Significant Improvements to LOGIST.Report as inadecuate

Significant Improvements to LOGIST. - Download this document for free, or read online. Document in PDF available to download.

The computer program LOGIST (Wingersky, Patrick, and Lord, 1988) estimates the item parameters and the examinee's abilities for Birnbaum's three-parameter logistic item response theory model using Newton's method for solving the joint maximum likelihood equations. In 1989, Martha Stocking discovered a problem with this procedure in that when the true item discriminations were used as starting values for the iteration procedure, item parameters were different from when the default starting value of one was used for the item discriminations. When a straight run to convergence was performed, the different initial starting values converged to the same item parameter estimates. This study investigated several methods for improving the automatic procedure, but when they failed to yield the necessary improvement, a method was devised that gives estimates nearly as good as those obtained from running to convergence. The method involves adding a step to get better initial parameter estimates for the automatic procedure. Abilities are grouped coarsely, and the grouped abilities are estimated iteratively, alternating between items and abilities until the maximum difference between the estimated item characteristic curves is less than some criterion. A new version of LOGIST, LOGIST7, has been produced. Four tables and 33 figures present analysis details. (Contains nine references.) (SLD)

Descriptors: Ability, Computer Software Development, Estimation (Mathematics), Item Response Theory, Maximum Likelihood Statistics

Author: Wingersky, Marilyn S.


Related documents