By Authors: Xu Li (London School of Economics and Political Science); Amandine Ody-Brasier (McGill University)
Interviewers: Yeaji Kim (University of Illinois at Urbana-Champaign); Lingyun “Lydia” Zhang (University of California, Irvine)
Article link: https://doi.org/10.1177/00018392251320898
We are excited to discuss your paper, “When Breaking the Laws Gets You the Job”, which offers a fascinating perspective on how lawbreaking can sometimes elicit community support. To start, we’d love to learn a bit more about the origin of the project. What initially sparked the idea, and when did it evolve from an interesting observation to a full research project?
Xu: We both did our PhDs at LBS and are good friends. I have long admired Amandine’s work, and we always wanted to work on something together. I think the starting point of this project was when I was an assistant professor at ESMT (European School of Management and Technology) in Berlin. I somehow ended up living in an area in Berlin with lots of clubs, which was frankly a bit annoying sometimes. And then, interestingly, one of my students was a DJ at the time, and he introduced me to the scene a little bit. Also, personally, at the time when I was going out to meet friends, I got the impression that in Berlin, there are two kinds of dominant professions, entrepreneurs and DJs. So, it always made me wonder how on earth do these DJs get jobs? How do they compete against each other? How do they stand out? We had some casual discussion about this when Amandine came to visit me, and I also introduced her to my student. We actually went to a few clubs, just to observe a little bit, and then we felt like there might be something worth exploring about the dynamics of this very unique community.
We started with a smaller sample of data. After that, we thought we should try to see if we could collect more data quantitatively, because we’re both mostly quantitative scholars, and then it all started from there. I would say it’s more of our curiosity about the scene, and to understand what happens, and then it started organically from there.
Amandine: Yes, it really started with an empirical puzzle. Xu reached out to me and talked about these amazing data that he had been collecting. He was interested in DJs, and was based in the world’s nightlife capital. We started looking at the quantitative data and then did interviews and talked to people in the field. Clearly, there seem to be different interpretations of breaking the law and bootlegging. So, we started wondering why. The question changed from why people do bootlegs to why people interpret bootlegs differently. Some supported bootlegs, whereas others talked about them in rather negative terms.
“We noticed a very interesting, puzzling empirical pattern, and noticed the variation in how people were talking about an illegal act. It’s only later that we connected it to the theory as we were working through the data, There were several possible explanations for why this might be happening.”
That’s interesting that the project was based on real-world observation. I think that gave us some answers about the second question, whether the research question originated from theory or from the context, and it was from the context.
Amandine: Yeah, 100%. We noticed a very interesting, puzzling empirical pattern, and noticed the variation in how people were talking about an illegal act. It’s only later that we connected it to the theory as we were working through the data, There were several possible explanations for why this might be happening. But most of them were connected to differences in skills, expertise, creativity, or potentially quality, and these were not sufficient to explain what we were observing. So, it pushed us to look more into the interpretative lens. To understand how people were making sense of engaging in this particular action – bootlegging.
Xu: I think that for us, the empirical surprise actually emerged very early in the research. When we think about why bootlegs would lead to more opportunities to perform, it is unusual. However, within both our data and our fieldwork talking to DJs, it seems to be something that happens organically, and was almost constantly being talked about by people in the field with different interpretations. So, it did not take us too much time to decide what to focus on. It was something that was surprising to us, and empirically, it was very robust.
The hand-collected dataset (covering 38,784 DJs from 2007 to 2016) is really impressive, and the project combines observational data with an online experiment, an expert survey, and interviews. Could you walk us through how long the project took overall? And was this mixed-method design part of the original empirical plan, or did it develop over time?
Amandine: Yeah, there is quite a lot going on. As Xu described, it really started with that very large quantitative dataset, which allowed us to establish the empirical pattern and show that something surprising was happening: the community was responding pretty positively to people engaging in illegal behavior. These DJs are getting hired more frequently, and that’s surprising. Based on this, we had already decided to do some interviews, since the idea was to understand the community and its specific norms, which would be really hard to capture just through secondary data.
The online experiment came through the review process. We had a good sense of what we thought was the mechanism, around the perception of a disinterested act versus not. But there were a bunch of other explanations that could also account for the results, and that’s where the experiments really helped us narrow the space of plausible explanations and check that the effect was not driven by skills, quality, or attention. It is truly about how people interpret that particular behavior.
Empirically, the logic was sort of sequential. We knew we wanted to begin with a large dataset and then draw on qualitative data to better understand the phenomenon from the perspective of people in the field. Through that process, we also included the experiment and a survey, which helped us rule out more straightforward explanations.
