Anjali Bhatt – Stanford Graduate School of Business
Listen below to the June installment of the ASQ Blog Podcast Series:
Transcript of Podcast:
Anjali Bhatt (AB): My name is Anjali Bhatt. I’m a third year Ph.D. student at Stanford GSB, and I have here with me today Professor JP Ferguson, also at Stanford GSB, to discuss with him his new ASQ article with Sarah Soule and Thomas Dudley entitled “Osmotic Mobilization and Union Support during the Long Protest Wave, 1960–1995.” JP, would it be fair to say that we can summarize this article as about social movement spillovers that cross the boundary of the firm?
John-Paul Ferguson (JPF): That’s as good a summary as we’re going to get. Thank you Anjali. It’s good to be here.
AB: This paper brings together two pretty neat datasets. Maybe you can tell me a little bit about these datasets and the motivation behind this paper more generally.
JPF: Sure. The first of those datasets is the one that Sarah worked with, with Susan Olzak, Doug McAdam, and John McCarthy, for many years in the late ’90s and early aughts. That is the dynamics of… they’re going to kill me if I don’t get this right, but I’m pretty sure it’s the dynamics of collective action, is the name of that dataset.
This was an attempt to get micro-level data on every social protest activity they could find in the United States over a long historical period, which they did in what I think of as a very old school, but probably the only way you could have done it, which is to say they hired an army of undergraduates to read and code The New York Times for decades and put together a massive database of every disruptive social protest event that tended to happen during that period.
At one point Sarah said to me, “The one thing it doesn’t have”—and they had to make a decision about this—”is labor activity.” It doesn’t have strikes, for example. That was partly a decision that strikes were already gathered by other parts of the government and partly that back in the day, there was so much labor activity, both out in the world and being covered in the newspapers, that it would have swamped a lot of the other things that they were looking at. Which, when you tell that to people today, always confuses them, because the youth are callow, and they don’t know their history.
It was a real achievement within that field, because pretty much all of the work up to that point in social movements relied on some aggregate measures of social protest activity. It might be the number of organizations in a given year, total number of protests that people could record, and so on.
In many ways, this resembled a problem that had existed in labor movements research, which is people tended to use things like aggregate strike activity, aggregate level of organizing, or what have you, because of where they were getting the data from. I have a background, having worked with data from the National Labor Relations Board, on unions and organizing and collective bargaining and the like. And so had a long history working with the micro-level data on union activity, and the genesis of the paper is fairly straightforward. I saw Sarah presenting out of that dataset, shortly after I came to Stanford, and said, “You’ve got some pretty neat peanut butter. I have jelly. The thing that’s missing from your data is something that I’ve worked with in the past. Maybe we should at some point think about putting these together.”
Then three or four years passed, and honestly, when Thomas joined our Ph.D. program as a first-year student, he was attached to me as an RA, and I like getting Ph.D. students as RAs, because it forces me to come up with new projects to break them on the wheel, as it were. I said, “Thomas, you need to learn how to munge data. You’re going to work on matching these two datasets for a while.” That idea, which had been simmering on the back burner, really came forward once we had someone… I make this sound like I put him through a lot of misery. I did, but I do want to emphasize that I only have my students do things that I’ve at least done once before, so I can make sure that they’re doing it right and teach them how to do it along the way. So medium misery.
AB: So it’s clear that there are a lot of areas of overlap between thinking about labor movements and social movements, but did you have the research question in mind when you were mulling it over, or did that come up more organically?
JPF: One of the things that I have always been fascinated by, as a researcher, is how available data tends to shape the questions that we ask. What are often explicitly understood as compromises early on, based on what the available data is, the next generation that picks up the ball and runs with it doesn’t realize why those compromises were being made, so often we end up ossifying our theories and our beliefs about the world because of empirical facts that we no longer think to examine.
