Authors:
Linus Dahlander – ESMT
Daniel A. McFarland – Stanford University

Interviewers:
Michael Mauskapf – Northwestern University
Kevin Gaughan – Northwestern University

Article link: http://asq.sagepub.com/content/58/1/69

Question 1. The core of this fascinating paper focuses on how factors such as organizational foci, actor homophily, and triadic closure affect tie formation and persistence in the realm of academic collaborations, but none of these predictors address the power dynamics inherent in most collaborative relationships. So we were curious—other than tenure status and # of citations, how might you measure the effect of power and politics on this process? For example, do you have any predictions about how author order might influence tie persistence (e.g., if a first author continues to be first author for subsequent collaborations, we would expect…)?

We also included grant resources, an important source of power within universities, as Pfeffer would argue. But you are right that there are many other sources of power that we were unable to observe.

Our current work explores author order in this context more, but it is somewhat tricky to deal with empirically. For instance, some disciplines have alphabetical order conventions (such as economics), whereas others have contribution-based order (such as psychology). It becomes even trickier as a growing number of journals have multiple first authors (Science being one example), and other journals ask the authors how much they contributed to different parts of the paper. So the assumption that the first person listed on the paper is the first author in the sense of contributing the most have become more questionable in the last years. To really tackle this question in a good way, one has to either look at one discipline where conventions are similar, or even better, look through each article if it lists how work was divided. There are even some interesting variation within journals where policies have changed that would be interesting for scholars to look further into. All these ideas about power and author order should culminate in a follow up paper soon about unequal collaborations. Invite us out for a talk and we can share it with you!

One final idea that we played with, was how a core of two people could continuously collaborate over time, but swapped out additional team members from one collaboration to the next. The imagery is familiar in the sciences where we often see two lab leaders appearing on publications over and over again, but the post docs and students come and go. We were unable to do this for this paper, but this seems like a natural way to extend our work for a deeper appreciation of network dynamics.

Question 2. In your explanation of the results, you mention an unexpected finding in regards to your status homophily prediction—specifically, that same-tenure status was not a significant predictor of tie formation or persistence. What are some possible explanations for this non-finding? Could one argue that different-tenure collaborations are more prevalent than their same-tenure counterparts because they serve complementary functions for both junior faculty (e.g., mentorship, status enhancement) and senior faculty (e.g., motivation to be productive, up-to-date knowledge of new methods and findings)?

One simple reason for this non-finding is that we have many additional controls that capture what researchers actually do, such as the amount of grant funding and their reference similarity. Once you start accounting for more alternative explanations, the effect of homophily becomes weaker. We know this from supplementary analyses that we pursued.

But yes, your explanation sounds feasible. Complementarity is sought after in these instrumental ties. We fully suspect that junior faculty are brought onto senior faculty grants and in exchange for funding the juniors give up expertise, updating the senior faculty’s methods and references into current literature. Our recent work with Craig Rawlings and Dan Wang in Social Forces points in this direction.

Question 3. After reading your paper, we remained curious about the role played by overall performance and/or reputation—is this addressed in your means-ends rationalization/outcome success measure? We weren’t sure whether “the annual number of forward citations per team member” refers to (A) the number of citations received by the collaborative project in question in year t, or (B) the total number of citations received by an author across all articles or grants in year t. Your answer might explain the lack of support for Hypothesis 6: if A, perhaps tie persistence is driven by a collaborator’s overall performance, rather than the performance of a particular project; if B, perhaps project-specific performance is what matters.

We measure it using approach (A) as we wanted to test whether the reputation of the collaboration affected whether it persisted. For grants, we counted the number of citations if they also had a publication published that received citations. You are right that we did not test the overall number of citations of the collaborators’ papers, so we cannot fully disentangle those two effects. Wish you were a reviewer!

Question 4. You test a diversity of predictive factors, but to what extent does liking—or emotions more generally—play a role in tie formation and persistence? You hint at the importance of these factors in your discussion, and they might be difficult or impossible to measure given your data, but what outcome(s) might you expect given your other findings? Perhaps one way to gather some indirect evidence would be to split your sample across tenured and non-tenured observations, and look for systematic differences in your results. For example, one might expect the magnitude of your effects to be lower for tenured faculty because they feel they can afford to make their collaboration decisions based on inter-personal liking, rather than the more objective, strategic, or structural factors your test. Ties formed and reinforced for emotional reasons might lead to less impactful collaborations, but they might also be viewed as more fulfilling by the collaborators in question. What do you think?

Our intuition is that emotions certainly play a role in tie formation and persistence. We are willing to forsake other opportunities if there is someone we really like as a person. Some people talk about the importance that potential collaborators pass the “beer test” – a person you can enjoy talking to outside work. After all, working with someone on a first draft, through multiple rounds of revisions in a journal such as ASQ takes a lot of time. If the relationship is purely professional and not multiplex to also surround other kinds of activities, then we are less likely to persist in the relationship. The effect of multiplexity is also captured in our models, even though it’s only within the realm of academia, and not relationships outside work.

The idea of the sample split is an interesting one, or potentially interact other variables with the tenure status to see how it changes the marginal effects. There are many possibilities to test such contingent effects that we simply did not have space enough for in the paper.

Question 5. What your next studies of academic collaboration?

We have several follow up papers in the works. The first is mentioned above and it looks directly at power differences in collaborations and how that can affect the flow of knowledge and activity in universities. A second paper we title “Network Careers” and it concerns how faculty build their networks in an attempt to survive as an academic. And then we have additional papers in the works where we are trying to extend these ideas and models to far larger data on say all of Thomson Reuters Web of Science. Simply put, we have a lasting collaboration here and will continue to find value in it!