How Can Social Science Become More Solutions-Oriented?

But what happens to a basic discipline without an applied counterpart? Social science aims to investigate how people and societies behave. But who uses its basic principles to understand and solve the many problems facing organizations and societies? And is this fundamentally a problem for the field?

Noshir Contractor, a professor of behavioral sciences at the McCormick School of Engineering, as well as a professor of management and organizations at the Kellogg School and communication studies in the School of Communication at Northwestern, recently sat down with Duncan Watts, principal researcher at Microsoft and an expert in how social influence spreads on networks. They discussed whether social scientists are doing enough to solve problems in the world around us—and what researchers and businesses can do to push the field forward.

This interview has been edited for length and clarity.

Noshir CONTRACTOR: I want to start by asking about an article you wrote last year, which argues that social science should be more practical and solution-oriented. You argued that some of the things that social scientists have held near and dear—like focusing on understanding and explaining phenomena in and of itself—is not going to get us far.

What prompted you to write this article? Why did this become an important issue for you?

Duncan WATTS: Well, the article reflects a frustration I’ve had for almost 20 years.

I come from outside of social science—physics and engineering—so to me the boundaries between economics and political science and psychology and sociology never really made a whole lot of sense. And so it seemed perfectly natural to me to read across all of these different disciplines. When I first started trying to understand social influence and contagion on networks, I started looking for articles that had those words in the titles. And sure enough, I found articles in economics journals, and I found articles in sociology journals and in psychology and political science journals.

And one thing that I found really perplexing and frustrating was that even though they purported to be about the same thing, and would often invoke the same examples, the content was unrecognizably different. It was partly stylistic, but even the mathematics would be impossible to reconcile!

As a scientist you’d like to be able to say, “Well, which mathematical model is better?” But I couldn’t even get to the point where I could express one model in terms of the other. And one makes an assumption that’s fundamentally incommensurable with the other one. So they could both be wrong, but they can’t both be right.

The point I make in the article is that social science has this very theory-oriented perspective on the world. And yet we have this mishmash of theories that don’t really add up. “Organizational behavior” is a perfect example of this. We have hundreds of theories of why organizations do what they do. Yet if you read that literature with the goal of making sense of it, it is really just headache-inducing.

Here’s an example. Take Microsoft, where I work now. A few years ago, Satya Nadella announced a major reorg. This is a multi-hundred billion dollar company. A hundred thousand full-time employees. Tens of thousands of people were moved around. Thousands of people lost their jobs. Thousands of other people got jobs. Everything about the company changed.

We have a hundred years of organization and management science. We have thousands and thousands of papers. You would think that somewhere in that vast volume of things with the word “science” at the end of them, there would be some instructions for Satya Nadella. How should he do this?

I don’t believe there’s an answer to that question in those thousands and thousands of papers. And if that’s not a question that we’re answering in organization science and management science, what are we answering?

CONTRACTOR: So rather than only being motivated by a certain theory, economists and sociologists should try to solve, not just understand, the same problem. And if that becomes the focus, then it follows that they will look anywhere they can for the relevant literature on that problem. That provides the incentive to navigate across disciplines.

But you make an important distinction between purely applied research and the kind you’d like to see more of.

WATTS: Right. What I’m talking about is where you use the applied problem as a way to generate new basic science.

For many years people would approach me after talks about how information spreads in social networks and ask, “How do I get my product to go viral?” And I would say, “That’s not the question we’re asking in sociology. We just kind of inquire about general mechanisms.” But after a while, I thought: maybe we should try to answer that question. That’s not a dumb question, that’s the question that anyone who’s not a social scientist would ask.

You’re trying to address the outsider and say, “Hey, social science is useful. We can actually tell you answers to your questions.” But in order to do that, we’re going to have to generate a lot of basic science ourselves. We don’t have an answer we can pull off the shelf.

CONTRACTOR: This might be a provocative question, but it’s one I’ve given some thought to. When basic science gets applied, we call it engineering. Is part of the issue that social science doesn’t have an equivalent?

Most of us would consider “social engineering” a four-letter word, but computational social science—which uses computational methods to investigate social phenomena—might be a platform for this. How can we use computational social science to address grand societal challenges? How do we accelerate innovation by assembling teams more effectively on the fly? How do we scale up global health solutions when we know the solutions exist, but have not been able to leverage networks well enough to propagate them?

In each of these cases, there’s a very vibrant research agenda. We don’t know enough about how teams are assembled. We don’t know enough about how things spread on the network. So there’s a lot of basic-science questions that are being addressed here. But in the process of addressing them, we’re also showing that we can make a difference. Even if we can’t give the best solution ever, it’s a better solution than the limited solutions we have today.

WATTS: It’s a really interesting point. One answer to the question of why social science isn’t more solution-oriented is that we’re missing this translational bit in the middle, as you say.

I think a second answer is more on the demand side. I wrote a whole book about how people think that social science is obvious, just common sense. Business leaders and politicians think they already know the answers.

Plus, they’re dealing with these incredibly complex problems, but they’re also in a rush. They don’t have time to wait for the research.

And research is expensive. Think of what it would cost to build a theory about how teams interact—a predictive “science of teams.” But if you want to do social science the way physicists do physics, you need your CERNS and LIGOs and Hubble telescopes.

CONTRACTOR: Yes, it would be very costly. But the fact that other areas are able to demand that kind of money says that we are not doing something right in selling our ideas. In other words, people literally don’t value what social science could do. That’s true in terms of funding agencies but also true in terms of businesses.

What do you want others to take away from this conversation?

WATTS: Well, a lot of the data that’s useful to computational social science is owned by industry, and yet a lot of the expertise to make sense of it is in academia. So we need to do more to facilitate industry–academic collaborations. Again, this is not a new idea in engineering. But it’s quite new to the social sciences. If we build collaborations around solving big problems, we can help businesses while also exploring fundamental scientific questions.

CONTRACTOR: The traditional social-scientist model is to go into a company and beg them to be nice enough to say, “Okay, we’ll share some data with you even though you’re not going to come back and help us.” And I think that model is broken. An industry–academic partnership has to be a win–win.