In other words, for me, this may sound like I’m an empiricist or methodologist, but it’s not really. It’s the way that the available empirics constrain our theorizing over time, and often there are things that we just think we know about the world that turn out to be wrong ’cause we haven’t gone back and reexamined them, and people often forget things that, empirical facts that were the meat that drove the original theory.
I’ll go briefly off the reservation for a second: population ecology is like this. It’s often considered a piece of high organizational theory, but it has built within it, for example, the idea that organizations are inert, and one of the things that causes inertia in organizations is the fact that routines allow organizations to be more efficient, but they require the tacit coordination of large numbers of people and so are hard to change.
Well, there’s an implication there that anything—any change, technologically, in terms of regulation or otherwise, within organizations—that makes coordination more flexible should reduce the level of inertia that organizations face. But the connection between the meso-level crap that’s happening inside—can we say crap on this podcast—crap that’s happening inside organizations and the theoretical constructs we work with often gets forgotten over time. It’s the same thing with things like institutional theory, that institutional theory, when we talk about decoupling, there’s an assumption that there’s a technical core of tasks and a bunch of other external-facing tasks that organizations do, and that we can decouple those.
Those kind of implications can get forgotten. Well, when it comes to labor organizing in particular, people tend to look at things like aggregate density, and they look at effects of things like social protests on aggregate union density. In fact, all the papers… Larry Isaac has a couple that we riff off of in designing this, in ASR and AJS, 2002 and 2006, respectively. He was looking at aggregate levels of social movements on aggregate levels of union density. The thing that we disliked about that kind of reasoning is what we really care about is a Coleman’s Boat issue. As we go from aggregates to aggregates, we really think what’s happening here is social movements cause mobilization or politicization of the population, they therefore engage in some form of costly collective action. That’s going to show up in those aggregate union density measures.
But in practice, a lot of things drive an aggregate like union density. People getting motivated by protests are going out and organizing unions, yeah that’s going to raise density. But like, factory closings and relocation of plants, non-union parts of the United States, and the disappearance of industries because of foreign competition, that’s all going to lower union density. That all goes into that outcome variable that people are using.
So imagine you try to look at the relationship between these aggregates and there’s no relationship or a negative relationship, and you conclude from that that, “Oh, the relationship between protest and mobilization for unions is non-existent or negative.” No, you just have an incredibly noisy aggregate measure that you’re trying to use.
The question that had been lurking in the back of my mind was, theory says there should be a positive spillover between different types of mobilization. I mean, that’s kind of what social movements as a research field is about, and maybe we aren’t finding it here because of the restrictive nature of the kind of data we’ve had to measure outcomes. If we could measure at an outcome that is more theoretically prior and closer to what we think is happening, maybe we see effects.
That’s the main finding in the paper, is there is indeed a positive relationship. Of course, it gets swamped by all these other changes, but the point of the paper was not to measure the exact level of union mobilization over these years. It’s to say, can we find any evidence of a relationship that theory would predict here?
AB: The other thing that you mention in the paper is that there are a lot of mechanisms that could actually link prior mobilization to future intrafirm mobilization. Do you see any of those as more important to focus on? I know you didn’t test those, but…
JPF: So rank speculation for a moment. I think that what are often referred to as the indirect channels of spillover between social movements are probably the more important ones. The direct measures are physical and concrete in various ways. This is the idea that you might have social movements sharing an office space, members who overlap between organizations. I could go on but you get the idea.
I think the indirect stuff tends to be the observation that a certain type of action is possible and acceptable in the world. I’m not myself taking part in a march, but perhaps I notice that people are able to take part in marches and it seems to have an effect in the world.
I don’t think this is independent of time. I think that today, those indirect effects are probably some of the most important ones. The demonstration effect that people manage to… Because my sense today is a lot of, partly because of the decline of things like the labor movement, people assume that they’re fairly powerless within organizations in which they have to spend a lot of their time.
I think that the demonstration effect of collective action is an important thing. Collective action that involves some way to protect yourself against retaliation is both feasible and desirable. Those kind of channels I think are probably the most important ones to think about going forward.
