[R] Anova interaction not tested

[Bert Gunter]

Gabe:

Don't be silly. lme is not appropriate here --you have only one stratum!.
Use lm and you'll see what's going on. It also looks like you should do some
reading up on linear models. V&R's MASS or Peter Dalgaard's INTRO to
Statistics with R might be places to start.

Incidentallyy, Version 2.8.1 of R gives for the output of aov:
########
aov(speed~brand*material)
Call:
   aov(formula = speed ~ brand * material)

Terms:
                   brand material Residuals
Sum of Squares  49.86190 13.50167  21.00500
Deg. of Freedom        2        2         2

Residual standard error: 3.240756 
4 out of 9 effects not estimable
Estimated effects may be unbalanced
Warning messages:
1: In model.matrix.default(mt, mf, contrasts) :
  variable 'brand' converted to a factor
2: In model.matrix.default(mt, mf, contrasts) :
  variable 'material' converted to a factor
########
Seems like this should pretty much tell you what's going on -- if you
understand linear models. (and if you don't, why not, since you're using
them?)

Cheers,
Bert


-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] On
Behalf Of Gabriel Murray
Sent: Wednesday, April 08, 2009 4:08 PM
To: Patrick Connolly
Cc: r-help at r-project.org
Subject: Re: [R] Anova interaction not tested

Playing around with this some more, it looks like it may purely be due to
the way that R handles unbalanced data. I will investigate lme for this.

Best,
Gabe

On Wed, Apr 8, 2009 at 3:24 PM, Gabriel Murray <p.editor at gmail.com> wrote:

> Here is some toy data to give an example. Imagine someone is test-riding
> several bicycles to decide which to buy, and measures their average speed
> during each test-ride. They want to assess the main effects and
interaction
> of brand and material on speed.
>
> Speed    Size    Brand        Material
> 20         small    giant        steel
> 25.1      medium    giant        steel
> 25.2      medium  trek        aluminum
> 26         small    cannondale    carbon
> 19.9      large    giant        aluminum
> 21         medium    trek        carbon
> 30         medium    cannondale    carbon
>
> This is the output I get:
>
> > fit = aov(Speed~Brand*Material, data=myData)
> > summary(fit)
>             Df Sum Sq Mean Sq F value Pr(>F)
> Brand        2 49.862  24.931  2.3738 0.2964
> Material     2 13.502   6.751  0.6428 0.6087
> Residuals    2 21.005  10.503
>
>
> I suppose I am missing something very obvious here as to why the
> interaction is not reported, but I don't see what it is.
> Cheers,
>
> Gabe
>
>
>
> On Wed, Apr 8, 2009 at 1:34 PM, Patrick Connolly <
> p_connolly at slingshot.co.nz> wrote:
>
>> On Wed, 08-Apr-2009 at 12:59PM -0700, Gabriel Murray wrote:
>>
>> |> I've noticed with certain datasets that when I try to do an anova and
>> test
>> |> for main effects and interaction for two explanatory variables,
>> sometimes
>> |> the main effect results are given but not the interaction results. For
>> |> example,
>> |>
>> |> ex1 = aov(Score ~ var1*var2, data=myData)
>> |> summary(ex1)
>>
>> Could it be that there are no degrees of freedom left?
>>
>> That's about my best guess without enough information.
>>
>>
>> |> ______________________________________________
>> |> R-help at 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.
>>
>>
>> Nothing you sent is reproducible.
>>
>>
>> --
>> ~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.
>>    ___    Patrick Connolly
>>  {~._.~}                   Great minds discuss ideas
>>  _( Y )_                 Average minds discuss events
>> (:_~*~_:)                  Small minds discuss people
>>  (_)-(_)                              ..... Eleanor Roosevelt
>>
>> ~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.
>>
>
>

	[[alternative HTML version deleted]]

______________________________________________
R-help at 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.