Jason Greenberg – Leonard N. Stern School of Business, New York University
Ethan Mollick – The Wharton School, University of Pennsylvania
Amy Ding Zhao – INSEAD
Hello ASQ readers and bloggers, my name is Amy Ding Zhao. I am a doctoral student at INSEAD and part of the organizing committee for the ASQ blog. This is a new installment of our ASQ podcast series. Today I’m interviewing one of the authors, Ethan Mollick, about their forthcoming ASQ article in the June issue, “Activist choice homophily and the crowdfunding of female founders.” Unfortunately, the other author, Jason Greenberg, cannot join us live today, but his comments will be incorporated in the written script of the interview. Ethan, thank you very much for joining us today.
EM: Thank you. I’m glad to be here.
Great. It would be nice if we can start by you telling me a little bit about the background. How did the idea come into place, and what motivates you to undertake this project?
EM: So actually, this started with as much an empirical puzzle as anything else. I’ve been doing work in crowdfunding for a while, and Jason and I had been talking about some data. There had been a repeated finding that women were outperforming men in crowdfunding, and that’s weird because women are unfortunately discriminated against in basically every way of raising funds for startups. Women actually have a very low rate of venture capital. Women make about 38% of business owners in the United States depending on how you calculate it. They make up about 3% of VC recipients, so it’s shockingly low numbers. That same kind of bias has been found in bank loans and angel investing, so this was a really shocking phenomenon that women were actually outperforming men. We wanted to figure out why that was the case. We started looking into this puzzle, and we found something weird which was that, we thought the reason why women would do better than men was because unlike other forms of fundraising which tend to be male dominated, there are quite a few women who fund projects on Kickstarter. We thought that in areas like fashion or children’s book publishing where there were more women than men doing funding, that’s why women would be succeeding because women would be helping each other out. What we found was the opposite—that women were actually succeeding where there were the least women. In technology and video games, where like 20% of the funders were female, women were actually outperforming men by the highest margin. That was a really interesting sort of empirical puzzle. That didn’t comport well with theory. We wanted to try to figure out why, and that’s how we started this exploration.
Oh, that explains why you introduced the concept of activist choice homophily. It’s because of the empirical patterns that didn’t really go well with the existing theories. Could you elaborate a little bit on that concept?
EM: Yeah. Homophily is one of the oldest and most grounded sociological approaches. There’s other words for it in economics and in psychology but the idea that, as Aristotle said, “Birds of a feather flock together.” People like people who look like themselves. One of the things that was interesting about homophily as it’s theorized it it’s very dyadic. My attraction to you is based on self regard. The more you are like me, the more I will like you or the more we overlap in traits, but dyadic, I’m spotting you, which is kind of interesting when you think about it because it’s not like we don’t have outside reference. Part of what activist choice homophily says is it says that when we look at someone else, we’re not just considering them, we’re also considering shared group membership. That’s influencing the degree to which we see homophily there. That makes complete sense when you think about it. I meet other Americans every day, but there’s not a shared disinterest or shared sense of discrimination that would force me to believe that I should be extra nice to other Americans as opposed to other people. The idea is that referent group matters, it’s not just a dyadic relationship. I think that that ends up being a really interesting point that sort of unlocks a lot of interesting angles on how to think about homophily.
In this paper you employed a mixed method using both lab experiments and large- scale quantitative analysis with real data. What motivates you with that decision?
EM: Increasingly . . . First of all, Jason’s not here but I give him credit. He’s done a lot of the . . . he and I really worked together on this set of stuff. I’ve been increasingly doing mixed methods stuff. I know Jason has too. I think that there’s a few reasons for it. One is, we’re under increasing pressure to show both mechanism in our work and impact. It’s hard to show both those things in one place. There’s a real world phenomenon that made us interested in this, so we wanted to be able to see that this reflects in the real world, but you can’t get mechanisms very easily even out of these launching of field data, unless you’re very careful. The lab lets us get at the mechanism, and then we can see if that mechanism is what explains patterns in the real world. We ended up running many, many experiment to make, they’re experiments that took a huge amount of time and are just footnotes now in this paper, but it was a really interesting process to try and nail this down with mixed methods.
You guys are both increasingly using mixed method, and you went through the whole process of writing about it and going through the R&R. Do you see any challenges and things that you want to mention in that process about mixed method studies?