AB: Yeah. I mean, I think a lot of the audience members for this podcast, Ph.D. students and otherwise, are probably interested in how these kinds of mechanisms apply to today’s world, with maybe new or semi-new social movements like Black Lives Matter and #MeToo which have roles both inside and outside the organizational settings. How do you think these apply? What do you think are some implications?
JPF: If you talk to anyone like me who has spent a lifetime… a lifetime, I’m 40 for God’s sake. Let me back that up. If you talk to anyone like me who’s spent more than a decade studying the labor movement, in 2018 in the United States, we’re treated like a recondite variety of historian. People don’t know anything about organized labor at this point or the labor movement generally in American history.
It’s a frustrating time to be alive, if you’re someone like me. Because a lot of social protest is both really important and frustrating insofar as it feels like people are rediscovering the wheel. People think about social protests, but they don’t think about how we would build structures that allow that protest to be institutionalized, how we could have countervailing power that would limit powerful people’s ability to retaliate against us and so on.
So much of that is like, “Yeah, that’s why we evolved to this thing called the labor movement.” A lot of problems that people are discovering anew… but I have to be careful when I say this for two reasons. When I say this, I mean “The labor movement already did that” or “That’s what the labor movement was for, you thick dolt.” You have to be careful about saying these things for two reasons. The first is that simple rule, when people suddenly become interested in something that you think is really important and have thought is important for a long time, you are supposed to greet their interest with joy and acceptance, not shout at them and say that you already knew that. There’s no better way to drive people away than to do that.
The second reason is you have to separate out the lessons we can learn from the organized labor movement from this specific historical instantiation of those in the American labor movement. People like me think that the labor movement or individual labor unions made plenty of mistakes historically, but there’s still ideas we can learn from that organizational form that are incredibly important.
The idea that a lot of types of change in organizations is very hard to do unless people feel like they are protected from retaliation, and that’s how you solve a lot of those problems of people taking risk, comes from the labor movement. Whether that comes exactly through something that looks like a labor union in the United States, as recognized under the National Labor Relations Act et cetera, is a separate question.
I mean, the striking examples are things like, think about discrimination in the workplace and diversity policies that firms have. We talk about some of this in the paper. Many of those were explicitly modeled on collective bargaining agreements that labor unions had developed over time. The idea that you have a grievance procedure and that you can appeal that, and that for example if you’re going to have arbitration, you want the arbitrator to be neutral—which among other things means can’t be paid by one party to avoid conflict of interest—this is all stuff that was developed through years of difficult negotiation between employers and labor unions.
People who are developing anti-discrimination and diversity policy learned from that and took those ideas away. Now, does that mean you exactly have to have a labor union for those things to work? No. But you always run into this problem: how do you put those things into place without an active actor, right? Without an organizational actor that pushes for them. So the need to have independent organizations that can lobby, that have a basis of power outside the firm as well as inside, is always incredibly important.
In terms of modern movements that are dealing with these issues, I think someone like me who’s spent a lot of time looking at labor unions would say, #MeToo is a hashtag that exists inside a lot of organizations without a larger grouping or organization in its own right beyond individual firms to lobby for a set of interests. Its lifespan is probably limited, or its effectiveness is to some extent limited. Because then you’re going to get progress in the organizations that are already sympathetic toward your goals.
This is just a general problem for organizational research. So many of the things that we like to study, they might be put into place by the firms that want to put them into place. The firms that cooperate with us and share their data are the firms that are probably more sympathetic to the policies that we’re trying to study.
As soon as you start to think about “How can we have diffusion of a practice that we think is important without simple self-selection by employers?”, you have to think about having independent organizations that can help push those things forward. Without that, you’re always going to have heterogeneous diffusion.
