Geek Answers Handbook

TukeyHSD after inside factors ANOVA

I run the following commands

a1 <- aov(Ratio~(Condition*Event) + Error(Participant/(Condition*Event)), Data) summary(a1)

This outputs the correct results (ANOVA tables, etc.).

When I run:

TukeyHSD (a1, Status)

I get: Error in UseMethod ("TukeyHSD"): there is no applicable method for "TukeyHSD"

Why does ANOVA work, but not TukeyHSD? There are lines inside the variable factors (the condition has 3 levels, and the event has 4 levels).

EDIT: When I repeat aov without Terminal Error, this works, however Tukey does not show a significant difference between any of the pairs (ANOVA was significant for Confidence). Does this mean that Tuki corrects multiple comparisons?

+4
source share
3 answers

aov() with Error() returns an object with class "aovlist" "listof" . TukeyHSD does not seem to have a method for any of these classes.

+5
source

As an alternative to traditional ANOVA repeated measurements for intra-articular design, you can use a linear approach using mixed effects. It is increasingly used in the scientific community and avoids some of the shortcomings of ANOVA, allowing you to create more complex error structures. With a continuous response variable, nlme is enough, but you can also use lme4 , which also allows you to deal with categorical response variables. For multiple comparisons (including Tukey post-hoc tests), the multcomp package (see glht() ) can be used with the mixed-effects models equipped with nlme::lme , as described here: Repeated ANOVA measurements using R.

One short note about your design: if your response variable (dependent), Ratio , is a fraction or a limited value, you might consider using another link function.

+4
source 
 library(laercio) some.aov <- aov(Ratio~(Condition*Event) + Error(Participant/(Condition*Event)), Data) anova(some.aov) LTukey(some.aov,"Condition")
+2
source

More articles:


All Articles