Tuesday, February 24, 2015

Can Applied Social Science Solve Important Problems? We'll See.

There's a move afoot to dramatically increase spending in the United States on social science research.  Right now, the National Science Foundation allocates less than 5% of its $7.5 billion annual budget on social science. Those who advocate a massive increase want to see additional money added to bring social science spending up to the same levels as biological, engineering and geoscience spending. Members of Congress have indicated that any increase in spending must "be in the national interest." Presumably, they want social science spending to lead to immediate economic improvements. That would be a mistake. We need social science to help us solve pressing social problems, not to help us make more money.

There are all kinds of useful applied social science research projects that could be undertaken immediately.  Social scientists could show parents how to help their kids succeed in school, that might not increase the Gross Domestic Product any time soon, but it would surely help more citizens lead happier, healthier and fulfilling lives.  Social scientists could determine which forms of democratic engagement (at each level of government) increase citizen confidence that their voices are being heard. This would create a more trusting electorate and maybe a more responsive government, even if it doesn't boost the economy.  Social scientists could help determine which patterns of urban development are more likely to reduce the stresses of everyday life and allow people to live together with less friction and greater tolerance. This would allow all of us to enjoy more peaceful lives, even if more jobs aren't created in the short term.  As the push for a massive increase in social science spending gains momentum, we need to think hard about how we want this to unfold.

Right now, some advocates of increased social science spending are thinking in terms of funding a single new center that would oversee everything.  In my view, this would be a terrible blunder. First, this would require spending a huge amount of money on bricks and mortar,  creating the equivalent of yet another college campus, diverting funds from the work that needs to be done. We have more than enough university buildings all over the country. Let's use those to house a decentralized network of repurposed social science research centers. Right now, NSF spends about $270 million a year on social science research. They would have to multiple that by almost 20 times to achieve parity with spending on other topics!   What if there were $5 billion to spend every year on social science?  How could we make sure that the additional spending leads to improvements in our lives?

Imagine that a new Applied Social Science Program were organized very differently from traditional research efforts. Instead of funding individual university-based social scientists who submit research proposals, several million dollars might be allocated to a pre-approved array of community-based research centers in the 300 or so American cities with 100,000 or more residents. This would encourage partnerships among government, industry and local universities.  Funds might be allocated every five years (after the initial round) to those centers which can demonstrate that they have successfully answered the questions they set out to address. Public opinion surveys in each city would be used to gauge whether the research results were valued by residents.  This would require each center to figure out a way of communicating its findings to the public. (A failure to do so would mean no continued funding.)  The national coordinators of such a program could distribute a list of questions they'd like all the community-based centers to consider, but the final choice would be up to each center. The national coordinators could also take responsibility for ensuring that work completed across the country was synthesized and shared.  If the point of social science research it to help solve problems, then we need to make sure that problem-havers and problems-solvers are working together.

What I'm suggesting, of course, challenges in at least four ways the prevailing logic of how social science research is currently done.  First, we would stop relying on individual university-based social scientists to decide which problems or questions should be given priority. I don't see why or how a small set of social scientists ensconced in major research universities (or running a national research program) can know which problems are most important. Second, I would not leave it to social scientists to judge the worthiness of their colleagues' performance. Social science research needs to be useful to those who have to take action. Unless social science scholars can show that their research is being used to make better decisions, they should not count on further government subsidies. So, we need to make sure that experts are working with the communities and groups that need their help. Third, I would put the burden on social scientists to figure out how to make their research findings understandable to the public-at-large. A failure to do so would mean the end of their government support. Finally, I would suggest that the usefulness of social science research findings ought to be judged in particular contexts (not in general). Context is everything in the social sciences. Efforts to generalize (in the way that makes sense in the natural sciences) don't make sense in the social sciences. Any expansion of government funding of social science research should, therefore,  be decentralized (because problems are defined differently in different places), place-based (so that public officials who need to take action are involved in defining the problems that need attention in their area), and action-focused. Applied social science researchers should be accountable for the usefulness of the work they produce.

Many social scientists want to make social science more like the physical or natural sciences. That is, they want to "do science" in a way that produces generalizable and irrefutable results.  To that end,  economics is currently moving toward randomized control trials (RCTs) -- experiments in which proof of the sort we expect in medical trials can be achieved through social experiments that control for everything except the one variable we want to study. For instance, if you want to know whether a certain approach to getting people to eat more responsibly is working, you have to find two populations, similar in almost all respects, and give one group the information and resources you think they need to eat in a better way while withholding the same information and resources from the other (matched) group.  Putting aside the ethics of withholding something important from half the subjects in such experiments,  the goal of RCTs is to strip away the importance of context. The goal is to prove things that are universally true.  Unfortunately, social science doesn't really lend itself to this kind of manipulation.  What's true in one context, at one point in time, from one standpoint, is not necessarily true across the board.

In my view, we shouldn't be talking about making huge new investments in applied social science until we are clear about WHY we think such additional expenditures are needed, especially who ought to benefit from the new work and HOW success ought to be measured.  There are a great many people in a great many places struggling with problems that social science can help them address. Helping these individuals in ways that they find useful should be the goal of government-funded applied social research. Success should be measured primarily through the eyes of those who need the help, not through the eyes of the scholars doing the work.  If the same scholars can repurpose the work they do to contribute to peer-reviewed journals, that would be great. But scholarly success shouldn't be the primary goal of social science research.