[R-lang] Re: aov in R

[João Veríssimo]

Hi Zhenguang,

Your aov() command seems right to me.
I haven't tried this in SPSS, but in R, I get the same F-value (37.08),
either using t.test(), aov(), or ezANOVA(); see code below.

Maybe check your SPSS analysis to see what's going on there?

Best,
João

## Import data
MT2F1 <- read.table("./MT2-F1.txt", header=T)
str(MT2F1)
## Convert Subject to factor
MT2F1$Subject <- as.factor(MT2F1$Subject)

## Paired t-test
(myttest <- t.test(MT2F1$Average~MT2F1$Perspective, paired=T))
myttest$statistic[[1]]^2

## Repeated-measures ANOVA (with aov())
summary(aov(Average~Perspective+Error(Subject/Perspective), MT2F1))

## Repeated-measures ANOVA (with ezANOVA())
library(ez)
ezANOVA(
  data = MT2F1
  , dv = Average
  , wid = Subject
  , within = Perspective
)

On Fri, 2014-04-04 at 17:50 +0100, Zhenguang Cai wrote:
> Hi,
> 
> I keep having radically different results for LME (especially when
> random slopes are included, due to Barr et al. 2013) between different
> versions of R (and supposedly different lme4 packages)  (I wonder
> anyone else encounter similar problems), so I am turning back to the
> good old ANOVA.
> 
> I had a repeated measure design, with Perspective as  two-level
> within-subject factor (Agent vs. Recipient). So I ran an ANOVA
> analysis in R, using
> 
> aov(Average~Perspective+Error(Subject/Perspective), MT2F1)
> 
> and the following are the results
> 
> ***********************
> Error: Subject
>           Df Sum Sq Mean Sq F value Pr(>F)
> Residuals 22  3.176  0.1444               
> 
> Error: Subject:Perspective
>             Df Sum Sq Mean Sq F value   Pr(>F)    
> Perspective  1 1.1414  1.1414   37.08 3.96e-06 ***
> Residuals   22 0.6772  0.0308                     
> ---
> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
> *************************
> 
> However, I found the F-value was much larger than the F-value (37.08)
> I observed in SPSS (5.92) (using the same data set). I wonder whether
> I used a wrong formula.
> 
> Also, I want to include a covariate "Pnum". I wonder the following is
> the correct formula:
> 
> aov(Average~(Perspective+Pnum)+Error(Subject/Perspective), MT2F1)
> 
> Thanks,
> 
> Zhenguang Cai
> 
> -- 
> Zhenguang Cai
> 
> Department of Cogntive, Perceptual and Brain Sciences
> University College London
> 
> https://sites.google.com/site/zhenguangcai/