[R] anova and lm results differ

[Iasonas Lamprianou]

So, should I avoid using the Anova option of the Rcmdr and use aov instead (as I have been doing)?

Dr. Iasonas Lamprianou


Assistant Professor (Educational Research and Evaluation)
Department of Education Sciences
European University-Cyprus
P.O. Box 22006
1516 Nicosia
Cyprus 
Tel.: +357-22-713178
Fax: +357-22-590539


Honorary Research Fellow
Department of Education
The University of Manchester
Oxford Road, Manchester M13 9PL, UK
Tel. 0044  161 275 3485
iasonas.lamprianou at manchester.ac.uk


--- On Tue, 31/8/10, Prof Brian Ripley <ripley at stats.ox.ac.uk> wrote:

> From: Prof Brian Ripley <ripley at stats.ox.ac.uk>
> Subject: Re: [R] anova and lm results differ
> To: "Iasonas Lamprianou" <lamprianou at yahoo.com>
> Cc: r-help at r-project.org
> Date: Tuesday, 31 August, 2010, 17:27
> You have confused 'Anova', 'aov' and
> 'anova'.  R has aov() and 
> anova(), package car (called from Rcmdr) has Anova(). 
> They are not 
> the same things: Anova computes (so-called, by SAS)
> 'type-II or 
> type-III analysis-of-variance tables', something
> deliberately not 
> available in base R.
> 
> See http://www.stats.ox.ac.uk/pub/MASS3/Exegeses.pdf for
> some 
> background and why many distinguished statisticians regard
> your 
> example of the use of Anova() as meaningless.
> 
> In any case, lm (in your subject line) is innocent ... it
> does not 
> compute any form of ANOVA table.
> 
> On Tue, 31 Aug 2010, Iasonas Lamprianou wrote:
> 
> > Dear all
> >
> > I have found that the two "equivalent" commands do not
> produce the same results.
> >    1. (I wrote this command by hand, this is
> what I would do usually)
> >    >summary(aov(eduyrs ~ cntry * edf,
> data=ESS1))
> >               
>   Df Sum Sq Mean Sq  F value   
> Pr(>F)
> >    cntry         
> 1    257  256.65  21.2251 4.243e-06 ***
> >    edf         
>   4  11010 2752.42 227.6296 < 2.2e-16 ***
> >    cntry:edf      4 
>    24   
> 6.03   0.4987    0.7367
> >    Residuals   3205 
> 38754   12.09
> >    ---
> >    Signif. codes:  0 '***' 0.001 '**'
> 0.01 '*' 0.05 '.' 0.1 ' ' 1
> >
> > 2. (I run this from Rcmdr, this is what my students
> would normally do)
> >    >AnovaModel.2 <- (lm(eduyrs ~
> cntry*edf, data=ESS1))
> >    > Anova(AnovaModel.2)
> >    Anova Table (Type II tests)
> >    Response: eduyrs
> >              Sum
> Sq   Df  F value  Pr(>F)
> >    cntry     
>    38    1   3.1158
> 0.07763 .
> >    edf       
> 11010    4 227.6296 < 2e-16 ***
> >    cntry:edf 
>    24    4   0.4987
> 0.73672
> >    Residuals  38754 3205
> >    ---
> >    Signif. codes:  0 '***' 0.001 '**'
> 0.01 '*' 0.05 '.' 0.1 ' ' 1
> >
> >    Unfortunately, the results are different.
> Could anyone, please, explain why?
> >
> >
> > Dr. Iasonas Lamprianou
> >
> >
> > Assistant Professor (Educational Research and
> Evaluation)
> > Department of Education Sciences
> > European University-Cyprus
> > P.O. Box 22006
> > 1516 Nicosia
> > Cyprus
> > Tel.: +357-22-713178
> > Fax: +357-22-590539
> >
> >
> > Honorary Research Fellow
> > Department of Education
> > The University of Manchester
> > Oxford Road, Manchester M13 9PL, UK
> > Tel. 0044  161 275 3485
> > iasonas.lamprianou at manchester.ac.uk
> 
> -- 
> Brian D. Ripley,           
>       ripley at stats.ox.ac.uk
> Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
> University of Oxford,         
>    Tel:  +44 1865 272861 (self)
> 1 South Parks Road,         
>            +44 1865
> 272866 (PA)
> Oxford OX1 3TG, UK           
>     Fax:  +44 1865 272595
>