R: how to compute differences based on a factor's levels?

Question

I have a data frame mytable with outcomes for two measurements, A and B, taken to a group of persons.

person measure outcome
1      A       3.6
2      A       2.3
3      A       4.2
1      B       3.9
2      B       3.2
3      B       2.7

I want to compute for each person the difference between the scores for A and B. That is, I want to obtain:

person outcome_diff
1      -0.3
2      -0.9
3       1.5

I searched for an answer, but I only found some concerning transformations within the levels of a factor, not across them.

I finally managed to work it out by doing:

mytable$outcome[mytable$measure=="B"] <- -1*mytable$outcome[mytable$measure=="B"]
outtable <- aggregate(outcome ~ person, data=mytable, FUN=sum)

Although it works, I wonder how to do it without messing up the original table. Furthermore, this solution is quite specific for computing a difference. What could be a more general way to achieve the same thing?

Answer 1

I would use plyr:

ddply(mytable, "person", summarize,
      outcome_diff = outcome[measure == "A"] -
                     outcome[measure == "B"])
#   person outcome_diff
# 1      1         -0.3
# 2      2         -0.9
# 3      3          1.5

Under the assumption that you always have exactly two measures A and B and in that order, you might also just do ddply(mytable, "person", summarize, outcome_diff = -diff(outcome)).