switch statement with multiple conditions in r

Question

I want to write a switch statement in r with three conditions but can't seem to get it to work. What am I doing wrong?

# assigning some values
test.type <- "p"
var.equal<- TRUE
  paired <- FALSE

# preparing text for which p-value adjustment method was used
test.description <- switch(
    EXPR = test.type & var.equal & paired,
    "p" & TRUE & TRUE = "Student's t-test",
    "p" & FALSE & TRUE = "Student's t-test",
    "p" & TRUE & FALSE = "Student's t-test",
    "p" & FALSE & FALSE = "Games-Howell test",
    "np" & TRUE & TRUE = "Durbin-Conover test"
  )
#> Error: <text>:10:23: unexpected '='
#> 9:     EXPR = test.type & var.equal & paired,
#> 10:     "p" & TRUE & TRUE =
#>                           ^

^{Created on 2018-11-08 by the reprex package (v0.2.1)}

A simpler version of this statement with just one condition does work-

# simpler switch
(test.description <- switch(
  EXPR = test.type,
  "p"  = "Student's t-test",
  "np" = "Durbin-Conover test"
))
#> [1] "Student's t-test"

^{Created on 2018-11-08 by the reprex package (v0.2.1)}

Should invalid combinations also produce a result? – Sven Hohenstein Nov 08 '18 at 16:12 — Sven Hohenstein, Nov 08 '18 at 16:12
Sure. It can default to some value like `"t-test"`. – Indrajeet Patil Nov 08 '18 at 16:14 — Indrajeet Patil, Nov 08 '18 at 16:14

score 4 · Answer 1 · answered May 29 '19 at 14:25

Another solution could be to use dplyr's case_when, which uses syntax more similar to your switch statements:

library(dplyr)

## initial dataframe
df <- data.frame(
  test.type = c("p", "p", "p", "p", "np", "np"),
  var.equal = c(TRUE, FALSE, TRUE, FALSE, TRUE, FALSE),
  paired = c(TRUE, TRUE, FALSE, FALSE, TRUE, FALSE)
  ) 

## add column test.description 
mutate(df,
  test.description = case_when(
      test.type == "p" & !var.equal & !paired  ~ "Games-Howell test",
      test.type == "p"                         ~ "Student's t-test",
      test.type == "np" & var.equal & paired   ~ "Durbin-Conover test",
      TRUE                                     ~ "Unknown combination"
     )
)

score 3 · Accepted Answer · answered Nov 08 '18 at 17:40

That's not how R's switch() function works. Syntactically, it's just a function call, so the selectors have to be things that can be treated as names, not expressions like "p" & TRUE & TRUE. So your first switch could switch on test.type, and then use if statements to choose values based on var.equal and paired. But it would probably look better as a sequence of if statements, like this:

test.description <- 
    if (test.type == "p" && !var.equal && !paired) "Games-Howell test" else
    if (test.type == "p")                          "Student's t-test" else
    if (test.type == "np" && var.equal && paired)  "Durbin-Conover test" else
                                                   "Unknown combination"

Some things to note here:

You can use if statements in an expression to produce a value; this is one big statement.
If the else keywords were moved to the next lines, it wouldn't work, because the code up to there is a complete statement, so the else clauses would be left dangling. (There are exceptions to this, but don't rely on them.)
You should almost always use the scalar && within an if test rather than the vector &.
Another way to format this is to put the values in braces, with the closing brace and the else on the next line. I like the formatting above a little better, but your preference may vary.

I find `if-else` statement really unappealing and prone to errors, so was looking for an alternative. I find `switch` statements much more intuitive. But there doesn't seem to be a way around `if-else` it seems. — Indrajeet Patil, Nov 19 '18 at 02:16

score 2 · Answer 3 · answered Apr 09 '22 at 13:57

Just do this:

test.type <- "p"
var.equal<- TRUE
  paired <- FALSE

test.description <- switch(
    EXPR = paste(test.type, var.equal, paired),
    "p TRUE TRUE" = "Student's t-test",
    "p FALSE TRUE" = "Student's t-test",
    "p TRUE FALSE" = "Student's t-test",
    "p FALSE FALSE" = "Games-Howell test",
    "np TRUE TRUE" = "Durbin-Conover test"
  )

Then:

test.description
[1] "Student's t-test"

switch statement with multiple conditions in r

3 Answers3