Toward a national information system for social science data files

Despite a breakthrough in library cataloging of files of machine‐readable social science data (directions are expected in the next Anglo‐American Cataloging Rules), the U.S. system for delivering information about data files still needs strengthening. While descriptive cataloging of locally held files by research libraries should prove helpful, suppliers' restrictions on “rediffusion” of data reduce the value of a national union catalog. Subject cataloging is also a problem, although a new form heading for data files would improve the usefulness of the Library of Congress Subject Headings. The greatest improvement would be inclusion of abstracts and indexing of the files in a national bibliographic service with on‐line search capability, especially one that could provide the codebooks required by those wanting to reuse others' data. With NTIS furnishing a prototype, ERIC is here proposed as a model of what is needed. ERIC has many desirable features as a “data information system” that would not disrupt present economic arrangements for delivering the files themselves.