model.tables

[Pravin Jadhav <pravinj@gmail.com>]

I am conducting a simple univariate analysis of a repeated measures design. Where the response for each subject is measured at 3 visits. Here is a simplified version of the data and the model.

 

>test<-data.frame(id=factor(rep(1:20,each=3)),time=factor(rep(1:3,20)),istrt=factor(rep(c(0,1),each=30)),resp=rnorm(60,10,2))   #20 subjects, 3 time points per subject

>test$resp<-ifelse(rnorm(60,0,1)>rnorm(60,0.7,0.7), "NA", test$resp) # missing  data (created empirically here)

>test1<-test[test$resp!="NA",]

>summary(aov(resp~time+istrt+Error(id), data="" #No interaction term between time and treatment

 

The results of this analysis seem reasonable.

 

Then, I was interested in obtaining the fitted means, I tried the following

 

>model.tables(aov(resp~time+istrt+Error(id), data="" type="means")

 

It gives the following error when the data are unbalanced. As I understand from the documentation, When type="effects" or type="means", the design must be balanced for standard errors to be computed. But I am NOT asking for se estimates at all. Please help!

Problem in print.array: No data to interpret as logical value: if(length(dn[[i]]) < d[i]) dn[[i]] <- paste(1:d[i])

Use traceback() to see the call stack

> traceback()

11: eval(action, sys.parent())

10: doErrorAction("Problem in print.array: No data to interpret as logical value: if(length(dn[[i]]) < d[i]) dn[[i]] <- paste(1:d[i])",

9: print.array(x, digits = digits, NA.fill = NA.fill, NaN.fill = NaN.fill, quote

8: print.table(ctable, digits = digits, NA.fill = NA.fill, NaN.fill = NaN.fill)

7: print(ctable, digits = digits, NA.fill = NA.fill, NaN.fill = NaN.fill)

6: print.tables.aov(yy)

5: print(yy)

4: source(auto.print = auto.print, exprs = substitute(exprs.literal))

3: script.run(exprs.literal = {

2: eval(_expression_(script.run(exprs.literal = {

1:

Thanks,

Pravin