This paper reviews the basic elements of the EM approach to estimating item parameters and illustrates its use with one simulated and one real data set, In order tc illustrate the use of the BILOG computer program, runs for 1-, 2-, and 3-parameter models are presented for the two sets of data. First is a set of responses from 1,000 persons to five items of the Law School Admissions Test. Second is a set of simulated data of 1,000 persons to 18 items. The examples bring into focus the degree to which item parameters in the 3-parameter model can be recovered. The review discusses an EM Algorithm for estimating item parameters; solution for item parameters when person ability values are knon; early computer program approaches; and the key elements of the Bock-Aitkin approach. Further described are extensions of the Bock-Aitkin approach, which include: (1) extension of the 3-parameter model; (2) prior distributions on item parameters; (3) estimation of the latent distribution; and, (4) different patterns of item attempts for different persons. (PN) *********************************************************************** Reproductions supplied by EDRS are the best that can be made from the original document. ***************************************w*******************************
[1]
R. D. Bock,et al.
Marginal maximum likelihood estimation of item parameters: Application of an EM algorithm
,
1981
.
[2]
E. B. Andersen,et al.
Estimating the parameters of the latent population distribution
,
1977
.
[3]
R. Darrell Bock,et al.
Fitting a response model forn dichotomously scored items
,
1970
.
[4]
E. B. Andersen,et al.
A goodness of fit test for the rasch model
,
1973
.
[5]
R. D. Bock,et al.
Marginal maximum likelihood estimation of item parameters
,
1982
.
[6]
David Thissen,et al.
Marginal maximum likelihood estimation for the one-parameter logistic model
,
1982
.