AB: In the paper you talk about how the two forms of organizations or organizational infrastructures that can enable that kind of change are labor unions, but also laws and legal systems, and that those laws and legal systems can sometimes take the place of the need for labor unions, potentially vice versa. What are the differences between those two forms that you think are important to consider today?
JPF: One of the classic arguments for why things like labor unions, or generally independent sources of worker power that could hold employers accountable inside firms, is a thing to be valued, is efficiency. Ironically, when you start saying unions and efficiency, people just blink at you, ’cause everyone today assumes that unions are an incredibly inefficient organizational actor.
Let me point out first as an organization theorist, you should take an open systems perspective on this. If you have an environment, as we do in the United States, that’s unbelievably hostile to unions and collective bargaining, think about what type of union is going to survive in that context. Eventually the only members of the population left are going to be the most ruthless and focused on self-preservation at all costs. Anyone more idealistic is going to get smashed. It’s not surprising to me that we have both a very anti-union climate in the United States overall and a lot of really ruthless and self-protective organizations. It’s kind of like in World War II when they had bombers coming back that were damaged, and they said, “We need to put armor on the spots where we don’t see damage” on the returning bombers, because the planes that came back are the ones that got hit in spots, but they made it back. The same thing holds: if you want more idealistic forms of work organization, make it easier for them to organize, full stop.
So when I say efficiency, people often just blink at me, ’cause today the labor union’s considered an extremely inefficient type of organization, but there’s a famous set of Supreme Court cases called the “steelworkers trilogy” from the 1970s. It’s essentially the series of Supreme Court cases where they started to say, through collective bargaining and arbitration, these private actors, labor unions and employers, have developed a type of case law. They’ve essentially, and the common law to some extent should defer to this stuff and should treat it as precedent when we’re trying to resolve conflicts in organizations.
I bring this up because the argument was, okay, let’s imagine we say that we’re going to put something in place like the Fair Labor Standards Act—maximum hours, minimum wages, et cetera. Then we want to enforce that. One version is we have a government agency that enforces this. They have to go out and you file claims and they come and they investigative, and depending on what they find, they decide whether they want to potentially press legal charges. This maybe has to go through the courts. Eventually you might choose to settle out of court, or a settlement might be decided, and the judge has to say what has to change. We all know how this system works. This is, by the way, mostly what we do with things like discrimination today.
Everyone agrees it’s pretty inefficient. The first thing any government regulator will tell you is how thinly spread they are relative to the scale of the problem. Anyone who’s tried to file a discrimination claim will talk to you about how difficult and drawn out that process is. One of the arguments with things like grievance procedures, arbitration, or anything else was, “Boy, it’d be a lot more efficient if we could just resolve this at the point where the problem occurs. We’d rather have enforcement directly within the workplace where we don’t have to rise to the level of bringing in outside regulators and so on.”
That feels like it would be faster, more flexible, and more efficient. Now, that can only happen if the people who are bringing the grievances forward feel like they have an actor inside the organization who will credibly represent their interests. If the government were to defer to arbitration where the arbitrator is entirely compensated by, kept in employment by the employer, obviously workers aren’t going to take that seriously, which is why we have a lot of push back today against things like mandatory employment arbitration.
But I think that this is one of the ways that we think it’s incredibly important to have independent organizations and not just rely on law and regulation. The distinction I would draw is there are things that unions negotiated through individual employers in the United States, historically—things like health insurance for their members. Health insurance, first off, it’s not like you have to arbitrate it and adjudicate it at every individual workplace. It’s a general social product that there’s a market failure where it doesn’t tend to be provided well otherwise. Most other countries solve this by legislation, right? They’ve got some form of universal healthcare, we’re done. “That’s how we’re going to take care of this,” and there it’s less efficient to try and bargain it with each individual employer and set it up in each workplace, never mind it’s also very incomplete. It leads to a lot of rigidity in the labor market, because people don’t want to change jobs, because they’ll lose their health insurance.
