Melissa Evanson
2010-Mar-09 22:21 UTC
[R] Tukey test for Mixed Effects Model with more than 1 fixed effect?
I am trying to decipher, via post hoc test (Tukey), which of my sites differ from eachother. I have 4 sites, 2 sets of In vs Out (MPA) in separate Regions. Therefore my Mixed Effects Model code has 2 fixed effects: CB.lme <- lme(AsinCB~ In_Out*Region, random = (~1| site.trans/Quadrat) , data = Subsampled_props, control = lmeControl(maxIter = 500, msMaxIter = 500, msMaxEval = 500)) When I run the summary and anova output, I still need to differentiate which categories are the same/different and the results only tell me for each subgroup. i.e., I know that there is a sig difference in coral branching In vs Out of the MPA but I don't know if this is in both regions or only one and if so, which one! I've got the following code to run a Tukey test, but it doesn't allow for more than one fixed effect and I have two : In_Out*Region. As is it only works for one or the other. summary(glht(CB.lme,linfct=mcp(In_Out="Tukey"))) Does anyone know how to incorporate more than one fixed effect in lme post hoc tests? Thanks! Melissa Evanson The summary output for the model: Linear mixed-effects model fit by REML Data: Subsampled_props AIC BIC logLik -1200.400 -1166.024 607.1998 Random effects: Formula: ~1 | site.trans (Intercept) StdDev: 0.06543219 Formula: ~1 | Quadrat %in% site.trans (Intercept) Residual StdDev: 0.1171801 0.04607132 Fixed effects: AsinCB ~ In_Out * Region Value Std.Error DF t-value p-value (Intercept) 0.2317220 0.02211590 967 10.477617 0.0000 In_OutOut -0.2061529 0.03131558 36 -6.583079 0.0000 RegionB -0.0390728 0.03131977 36 -1.247544 0.2203 In_OutOut:RegionB 0.1460696 0.04433322 36 3.294813 0.0022 Correlation: (Intr) In_OtO ReginB In_OutOut -0.706 RegionB -0.706 0.499 In_OutOut:RegionB 0.499 -0.706 -0.706 Standardized Within-Group Residuals: Min Q1 Med Q3 Max -1.01939542 -0.20739978 -0.03831860 0.23440502 1.45546317 Number of Observations: 1007 Number of Groups: site.trans Quadrat %in% site.trans 40 1007 My anova output is: numDF denDF F-value p-value (Intercept) 1 967 172.90334 <.0001 In_Out 1 36 36.13478 <.0001 Region 1 36 2.32907 0.1357 In_Out:Region 1 36 10.85579 0.0022 [[alternative HTML version deleted]]