String pulled directly from source data seems to not match string in source data - r

Question

I have a string that is failing to evaluate as a match with itself. I am trying to do a simple subset based on one of 8 possible values in a column,
out <- df[df$`Var name` == "string",]
I've had it work multiple times with different strings but for some reason this string fails. I have tried to get the exact string (thinking there may be some character encoding issue) from the source using the four below avenues but have had no success. Even when I make an explicit call to a cell I know contains that string and copy that into an evaluation statement it fails
> df[i,j]
[1] "string"
df[i,j]=="string" # pasted from above line
I don't understand how I can be explicitly pasting the output I was just given and it not match.
## attempts to get exact string to paste into subset statement
# from dput
"IF APPLICABLE – Which of the following best characterizes the expectations with"
# from calling a specific row/col (df[i, j])
[1] "IF APPLICABLE – Which of the following best characterizes the expectations with"
# from the source pane of rstudio
IF APPLICABLE – Which of the following best characterizes the expectations with
# from the source excel file
IF APPLICABLE – Which of the following best characterizes the expectations with
I don't have a clue what could be going on here. I am explicitly drawing the string straight from the data and yet it still fails to evaluate as true. Is there something going on in the background that I'm not seeing? Am I overlooking something ridiculously simple?
edit:
I subset based on another way, below is a dput and actual example of what I'm doing:
> dput(temp)
structure(list(`Item Stem` = "IF APPLICABLE – Which of the following best characterizes the expectations with",
`Item Response` = "It was required.", orgchar_group = "locale",
`Org Characteristic` = "Rural", N = 487, percent = 34.5145287030475,
`Graphs note` = NA_character_, `Report note` = NA_character_,
`Other note` = NA_character_, subsig = 1, overall = 0, varname = NA_character_,
statsig = NA_real_, use = NA_real_, difference = 9.16044821292665), .Names = c("Item Stem",
"Item Response", "orgchar_group", "Org Characteristic", "N",
"percent", "Graphs note", "Report note", "Other note", "subsig",
"overall", "varname", "statsig", "use", "difference"), row.names = 288L, class = "data.frame")
> temp[1,1]
[1] "IF APPLICABLE – Which of the following best characterizes the expectations with"
> temp[1,1] == "IF APPLICABLE – Which of the following best characterizes the expectations with"
[1] FALSE

Answer 1

Turns out it was in fact a non-printable character, shoutout to the commenters for helping me figure it out by 1) suggesting it and 2) showing that it worked for them.
I was able to figure it out using insights from here (& here) and here.
I used a grep command (from #Tyler Rinker) to determine that there was in fact a non-ASCII character in my string, and a stringi command (from #hadley) to determine what kind. I then used base solution from #Josh O'Brien to remove it. Turns out it was the heiphen.
# working in the temp df
> x <- temp[1,1]
> grepl("[^ -~]", x)
[1] TRUE
> stringi::stri_enc_mark(x)
[1] "UTF-8"
> iconv(x, "UTF-8", "ASCII", sub="")
[1] "IF APPLICABLE Which of the following best characterizes the expectations with"
# set x as df$`Var name` and reassign it to fix
df$`Var name` <- iconv(df$`Var name`, "UTF-8", "ASCII", sub="")
Still don't understand it enough to explain why it happened but it's fixed now.