[R] ftable as latex (with Hmisc?)

[Dieter Menne]

> Dieter Menne wrote:
>> I tried to create an ftable lookalike of the following data set in 
>> LaTeX/Sweave
>> with summary(formula,..), but I could not get it to work for count tables; 
>> numeric tables work fine. 
>> 
> > summary(formula,...,fun="table") does not give the full cross breakdown.
> 

Frank E Harrell Jr <f.harrell <at> vanderbilt.edu> writes:

> See if fun=table (no quotes) or method='cross' work for you.
> Frank
> 

No, it does not, that's why I asked. I had tried both, fun=table (with or
without "") give a individual breakdown by factor, and method="cross" an 
error message (works find with numerical data).

Dieter

> df = data.frame(drug=sample(c("P","V"),100,TRUE),
+                 Hist1=sample(c("Pos","Neg"),100,TRUE),
+                 Hist2=sample(c("Pos","Neg"),100,TRUE))
> ftable(drug~Hist1+Hist2,data=df)                
            drug  P  V
Hist1 Hist2           
Neg   Neg        14  6
      Pos        11 14
Pos   Neg        13 15
      Pos        14 13
> summary(drug~Hist1+Hist2,data=df,fun=table)
drug    N=100

+-------+---+---+--+--+
|       |   |N  |P |V |
+-------+---+---+--+--+
|Hist1  |Neg| 45|25|20|
|       |Pos| 55|27|28|
+-------+---+---+--+--+
|Hist2  |Neg| 48|27|21|
|       |Pos| 52|25|27|
+-------+---+---+--+--+
|Overall|   |100|52|48|
+-------+---+---+--+--+
> summary(drug~Hist2+Hist2,data=df,method="cross")
Error in Y %*% rep(1, ncol(Y)) : requires numeric matrix/vector arguments
In addition: Warning message:
In mean.default(newX[, i], ...) :
  argument is not numeric or logical: returning NA
> 
>