Date: Tue, 6 Feb 1996 04:16:20 -0800 To: halebsky@ssc.wisc.edu, "socgrad@ucsd.edu"@ssc.wisc.edu From: lichter@nicco.sscnet.ucla.edu (Michael I. Lichter) Subject: Re: what has quant. meth. accomplished? At 12:48 AM 2/6/96, halebsky@ssc.wisc.edu wrote: >Let me rephrase part of my >question: What successful social policy could not have been >successful without the use of high level statistical methods ... ? 1. How do you define "success" and how do you define "social problem"? These are not objectively given. Who defines the terms and sets the agenda? I would argue that social problems are more likely to be solved through redefinition than through action. 2. What is a "high level statistical method"? Crosstabs? Logistic Regression? Time Series Analysis? Why does it matter? 3. You are assuming a process here where a researcher chooses a method, this method determines the outcome of a line of research, the research determines a course of public action, and the public action solves the problem. But the method does not determine the analysis, the results of the analysis do not give a uniquely determined course of action, and the courses of action that are possible and likely to be successful are bounded by a complex configuration of values, politics, resources, etc. Don't be coy, here, Habebsky. Why you're really asking is "since quantitative methods have been useless in solving social problems, why don't we toss them and stop forcing poor, innocent grad students to go through those horrible stat classes." But why aren't you asking "why should we bother studying sociology if it hasn't solved any of our collective problems?" First, sociology is more oriented towards producing understanding than it is towards producing solutions. There is a value in understanding social phenomena, e.g. prejudice, even when we're unlikely to be able to "solve" them. Second, the suggestion of solutions does not in itself solve problems. Somebody has to take your solutions seriously and implement them. As Harvey Molotch has argued, this country is singularly hostile to sociological ideas. Sociologists have very little influence over public policy. Marx argued that a social *science* has to test its propositions by putting them into practice. "Fixing" social problems is not something we've had a lot of practice at, and so it is not surprising that "how to" is not well understood. At 12:48 AM 2/6/96, halebsky@ssc.wisc.edu wrote: >Assuming that a major social problem has been >solved, my question is this: was Brown based on high-level >quantitative methods, or was it based essentially on quantitative >*data*? Did it require a lot manipulation to come to the >conclusion that separate was not equal, or was it a matter of >getting the data (which I agree is critical) and then laying out >various descriptive statistics, comparisons of percentages, etc.? The relevant question is "do complex multivariate statistical techniques add to our knowledge about a topic?" The answer is "it depends on the problem". In THE TRULY DISADVANTAGED, W.J. Wilson has shown that you can make a pretty provocative and perhaps even persuasive argument about contemporary social phenomena without resorting to anything more complex than crosstabs. On the other hand, if you want to capture the whole interaction between a number of variables, crosstabs are probably going to be somewhere between impractical and useless. It is perfectly reasonable, when looking at a piece of research, to ask "is the use of this method the most appropriate to the research question?" You should ask this whether the method used is quantitative or qualitative, complex or straightforward. Let the method fit the problem. Michael -- Michael Lichter UCLA Department of Sociology