Well, the F value is just MS_effect/MS_error. So you run the original model, run the subset models, but disregard the F values. To calculate the correct F values, divide MS_effect from the subset model by the MS_error term from the original (full) model. It's not really a matter of learning how to do it in R -- you're just doing simple division.
-Ista On Sun, Dec 20, 2009 at 12:28 PM, Or Duek <ord...@gmail.com> wrote: > Thank you very much Ista. > Can you please be a bit more specific as to using the overall error. I don't > know how to actually do it in R. > thank you, > Or. > > On Sun, Dec 20, 2009 at 7:09 PM, Ista Zahn <istaz...@gmail.com> wrote: >> >> Hi Or, >> I understand your question (and am not sure what the confusion is >> about actually). Just like you said, you want to know the effect of A >> at B=1, and the effect of A at B=2. In this case, you want to know if >> drug has a significant effect for those with strain one, and whether >> drug has a significant effect for those with strain two. So the good >> part is, I understand the question. >> >> >> The bad part is that unfortunately I don't know how to do it with >> aov(). Also, I'm not a stats guru and could easily be wrong about the >> following advice. You've been warned! >> >> If you didn't have a mixed model, I would tell you to run the model >> using lm() twice, setting drug = 1 as the reference group the first >> time, and drug = 2 as the reference group the second time, but this >> won't work in your case. The best I can offer is a suggestion to run >> the model separately for each level of drug. Note that you can and >> should use the error term from the overall model though -- you will >> have to do this "by hand" (just divide MS_effect from the subset >> model by MS_error from the full model, and evaluate using df_error >> from the overall model). So basically, I'm suggesting that you do >> >> Data.drug1 <- subset(Data, drug == "1") >> aov.model.drug.1 <- aov(dependent~(exposure*strain) + >> Error(subject/exposure) + (strain), data=Data.drug1) >> summary(aov.model.drug.1) >> >> Data.drug2 <- subset(Data, drug == "2") >> aov.model.drug.2 <- aov(dependent~(exposure*strain) + >> Error(subject/exposure) + (strain), data=Data.drug2) >> summary(aov.model.drug.2) >> >> Good luck! >> >> -Ista >> >> On Sun, Dec 20, 2009 at 11:35 AM, Or Duek <ord...@gmail.com> wrote: >> > For some reasion I wasn't able to use TukeyHSD - I think because I need >> > to >> > set the different levels under a second variable. >> > Tukey only helps me when I have more than 2 levels of same variable. >> > Thanl you. >> > On Sun, Dec 20, 2009 at 6:32 PM, S Devriese <sdmaill...@gmail.com> >> > wrote: >> > >> >> On 12/20/2009 04:56 PM, Or Duek wrote: >> >> > I don't have missing data. >> >> > about what I need. >> >> > Lets say the drug*strain interaction is significant - now I want to >> >> > check >> >> > for drug under the levels of strain - compare drug 1 and 2 only on >> >> > strain >> >> 1 >> >> > and then only on strain 2. >> >> > Or I'd like to compare the strains under levels of exposure. >> >> > This is the kind of data I fail to see in summary() but it is >> >> > important >> >> to >> >> > understand the interactions. >> >> > thank you. >> >> > >> >> >> >> Do you main pairwise multiple comparison tests like Tukey Honest >> >> Significant Difference tests? Then you could use TukeyHSD in the stats >> >> package or see the DTK package (Dunnett-Tukey-Kramer Pairwise Multiple >> >> Comparison Test Adjusted for Unequal Variances and Unequal Sample >> >> Sizes) >> >> >> >> Stephan >> >> >> > >> > [[alternative HTML version deleted]] >> > >> > ______________________________________________ >> > 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. >> > >> >> >> >> -- >> Ista Zahn >> Graduate student >> University of Rochester >> Department of Clinical and Social Psychology >> http://yourpsyche.org > > -- Ista Zahn Graduate student University of Rochester Department of Clinical and Social Psychology http://yourpsyche.org ______________________________________________ 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.
