collaboration case study diversity United States

Case studies

Collaborating with people like me

Ethnic co-authorship within the US

Cognitive psychology tells us that each of us has a bias towards homophily, i.e. to have an affinity for those who are more similar to us than different. By Juliet Bourke - Consulting, Partner.

Ethnicity is likely to be one of the factors we each take into account in assessing another person’s similarity or difference. But does it really matter? Does collaborating with people with diverse perspectives really open our lens of understanding in a tangible way such that we will be able to generate demonstrably better outcomes?

To explore this question recent research compared whether collaboration between ethnically diverse researchers of scientific papers produced better papers than those developed by ethnically similar researchers. Professor Freeman (National Bureau of Economic Research, USA) and Huang (Harvard University and National Bureau of Economic Research, USA) found that people of similar ethnicity co-author together more often than predicted by chance, and that such homophily is associated with weaker scientific contributions. They also found that papers with more authors from more locations across the USA tended to be published in higher impact journals and receive more citations in subsequent journal articles.

In essence, the researchers found that diversity in scientific paper co-authorship is associated with higher scientific impact, suggesting that diversity of perspective expands a team’s thinking pool and therefore idea generation and exploration, let alone connectivity with more people.

US working paper, April 2014  

Aim

This research explored the changing ethnic and national origin of US-based scientists and engineers from the 1970s to the 2000s. It sought to examine whether researchers disproportionately worked with people of the same ethnicity and whether such ‘homophily’ (love of same) was associated with more or less valuable scientific work.

 

Method

The authors used Thomson-Reuters’ Web of Science database for the years 1985 to 2008 to identify the papers for which all co-authors had US addresses (~1.5 million papers) then classified the authors’ names into one of nine categories (e.g. English, Chinese, European, Hispanic/Filipino, etc.). They analysed the distribution of ethnic name origin of the authors and the number of citations of these papers in combination with the impact factor of the journals in which they are published.

Findings

Freeman and Huang’s two main findings re that: (1) people of the same ethnicity co-author more frequently than predicted by chance, and (2) papers authored by people with greater ethnic similarity tend to be published in less prestigious journals and receive fewer citations.

  • Not by chance

    The authors found high levels of homophily that were statistically significant for all nine ethnic groups and the homophily was largest for the smallest ethnic groups (Japanese, Korean, Vietnamese). They were unable to determine what factors and preferences led to this finding. It could be the result of all ethnic groups preferring to work with persons of their ethnicity, or some ethnic groups preferring to work with persons of their ethnicity, or differential preferences for homophily between the ethnic groups.
  • Diversity trumps homophily

    The second finding indicates that homophily of co-authors is associated with publication in weaker journals and results in fewer citations. The authors suggested that this outcome reflects a constellation of factors, including the possibility that researchers with weaker publication records will have a smaller network of research collaborators with whom to research and publish.

    Conversely, greater ethnic diversity between co-authors results in publication in more prestigious journals, and this is likely to reflect the significance of a study’s contribution to knowledge. This finding suggests that the ethnic diversity of collaborators contributes to idea generation, elaboration and exploration, leading to breakthrough insights worthy of prestigious publication and citation.

    Freeman and Huang explored a number of aspects including the effect of ethnic distribution of researchers across the US, multiple authorships of an article, prior publications by authors, and cross-checking with a different data source. These did not affect their conclusions.

    In summary, the research found that ethnic diversity in inputs into research papers led to greater contributions to science, as measured by impact factors and citations.

Implications

The direct implication of this research is to confirm the benefit of diversity achieving more impactful outcomes - in the context of ethnicity and US-based scientific research. This conclusion is consistent with many other studies on diversity and outcome quality. Thus, for researchers, greater impact is likely to be achieved by undertaking work and publishing as part of larger, more diverse teams.

More generally, two areas to consider are the ability to act on these findings and consideration of the cost/benefit. Firstly, increasing the diversity of teams should increase the quality of the outcomes they achieve. The research reviewed in this article doesn’t consider whether higher ethnic diversity was deliberate or accidental. However, it is logical to predict that deliberately increasing diversity will lead to better outcomes.

Secondly, the article highlighted the trade-off between increased short term productivity driven by easier communications from “one’s own group” against the value of complementary research skills and knowledge. This tension of speed vs. quality illustrates that multiple decision criteria may need to be considered when consciously adjusting the diversity mix of teams. In other words, is a potentially slower, more expensive and more difficult process worth the higher quality outcome? Or state this more provocatively, are there circumstances where diversity is not worth it? In the case of scientific research, publication and citation only track outputs and benefits in terms of the significance to scientific knowledge, not the costs and inputs.

To read the full article, see Richard B. Freeman and Wei Huang (2014) “Collaborating with people like me: Ethnic co-authorship within the US”, NBER Working Paper No. 19905, February 2014.

Did you find this useful?