I am trying to test for fixed factor main effects in an unbalanced mixed effects model but when I fit the reduced model for "mic" factor effects, the extra degrees of freedom are being allocated to a nested term rather than the residuals. The model has inc, mic and spp are independent variables and vial nested within spp. inc and spp are already coded as factors since they were entered as text in the dataframe "t".
##Full Model
fit<-lm(cmm~inc+factor(mic)+spp/factor(vial),t) anova(fit)
Analysis of Variance Table Response: cmm Df Sum Sq Mean Sq F value Pr(>F) inc 1 2099.3 2099.3 387.6854 < 2.2e-16 *** factor(mic) 2 502.3 251.2 46.3823 < 2.2e-16 *** spp 2 7947.1 3973.5 733.8008 < 2.2e-16 *** spp:factor(vial) 94 3746.9 39.9 7.3612 < 2.2e-16 *** Residuals 1112 6021.5 5.4 ## Reduced Model
fitredmic<-lm(cmm~inc+spp/factor(vial),t) anova(fitredmic)
Analysis of Variance Table Response: cmm Df Sum Sq Mean Sq F value Pr(>F) inc 1 2099.3 2099.3 387.6854 < 2.2e-16 *** spp 2 8149.7 4074.9 752.5124 < 2.2e-16 *** spp:factor(vial) 96 4046.6 42.2 7.7843 < 2.2e-16 *** Residuals 1112 6021.5 5.4 ##So the degrees of freedom for the comparison are 0:
anova(fit,fitredmic)
Analysis of Variance Table Model 1: cmm ~ inc + factor(mic) + spp/factor(vial) Model 2: cmm ~ inc + spp/factor(vial) Res.Df RSS Df Sum of Sq F Pr(>F) 1 1112 6021.5 2 1112 6021.5 0 -9.0949e-13 Is this meaningful? How can I fix this? Thank you! Wes ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
