random numbers, random number generation,

An item bank typically contains items from several tests that have been calibrated by administering them to different groups of examinees. The parameters of the items must be linked onto a common scale. A linking technique consists of an anchoring design and a transformation method. Four basic anchoring designs are the unanchored, anchor-items, anchor-group, and double-anchor designs. The transformation design con sists of the system of equations that is used to trans late the anchor information and put the item parame ters on a common scale. Several transformation methods are discussed briefly. A simulation study is presented that compared the equivalent-groups method with the anchor-items method, using varying numbers of common items, applied both to the situation in which the groups were equivalent and one in which they were not. The results confirm previous findings that the equivalent-groups method is adequate when the groups are in fact equivalent. When the groups are not equivalent, accurate linking can be obtained with as few as two common items. Linking using a more efficient interlaced anchor-items design can provide accurate linking without the expense of including ex plicit common items in each of the tests.

A computerized system was developed for generating narrative interpretations of scores from a battery of personnel screening tests. The report structure and interpretive statement library were designed to capture the test expertise and interpretive strategies of a panel of testing experts. This was accomplished by enumerating the questions that the experts believed the battery could answer, developing answers to these questions, and devising rules for selecting the appropriate answers based on test‐battery scores. The accuracy, thoroughness, readability, and coherence of the computer‐generated reports were evaluated in comparison to reports generated by human experts for the same examinees. Results of the evaluation showed the computerized reports to be more accurate and thorough, as readable, and somewhat less coherent than interpretations generated by the typical human expert. The computerized system development and validation strategies described are useful for other applications in which numbers are interpreted in a narrative report format.

'4 A simulation study to determine appropriate linking methods for adaptive testing items was designed. Responses of examinees of three group sizes for four test lengths were simulated. Three basic data sets were created: (a) randomly sampled data set, (b systematically sampled data set, and (c) selected data set. Three categories of evaluative criteria were used: fidelity of parameter estimation, asymptotic ability estimates. root-mea n-square error of estimates, and the correlation between true and estimated ability. Test length appeared to be relatively more important to calibrition effectiveness than was sample size. efficiency analyses suggested that increases in test length were at least three to four times as effective in improving calibrafion efficiency as proportionate increases in calibration sample sizes. ability estimation (an equivalent-groups procedure) were somewhat more effective than the others and that the equivalent-tests method was typically no better than not linking at all. Analyses using the relative efficiency criteria suggested that the equivalent-groups procedures were superior to the equivalent-tests procedures and that those using Bayesian scoring prucedires were slightly superior to the others tested. Efficiency loss due to linking error was always less than that due to item calibration error and although test length and sample size had a definite effect on calibration efficiency. no strong effects appear with respect to linking efficiency. For the systematically sampled data set. the anchor-test and anchor-group methods were considered along with the equivalence methods. In terms of linking efficiency, the anchor-test method produced the most efficient item pools. The anchor-grou 4 method resulted in efficiencies equivalent to those of the anchor-test procedure if large groups were used. but with smaller groups the efficiencies dropped somewhat. The equivalence methods were somewhat less efficient than either of the anchor methods. Bayesian scoring was preferred over the maximum-likelihood scoring procedure. An application of the .1, results of this research to a practical linking problem was described with equivalent-groups linking. An anchor-test linking method was suggested for adding items at later times. I

An adaptive test tailors the difficulty of the items to the ability of the examinee being tested. This article describes the general principles and models of adaptive testing beginning with the early Binet tests and continuing through state‐of‐the‐art techniques. Such computerised adaptive tests (C.A.T.s), based on item response theory (I.R.T.), are useful both for measurement and classification applications. I.R.T. provides a means by which different sets of items, as administered in a C.A.T., can be scored on a common scale. The principal advantages of CAT. are efficiency and control over either the precision of measurement or the accuracy of classification. C.A.T. procedures can be easily implemented without writing programs using commercially available software (e.g. MicroCAT). Un test “adaptatif” accorde la difficulté des items à la capacityé des candidate aux tests. Cet article décrit les principes généraux et les modèles de tests “adaptatifs” depuis les tests de Binet jusqu'aux techniques de pointe actuelles. Tels tests informatisés (C.A.T.s), basés sur la théorie des réponses par item (I.R.T.), sont utiles à la fois pour la mesure et la classification. I.R.T. procure les moyens par lesquels différents sets ďitems, tels qu'ils sont administrés dans un C.A.T., peuvent être notés sur une échelle commune. Les principaux avantages du C.A.T. sont ľefficacité et le contrôle de la précision de la mesure et de la fidélité de la classification. Le protocole du C.A.T. peut être aisément utilisé sans texte écrit en se servant des programmes informatisés disponibles dans le commerce.