| This dissertation conceptualizes the citation process as a dyadic relationship that simultaneously depends on characteristics of both citing and cited articles. Using data on two research areas, one in astrophysics and one in economics, this research develops a network-analytic approach to analyze all of the citation links from later papers to earlier papers in an area. Specifically, regression analyses examine the extent to which paper, author, and journal characteristics of both potentially citing and potentially cited papers influence the probability that a citation between the papers exists. Substantively, this project extends our understanding of stratification in science by assessing the relative worth of competing arguments on the use and functions of citations. Furthermore, a network approach permits the inclusion of variables indicating various relationships between citing and cited authors and papers so that I can test the argument that social ties between scientists influence their citation decisions. Finally, by adopting a comparative framework, this dissertation tests hypotheses about variation in the allocation of citations across disciplines differentially located along a hard-soft dimension.; Results from both generalized least squares and logistic regressions identify strong effects of cited article quality and content but only weak or insignificant effects of functionally irrelevant characteristics of cited author such as institutional prestige or eminence. These findings tend to support a normative interpretation of the allocation of citations in which citations reflect payment of intellectual debt rather than a social constructivist interpretation in which citations are rhetorical tools of persuasion. Furthermore, the lack of effects of social ties between citing and cited authors provides little support for the argument that authors who know one another are more likely to cite one another's work. Finally, consistent with much of the literature on disciplinary differences, my results suggest that the reward system is more universalistic and less dependent upon ascriptive characteristics in astrophysics than in economics. However, results from these two areas provide only weak support for the argument that the natural sciences differ from the social sciences in their pattern of rapid incorporation of new knowledge and cumulative building upon recent work. |
