R help - Dec 2009 - R ANOVA question

Hi All,

I'm fitting an intercept ANOVA model using R. I'm using following
contrast
statement, so my first level is considered as base level.

options(contrasts = c("contr.treatment", "contr.treatment"))
My data has two factors are "Month" and "WeekDay". Due to
default alphabetic
ordering, the Month(April) and WeekDay(Friday) are considered as base levels
and hence their coefficient is not shown in the ouput. Other month
coefficient's are printed after subtracting April's coefficient from
them;
similarly this holds for weeks' coefficients.

My Question is: How to get back the original month coefficients (including
April's coef.). So that I can find seasonal effects (for e.g. I'll
average
Feb, Mar, Apr to get an estimate for "SPRING" season). Any suggestions
on
how can I do this?

Thanks,
Xin

	[[alternative HTML version deleted]]

On Dec 6, 2009, at 1:14 PM, Xin Ge wrote:
> Hi All,
>
> I'm fitting an intercept ANOVA model using R. I'm using following  
> contrast
> statement, so my first level is considered as base level.
>
> options(contrasts = c("contr.treatment",
"contr.treatment"))
> My data has two factors are "Month" and "WeekDay". Due
to default
> alphabetic
> ordering, the Month(April) and WeekDay(Friday) are considered as  
> base levels
> and hence their coefficient is not shown in the ouput. Other month
> coefficient's are printed after subtracting April's coefficient
from
> them;
> similarly this holds for weeks' coefficients.
You can specify no-intercept models:

?formula
------

?weekdays
?months
?quarters

You can specify the order of your levels with factor:

wkfac <- factor(wkfac, levels=weekdays(Sys.Date()+1:7) )
>
> My Question is: How to get back the original month coefficients  
> (including
> April's coef.). So that I can find seasonal effects (for e.g. I'll
> average
> Feb, Mar, Apr to get an estimate for "SPRING" season). Any  
> suggestions on
> how can I do this?
  You seem to need further reading about models that incorporate  
seasonality. RSiteSearch should help provide some worked examples.
>
> Thanks,
> Xin
>
--

David Winsemius, MD
Heritage Laboratories
West Hartford, CT