That’s the kind of things that unions in the United States did, that someone like me says, “Well, now that in a sense unions are kind of gone in many ways, and if we want to bring something like that back, don’t focus on problems like that one where the easiest solution”… ‘Cause notice, that’s also really efficient to enforce at a fairly centralized federal level. It’s things like your autonomy at the workplace, your freedom from retaliation, your specific negotiation for the conditions of working benefits—that’s a lot more efficient to negotiate at the workplace. Those are important issues. I would rather solve those through the empowering of independent organizations to negotiate them than to rely on regulation, to both enforce the general idea and all the different applications.
AB: It’s a fascinating discussion that we could go on forever about. But returning to the paper and the project specifically, I’m curious about how you thought about framing this paper and what other framings you considered.
JPF: I’ll nip at the hand that feeds me, as I talk about this. Sorry ASQ, you’re a good journal. I think the main contribution of this paper is to say people tended to… I think the theoretical question is the first and most important one: we think that there should be a positive relationship between these large social movements and this large organizational residue of a previous social movement upsurge, i.e., the labor movement. And the fact that people haven’t found one is kind of weird, and I think that’s because they were using the wrong type of data, and we need to look at that, and lo and behold there is. And there’s I think an important contribution there in noting we can improve upon past findings by saying if we take the research question very seriously and we focus on are we measuring this correctly, maybe we get very different results.
That kind of stuff is where I think a lot of advances can lie. There’s a tendency—and it’s not just an ASQ problem, this exists with journals in our field—that people want to know what the novel or broad theoretical contribution is. The reason why this makes me grumpy, and you’ll hear this in the podcast, is because then we get the question like, “How does this generalize, or what’s the broader implications of this question?”
Part of me, of course, wants to say, “Really we’re talking about the largest wave of social mobilization in American history, and the largest organizational product of working class mobilization in American history, and how they related to one another over a couple of decades, and seriously, you want to ask me how does this generalize to other settings?” It’s like, I’ve read this journal. I know you’ve got single-firm studies that look at something in two food co-ops. Give me a break.” That’s the nipping.
Now, there’s a moment when it’s like, “Well, how does this generalize to other settings?” Jesus. Ultimately, if you push me for how do we take a giant important social objects and say, “How do they generalize?” I’m going to say well, the main thing here that’s different from other social movements work that I have seen is we ended up looking at the, as opposed to… People have often tended to look at the activity of people outside of organizations and then outcomes inside organizations. We said, “Well, the difference here is we’re looking at activity outside organizations and the founding of new movements inside organizations.” Fine. I’m happy to talk about that to some extent.
In practice, of course, there are cases beyond protest movements that really took off in the 1960s and organized labor. More than a decade ago, Michael Piore and Sean Stafford had a great piece in Industrial Relations, when they talked about shifting axes of mobilization away from class- or identity-based issues, and were pointing out the rise of identity groups within firms. You might have a collection of Indian engineers, female accountants, gay supervisors. As the case might be, in many organizations we start to see such interest groups forming.
Obviously that bears some relationship to public mobilization around those identity issues. It’s important to think about it, and also the thing that we really liked about looking at labor organizing is you could look at heterogeneity in that diffusion, because what we so often see is we have a social movement and then an issue that people mobilize around later on: we want to green the WTO, we want to focus on the adoption of same-sex marriage provisions. But it’s like we only have one social movement and then maybe a couple of firms. We don’t have multiple social movements and multiple possible targets inside those firms.
The great thing about looking at the protest movements of the 1960s and then looking at their effect on the foundation of labor unions is there’s actually tremendous heterogeneity among labor unions. If you wanted to look at spillover, you could focus not just on the characteristics of the firms but also the organizations that might be formed. That we thought was a really interesting… Indeed, that’s one of the findings of the paper is—I mean, it shouldn’t shock us, but it’s nonetheless relieving to see that it conforms with what we might have predicted—that unions that had in the past taken fairly costly, but we think the right, stances on issues like civil rights or opposition to the Vietnam War are the ones that tended to benefit more from social mobilization around those issues.
