Rethinking the Science Question in Social Science and Social Work:
Practice Wisdom, Local Knowledge, Metis and Phronesis
Prepared by Anne Dalke
Additions, revisions, extensions are encouraged in the Forum
Participants: Sharon Burgmayer, Laura Cyckowski, Anne Dalke, Wil Franklin, Paul Grobstein, Michael Pfeiffer, Megan Rowley, Sandy Schram.
Sandy began by explaining his position that, "by offering "universal, decontexualized, top-down causal knowledge of the local world," the social sciences are "making the mistake of trying to emulate the methodology of natural science." Social scientists have a "different subject matter--human beings"--and are "better positioned to offer bottom-up analysis, aka "informed local knowledge" or practice wisdom." There is actually now a movement in political science, called "perestroika," which advocates these views on panels, at conferences, and in books such as Perestroika!: The Raucous Rebellion in Political Science and Making Political Science Matter: The Flyvbjerg Debate and Beyond (the latter was edited by Sandy with Brian Caterino).
One of the confusions that emerges when perestroikans present their work is the challenge that they are "a bunch of philosophers/theorists/interpretists opposed to quantitative, statistical, empirical research, who want to replace it with in-depth qualitative research that highlights subjective experience of social relations." Nope. Perestroikans are interested in problem-driven research in social science: i.e., they let real-world problems frame the inquiry, and then choose the best methods of data-collection for solving those problems. They don't let the theory or the method drive the research, but work in the other direction. There's lots of emphasis, in the work of the perestroikans, on methodological pluralism: they engage in hybrid methods, drawing on different kinds of work to inform local, contextualized, situated problem-solving.
Their real assault is on the positivistic, causal modeling that has dominated political science for the past few decades. They are particularly critical of rational-choice theory imported from economics, the sort of technical, abstract, mathematical, reductive models used to perform a cost-benefit calculus of maximized utility, demonstrated (for example) by the overly decontextualized and "rationalistic" work of John Nash on social decision-making and Thomas Schelling on deterence. "Life is NOT a game and we are NOT rational actors in it." Social science should not overlook the experience of social relations in context, occluding the subjective, emotional dimensions in human interaction in attempts to create predictive models.
The alternative form of political science Sandy practices is very pluralistic; it involves mixing methods and using quantitative information. The work he is doing now (and will present at a Symposium on In/Dependence: Disability, Welfare and Age @ UWisconsin in April) is on Uncaring Neoliberal Paternalism: A Compassionate Response to the Punitive Turn in Poverty Management. This is problem-driven research, using Florida as a case study of how welfare reform is going forward. It mixes quantitative and qualitative methods to demonstrate the consequences for recipients in the current radical transformation of welfare.
Sandy argues that we are now moving from a Keynesian to a neoliberal paternal state. John Maynard Keynes was a social democrat who greatly influenced the formation of the welfare state after World War II (As Nixon said, "We're all Keynsians now"). Keynes encouraged government to take a role in monitoring and subsidizing income, and in instituting wage and price controls. The Keynesian state was designed to be counter-cyclical: it pumped money into the economy and subsidized standards of living as needed.
In the current era of globalization, we have a tendency to say we have free trade, with open borders enabling the free flow of commodities and people. That may indeed be what we have, at the top. But at the bottom we have a punitive disciplinary state for those who don't meet the requirements for participation. The government has become very punitive in its treatment of people who can't fit into the new order. People @ the top can earn positive credentials (like a Ph.D.). Those on the bottom get negative degrees (such as felony records, bad credit ratings, welfare sanctions, the reduction or lose of access to benefits, etc.)
Quantitative information helps us study this big change. Between 1990-2001, for instance, when cash assistance to dependent families was reduced by 50%, there was a reverse proportional increase in prison population. "We became more punitive" (="more interested in locking them up"). There are lots of different theories to explain such correlations; in Regulating the Poor (1971), for instance, Piven and Clower argued that social welfare policy in the US tends to be cyclical. In times of instability, welfare grows, offering more benefits to the poor, in order to "make them less rebellious." During calmer times, welfare becomes more restrictive and disciplinary. Sandy himself published an essay in American Sociological Review that shows statistical correlations between the greatest amount of welfare growth and the highest levels of rioting, deaths, property damage and largest numbers of troops deployed domestically.
