Despite their professional training and study in the development of research collections in academic settings, librarians often consult with or even defer to faculty in selecting materials. Faculty often use various methods of evaluation that tend to emphasize qualitative data or even anecdotal evidence. Bibliometric analysis offers emerging tools to quantify these decisions, reflecting fundamental principles of library science. This study compares faculty choices of serials subscription cancellations to the choices that would have been predicted using a bibliometric tool, the California Digital Library Weighted Value Algorithm (CDL-WVA). Faculty choices differed significantly from the decisions predicted by CDL-WVA. However, as the bibliometric score increased, so did the rate of match between faculty choice and decisions predicted by CDL-WVA. Implications of these findings for collection development are discussed.