How to avoid dropping levels with 0 frequencies - r

Question

I am trying to compare two questions (columns Q1_b and Q2_b) and barplot them next to each other (in the same barplot), the answer options are 1-6. The problem is that noone answered with 4 for Q1_b, so the barplot skips to displaying 5 where 4 should be for Q1_b, next to the percentage of people who answered 4 for Q2_b. How can I make sure R doesn't do this and automatically enters a 0% column if there weren't any answers for a specific option?
alldataset<-structure(list(Q1_b = c(6L, 1L, 5L, 3L, 5L, 6L, 6L, 2L),
Q2_b = c(1L, 2L, 2L, 5L, 4L, 3L, 6L, 1L)),
.Names = c("Q1_b", "Q2_b"),
class = "data.frame",
row.names = c(NA, -8L))
Qb<-table(alldataset$Q2_b)
Qf<-table(alldataset$Q1_b)
nrowFUP<-NROW(alldataset$Q1_b)
nrowBL<-NROW(alldataset$Q2_b)
options(digits=6)
newbl <- transform(as.data.frame(table(alldataset$Q2_b)),
percentage_column=Freq/nrowBL*100)
newfup <- transform(as.data.frame(table(alldataset$Q1_b)),
percentage_column=Freq/nrowFUP*100)
matrixQ1<-cbind(newbl$percentage_column, newfup$percentage_column)
matrixQ1dataframe<-data.frame(matrixQ1)
rmatrixQ1<-as.vector(t(matrixQ1dataframe))
roundedrmatrix<-round(rmatrixQ1, digits=0)
barplotmatrix<-matrix(roundedrmatrix)
par(mar=c(7.5,4,3,2), mgp=c(2,.7,0), tck=-.01, las=1, xpd=TRUE)
b<-barplot(matrix(roundedrmatrix, nr=2),
beside=T, xlab="",
ylab="Percentage",
cex.lab=0.9,
main="Comparison",
cex.main=0.9, ylim=c(0,70),
col=c("black","yellow"),
names.arg=c(1:6),
legend=c("Q2_b","Q1_b"),
args.legend=list(x="bottomleft",
cex=0.8,
inset=c(0.4,-0.4)))
text(x=b, y=roundedrmatrix,labels=roundedrmatrix, pos=3, cex=0.8)
R also warns me this will happen by displaying:
Warning message:
In cbind(newbl$percentage_column, newfup$percentage_column) :
number of rows of result is not a multiple of vector length (arg 2)
I have been trying for ages to sort this out but I am not getting anywhere. Can anyone help?

Answer 1

The problem is that you never told R that you vectors represent categorical responses with potential values of 1-6, so it does not know to include the 0 counts (you would not want it to include a 0 for 7, 8, 1 million, etc.).
Try replacing your 1st 2 lines with:
Qb<-table(factor(alldataset$Q2_b, levels=1:6))
Qf<-table(factor(alldataset$Q1_b, levels=1:6))
or run somethingn like:
alldataset$Q1_b <- factor(alldataset$Q1_b, levels=1:6)
alldataset$Q2_b <- factor(alldataset$Q2_b, levels=1:6)
before the table commands.

Answer 2

You need to tell table to use all values from one to six with table(factor(x, seq.int(6))).
Here is an improved version of your code:
dat <- t(round(sapply(rev(alldataset),
function(x) table(factor(x, seq.int(6)))) /
nrow(alldataset) * 100))
par(mar=c(7.5,4,3,2), mgp=c(2,.7,0), tck=-.01, las=1, xpd=TRUE)
b <- barplot(dat, beside=T,xlab="", ylab="Percentage", cex.lab=0.9,
main="Comparison", cex.main=0.9, ylim=c(0,70),
col=c("black","yellow"), names.arg=c(1:6), legend=names(dat),
args.legend=list(x="bottomleft", cex=0.8, inset=c(0.4,-0.4)))
text(x=b, y=dat,labels=dat, pos=3, cex=0.8)