s-news
[Top] [All Lists]

three questions

To: s-plus news <s-news@lists.biostat.wustl.edu>
Subject: three questions
From: jose Bartolomei <surfprjab@hotmail.com>
Date: Tue, 4 Mar 2008 14:36:40 +0000
Importance: Normal

Dear S-plus experts

 

I have three questions

 

1)      It is possible to manipulate or use the crosstabs (ct) result values as separate entities (eg., a<- first row, last column)? I will like to automate the process of adding a ct value to the parameter of another function.

 

2)      Does a weight function to be use with ct exist?  I need to use a vector that represents a weight for my data.

 

3)      Currently I want to conduct a binary logistic model.

 

My dependent variable is a factor with 3 categories 1, 2 & 7.

 

To have only two (2) categories I conducted

 

  Adult.HistoryB<-(Adult.History[!is.element(asthma2, c("7")),])

 

My problem is that when I run:

 

attach(Adult.HistoryB)

ddist<-datadist(asthma2, age, sex, income2)

options(datadist='ddist')

lrm(asthma2~sex, data="">

 

I obtain:

 

Warning messages:

  Length of longer object is not a multiple of the length of the shorter object in: e1 != e2

Problem in fitter(X, Y, penalty.matrix = penalty..: program logic error 1

 

If I run the function table in my dependent var the category 7 is still registered in the table.

 

 table(asthma2)

   1       2          7

 776     3341    0

 

Could be that the problem?

How can I solve the warning message?

 

I ran lrm with the three categories and it run adequately.

 

Thanks very much,

 

Jose



Helping your favorite cause is as easy as instant messaging. You IM, we give. Learn more.
<Prev in Thread] Current Thread [Next in Thread>