Given an estimated model compute the expected test score. Returns the expected values in the same form as the data used to estimate the model.
Usage
expected.test(
x,
Theta,
group = NULL,
mins = TRUE,
individual = FALSE,
which.items = NULL,
probs.only = FALSE
)Arguments
- x
an estimated mirt object
- Theta
a matrix of latent trait values (if a vector is supplied, will be coerced to a matrix with one column)
- group
a number or character signifying which group the item should be extracted from (applies to 'MultipleGroupClass' objects only)
- mins
logical; include the minimum value constants in the dataset. If FALSE, the expected values for each item are determined from the scoring 0:(ncat-1)
- individual
logical; return tracelines for individual items?
- which.items
an integer vector indicating which items to include in the expected test score. Default uses all possible items
- probs.only
logical; return the probability for each category instead of traceline score functions? Only useful when
individual=TRUE
References
Chalmers, R., P. (2012). mirt: A Multidimensional Item Response Theory Package for the R Environment. Journal of Statistical Software, 48(6), 1-29. doi:10.18637/jss.v048.i06
Examples
if (FALSE) { # \dontrun{
dat <- expand.table(deAyala)
model <- 'F = 1-5
CONSTRAIN = (1-5, a1)'
mod <- mirt(dat, model)
Theta <- matrix(seq(-6,6,.01))
tscore <- expected.test(mod, Theta)
tail(cbind(Theta, tscore))
# use only first two items (i.e., a bundle)
bscore <- expected.test(mod, Theta, which.items = 1:2)
tail(cbind(Theta, bscore))
# more low-level output (score and probabilty elements)
expected.test(mod, Theta, individual=TRUE)
expected.test(mod, Theta, individual=TRUE, probs.only=TRUE)
} # }