To what extent does the citation advantage of collaboration depend on the citation counting system?

This paper investigates the extent to which the apparently higher productivity and citation impact of authors with high levels of co-authorship depend on the way in which citation credit is assigned. For articles in physics it compares two methods of allocating credit, the standard (whole) counting system and the fractional counting system. Although numerous studies have found that when the whole counting system is used, more highly coauthoring researchers are more productive and publish more highly cited articles, no study has established whether this is true when the fractional counting system is used. This is particularly important as findings on co-authorship can be used to support the recent encouragement of co-authorship by research policy. This study finds: (a) when whole counting is used, the productivity and citation level increases with level of co-authorship; but (b) when fractional counting is used, the productivity and citation level decreases with level of co-authorship. This casts doubt on the received wisdom about the advantage of co-authored research.