This study evaluates two databases, "Historical Abstracts" and REESWeb, to determine their effectiveness in supporting academic social science research. While many performance evaluations gather quantitative data from isolated query and response transactions, this study is a qualitative evaluation of the databases in the context of actual research. The evaluation tool was based on research done on the information goals and behaviors of academic social scientists. The information goals were divided into specific information tasks against which the databases were evaluated in the context of a simulated research project. The most notable project limitation was the linear design of the evaluation study itself. Although the databases were initially chosen for their differences (one is print, the other is electronic; one is an index, the other is a directory) the study indicates that both databases show strengths in the same task, that of extracting. General conclusions concerning databases designed to support scholarly research include: (1) in order for librarians to more efficiently support academic research it is important that they understand the stages of research; (2) every task should be matched to at least one tool--tools should be equipped to aid the academic researcher at each stage of his research; and (3) tools should be designed to allow maximum efficiency in research by performing multiple, related tasks, for example, extraction and selection. Appendices include overview questions, identification of key sources, REESWeb homepage, and lists of resources found in the second stage of evaluation of both databases. (Contains 32 references.) (Author/SWC)