Ethnic Patterns in Scientific Collaboration

08/01/2014
Featured in print Digest

...papers written by ethnically similar co-authors are cited less frequently and are likely to be published in lower-impact journals...

Scientific researchers based in the United States are more likely to co-author papers with people of similar ethnicity, according to Richard Freeman and Wei Huang in Collaborating With People Like Me: Ethnic Co-Authorship Within the U.S. (NBER Working Paper No. 19905). In addition to discovering ethnic patterns of collaboration, the study also finds that papers written by ethnically similar co-authors are cited less frequently and are likely to be published in lower-impact journals than papers by more heterogeneous teams of authors. Most of this effect is attributable to the fact that ethnically similar co-authors tend to have weaker prior publication records than co-authors who publish with scholars from other ethnicities, but some of the effect may also reflect the way interactions of people with different perspectives produce ideas.

The researchers analyze the ethnic identity of co-authors with U.S. addresses for more than 1.5 million scientific papers in the Thomson Reuters Web of Science database. They use a name-ethnicity matching program to assign ethnicity, which is divided into nine categories. Between 1985 and 2008, the proportion of names from all categories except Anglo-Saxon/English and European increased. Chinese names increased the most, from almost 5 percent to more than 14 percent. Analysis of first names suggests that an estimated 85 percent of the new Chinese co-authors were born in China.

Freeman and Huang compare the observed distribution of the ethnic composition of co-author teams with the expected distribution if co-author teams were randomly drawn from all authors in the database. For scientific papers with two authors, the ethnic groups for which the ratio of the actual number of ethnically similar co-authors to the number that would be expected based on random matching was closest to one are Anglo-Saxon/English and European, for which the ratios are modestly greater than one. For other ethnic groups, the ratios are much higher, in part reflecting the different number of researchers in these groups. Freeman and Huang note that they cannot reject the hypothesis that researchers in all groups have the same heightened affinity for working with other researchers from their own group. They find strong evidence of similar patterns of ethnic group collaboration for research papers with larger numbers of authors as well.

Co-authors with a larger number of previous papers were less likely to co-author a paper with someone of the same ethnicity, suggesting that they are able to draw on a larger network of potential collaborators outside their ethnic group. Freeman and Huang also find that research papers involving cross-university or cross-research group collaborations have higher impact factors and more citations. They suggest that diversity in location as well as in ethnic mix produces more influential science.

-- Linda Gorman