| An increase in the number of hospital libraries and their expanded role within their parent organizations along with the economic restraints caused by inflation, government regulations, and increased number of available health related materials has contributed to health sciences libraries looking for ways to share resources. Given that a library's collections of serials and monographs represent the greatest contribution to cooperation, an increased understanding of the extent to which each library's collection is unique can help make coalitions more successful. This study attempts to quantify and describe the extent of collection overlap and diversity among eight health sciences libraries, members of a consortium of over twenty-five midwest libraries, and investigates how various factors such as library size, subject of material, and the Brandon/Hill List relate to the amount of overlap. Both monographs and serials have been studied. A possible measure of serials holdings overlap was defined in this study. A basic method of comparison was developed using health sciences libraries. Theoretical beliefs about two factors which could predict overlap of monograph collections, library size and subject, were confirmed. Major differences assumed to exist and borne out in earlier diversity and strengths were found in the small and medium sized libraries. A computerized means of analyzing holdings would be a useful tool for more extensive research as a year by year comparison by hand for each title among each paired set of libraries severely limits the size of study feasible. Given the scope of this study, generalizations about serials would be premature. However, a few patterns emerged that should be tested further. Overall, serials holdings overlap tended to increase as the size of the library increased. Subject also seemed to be an indicator of overlap among serials similar to monographs. For both monographs and serials, the selection tool was a strong predictor of overlap. In future studies other types of libraries should be studied to determine if selection tools specific to their libraries are useful in predicting overlap. |
