Jason P. Davis – Massachusetts Institute of Technology
Kathleen M. Eisenhardt – Stanford University
Johnathan Cromwell – Harvard Business School
Paola Zappa – University of Lugan
Article link: http://asq.sagepub.com/content/56/2/159.short
Question 1. In your findings, you found a strong pattern that rotating leadership between organizations had a positive effect on performance. However, when developing innovation it’s important to have both structure and flexibility, and it’s conceivable to think that too much rotation could result in too much flexibility and not enough structure. Was there any evidence from your data to suggest that perhaps too much rotation could be detrimental to performance? If you were giving advice to companies to pursue a rotating leadership strategy, what caveats would you provide?
First of all, thank you for reading our paper so closely, and the opportunity to talk about it in this forum. The ASQ blog is becoming a great new institution that provides authors an opportunity to reflect on their work. And thanks for noticing a link between our research about rotating leadership in collaborations and flexibility, two topics that we’ve spent considerable effort thinking about in both this 2011 ASQ paper, and in our 2009 ASQ paper where we examined the tension between efficiency and flexibility and its link to the amount of organizational structure. So of course, we agree that it possible that too much rotation could hinder efficiency, although, to answer the question directly, we actually didn’t find any evidence of this in our data. One reason this may be is that generating leadership rotations is actually fairly costly and even disruptive, and involves significant managerial attention to be done effectively. That is, alternating control is hard and few managers would attempt to do an extreme number of rotations, so its not surprising that there would be few data points on the “too much rotation” side of an inverted U-shaped curve, even though it is theoretically possible.
Given this cost and potential disruption, probably the biggest caveats we would suggest to companies seeking to pursue a rotating leadership strategy would be to think about whether this approach is a good fit for their collaborative objectives and environment. As we note in the paper, rotating leadership is most useful in highly interdependent ecosystems like computing and communications where there are natural complementarities that can be productively combined into innovations. We speculate that consensus leadership styles may work just as effectively as rotating leadership in collaborations not involving recombinant innovation (e.g., joint marketing alliances), since they are simple enough that most issues can be resolved using consensus-based processes. And domineering leadership is clearly effective in straightforward technology licensing arrangements where a larger player is trying to efficiently extract value from IP. That said, innovative collaboration is becoming increasingly common in industry ecosystems, so we think rotating leadership will maintain its relevance.
Question 2. You found that a rotating leadership model was highly valuable to organizations that had complementary assets and were pursuing innovation. However, many alliances are established between organizations that have similar assets and pursue different goals such as economies of scale or economies of scope. Would you expect rotating leadership to be equally valuable in cases different from those examined in the paper?
This is an interesting way to put the question. Above, we discussed the managerial attention required to implement rotating leadership. But we do believe that the process could be equally effective in scale and scope focused alliances, although it is costly. In fact, one thing to keep in mind is that many “simple” alliances focused on exchange can easily morph into “complex” alliances focused on innovation, in which case using a rotating leadership process can be more useful. In fact, this is a common story in the computer industry, such as the case of Intel and Microsoft who began their 30-year symbiotic relationship as a series of exchange alliances. But as the two firms became more interdependent, they discovered opportunities for technology collaboration that led to the development of the Wintel PC platform. Their latest challenge is figuring out whether they should “enlarge” these relationships to collaborate on common innovation objectives with multiple other prominent firms like Cisco or IBM (so called “multi-partner alliances”), a topic that Jason explores in a recently provisionally accepted ASQ paper that will probably appear in the first half of 2016.
Question 3. One aspect that makes your study so impressive is that you worked with sixteen organizations (i.e. 8 collaborations) all from the same industry, and you were studying processes that could be considered highly sensitive. How did you manage the competing tensions of providing confidential value to each organization while also recruiting several competitors so you could develop generalizable insights for the study?
Thanks for the compliment, although we should probably correct that we actually studied 10 organizations participating in 8 alliances – that is, a few companies participated in multiple collaborations in our sample. That was actually by design, because we wanted to understand the locus of collaborative approaches – is it the firm or the relationship or the dyadic processes themselves? Each explanation is present in the alliance literature. Research designs like these are one way that multiple case research tries to achieve experimental control.
