wizirt: Estimate an IRT model

In Pflegermeister/wizirt: A United API for IRT

Description

Estimate an IRT model using various engines

Usage

wizirt(
  data,
  rownames = NULL,
  item_type = "Rasch",
  engine = "mirt",
  tol = 1e-05,
  abs_fit = T
)

Arguments

data	An Person x Items matrix or dataframe of dichotomous response values (e.g. correct/incorrect). Rows are persons and columns are items, one row per person, one column per item. No other information allowed.
rownames	Optional unique row IDs for the data (i.e. examinee IDs). If omitted, uses 1:nrow(data).
item_type	Character. Must be one of "Rasch", "1PL", "2PL" or "3PL".
engine	Character. Currently supported engines are "mirt" and "ltm" for Rasch, 1PL, 2PL, and 3PL models. "eRm" is supported for Rasch models only.
tol	Numeric. Convergence criterion. Currently only implemented when engine is mirt.
abs_fit	Logical. Should absolute fit statistics be calculated? Increases time for estimation. Currently supported engine is "mirt".

Value

Returns a list of class wizirt. spec is a list of information for the parsnip backend. Printing spec prints a summary of the model run. elapsed contains the time it took the model to run. fit contains the model information:

• data is the data passed to the model

• model contains model fit information, including:

• engine a list with values pkg (the package used for estimation), ver (pkg version), func (functiong used from pkg), and call (call made to pkg)

• n_factors the number of factors estimated

• item_type the item type passed to wizirt (Rasch, 1PL, 2PL, or 3PL).

• estimation a list with information related to convergence. convergence a T/F value of whether the model converged, method the estimation method, criteria the convergence criteria, iterations the number of iterations it took for the model to converge, log_lik the loglikelihood at convergence, abs_fit the absolute fit of the model, df the number of parameters estimated.

• parameters a list of estimated parameters. coefficients is a data frame of estimated item-statistics and persons is a data frame of estimated person statistics.

• original_object is the object returned from the engine.

Examples

data("responses")
my_model <- wizirt(data = responses[, -1], item_type = "2PL", tol = 1e-4, engine = "mirt")
print(my_model, type = "tech")
print(my_model, type = "desc")
print(my_model, type = "item")
print(my_model, type = "person")
print(my_model, type = "na_item")
anova(my_model)

Pflegermeister/wizirt documentation built on Nov. 22, 2020, 8:27 p.m.