Hambleton, R.K., University of Massachusetts, USA
The next generation of educational and psychological tests will be administered and scored with a computer, and the computer will increase the validity of scores from these tests by (1) permitting the expanded use of new and helpful item formats, (2) permitting new test designs such as multi- stage and computer-adaptive tests, (3) permitting full information scoring algorithms, and (4) creating new possibilities for score reporting. The impact of computer-based testing (CBT) on testing practices in the future is certain but at the same time, for every promise of CBT, there are problems that must be overcome for CBT to achieve its full potential. A full list of promises of CBT will be presented (e.g., shorter testing time, increased test security, immediate reporting of scores, testing on demand, extended use of video and audio), and for each promise, a set of related problems will be described and how these problems might be overcome or addressed with appropriate research. For example, one promise of computer- adaptive testing is shorter tests, but also computer-adaptive tests do not permit candidates to omit questions or return to questions and change their answers. As a result, some candidates question the validity of their test scores. One possible solution is to use test designs that shorten testing time, but permit omits and changing answers. Multi-stage testing is one promising possibility. In total, 10 to 12 problems will be presented. In introducing 10 to 12 problems, special attention will be given to technical advances in the areas of test design and new item formats, and technical problems such as handling item exposure and field-testing new items. In a final section of the paper, several directions for future research will be described.