Reformat categorical data in R

Question

I have a categorical dataset that I am trying to summarize that has inherent differences in the nature of questions that were asked. The data below represent a questionnaire that had standard close-ended questions, but also questions where one could choose multiple answers from a list. "village" and "income" represent close-ended questions. "responsible.1"...etc... represent a list where the respondent either said yes or no to each.

VILLAGE  INCOME         responsible.1   responsible.2   responsible.3   responsible.4   responsible.5
   j     both           DLNR             NA              DEQ              NA           Public
   k     regular.income DLNR             NA              NA               NA           NA
   k     regular.income DLNR             CRM             DEQ              Mayor        NA
   l     both           DLNR             NA              NA               Mayor        NA
   j     both           DLNR             CRM             NA               Mayor        NA
   m     regular.income DLNR             NA              NA               NA           Public

What I want is a 3-way table output with "village" and the suite of of "responsible" responsible variables wrapped up into a ftable . This way, I could use the table with numerous R packages for graphs and analyses.

        RESPONSIBLE             
VILLAGE INCOME          responsible.1   responsible.2   responsible.3   responsible.4   responsible.5
j       both            2               1               1               1               1
k       regular income  2               1               1               1               0
l       both            1               0               0               1               0
m       regular income  1               0               0               0               1

as.data.frame(table(village, responsible.1) would get me the first, but I can't figure out how to get the entire thing wrapped up in a nice ftable .

Answer 1

> aggregate(dat[-(1:2)], dat[1:2], function(x) sum(!is.na(x)) )
  VILLAGE         INCOME responsible.1 responsible.2 responsible.3 responsible.4 responsible.5
1       j           both             2             1             1             1             1
2       l           both             1             0             0             1             0
3       k regular.income             2             1             1             1             0
4       m regular.income             1             0             0             0             1

I'm guessing you actually had another grouping vector , perhaps the first "responsible" column?

I don't really understand the sorting rules but reversing the order of the grouping columns may be closer to what you posted:

> aggregate(dat[-(1:2)], dat[2:1], function(x) sum(!is.na(x)) )
          INCOME VILLAGE responsible.1 responsible.2 responsible.3 responsible.4 responsible.5
1           both       j             2             1             1             1             1
2 regular.income       k             2             1             1             1             0
3           both       l             1             0             0             1             0
4 regular.income       m             1             0             0             0             1