EM: Yeah. Without seeming too pedantic, though I guess I’m an academic so that’s what we do, mixed methods are risky in some ways, because what happens is that no one is comfortable with all methods. You have reviewers—and we had great reviewers on this paper and our editor was terrific so this is not criticism there—but just in general mixed methods, part of the problem is that you have to make everybody happy with every method. What you’d hope is a mixed method paper is about accumulation of evidence in some ways. It’s hard to get . . . This paper we happen to have some pretty solid results everywhere, but a lot of cases, you’re sort of circling around a point or eliminating alternatives. As a result, you’re using a bunch of results that are open to interpretation or discussion. You’d hope that you get more credit for doing mixed methods and having a lot of answers that are all sort of pointing the same direction, but instead you need to make sure you’ve really nailed each part of the mixed methods. Reviewers might know experiments but might not know large-scale matching approaches. It ends up being, it could be difficult to navigate, but again, our editor was terrific on this, and it was very a helpful process to get through it, but I would say the mixed approaches are often more difficult for that reason.
Do you see a trend of people embracing it more and more?
EM: I think so. I think especially for those of us who do, who are not economists who publish in econ journals, we’re still being touched by the identification revolution on one end, by experiments on the other. If you’re doing this economic sociology work or entrepreneurship work, or there’s other work like that, I think that the way you maintain the ability to do identification and get clean results and still talk about significance, I think these mixed methods are increasingly going to be the way to go. I also think the truth is we’re entering a world where we’ll be drowning in data pretty soon. Just showing correlations and stuff is not going to be enough. I think we need to train up in these kind of approaches.
I’m interested also in the real-world data part of the paper. The Kickstarter data you used seem to be a really rich data set, and apparently you have devoted a lot of time and energy in putting it together and cleaning it. I feel like the digital era has provided us researchers with all this access to information and data. What do you see as the challenges and opportunities of using these large-scale, very detailed and web-scraping data? Do you have suggestions to the aspiring scholars?
EM: Yeah. I think we’re on the cusp of some interesting stuff. I’m just starting to play around with machine learning myself. The number of big data sets you get access to is increasing on a regular basis. I ask the question mentally, “Does Facebook already know the answer to any of the things I’m checking, and if they do, do I really . . .” You know, if Facebook could push a button and tell you for the one billion people on Facebook it turns out this is how you make friends, this is how relationships work, they can answer those questions. We need to make sure . . . In some ways it’s both a response that we need more practical theory that we can do that has results to guide this set of stuff, but also I think that we have an obligation to think about these methods and think about what they mean. I’m a big fan of data sets that are exciting that you can actually understand really well. I’ve backed 65 Kickstarter projects. You can’t see it because you’re listening to a podcast, but some of my . . . I’ve got Kickstarter stuff all over my desk of fun things that I’ve gotten on Kickstarter. Before that I studied video games, and I collected my own survey data. I think that there’s a sense of making sure you know what your data set is, building up a rich data set that you can ask lots of questions. I am a big survey person. That’s been a very helpful approach to use also. I think collecting a unique data set no one else has and getting up to speed on those methods and figuring out how you can ask theoretical questions that are not necessarily unattached to data but help guide interesting conversation about mechanisms is important.
The crowdfunding platforms represent an example of new contexts we see today. In your experience, how are these new contexts perceived in the publishing process in our field?
EM: I like weird data sets a lot. I think there’s advantages and disadvantages. The advantage is it’s intriguing and hopefully relevant—not that 17th-century bridge building isn’t important also . . . or the patent database or anything else—but it’s nice to have something that no one else has, gives you an advantage, makes your work more interesting. There’s a weird factor in publishing I’ve seen though which is that, having a really interesting data set can give you sort of points that count toward getting a first-round R&R, because people are excited to see more about the data, but actually can hurt you later on because it can make people intrigued by your context but then disappointed you can’t deliver enough interesting results. Getting caught up in the novelty of the context can actually end up slowing you down because you get more first rounds and less second and third rounds because you end up with reviewers who might be interested in the context but not in the work. The good thing about the crowdfunding stuff is I’m lucky enough to be the throwaway cite. That’s the goal of every early-stage scholar is you get the throwaway cite, where people dismiss a field out of hand, they cite you: “Also you can raise money through crowdfunding (Mollick, 2012)”. It’s nice to be in a field where I think there is some validity in it, but I like intriguing data. I know this doesn’t work for everyone, but for me, working in data that I care about, and I know really well . . . it could be depressing too. I was telling you earlier that I did some stuff on video games that involved using this giant database of every PC game ever made. There’s like 5,000 or 6,000 entries I was working with. I realized I had played enough of them that I could hand-check the data set. I realized, “Oh, wow, I’ve played too many games in my life.” But there’s a sense if you know the data well, then patterns that other people wouldn’t get make sense to you, like “why is this thing showing up strangely? Oh, I understand the data really well.” So I think weird contexts work really well, novel things work well if they’re relevant and if you know a lot about them.
