Further evidence for adjusting for multiple authorship

Conclusion

With the research byLong, Allison andMcGinnis, Cole andCole, an-Garfield an initial understanding of the consequences of multiple authorship for empirical studies of science is emerging.⁹ It is too early to suggest, asLong andMcGinnis do, that indices which disgegard multiple authorship are without serious short-comings. The best advise would be to use measures which take multiple authorship into account, compare them with others that do not, and examine any differences that might emerge. At least for the data will which I have worked, multiple authorship makes a considerable difference. Although little difference is found with the data collected by long and his colleagues when examined within an unclear regression analysis framework, it would seem premature to recommend unadjusted counts. The best advice to the prospective researcher would be to collect complete information on the number of authors for all published work and all citations assuming that difference between adjusted and unadjusted counts will emerge and be comfortable with the knowledge that if it does, the measures used can properly treat the data.

The major point I made earlier, that the failure to adjust for multiple authorship is the most serious error in empirical judgment made in the sociology of science, emerges more forcefully. The error will require substantial redevelopment of previous empirical literature constructed with these faulty measures.

Finally, the point should not be lost that the most serious problem remains and that is the validity of the “paper model” of science with examines article and citation counts to understand this complex dynamic project of human inquiry.¹⁰