Re: Alternative to model.tables.aov?

[Pravin Jadhav <pravinj@gmail.com>]

Dave,

 

In fact, I tried se.contrast as well. Unfortunately, I could not get it work. A friend suggested using the lme procedure instead of aov procedure and helped me to set up the mixed model to obtain similar results. It enabled me to obtain the estimates and SEs fairly easily. I don't know the details about your model, but consider using lm/lme to do the job.

 

Pravin 

 

On 1/26/06, Thompson, David (MNR) <david.thompson@mnr.gov.on.ca > wrote:

Thank you Pravin,

I now get the means tables but, no standard errors for unbalanced
design.

The output included the comments:
   "Standard error information not returned as design is unbalanced."
   "Standard errors can be obtained through se.contrast.",

but I don't know how to setup the required contrast.obj in
   se.contrast.aov(object, contrast.obj).

I have a full set of contrasts on two factors (dfs are 5 and 3) so, it
appears
that I need the matrix form. If I understand the help on
se.contrast.aov ,
   ". . . Multiple contrasts are specified in the matrix form;
   each column of the matrix defines a contrast.",

I will need to expand the matrices that I assigned to the data.frame
as contrasts to the full length of my data? Is this correct?

Thanks again, DaveT.

-----Original Message-----
From: Pravin Jadhav [mailto:pravinj@gmail.com ]
Sent: January 26, 2006 1:39 PM
To: Thompson, David (MNR)
Cc: s-news
Subject: Re: [S] Alternative to model.tables.aov?


Dave,

I had similar problem last week with model.tables  and I learned that
the problem is a result of the line of code 'oldClass(ctable) <-
"table"' in print.tables.aov():
..
names(dimnames(ctable)) <- names(dimnames(table))
assign("ctable", ctable, where = 1)
#PROBLEM WITH LINE BELOW
oldClass(ctable) <- "table"
print(ctable, digits = digits, NA.fill= NA.fill, NaN.fill = NaN.fill,
...)
..
(supposedly this a bug)
I could resolve it after contacting the customer service. If you comment
out that line, the code will run.

Pravin



On 1/25/06, Thompson, David (MNR) < david.thompson@mnr.gov.on.ca > wrote:

Hello,

How does one extract means and standard errors from a
split-plot model that happens to be unbalanced?

Thanks, DaveT.
**********************************************************
Silviculture Data Analyst
Ontario Forest Research Institute
Ontario Ministry of Natural Resources
Sault Ste. Marie, Ontario, Canada
david.thompson@mnr.gov.on.ca
**********************************************************
--------------------------------------------------------------------
This message was distributed by s-news@lists.biostat.wustl.edu.  To
unsubscribe send e-mail to s-news-request@lists.biostat.wustl.edu with
the BODY of the message:  unsubscribe s-news
--------------------------------------------------------------------
This message was distributed by s-news@lists.biostat.wustl.edu .  To
unsubscribe send e-mail to s-news-request@lists.biostat.wustl.edu with
the BODY of the message:  unsubscribe s-news