AB: Let’s talk a little bit about the research process itself. With this paper there are a senior faculty member, a junior faculty member, and a Ph.D. student involved. I’m curious about what the collaboration was like, what you learned from it, and how it might affect your thinking about collaborative projects going forward?
JPF: Sure. This is the only paper I’ve ever written with a senior colleague. Read into that what you will. I think it mostly means I was lazy in grad school and didn’t write more papers with my advisors, whereas I’ve collaborated several times with graduate students on papers. And the collaboration between Sarah and me really had its start, like I said, it was partly that, “Okay, we each have a dataset that would be useful.” But Sarah and her collaborators on that project were very explicit with folks like, “We want this to be a dataset that people can download and use for other projects.” It’s not like her participation was a quid pro quo or anything like that. Rather, I said, “Sarah, I know the literature on union organizing well. I know a lot of the history of social movements in the 1960s and ’70s very well. But compared to you, I’m no expert on the history of social movement research and the way that social movement scholars work on this.”
The wonderful part about Sarah is I could say, “Sarah, we need six paragraphs here… if you could organize the various mechanisms by which social movements can spill over and affect one another.” It would take me a couple of months to read up on the literature, understand how that fits together. And she’d say, “Yeah, I’m flying somewhere tomorrow. I’ve got 45 minutes on the plane, I’ll write it and I’ll send it to you.” There it would be, right? Then I would edit that and we’d put it in. I’d show it back to her, “Did I get this right?” And so on.
I think for a lot of the first round of the paper, the fact that we could contribute our separate bodies of knowledge in a sense to the lit review was incredibly important. Because that gave us a way… Those were the parts where our contributions were additive, and it was very easy to do that.
On the research method, I insist to this day that I’m a fairly bad econometrician. I try to focus on research designs that don’t require high powered statistical knowledge, because I lack it. Whereas Thomas, in his spare time, earned a master’s in statistics at Stanford. I spent a lot of time, he and I would go back and forth about what type of model should we do? Do we think this is the right way to specify this? Whereas Thomas was at the time getting up to speed in organizational theory, generally, and the history of social movements and particularly organized labor isn’t something that’s taught to kids that were born in the late 1980s, in the United States, at least as far as I can tell.
He could contribute a lot of methodology that I was rusty on and Sarah, being older than me, was rustier still, right? I don’t want to simply imply that age is correlated with statistical expertise, but in our small case it was. The other great advantage of having someone like Thomas who wasn’t a subject expert, either in labor movements or in social movements, was then you have a third set of eyes reading the paper, who can say, “This doesn’t follow” or, “You’re making a bunch of assumptions about how I think the world works that aren’t actually correct.” Or, “You’re referring to things that I have no idea what they are, and I have to duck over to Wikipedia to make sure I understand what you’re talking about,” which neither Sarah nor I was going to notice in some of those cases that we had done that.
It sounds so patronizing to say he lent a really useful element of naïveté to the process, but there’s truth. In fact, this is what we do when we write papers anyway, is we write our paper, and then the reason you workshop your paper is your smart colleagues nonetheless have no idea what you’re talking about, and that’s when you realize, “Oh, I’ve taken a bunch of things for granted that I thought my audience was going to know and they don’t,” right?
There are times when having someone like that as a co-author just speeds that process up in a really important way. Frankly, professors are bad at explaining to graduate students that they can be useful for this. Graduate students are really scared to do it, because a graduate student never wants to tell a professor, “I don’t understand what you’re talking about in this paragraph.” ‘Cause the answer they’re afraid the professor’s going to give is, “That’s because you’re dumb and a failure and should leave the program.”
