This book offers a comprehensive introduction to the latest developments in the theory and practice of CAT. It can be used both as a basic reference and a valuable resource on test theory. It covers such topics as item selection and ability estimation, item pool development and maintenance, item calibration and model fit, and testlet-based adaptive testing, as well as the operational aspects of existing large-scale CAT programs.
Over the last 20 years there have been a large number of technical advances and changes in the field of educational and psychological testing. According to Anne Anastasi, The decade of the 1980's has been a period of unusual advances in ,psychological testing. Technological progress, theoretical sophistication, and increasing pro fessional responsibility are all evident in the fast-moving events in this field (A. Anastasi, Psychological Testing, Sixth Edition. New York: Macmillan, 1988). On the psychometric front, advances in topics such as item response theory, criterion-referenced measurement, generalizability theory,· analy sis of covariance structures, and validity generalization are reshaping the ways that ability and achievement tests are constructed and evaluated, and that test scores are interpreted. But \Jsychometric advances, as substantial and important as they have been, are only a fraction of the major changes in the field of testing. Today, for example, the computer is radically chang ing the ways in which tests are constructed, administered, and scored. Computers are being used to administer tests "adaptively." That is, the sequence of questions an examinee is administered depends upon his or her performance on earlier administered items in the test. Tests are "adapted" to the ability levels of the examinees who are being assessed. One result is shorter tests with little or no loss in measurement precision. Computers are also being used to store or bank test items. Later, items of interest can be selected, and the computer is used to print copies of the test.
The arrival of the computer in educational and psychological testing has led to the current popularity of adaptive testing---a testing format in which the computer uses statistical information about the test items to automatically adapt their selection to a real-time update of the test taker’s ability estimate. This book covers such key features of adaptive testing as item selection and ability estimation, adaptive testing with multidimensional abilities, sequencing adaptive test batteries, multistage adaptive testing, item-pool design and maintenance, estimation of item and item-family parameters, item and person fit, as well as adaptive mastery and classification testing. It also shows how these features are used in the daily operations of several large-scale adaptive testing programs.
Unlike other forms of adaptive testing, multistage testing (MST) is highly suitable for testing educational achievement because it can be adapted to educational surveys and student testing. This volume provides the first unified source of information on the design, psychometrics, implementation, and operational use of MST. It shows how to apply theoretical statistical tools to testing in novel and useful ways. It also explains how to explicitly tie the assumptions made by each model to observable (or at least inferable) data conditions.
The goal of this guide and manual is to provide a practical and brief overview of the theory on computerized adaptive testing (CAT) and multistage testing (MST) and to illustrate the methodologies and applications using R open source language and several data examples. Implementation relies on the R packages catR and mstR that have been already or are being developed by the first author (with the team) and that include some of the newest research algorithms on the topic. The book covers many topics along with the R-code: the basics of R, theoretical overview of CAT and MST, CAT designs, CAT assembly methodologies, CAT simulations, catR package, CAT applications, MST designs, IRT-based MST methodologies, tree-based MST methodologies, mstR package, and MST applications. CAT has been used in many large-scale assessments over recent decades, and MST has become very popular in recent years. R open source language also has become one of the most useful tools for applications in almost all fields, including business and education. Though very useful and popular, R is a difficult language to learn, with a steep learning curve. Given the obvious need for but with the complex implementation of CAT and MST, it is very difficult for users to simulate or implement CAT and MST. Until this manual, there has been no book for users to design and use CAT and MST easily and without expense; i.e., by using the free R software. All examples and illustrations are generated using predefined scripts in R language, available for free download from the book's website.
