Long to wide w/ two repeated measures

Question

I know this has been asked numerous times on here under the rubric of "long to wide" but I've run into a situation where I have two value variables that are repeated measures.

          id sex  time      score1    score2
1  subject 1   m Time1 -0.20926263 0.2499310
2  subject 2   m Time1  0.17147511 3.2708905
3  subject 3   m Time1 -0.82619584 0.5993917
4  subject 4   f Time1 -0.95568823 4.4729726
5  subject 5   f Time1 -2.29939525 8.0101254
6  subject 1   m Time2 -0.37914702 3.6387589
7  subject 2   m Time2  0.26759909 4.9027533
8  subject 3   m Time2  0.07727621 2.1848642
9  subject 4   f Time2 -0.08613439 5.8747074
10 subject 5   f Time2 -0.02743044 4.3963938
11 subject 1   m Time3  0.07176053 3.7959496
12 subject 2   m Time3  0.46463917 5.2494579
13 subject 3   m Time3 -0.68764512 2.2639503
14 subject 4   f Time3 -0.56670061 2.3361909
15 subject 5   f Time3  1.70731774 5.8345116

Quick way to reproduce the data frame (DF).

DF<-data.frame(id=rep(paste("subject", 1:5, sep=" "), 3),              
               sex=rep(c("m","m","m","f","f"), 3),                     
               time=c(rep("Time1",5), rep("Time2",5), rep("Time3",5)), 
               score1=rnorm(15), score2=abs(rnorm(15)*4))              

I can solve the issue of long to wide for two measured repeated measure variables using the reshape function from base but I was hoping for a plyr or reshape2/1 answer, as these packages are generally much more intuitive to me. If you have any other solutions go ahead and provide them as the learning would be great.

Solution from base:

wide <- reshape(DF, v.names=c("score1", "score2"), idvar="id",         
                timevar="time", direction="wide")                      
wide   

Answer 1

I think this will do it:

library(reshape)
m <- melt(DF)

Simplest, but time and score are in the opposite order from your example (in case it matters)

cast(m,id+sex~...)

Or more explicitly:

cast(m,id+sex~variable+time)

You can cut this down to a one-liner:

recast(DF,id+sex~...)

If you like you can use the newer reshape2 package instead of reshape, replacing cast with dcast (the version of recast included in reshape2 doesn't give the desired result.)