We appreciate your question about informants. We get this a lot, and the answer is often surprising to those who ask us. In most cases, it actually wasn’t that difficult to get these corporate executives to talk to us. We were probably aided by the fact that these kinds of alliances are both important and not well-understood by the executives involved. So they were very interested in knowing what works and what does not. So most executives quickly agreed to chat (although scheduling these conversations sometimes took time). We promised them all that we’d report back our findings (without violating any confidentiality, of course) – essentially, we gave them our ASQ paper . We also told them that we would disguise names. But, of course, some informants were worried that their competitors (or bosses) would guess their identities. To this, we often reminded them that the academic timeline of publishing papers is substantial such that the specific company information (but not the findings!) would be out-of-date before the paper was published. Ultimately, we think many of them were motivated by a desire to support the production of scientific knowledge, and to help a struggling grad student with his dissertation. Many students are scared away from inductive research because they think access is impossible – in fact, being a doctoral student is one of the most privileged positions to be in, since so many people want to help you, if only you’d ask.
Question 4. Sample selection is crucial for comparative case study research, especially to derive theoretically important arguments. What was your process in selecting this sample and developing your theoretical insights? What guidelines to selecting organizations would you give students who are interested in conducting similar research?
That’s right. Sampling is essential for drawing correct inferences from inductive studies. We use theoretical sampling where the goal is to capture important variation but control for alternative explanations. One way to think about it is like this: First, you consider research focus and generalizability. We knew that we wanted a range of technologies to avoid just a “software” or “semiconductor” story – i.e., something broader about collaborative processes. Second, you think about alternative explanations. So, for example, since having socially embedded relationships and strategic interdependence is clearly important in the alliance literature, we choose collaborations that all had these features. Going back and forth between the literature and our pilot study informants was helpful in figuring this out. The key insight from these executives is that having good contracts, trust, frequent interactions, complementary technologies, etc. were “the basics” of collaborating, but that the real challenge was the process of managing innovative collaborations. This confirmed that the prior studies were capturing a modest amount of the performance variation, but also that there was a wide gap between what the focus of prior studies and the challenges of collaboration – so a good chunk of variation remained to be explained.
Question 5. When writing qualitative research, quantitative data is often included to bolster findings, but it usually plays a secondary or supporting role for the qualitative data. In your paper, the quantitative data seem to play a more central role – for example, you calculated network cascades in each organization to show how diverse members are mobilized across different cases. When do you find this approach to be more appropriate as opposed to a more purely qualitative approach? What are the major challenges when going through the review process?
This is one of the biggest misconceptions about multiple-case research. We don’t see ourselves as doing “qualitative research” – instead, we’re doing “inductive research” where we are trying to build new theories, and will use whatever data best serve that purpose – whether they involve words, numbers or both. Of course, qualitative data are the backbone of each case, but quantitative data are especially useful for comparing across cases (given the right measures). It is also possible that our backgrounds inform this: we are engineers and scientists by training, and so we are in our comfort zone with numbers. But there is another critical reason: we try to keep in mind that we are actually participating in a broader scientific process that involves developing and testing theories. Inductive researchers develop theories, and deductive researchers test them. That is, we see ourselves in conversation with deductive researchers because, ultimately, we want them to test our theories. Developing constructs that can be measured numerically helps others to replicate and test our work with larger samples. We think that using mixed measurement methods actually helps in the review process, if only because inductive papers are likely to pull diverse reviewers – in this study, ranging from case methods experts to network theorists to scholars of organizational change. Different scholars prefer different approaches, and so producing convergent measures can help to convince them of the emergent findings because they can all find something they prefer.
Finally, as with nearly any ASQ paper process, the greatest challenge was theoretical development, yet we’ve come to see that as a normal and useful way to improve thinking. The ASQ community has deep expertise, and we were lucky to draw on it. We are enormously grateful for the challenging and insightful comments of our three reviewers and editor, Phil Anderson.
Again, we’d like to thank you for the opportunity to stroll down memory lane, and reflect on our rotating leadership paper. Thanks for all you do in running the ASQ blog – it’s a great service to the community.
Jason and Kathy