A method based on 0-1 linear programming (LP) is presented to stratify an item pool optimally for use in "alpha"-stratified adaptive testing. Because the 0-1 LP model belongs to the subclass of models with a network-flow structure, efficient solutions are possible. The method is applied to a previous item pool from the computerized…

Presented at the American Educational Research Association (AERA) Conference in San Diego in May 2005. This presentation explores the issues and problems with declining response rates to the questionnaire once test registration moved from paper format to a web platform. A decline would affect research over time so interventions were put in place…

Describes computer programs designed to help students prepare for such tests as the Scholastic Aptitude Test (SAT) and the American College Testing Program's ACT Assessment. Includes list of special benefits the computer can provide in several areas as well as reviews of several programs. (JN)

Children aged 7 to 14 years were administered a computerized version of the Raven's Coloured Progressive Matrices test. Computer and traditional version performance was found to be similar in terms of total mean score, correlation with the WISC-R, Raven's subscale intercorrelations, and Raven's total mean score composition. (Author/RD)

Citing the importance to students of immediate feedback on multiple choice test items, the author discusses four electronic or mechanical devices for rapid test scoring. (SJL)

Describes tailored testing (in which a computer selects appropriate items from an item bank while an examinee is taking a test) and shows it to be superior to paper-and-pencil tests in such areas as reliability, security, and appropriateness of items. (IRT)

The primary advantages of this method are qualitative and appear as improved student satisfaction, increased learning effectiveness, and increased academic quality. (Author/IRT)

Developed a multilevel item response (IRT) model that allows for differences between the distributions of item parameters of families of item clones. Results from simulation studies based on an item pool from the Law School Admission Test illustrate the accuracy of the item pool calibration and adaptive testing procedures based on the model. (SLD)

A computer testing program using HyperCard was tested for equivalence with written formats of a test with adult basic education students. Results of the comparative analysis indicated no significant differences between test scores but did indicate a student preference for the computer-based test. (Author/LRW)

The development and scoring of a recently introduced computer-based writing skills test is described. The test asks the examinee to edit a writing passage presented on a computer screen. Scoring difficulties are addressed through the combined use of option weighting and the sequential probability ratio test. (SLD)

Compared student performance on computer-based assessment to machine-graded multiple choice tests. Found that performance improved dramatically on the computer-based assessment when students were not required to scroll through the question paper. Concluded that students may be disadvantaged by the introduction of online assessment unless care is…

Outlines a series of security steps that might be taken by researchers or organizations that are contemplating Web-based tests and performance assessments. Focuses on what can be done to avoid the loss, compromising, or modification of data collected by or stored through the Internet. (SLD)

Describes work in progress at SRI International in the Performance Assessment Links in Science (PALS) project, which is developing an online, standards-based, interactive resource bank of science performance assessments. Coupled with the development of the resource bank is a program of research on effective use of these resources. (SLD)

Studied the relationship between test mode (paper and pencil or computerized with and without editorial control) and computer familiarity for 222 undergraduates. Results emphasize the importance of evaluating time constraints when converting exams from paper to computer delivery. (SLD)

Parameter recovery in the graded-response model was investigated using the MULTILOG computer program under default conditions. Results from 36 simulated data sets suggest that at least 500 examinees are needed to achieve adequate calibration under the graded model. Sample size had little influence on the true ability parameter's recovery. (SLD)