Texas and Florida lead the country in prison construction, in imprisonment, and in proportion of state expenditures spent on prisons. But the states which have the most restrictive criminal justice and sentencing policies (such as the "three strikes you're out" laws) also have the highest crime rates. The states which are most innovative in making welfare more punitive turn the work over to local boards, who administer it in business fashion. There are parallel activities across all social welfare domains: using sanctions to prod people to return to work, for instance, or removing children more quickly from their traditional biological families. Privitizing, which results in more punitive actions towards the poor--using penalties to discipline, rather than to support them--constitutes the bottom part of neo-liberal paternalistic state.
Sandy then reported on the data he gathered in his study of welfare reform in Florida from January 2000-December 2004. He performed a statistical analysis, using a "Hazard model," which calculated survival rates by analyzing the data month by month. The major finding was that devolution--turning the administration of welfare over to local counties--matters. Welfare recipients are much more likely to be sanctioned in conservative counties. (Conservativism is measured by voting patterns on state propositions.) Welfare is so punitive, and sanctions so likely, that the average welfare recipient survives (=remains on welfare) only 8 months. The odds ratio of black clients being sanctioned increases with time spent on welfare. Being black is a real disadvantage in conservative political environments; devolution has racial consequences in conservative areas, where it has the effect of increasing the disparity between black and whites.
More general conversation began when Sandy was asked how he got such smooth (consistent) data. Surely, "the human element is never that smooth." How sensitive--from the question of statistical analysis--are the strong underlying rules? Wasn't he "using the model he didn't like to use"? In trying to simplify and condense the data, to make his point, wasn't the "way he presents it still colored by the old paradigm"? He was challenged on his dismissal of "decontextualized pursuit of predictability," when his own work seemed so decontextualized...in pursuit of what, if not predictability?
He's "not against" statistics, probabilities, making predictions, or generally simplifying activities that make the data seem more systematic than it is. He is doing "hybrid research," actually using a variety of methods to try and show some systematic patterns in Florida. His charts don't represent a real person, but a hypothetical. His work draws on probablility theory, not in order to generalize, but rather in order to learn what we can about how adequate the models are for describing real data. These are snapshots, based on the real data.
One of the questions he explores, for instance, is to what extent case managers exercise (what Herbert Simon, the Nobel laureate in Economics, called) "bounded rationality." Since there are limits in how much information they can handle, and how much time they have to handle it, case managers are forced to engage in racialized, bounded rationality. At an unconscious level, they engage in racial profiling, and are encouraged to sanction non-whites. Their assumptions don't take into account that some are in better position to be a "good client." This is institutionalized racism.
It is also striking to see how the probability of sanctions shadow the tourism rate; is this a causal law? A possible explanation is that, during tourism season, case managers and welfare providers are under the impression that more low-wage jobs are available, so they give their clients less leeway. Research has actually uncovered seasonal sanctioning of welfare recipients in many states, often tied to migrant field work. (In Washington state, this was known as the "spring slam": during planting season, recipients were "kicked off the welfare rolls and forced into the fields").
Sandy's research also examined transition assistance (two years of child care coverage, for instance, if recepients were willing to undergo extra monitoring, such as submitting their pay stubs). Sanctioned people have lower incomes; after they are off welfare, they fall even more behind. We tend to sanction the worse off, and make them even worse off. In new system of neo-liberal paternalism everything matters: where (whether the area is conservative), who (whether the recipients are black or Hispanic), when (what season), and to what effect (whether recipients are worse off). The world of welfare is a punitive system that punishes people for being poor.
Members of the audience weren't convinced that what Sandy was doing was "very different from what scientists are doing"; @ an earlier presentation of the paper, sociologists had observed, similarly, that he was "just doing sociology." There are, however, a couple of differences. The work is more problem-driven and interpretive than conventional political science. The statistical models are very much influenced by extensive field work. Doesn't that make it "more like science," then? With projections that should be more valid? "But science does not get to monopolize statistics and probability estimates."
Science is an attempt to create a cumulative body of laws about how world works. Sandy's work involves no effort to make general statements about human behavior. Rather, it uses techniques to describe what is going on now. Further, it makes claims about how welfare systems should operate. The intent, finally, is to have political impact. The goal is not just to present what is; the work is self-consciously designed to be proactive, to provide guidelines for a political agenda. Political science is framed as a form of problem-solving: racializing poverty in the U.S. This study is not divorced from the way policy is implemented. It is not purely academic, scholarly research, but applied, political work.
The discussion of "rethinking science education" continues on the on-line forum, and will resume in person on March 24 when Peter Briggs and Al Albano will talk about "Promoting Cross-Disciplinary Discussion among Freshmen."
Return to Brown Bag Series on Rethinking Science Education