Xu: Yeah, it’s a long process, but I think it’s rewarding in that it allows us to really narrow down and nail the mechanisms. As Amandine mentioned, in the very first version of the paper, we already had quite extensive quantitative analysis based on the field data, and we had also done tons of interviews, so we were confident about what was happening there. But of course, in the review process, there were a lot of questions about more precisely ruling out alternative explanations and nailing down the precise mechanism. In response, we included an expert survey and an additional experiment, which were all done in the first round of the revision process. These additions were very beneficial, as they made it so clear what was not happening and therefore may not explain our findings.
What was the biggest challenge in the data collection and analysis process?
Xu: In terms of data collection, Amandine and I were trained in similar ways, and we work in very similar ways. I remember we started by collecting a smaller batch of data just to get a sense of what we could obtain from this particular online platform. I also had help from some RAs I always work with. We first collected a couple of years of data, and it seemed to be well structured and reliable in terms of accuracy. After seeing that, we decided to expand it to cover the platform from the very beginning. We were quite lucky that the platform had not yet been restructured at that point. Afterward, it underwent changes, but we had already captured a particular period that was very interesting to us and well suited for testing our idea theoretically.
The data collection and cleaning process did not take too much time. But we spent more time trying to understand the context through field visits and by talking to experts. That’s a very rewarding process, as it makes us feel confident about what we are discovering quantitatively.
Amandine: In terms of the timeline, we started the project more concretely after Xu and I met in Berlin back in 2019. So overall it was a long process.
The research has a multi-method approach, and seems to have taken a long time, which would be a very difficult process. I think it also depends heavily on collaboration. Although you touched upon this, could you share more about how your collaborations began, and how your different backgrounds shaped what each of you saw as theoretically important or empirically feasible?
Amandine: To start, one reason we worked on this project together is that we knew each other and liked each other’s work. We overlapped at LBS as PhD students, and I was hoping for an opportunity to work with Xu. But we needed to find a project that made sense for both of us. Then Xu approached me with the data and said, “Maybe we can start looking at the career trajectories of DJs.” And there is this fascinating community with amazingly complex and rich norms, which is what I’m excited about and what I like to study. My prior work was on the Champagne grape growing community, which also had very specific norms and practices, so it felt like the perfect project for us to work on together. So, it just felt like a natural intersection: Xu had really good data, which he was excited about, and I had an interest in this type of communities, so we complemented each other really well.
“My prior work was on the Champagne grape growing community, which also had very specific norms and practices, so it felt like the perfect project for us to work on together.”
Xu: Amandine is a few years ahead of me professionally, and when I started my first job, I felt she was stronger and more precise in theorizing than I was. I had more of an economics and engineering background, so I could be very efficient with data. It was a perfect combination from the beginning, and I do feel that I’ve learned a lot working with her.
We work together very efficiently, and we’re still working together. It’s a very pleasant collaboration. And I think both of us always have rather interesting and unique contexts in our studies. For example, she studied champagnes, and I studied Chinese medicine. A lot of these contexts are more historical, unique, and have very interesting norms to begin with. So I think that this particular context, EDM, speaks to both of our interests. On top of that, when you have something so interesting and unique, it does make you feel intrigued all the time working on it. You don’t feel bored.
I think one of those intriguing moments could be when you can see the different interpretations or opinions on the same thing. So, were there any moments when you disagreed on interpretation, and if so, how did you work through those differences?
Amandine: I don’t remember a lot of disagreements, because I think we were learning, discovering, and making sense of the context together. I remember at some point I was wondering a little bit more than Xu, probably because he had a better understanding of the setting, about the role of attention. I was wondering whether some of the effects that we were uncovering might not be about disinterestedness, but more about the extent to which remixes were able to grab the audience’s attention. We just looked at the empirical data. We ran the experiment, and the data basically helped resolve the disagreement. The data always came back clear that it was not about attention, or perceived quality. And that was very reassuring. So knowing that this is what the data say, we could focus on what we think this allows us to say, and what an alternative direction is to explain the results.
We also wanted to learn a bit more about the review process. You mentioned that the feedback on the first version of the manuscript led you to think about adding experiments. What other pieces of feedback made you rethink the framing or the overall structure of the paper?
Amandine: One aspect of the paper that definitely changed based on the reviewers’ feedback was how we were thinking about the dependent variable. One concern readers could have was the extent to which getting a gig was really related to support from other DJs. We originally considered using the full sample set of gigs an artist gets—and we still report those results in complementary analyses. But as we went through the review process, the team suggested that opening gigs were probably those where artists benefit most from the support of other DJs and the community, because these artists are just starting their career and probably still rely a lot on their network.
