Function that checks if a variable is time-invariant for each subject.

check.timeinvar(x, id, data, out=1, na.rm=TRUE)

Arguments

x

argument to specify the variable to check.

id

argument to specify a subject id variable.

data

optional data frame that contains the variables specified above.

out

either a string or an integer (1 = "logical", 2 = "id", 3 = "data") indicating what information should be returned in case there are subjects where the variable is not time-invariant.

na.rm

logical indicating whether missing values should be removed before checking (default is TRUE).

Details

The function checks if the values of a variable are constant (i.e., time-invariant) for each subject.

When na.rm=TRUE (the default), missing values are ignored. When setting na.rm=FALSE, then missing values are treated as distinct values from any non-missing values. See ‘Examples’.

Value

When out = 1 or out = "logical", the function simply returns a logical (i.e., TRUE or FALSE), depending on whether the variable is time-invariant within each subject.

When out = 2 or out = "id", the function returns a vector with the ids of those subjects where the variable is not time-invariant.

When out = 3 or out = "data", the function returns the data for those subjects where the variable is not time-invariant.

Author

Wolfgang Viechtbauer wvb@wvbauer.com

See also

Examples

# illustrative dataset
dat <- data.frame(subj=rep(1:4, each=5),
                  obs = 1:5,
                  age = rep(c(20,31,27,22), each=5),
                  stress = c(2,3,1,4,2, 3,3,3,3,3, 1,1,2,6,4, 1,2,1,3,1))
dat
#>    subj obs age stress
#> 1     1   1  20      2
#> 2     1   2  20      3
#> 3     1   3  20      1
#> 4     1   4  20      4
#> 5     1   5  20      2
#> 6     2   1  31      3
#> 7     2   2  31      3
#> 8     2   3  31      3
#> 9     2   4  31      3
#> 10    2   5  31      3
#> 11    3   1  27      1
#> 12    3   2  27      1
#> 13    3   3  27      2
#> 14    3   4  27      6
#> 15    3   5  27      4
#> 16    4   1  22      1
#> 17    4   2  22      2
#> 18    4   3  22      1
#> 19    4   4  22      3
#> 20    4   5  22      1

# check the age variable
check.timeinvar(age, subj, data=dat)
#> [1] TRUE

# check the stress variable
check.timeinvar(stress, subj, data=dat)
#> [1] FALSE

# for which subjects is stress non-constant
check.timeinvar(stress, subj, data=dat, out=2)
#> [1] "1" "3" "4"

# show the data for those subjects
check.timeinvar(stress, subj, data=dat, out=3)
#>    subj obs age stress
#> 1     1   1  20      2
#> 2     1   2  20      3
#> 3     1   3  20      1
#> 4     1   4  20      4
#> 5     1   5  20      2
#> 11    3   1  27      1
#> 12    3   2  27      1
#> 13    3   3  27      2
#> 14    3   4  27      6
#> 15    3   5  27      4
#> 16    4   1  22      1
#> 17    4   2  22      2
#> 18    4   3  22      1
#> 19    4   4  22      3
#> 20    4   5  22      1

# by default missings are ignored
dat$age[2] <- NA
check.timeinvar(age, subj, data=dat)
#> [1] TRUE

# treat NAs as distinct values
check.timeinvar(age, subj, data=dat, na.rm=FALSE)
#> [1] FALSE
check.timeinvar(age, subj, data=dat, na.rm=FALSE, out=2)
#> [1] "1"
check.timeinvar(age, subj, data=dat, na.rm=FALSE, out=3)
#>   subj obs age stress
#> 1    1   1  20      2
#> 2    1   2  NA      3
#> 3    1   3  20      1
#> 4    1   4  20      4
#> 5    1   5  20      2