discussion" From: "Ingrid K. Friberg" Subject: Contrast in Proc Mixed (GLIMMIX) Hi, I am trying to learn about contrasts. I have a variable (woodfinal) with three levels (0,1,2). And am trying to use the contrasts to replicate the F-test from the type 3 test of fixed effects to help me understand and eventually gat joint F-test for two or more variables (as I did in NLMIXED). The contrast that I used is below, but I need two degrees of freedom, and can't figure this out. I saw soemthing about generating an equation to think it through, and that gave me the idea 3, -3, 0, but that gave me the same answer as this one. Or is this not doing it because I have a random effect? Thanks... contrast 'wood' woodfinal 1 -1 0; Effect woodfinal Estimate Error DF t Value Pr > |t| Intercept -1.0541 0.1748 183 -6.03 <.0001 babyagedays 0.01269 0.002082 2288 6.09 <.0001 babyagedays1 -0.02900 0.007147 2288 -4.06 <.0001 babyagedays2 0.05608 0.01870 2288 3.00 0.0027 woodfinal 0 -0.2243 0.1267 2288 -1.77 0.0767 woodfinal 1 0.2923 0.2186 2288 1.34 0.1812 woodfinal 2 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F babyagedays 1 2288 37.13 <.0001 babyagedays1 1 2288 16.47 <.0001 babyagedays2 1 2288 9.00 0.0027 woodfinal 2 2288 3.44 0.0322 Contrasts Num Den Label DF DF F Value Pr > F wood 1 2288 5.59 0.0181 ```

Back to: Top of message | Previous page | Main SAS-L page