After completing this project, did your other work also evolve in some certain ways inspired by the findings and experience in the study?
EM: Yeah. First of all, again . . . my co-author is also doing work on this set of stuff, but I remain really interested in this gender and disadvantaged piece in general. I study entrepreneurship, their systematic biases against many different classes of individuals based on sociodemographic, geographic, all kinds of other characteristic human capital. If you think about it, if entrepreneurship is the engine for innovation and the engine for economic uplift and economic growth as a society, then the fact that we’re leaving people behind who would otherwise be good at doing this is really tragic. I’m involved in a bunch of projects trying to think about why these causes are happening and focusing on that phenomenon. I’ve done a bunch of crowdfunding work, but crowdfunding is interesting just because it’s not left censored. Crowdfunding is cool in and of itself, but as a data set, it’s a really cool data set because it isn’t left censored. I can watch failures at a very early stage. Now I’ve surveyed 50,000 crowdfunding projects and 400,000 people so I have this really rich data set I’m now working with on this set of stuff trying to get even deeper.
Last but not least, I’m very curious about your collaboration on this paper. Could you share with us a little bit of when you decided to engage in this joint project, and how did you work as a team throughout the process?
EM: Jason and I were in grad school together. We’ve got that natural connection. I don’t remember the exact origins, I think I wanted to approach him about these patterns. This was incredibly even lift on these sets of things. We both worked really hard on this stuff. I’ve had papers where the theory building was very divided from the empirics or vice versa, but this was a much more blended effort. It helped that we went to grad school together so we knew some of the same literature and some of the same people. It helped that . . . I think Jason taught me some empirical techniques. I think he probably took the lead on the large data set stuff and then I had a lot of, I was lucky enough to work with a couple of people here like Nancy Rothbard and Sigal Barsade and people like that who helped teach me some of the experimental techniques that I hadn’t learned. I maybe took a little bit more lead on that, but we both learned a lot of stuff and then taught each other. This was a really ideal collaboration. But of course you say that after the fact once it’s published so we are selected on the dependent variable, right? (laugh)
It’s a great pleasure to have you for this ASQ podcast. Thank you very much again for joining us.
EM: Awesome, thanks for letting me talk to you guys.
Comments from Jason Greenberg
Dear ASQ Readers and Blog Listeners:
Unfortunately, my teaching, administrative, research, and familial (I have a newborn son, my second child…) obligations made it impossible to participate in the podcast concerning the activist choice paper with Ethan. Please accept my sincerest apologies.
I do, however, want to take this opportunity to discuss one element of the publication process—working with a co-author. Fortunately, as Ethan mentioned, we had history going back to our graduate school days at MIT that made the initial stages of the process far smoother than is the case with strangers or acquaintances (much like the literature on entrepreneurial teams suggests).
If I recall correctly, this paper, or more precisely, the idea to work on a paper together, was conceived of at a Tuck Summer Camp several years ago. Ethan told me about some cool data he had and we then began talking about various ideas we could pursue with them.
Like many research papers, the road was winding, and there were ups and downs. Fortunately, we worked well together and Ethan’s “can-do” attitude helped us overcome and move forward. I think it also helped that we have a similar mindset and timeline for getting work done, which minimized one potential source of tension that can arise when, for example, working with very senior (or very junior) colleagues that have a different time horizon. It was also nice working with a co-author that served as an invested and motivated source of social support and encouragement. Conducting research is a fantastic privilege and job, but at times you can lose your way or interest in a topic given how long of a process it is. A (good) co-author helps you through that.
Thank you. To the ASQ blog community, thanks again for listening. Ethan and Jason’s activist choice homophily article we discussed today will be featured in the June issue of ASQ and is already available online. The next installment of our ASQ podcast series will be coming out in August 2017 when the September issue of ASQ is published, so stay tuned on the blog. Thanks!
Ethan Mollick. (2012) “People and Process, Suits and Innovators: The Role of Individuals in Firm Performance,” Strategic Management Journal, 33(9), 1001–1015.