plink (version 1.5-1)

read.bilog: Import Parameters from IRT Software

Description

This function imports item and/or ability parameters from BILOG-MG 3, PARSCALE 4, MULTILOG 7, TESTFACT 4, ICL, BMIRT, and ltm.

Usage

read.bilog(file, ability = FALSE, pars.only = TRUE, as.irt.pars = TRUE)
read.parscale(file, ability = FALSE, loc.out = FALSE, pars.only = TRUE, 
  as.irt.pars = TRUE)
read.multilog(file, cat, poly.mod, ability = FALSE, contrast = "dev", 
  drm.3PL = TRUE, loc.out = FALSE, as.irt.pars = TRUE)
read.testfact(file, ability = FALSE, guessing = FALSE, bifactor = FALSE, 
  as.irt.pars = TRUE)
read.icl(file, poly.mod, ability = FALSE, loc.out = FALSE, 
  as.irt.pars = TRUE)
read.bmirt(file, ability = FALSE, sign.adjust = TRUE, loc.out = FALSE, 
  pars.only = TRUE, as.irt.pars = TRUE)
read.erm(x, loc.out = FALSE, as.irt.pars = TRUE)
read.ltm(x, loc.out = FALSE, as.irt.pars = TRUE)

Arguments

file

filename of file containing the item or ability parameters

ability

if TRUE, file contains ability parameters

pars.only

if TRUE, only the item/ability parameters will be imported (i.e., any other information like standard errors will be dropped).

loc.out

if TRUE, the step/threshold parameters will be reformated to be deviations from a location parameter

as.irt.pars

if TRUE, the parameters will be output as an object (this is only applicable to item parameters)

cat

vector with the number of response categories for each item. For multiple-choice model items, cat is the number of response categories plus one (the additional category is for 'do not know')

poly.mod

a object. See the documentation for the function as.irt.pars for more information on creating this object.

contrast

an object identifying the type of contrast(s) used to estimate the various parameters for each item. See below for more details.

drm.3PL

logical value indicating whether the dichotomous items (if applicable) were modeled using the three parameter logistic model (3PL)

guessing

logical value indicating whether a guessing parameter was modeled

bifactor

logical value indicating whether the bifactor model was used to estimate the item/ability parameters

sign.adjust

logical value indicating whether the difficulty/step parameters should be multiplied by -1 to make them consistent with common formulations of multidimensional response models

x

output object from one of the following functions in the eRm package: LLTM, LPCM, LRSM,PCM, RM, or RSM or one of the following functions in the ltm package: rasch, ltm, tpm, grm, or gpcm

Value

Returns a data.frame or an object of class if as.irt.pars = TRUE.

Details

The file extensions for the item parameter and ability files respectively are as follows: .par and .sco for BILOG-MG, PARSCALE, and MULTILOG, .par and .fsc for TESTFACT, and .par and .ss for BMIRT. For ICL, the file extensions are specified by the user, and for ltm, the name of the output object is specified by the user. When item parameters are estimated in MULTILOG for models other than the 1PL, 2PL, and GRM, the program estimates (and returns) contrast parameters. MULTILOG implements three types of contrasts: deviation, polynomial, and triangle (see Thissen & Steinberg, 1986 for more information). A single type of contrast can be used for all parameters (a, b, and c) for all items or different contrasts can be specified for individual parameters and individual items. If a single type of contrast is used for all parameters for all items, a character value can be specified for the contrast argument: "dev", "poly", or "tri" for the three types of contrasts respectively. When different contrasts are used, contrast should be a list of length nine. The list elements should be ordered as follows "dev.a","poly.a", "tri.a","dev.c","poly.c","tri.c","dev.d", "poly.d","tri.d" where the first three elements correspond to the various contrasts for the slope parameters, the next three elements correspond to the contrasts for the category parameters, and the last three elements correspond to the contrasts for the lower asymptote (guessing parameters). There are two approaches that can be implemented using this list 1) character vectors with the model names "drm", "grm", "gpcm", "nrm", and "mcm" indicating that the given parameters for all items associated with the given model should be transformed using the specified contrast. In instances where a model is not included for a given parameter (for any of the contrasts) the parameters will be transformed using deviation contrasts. 2) numeric vectors identifying the contrasts used for given parameters for given items can be specified. It is only necessary to include item numbers for the various parameter/contrast combinations when deviation contrasts are not used. See below for examples of how to formulate this argument.

References

Hanson, B. A. (2002). IRT command language [Computer Program]. URL http://www.b-a-h.com/software/irt/icl/ Mair, P & Hatzinger, R. (2007). Extended Rasch modeling: The eRm package for the application of IRT models in R. URL http://www.jstatsoft.org/v20/i09 Muraki, E. & Bock, R. D. (2003). PARSCALE 4: IRT item analysis and test scoring for rating scale data [Computer Program]. Chicago, IL: Scientific Software International. URL http://www.ssicentral.com Rizopoulos, D. (2006). ltm: An R package for latent variable modeling and item response theory analyses. Journal of Statistical Software, 17(5), 1-25. URL http://www.jstatsoft.org/v17/i05/ Thissen, D. (2003). MULTILOG 7: Multiple, categorical item analysis and test scoring using item response theory [Computer Program]. Chicago, IL: Scientific Software International. URL http://www.ssicentral.com Thissen, D. & Steinberg, L. (1986). A taxonomy of item response models. Psychometrika, 51(4), 567-577. 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/ Wood, R., Wilson, D. T., Muraki, E., Schilling, S. G., Gibbons, R., & Bock, R. D. (2003). TESTFACT 4: Test scoring, item statistics, and item factor analysis [Computer Program]. Chicago, IL: Scientific Software International. URL http://www.ssicentral.com Yao, L. (2008). BMIRT: Bayesian multivariate item response theory [Computer Program]. Monterey, CA: CTB/McGraw-Hill. Zimowski, M. F., Muraki, E., Mislevy, R. J., & Bock, R. D. (2003). BILOG-MG 3: Multiple-group IRT analysis and test maintenance for binary items [Computer Program]. Chicago, IL: Scientific Software International. URL http://www.ssicentral.com

Examples

# Illustration of how to formulate the contrast argument. Say that we 
# have 20 items where the first 15 are modeled using the 3PL and the 
# last five are modeled using the GPCM.  For the 3PL items, deviation 
# contrasts are commonly used for all of the parameters, whereas with 
# the GPCM items, polynomial contrasts are typically used for the slope 
# parameters and triangle contrasts are used for the category parameters. 
# The contrast argument could be specified as follows

contrast <- vector("list",9)
# Note: the list elements do not need to be named for read.multilog
names(contrast) <- c("dev.a","poly.a","tri.a","dev.c","poly.c","tri.c",
  "dev.d", "poly.d","tri.d") 
contrast$poly.a <- 16:20
contrast$tri.c <- 16:20

# The object could alternatively be formatted as follows
contrast <- vector("list",9)
names(contrast) <- c("dev.a","poly.a","tri.a","dev.c","poly.c","tri.c",
  "dev.d","poly.d","tri.d") 
contrast$dev.a <- 1:15
contrast$poly.a <- 16:20
contrast$dev.c <- 1:15
contrast$tri.c <- 16:20
contrast$dev.d <- 1:15