Question

Example:

x <- c( 1, NA, 0, 1)
y <- c(NA, NA, 0, 1)
table(x,y, useNA="always") # --->
#       y
# x      0 1 <NA>
#   0    1 0    0
#   1    0 1    1
#  <NA>  0 0    1

My question is:

a <- c(NA, NA, NA, NA)
b <- c(1, 1, 1, 1)
table(a, b, useNA="always") ## --> It is 1X2 matrix. 
#       b
# a      1 <NA>
#   <NA> 4    0

I want to get a 3X3 table with the same colnames, rownames and dimensions as the example above.. Then I will apply chisq.test for the table.
Thank you very much for your answers!

Answer 1

You can achieve this by converting both a and b into factors with the same levels. This works because factor vectors keep track of all possible values (aka levels) that their elements might take, even when they in fact contain just a subset of those.

a <- c(NA, NA, NA, NA)
b <- c(1, 1, 1, 1)

levs <- c(0, 1)

table(a = factor(a, levels = levs), 
      b = factor(b, levels = levs), 
      useNA = "always")
#       b
# a      0 1 <NA>
#   0    0 0    0
#   1    0 0    0
#   <NA> 0 4    0