estip: Estimate item parameter for binary0,1 response data.

In takuizum/irtfun2: Some useful functions for IRT

Description

1PL,2PL,3PL,Bayes1PL,Bayes2PL and multigroup estimation is avairable now. U must install C++ compiler(Rtools for windows or Xcode for Mac)in your PC or Mac.

Usage

estip(
  x,
  model0 = as.character(c("2PL")),
  N = 31L,
  bg0 = 1L,
  fc0 = 2L,
  ng = 1L,
  gc0 = 2L,
  eMLL = 1e-06,
  eEM = 1e-04,
  eM = 0.001,
  emu = 0.001,
  esd = 0.001,
  D = 1.702,
  ic = 1/5,
  max = 6,
  min = -6,
  mu = 0,
  sigma = 1,
  Bayes = 0L,
  method = "Fisher_Scoring",
  mu_a = 0,
  sigma_a = 1,
  mu_b = 0,
  sigma_b = 2,
  mu_c = 1/5,
  w_c = 5,
  fix = 1L,
  print = 0L,
  min_a = 0.1,
  maxabs_b = 20,
  maxiter_em = 200L,
  maxiter_j = 20L,
  maxskip_j = 0L,
  rm_list = as.character(c("NONE")),
  thdist = "normal",
  estdist = 1L
)

Arguments

x	an item response data which class is data.frame object.
model0	Character.U can select which one, "1PL","2PL","3PL".
N	the number of nodes in integration.
bg0	the number of base grade.
fc0	a column of first item response.
ng	the number of groups
gc0	a column of group. the element must be integer and the minimum number must be 1.
eMLL	a convergence criteria(CC) for marginal log likelihood.
eEM	a CC in EM cycle.
eM	a CC in M step.
emu	a CC for population distribution mean.
esd	a CC for population distribution standard deviation.
D	a scaling constant.
ic	initial value for guessing parameter.
max	maximum value of theta in integration.
min	minimum value of theta in integration.
mu	hyperparameter for theta dist.
sigma	same as above.
Bayes	If 1, marginal Bayesian estimation runs.
method	Optimising method in M step. "Fisher_Scoring" or "Newton_Raphson" can be selected.
mu_a	hyperparameter of log normal dist for slope parameter.
sigma_a	same as above.
mu_b	hyperparameter of normal dist for location parameter.
sigma_b	same as above.
mu_c	hyperparameter for lower asymptote parameter.
w_c	weight of a Beta dist.
fix	Don't use. If 1, fix population distribution mean and sigma each EM cycle.
print	How much information you want to display? from 1 to 3. The larger, more information is displayed.
min_a	minimum value of slope parameter.
maxabs_b	maximum absolute value of location parameter.
maxiter_em	maximum iteration time for EM cycle.
maxiter_j	maximum iteration time for Newton Raphton in M step.
maxskip_j	Dont use.
rm_list	a vector of item U want to remove for estimation. NOT list.
thdist	Which distribution do you want normal or empirical for E step.
estdist	If 1, calculate esimated population distribution via EM argorithm.

Examples

res <- estip(x=sim_data_1, fc=2)
# check the parameters
res$para
res$SE

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