As a result, a pretty consequential change we made to realign the theory with the empirics was to shift from analyzing all gigs to focusing on opening gigs. There were several other changes, of course, but this was likely the most consequential one. It was driven by the review process, and I don’t think we would have made that change without that suggestion.
Xu: Yeah, I agree. Now that I recall the process, that was perhaps the fundamental change we made. But the thing is that we already had that analysis in the paper. It was just a change of sequence in presenting the results. We were aware that opening gigs could serve as a potential proxy that may better capture peer support. During our interviews, for example, we once saw a booking sheet from one of the DJs, and there was clear information indicating headliners’ influence on a gig’s lineup. Moments like this made us feel that opening gigs speak more to the mechanism.
I remember that even in the earlier version, we had already replicated the main regressions using opening gigs. But I think it was the review team who really wanted us to be very precise about the tests. If we’re going to talk about support and the elements that we now discuss in the paper, maybe we can reverse and further highlight the opening act as the main analysis, with all gigs as an additional analysis. I do think this makes the empirics cleaner in supporting the theory. I agree that this was the fundamental change, but the good news is that, for us, practically, it wasn’t such a big hassle to change, because we already had that to begin with.
I wanted to follow up on the overall timeline. You mentioned that you started collecting the data in 2018 and then moved from observational data and interviews into analysis. Given that the empirical pattern seemed quite clear early on, I’m curious whether the manuscript development and review process felt relatively “smooth”, or if there were points where things became more challenging.
Amandine: It was actually challenging, not so much in terms of each component, but in maintaining some discipline when putting all the different sources of data together. Structuring everything in a way that made sense and helped readers follow the argument turned out to be the biggest challenge.
It was definitely not smooth. It took a lot of trial and error. Although we had many useful pieces of information, they did need to be structured in a coherent way, which took a lot of time.
Xu: We submitted the paper in 2022. The review process took over 2.5 years, which I guess is not unusual for ASQ, given the bar is so high. So, I agree with Amandine. In the end, the challenge is how to tell a coherent story effectively. We had so much information and different methods, and the paper could go on and on—it could get very long—but we want to find a way to best convey the main message.
To wrap up, we’d love to ask you to reflect a bit more broadly on the project. Looking back, what did you enjoy most about working on this project? And if you were starting this project from the beginning, is there anything that you would do differently?
Amandine: I think the part that I enjoyed most was certainly talking to people from the community. It was inspiring to talk to individuals who were so dedicated to their craft, and so willing to give their time and energy, both to us as researchers and to the community, which shows up in the qualitative section of the paper. The extent to which they care, the complexity of the norms, how they think about music, and what it means to contribute to music, and how they show respect to other artists were all fascinating. I had done a lot of interviews with people who are passionate about their industry, like the people in Champagne. It was very nice to see that again, but in a completely different context. So, for me, that was the most motivating part.
As for what I would do differently, I’m not entirely sure. I think maybe we should have thought about the experiment and survey quicker, since obviously, we now see them as important for ruling out potential alternative explanations. But I also think that the sequencing played a role. It made sense to do things at a particular point in time. It helped us through our thinking.
Xu: I completely agree. I’ve certainly learned a lot from talking to people in the field. One of the reasons that convinced me to pursue this seriously was the passion these artists showed and their own enthusiasm for trying to understand what is happening with such a unique phenomenon. If you think about it, we’re talking about illegal behavior, but there’s really no reservation in sharing how they think about it. This made us wonder why there seemed to be a gap between how the formal institution, the government, thinks about certain behavior and how people within the community understand it, and even the varied understanding of people within the community itself, which is very inspiring to us.
Also, I didn’t expect how thoroughly these DJs were thinking about all these different elements, so it’s fascinating that they have thought about all these themselves already, sometimes even wider implications. When the paper finally came out, a few DJs reached out to us, and they were very pleased with the outcome.
Interviewer Bios:
Yeaji Kim is a PhD candidate in Organizational Behavior (Macro) at the University of Illinois Urbana-Champaign. Her research examines audience evaluation in markets, with particular attention to how social structure and cultural meanings shape how products and producers are interpreted. Using quantitative approaches, she studies contexts such as online review communities and creative industries, including tea, perfume, and craft chocolate.
Lingyun (Lydia) Zhang is a PhD student in Organization and Management at the Paul Merage School of Business, UC Irvine. Her research broadly focuses on culture, creativity, careers, and labor markets. Methodologically, she uses mixed methods, including natural language processing, experiments, and interviews.