The general theme of this book is to present the applications of artificial intelligence (AI) in test development. In particular, this book includes research and successful examples of using AI technology in automated item generation, automated test assembly, automated scoring, and computerized adaptive testing. By utilizing artificial intelligence, the efficiency of item development, test form construction, test delivery, and scoring could be dramatically increased. Chapters on automated item generation offer different perspectives related to generating a large number of items with controlled psychometric properties including the latest development of using machine learning methods. Automated scoring is illustrated for different types of assessments such as speaking and writing from both methodological aspects and practical considerations. Further, automated test assembly is elaborated for the conventional linear tests from both classical test theory and item response theory perspectives. Item pool design and assembly for the linear-on-the-fly tests elaborates more complications in practice when test security is a big concern. Finally, several chapters focus on computerized adaptive testing (CAT) at either item or module levels. CAT is further illustrated as an effective approach to increasing test-takers’ engagement in testing. In summary, the book includes both theoretical, methodological, and applied research and practices that serve as the foundation for future development. These chapters provide illustrations of efforts to automate the process of test development. While some of these automation processes have become common practices such as automated test assembly, automated scoring, and computerized adaptive testing, some others such as automated item generation calls for more research and exploration. When new AI methods are emerging and evolving, it is expected that researchers can expand and improve the methods for automating different steps in test development to enhance the automation features and practitioners can adopt quality automation procedures to improve assessment practices.
This book traces the development of computerized adaptive testing (CAT) from its origins in the 1960s to its integration with the Armed Services Vocational Aptitude Battery (ASVAB) in the 1990s. A paper-and-pencil version of the battery (P & P-ASVAB) has been used by the Defense Department since the 1970s to measure the abilities of applicants for military service. The test scores are used both for initial qualification and for classification into entry-level training opportunities. /// This volume provides the developmental history of the CAT-ASVAB through its various stages in the Joint-Service arena. Although the majority of the book concerns the myriad technical issues that were identified and resolved, information is provided on various political and funding support challenges that were successfully overcome in developing, testing, and implementing the battery into one of the nation's largest testing programs. The book provides useful information to professionals in the testing community and everyone interested in personnel assessment and evaluation. (PsycINFO Database Record (c) 2004 APA, all rights reserved).
This new text provides the most current coverage of measurement and psychometrics in a single volume. Authors W. Holmes Finch and Brian F. French first review the basics of psychometrics and measurement, before moving on to more complex topics such as equating and scaling, item response theory, standard setting, and computer adaptive testing. Also included are discussions of cutting-edge topics utilized by practitioners in the field, such as automated test development, game-based assessment, and automated test scoring. This book is ideal for use as a primary text for graduate-level psychometrics/measurement courses, as well as for researchers in need of a broad resource for understanding test theory. Features: "How it Works" and "Psychometrics in the Real World" boxes break down important concepts through worked examples, and show how theory can be applied to practice. End-of-chapter exercises allow students to test their comprehension of the material, while suggested readings and website links provide resources for further investigation. A collection of free online resources include the full output from R, SPSS, and Excel for each of the analyses conducted in the book, as well as additional exercises, sample homework assignments, answer keys, and PowerPoint lecture slides.
This book introduces computer-based testing, addressing both nontechnical and technical considerations. The material is oriented toward practitioners and graduate students. The practical emphasis will be useful to measurement professionals who are or will be responsible for implementing a computerized testing program. The instructional information is also designed to be suitable for a one-semester graduate course in computerized testing in an educational measurement or quantitative methods program. While certain theoretical concepts are addressed, the focus of the book is on the applied nature of computerized testing. For this reason, the materials include such features as example applications, figures, and plots to illustrate critical points in the discussions. A wide range ofnontechnical issues need to be considered in implementing a computer-based testing program. Separate chapters are provided on test administration and development issues, examinee issues, software issues, and innovative item types. Test administration and delivery issues include the location of exam administration, selection of hardware and software, security considerations, scheduling of administration frequency and time limits, cost implications, and program support as well as approaches for addressing reliability, validity, comparability, and data analysis. Examinee issues include the influence ofexaminees' reactions to adaptive testing, the effect ofcomputer based task constraints, and the impact of examinees' prior computer experience. Software issues include usability studies and software evaluation as tools in selecting and developing appropriate software, based on the test program needs.
Introduction to measurement -- The one-parameter model -- Joint maximum likelihood parameter estimation -- Marginal maximum likelihood parameter estimation -- The two-parameter model -- The three-parameter model -- Rasch models for ordered polytomous data -- Non-Rasch models for ordered polytomous data -- Models for nominal polytomous data -- Models for multidimensional data -- Linking and equating -- Differential item functioning -- Multilevel IRT models.
