Find rows that are not equal/identical in two columns including NAs [duplicate]

Question

I want to create a new dataframe in R that contains all the rows that do not have identical values in two given rows including those that are NA in one row and not in another.

So I want to subset this

person,opinion_1, opinion_2
a,agree,agree
b,disagree,agree
c,disagree,
d,disagree,

 df_example <- structure(
      list(
        person = c("a", "b", "c", "d"),
        opinion_1 = c("agree",
                      "disagree", "disagree", "disagree"),
        opinion_2 = c("agree", "agree",
                      NA,NA)
      ),
      row.names = c(NA, -4L),
      class = c("data.table", "data.frame")
    )

into this

person,opinion_1, opinion_2
b,disagree,agree
c,disagree,
d,disagree,

I have tried using x <- df[which(df$opinion_1 != df$opinion_2),] but this only returns

person,opinion_1, opinion_2
b,disagree,agree

Is there a solution so that the subset will include mismatched NAs?

Answer 1

We can do this using subsetting, the code below represents keep every row in which the opinion differs and additionally keep those rows that only have one opinion recorded:

dfn <- df[(df$opinion_1 != df$opinion_2) | is.na(df$opinion_1) | is.na(df$opinion_2),]
# results in:
  person opinion_1 opinion_2
2      b  disagree     agree
3      c  disagree      <NA>
4      d  disagree      <NA>

Data

df <- read.table(text = "person opinion_1 opinion_2
a agree agree
b disagree agree
c disagree NA
d disagree NA", header = T, stringsAsFactors = F)