Use of the Rasch model for vertical equating of tests is discussed. Although use of the model is promising, empirical results raise questions about the adequacy of the Rasch model. Latent trait models with more parameters may be necessary. (JKS)

Ten item discrimination indices, across a variety of item analysis situations, were compared, based on the validities of tests constructed by using each of the indices to select 40 items from a 100-item pool. Item score data were generated by a computer program and included a simulation of guessing. (Author/CTM)

Reasonable and practical solutions to two major problems confronting the developer of any test of educational achievement (what to measure and how to measure it) are proposed, defended, and defined. (Author/PN)

A set of guidelines for evaluating criterion-referenced tests is presented. Additionally, 11 sets of extant criterion-referenced tests are evaluated using these guidelines. (JKS)

In studying the effectiveness of different instructional programs, a criterion measure can favor one of the programs because there is greater overlap between the content covered on the test and in that program. This overlap can be measured using teacher estimates or teacher estimates combined with curriculum analysis. (Author/BW)

The purpose of this study was to compare the effects of four item selection rules--(1) Fisher information (F), (2) Fisher information with a posterior distribution (FP), (3) Kullback-Leibler information with a posterior distribution (KP), and (4) completely randomized item selection (RN)--with respect to the precision of trait estimation and the…

Teacher-selected test level resulted in: (1) most Educable Mentally Retarded students tested responding above chance levels on all subtests of the Metropolitan Achievement Test, (2) reliability coefficients comparable to those of the standardization sample, and (3) moderate to high positive point-biserial correlations for all subtest-level…

An experiment investigated the extent to which the results of out-of-level testing may be biased because the child given an out of level test may have had a significantly different curriculum than the children given in-level tests. Item analysis data suggested this was unlikely. (CTM)