estGip: Generalized Item Response Theory parameter estimation

In takuizum/irtfun2: Some useful functions for IRT

Description

Generalized Item Response Theory parameter estimation

Usage

estGip(
  x,
  fc = 3,
  Gc = NULL,
  bg = 1,
  IDc = 1,
  D = 1.702,
  Ntheta = 31,
  Nphi = 10,
  method = "BFGS",
  th_dist = "normal",
  rm_list = NULL,
  phi_dist = "uniform",
  v = 3,
  tau = 1,
  mu_ph = 0,
  sigma_ph = 0.25,
  min_ph = 0.01,
  max_ph = 2,
  paramab = c(1, 4),
  mu_th = 0,
  sigma_th = 1,
  min_th = -4,
  max_th = 4,
  eEM = 0.001,
  eMLL = 1e-06,
  eDIST = 1e-04,
  maxiter_em = 100,
  print = 0,
  esteap = FALSE,
  estdist = FALSE,
  initial = NULL
)

Arguments

x	DataFrame.
fc	the first column.
Gc	the grade column.
bg	a mumber of base grade.
IDc	the ID column.
D	scale constant
Ntheta	the number of the nodes of theta dist.
Nphi	the number of the nodes of phi dist.
method	the method of optimiser. Default is "Fisher_Scoring", but optim function also be able to use.
th_dist	a type of theta dist."normal" or "empirical"
rm_list	a vector of item U want to remove for estimation. NOT list.
phi_dist	a prior distribution of phi. invchi is inverse chi distribution. lognormal is log normal distribution.
v	a hyper parameter of invchi for phi
tau	a hyper parameter of invchi for phi
mu_ph	a hyper parameter of lognormal dist for phi
sigma_ph	a hyper parameter of lognormal dist for phi
min_ph	a minimum value of phi distribution
max_ph	a maximum value of phi distribution
paramab	a hyper parameter of generalized beta distribution
mu_th	a hyper parameter of normal dist for theta
sigma_th	a hyper parameter of normal dist for theta
min_th	a minimum value of theta distribution
max_th	a maximum value of theta distribution
eEM	a convergence criterion related to item parameters in EM cycle.
eMLL	a convergence criterion related to negative twice log likelihood in EM cycle.
eDIST	a convergence criterion related to the parameter of theta distribution in EM cycle.
maxiter_em	the number of iteration of EM cycle.
print	How much information you want to display? from 1 to 3. The larger, more information is displayed.
esteap	logical. If TRUE, estimate subject theta & phi EAP.
estdist	logical. If TRUE, estimate population distribution theta & phi.
initial	if use specify initial value, set here data.frame.

Value

the output is a list that has item parameter and person parameter.

Examples

res <- estGip(x=sim_dat_girt,fc=2, Ntheta=10, Nphi = 5, min_ph = 0.001, max_ph = 2)
# check the parameters
res$item
head(res$person)

takuizum/irtfun2 documentation built on May 10, 2020, 8:30 a.m.