plink (version 1.5-1)

as.poly.mod: poly.mod objects

Description

This function attempts to turn the given values into a object that associates each item with a specific unidimensional or multidimensional item response model.

Usage

as.poly.mod(n, model = "drm", items = NULL)

Arguments

n

total number of items

model

character vector identifying the IRT models used to estimate the item parameters. The only acceptable models are drm, gpcm, grm, mcm, and nrm. See below for an explanation of the codes.

items

list identifying the item numbers from a set of parameters that correspond to the given model in model.

Value

Returns an object of class

Details

When creating a poly.mod object, there is no difference in the specification for unidimensional versus multidimensional item response models. If all the items are dichotomous, it is only necessary to specify a value for n. If all the items correspond to a single model (other than drm), only n and model need to be specified. The IRT models associated with the codes:

drm:

dichotomous response models (includes the 1PL, 2PL, 3PL, M1PL, M2PL, and M3PL)

gpcm:

partial credit model, generalized partial credit model, multidimensional partial credit model, and multidimensional generalized partial credit model

grm:

graded response model and multidimensional graded response model

mcm:

multiple-choice model and multidimensional multiple-choice model

nrm:

nominal response model and multidimensional nominal response model

References

Weeks, J. P. (2010) plink: An R package for linking mixed-format tests using IRT-based methods. Journal of Statistical Software, 35(12), 1--33. URL http://www.jstatsoft.org/v35/i12/

See Also

Examples

# Ten dichotomous items
as.poly.mod(10)

# The first ten items in the set of associated (not present here) item 
# parameters are dichotomous and the last five were estimated using the 
# generalized partial credit model
as.poly.mod(15, c("drm", "gpcm"), list(1:10,11:15) )

# Ten multidimensional graded response model items
# Note: This same specification would be used for a unidimensional
# graded response model
as.poly.mod(10, "grm")