Standing on Academic Shoulders: Measuring Scientific Influence in Universities

This article measures scientific influence using citations to academic papers. The data source is the Institute for Scientific Information; institutions include top U.S. research universities. The fields represented span science; and the time period is 1981-1999. The database includes 2.4 million papers and 18.8 million citations that account for much of the basic research conducted in the United States in the late 20th century. We use the citation probability, or actual citations divided by potential citations, to capture utilization of the literature by individual articles. Within fields the mean citation probability is roughly 10^-5. Cross-field probabilities are less than a one-tenth as large and are significant in less than a fourth of the possible cases. Field restricts citation, and this fact suggests limits on scientific influence. Cross-field probabilities are symmetric for mutually citing fields. However, ranked by quality of institution, citation probabilities are asymmetric within fields. Citation probabilities from lower to higher ranked institutions exceed the reverse citations, though the latter are significant. Higher ranked institutions are more often cited by peers than lower ranked institutions. This suggests that knowledge flows from peers increase with rank of institution. Overall the results suggest that spillovers in basic science are important but bounded, limiting the knowledge that spills over between one scientist and another.