by R.S. Renner

This thesis proposes hypertext technology as a design approach for improving the functionality of existing experimental management systems. Traditional experimental management systems exhibit one of two weaknesses, they are either 1) restricted to a specific problem domain, or 2) designed in a linear fashion. A linear design provides only sequential or random presentation of stimuli. Such an approach supports standardized general testing but does not allow the system to adapt or conform to be user-specific.

User-specificity is especially appealing in behavioral psychology, where much emphasis is placed on stimulus-response. For this reason, providing an experimental system with the capacity to present stimuli based upon user response is an appealing concept. This concept becomes even more appealing when the stimuli is allowed to cross multiple problem domains.

In a hypertext system the user navigates through the database, requesting only the information relevant to some chosen topic. In this way, the user determines the order of information retrieval by direct manipulation or selection. This means that information is presented in a logical order, rather than the traditional sequential or random approach. In hypertext, this is referred to as branch-logic routing.

Applying the principles of hypertext to experimental management systems provides a means by which the presentation of stimuli can follow a logical ordering, based on subject response. In other words, experiments can be constructed in such a way as to allow different user responses to take different routes through the knowledge base. The knowledge base domain represents the stimuli, a question set for instance.

A prototype system is presented to support the proposed concept. The system, called HyExpmt (Hypertext Experiment Executor) is designed to allow construction of experiments as well as experiment execution. The system supports adaptive testing through the design of a hypertext engine. Other features of the system include full support for multiple paradigms and problem domains, a tracking facility, a timing function, provisions for storage of results, formatting of results for external data analysis, re-usability of experiments, and internal editor, explicit error messages, and numerous customizable parameters.

Within the scope of this thesis are issues concerning experimental management systems, hypertext technology, and the HyExpmt prototype. A brief history of computer-based experimentation is provided, including a discussion of early and recent systems for psychological assessment. A detailed history and description of hypertext technology is also given, followed by a discussion of current hypertext tools and applications. Next the HyExpmt system prototype is explained, with a complete description of its goals, features, and functionality. And finally, conclusions are drawn regarding the proposed design alternative and its implications. Problems, limitations, and future research ideas are discussed, to provide avenues for the continuation and promotion of te proposed concept.

Thanks to C.J. Kacmar and J.S. Chiou for their contributions to this research, and for making the FSU HyExmpt prototype available.