ABSTRACT
This chapter addresses the item development, test construction, and scoring of open-ended responses with the goal of moving them from art to more formal science. This is moving educational and psychological measurement further toward being a science rather than an art form. The chapter is to show how the n-layer model can be used to generate test items in different languages. The basic concept that is presented for computerized generation of items is the item model. The item model is essentially a template for the placement of component parts of the item. While being very excited about this work from a scientific perspective, author has not yet been sold on the practicality of this type of methodology for large-scale, high-stakes testing programs. Automated test assembly (ATA) is a clear advance over human construction of tests. The literature on ATA describes applications with tens or even hundreds of constraints on the test construction process.
