The relevant reference is "Exegeses on Linear Models" by Bill Venables,
http://www.stats.ox.ac.uk/pub/MASS3/Exegeses.pdf
and for you case particularly section 5.1. on "'Type III' sums of
squares"
Best,
Bendix
----------------------
Bendix Carstensen
Senior Statistician
Steno Diabetes Center
Niels Steensens Vej 2
DK-2820 Gentofte
Denmark
tel: +45 44 43 87 38
mob: +45 30 75 87 38
fax: +45 44 43 07 06
[log in to unmask]
www.biostat.ku.dk/~bxc
----------------------
> -----Original Message-----
> From: A UK-based worldwide e-mail broadcast system mailing
> list [mailto:[log in to unmask]] On Behalf Of Evangelos
> Kontopantelis
> Sent: Friday, April 28, 2006 10:09 AM
> To: [log in to unmask]
> Subject: QUERY: ANOVA and interaction of factors
>
>
> Hi all
>
> I have a problem when executing a 2-way ANOVA. Let's say I
> want to find out about the significance of factors "year" &
> "practice" regarding variable "work". When I execute a 2-way
> ANOVA without including the interaction effect
> ("year"*"practice"), both factors seem significant. However,
> when I add the interaction only "practice" turns out to be
> significant - I was expecting either "year" or the
> interaction be significant as well (Homogeneity was not
> rejected btw). Anyway, can somebody point to a couple of
> references that have come across something like that? What is
> the "proper" way to confront this issue? Include the
> interaction in step A and since it's not significant remove
> it from the model and re-execute in step B? Which results
> should I trust?
>
> Thank you in advance
>
> Evan
>
>
> Evangelos Kontopantelis
> Research Associate in Statistics
> National Primary Care Research & Development Centre
> 5th Floor, Williamson Building
> University of Manchester
> Oxford Road
> Manchester
> M13 9PL
> 016127-57646
> [log in to unmask]
> <mailto:[log in to unmask]>
>
|