genItembank: Generate itembank.
In Karel-Kroeze/MCAT: What the package does (short line)

Description Usage Arguments Details Value

Generates a Q-dimensional itembank under the Graded Response Model (GRM), Sequential Model (SM), Generalized Partial Credit Model (GPCM) or Generalized Three Parameter Logistic Model (3PL).

1
2
3

genItembank(Q = 2, K = 50, model = "GPCM", covar = diag(Q),
  a = list(method = "normal", a = 0, b = 1), b = list(method = "normal", a =
  0, b = 1), c = list(a = 0, b = 0), M = list(min = 2, max = 4))

`Q`	integer; number of dimensions
`K`	integer; number of items
`model`	character; IRT model used to generate item parameters (can be either 'GRM', 'SM', or 'GPCM").
`covar`	matrix; Covariance matrix for alpha (discrimination parameter), defaults to identity matrix of rank Q. See details.
`a`	list of options for sampling discrimination parameters. See details.
`b`	list of options for sampling difficulty parameters. See details.
`c`	list(min, max); named list of distribution parameters for the guessing parameter, c. Guessing parameters are drawn from a continuous uniform distribution with minimum `a` and maximum `b`. Default is list(min=0,max=0) for all `c` = 0. Only used for 3PL, ignored in all other models.
`M`	list(min, max); named list of two integer values for the minimum and maximum values for a discrete uniform draw of number of categories per item. Forced to 1,1 in 3PL.

Alpha parameters may be correlated, and are > 0. This is accomplished by an initial draw from a multivariate normal distribution with the given covariance matrix. We then take the exponent of these values, and for each item set the sum of exponents to 1. This does mean that the observed covariance is not the same as the generating covariance matrix.

Itembank. A list with class MCAT.items of parameters simulated under the given model. Additionally the itembank includes a list with options used to create it.

Karel-Kroeze/MCAT documentation built on May 8, 2019, 4:50 p.m.