In practice, what it probably means is that’s because I wrote it fast and didn’t think about writing clearer sentences. I’m not going to lie, there are professors who are insecure and will cover up the fact that they wrote something poorly and unclearly by attempting to pass the blame onto the student, but they’re wrong and they should learn from the fact that… One of the best pieces of advice I got in graduate school was confusion is a data point. Whether or not you end up being able to explain it to the person, you have to remember the fact that they were confused is something you should notice, and frankly, if more than one person is confused, the odds are extremely good it’s ’cause you wrote something confusing, not because two dullards happened to read your work.
Yeah, I find that incredibly useful. The same point holds, Sarah would write things about social movements that I didn’t recognize and needed her to explain in more detail. I would write things about labor movements that vice versa.
AB: My last question for you is how has this specific project influenced your research agenda and what questions would you like to study next?
JPF: Well first I’ll mention a specific one related to this project, and then the general. More broadly in my research, I do a lot on employment segregation, and I’ve done many papers using data from the Equal Employment Opportunity Commission, for example, and one of the findings in this paper is that progressive unions did better, particularly in the wake of protest related to civil rights. It follows that if we knew… and there’s an implied chain of mechanisms there, that in particular, African American workers during the 1960s and 1970s recognize both the effectiveness, feasibility, and desirability of collective action to improve their situation in the world, politically and economically. We know that historically black workers have been more likely to join unions, and that only accelerated particularly in the public sector in the wake of the civil rights movement. Jake Rosenfeld and other people have done excellent work on many of these points. It follows that if we knew the employment composition of the places where the union organizing drives that we study in the paper, effects should be stronger in workplaces where the workplace was less white.
Now, to figure that out requires matching some, the historical data from the Equal Employment Opportunity Commission with the organizing data from the National Labor… So basically tying a third gigantic dataset together to look at this. But my strong suspicion is the effects we found in the paper should be stronger in more, I’m going to say generally more diverse, but particularly less white workplaces. I say that because white Americans vote for unions at lower rates than non-white Americans and this is a pretty stable fact historically. Don’t ask me why, it breaks my heart.
That’s a specific follow-up with this paper that I would like to see happen. I think a general point about the research process, though, is this was one of two papers I was working on around this time where people kept asking me as I was putting the dataset together like, “What are you hoping to find?” I.e. is there a positive result? Is there a negative result? What have you.
I said, “I expect to find a positive result, but I think this is the right way to be measuring this, and I’d be just as interested if I found a negative result.” Insofar as it’s not a perfect research design, I’d be a little more frustrated with a null result in this case, because it’s first off, it’s just wrong to say you can’t learn from a null result. You can’t learn well from a null result when you have observational data. The distinction should be clear. You can learn all sorts of things from null results when you have a good experiment; that’s what an experimental process is all about.
But what was striking there was I realized, “Okay, one way or the other I’m interested in this.” I’ve often said to Ph.D. students here that at some point as an assistant professor, I decided I just never again wanted to work on a research project where if I didn’t get a certain result I was going to feel like the project was a waste. Because I feel like it’s not just that that’s… The conspiratorial version of this is that’s what leads people toward scientific fraud. I don’t necessarily think that’s the case, but I do think that’s what leads us toward hypothesizing after the results are known, that’s what leads us to endless reframing. It’s what leads us to mining the data and building a story around a collection of opportunistic, significant conclusions.
All of this partly stems from that idea, and no, we should be designing research questions where at the outset, we agree that the results, whatever they are, are interesting. We should not be starting research projects implicitly with the idea that we have to find a certain result or it’s over. Because that’s not actually how science works, and with this paper, I was pleased in a sense that I got the positive result, but a negative one would have been just as interesting. I think that’s the direction you have to go with research.
AB: JP, thanks for your time. Is there anything that I should have asked you that I didn’t?
JPF: Gosh, there’s so many ways to go with that question. How awesome I am, when I expect to be handed my Nobel for this work, those kind of things. But I guess we can just defer those for another chat.
AB: JP, thanks so much for your time.
JPF: Thank you.
AB: Hope everyone gets a chance to read their new piece in ASQ.