Monday, September 7, 2009

False debate: Quantitative vs. Qualitative Learning

Hal Pepinsky
September 7, 2009

My eighteen-month-old granddaughter Mila keeps driving home to me how fundamental yeses and noes are to human communication. Mila used sign language for no by the time she said ma. Not long after, as I kept shaking my head back and said the y word while Mila vigorously shook her head in squealing delight, she also learned to say “yes!”
When I was with her in Durango a week ago, Mila kept asking new words for things and stringing two and three words together--a voracious listener, looker and talker. Inadvertently in one moment, I played into what theoretical physicists might call a critical experiment: I said “no” to Mila’s face when I thought she was about to swing a picture off the wall over her head. Thinking back, all I needed to have done, as pretty much all the adults in her life I had seen do, was to pull her hand gently and firmly away, by sign and word said something like “hurts,” and distracted her. When instead I said “no,” Mila virtually collapsed in tears and wander off to be by herself.
In this one incident, an infant taught a retired criminologist a socially significant lesson in how NOT to achieve social control, and led me reconsider data on how I had seen Mila treated in similar circumstances, thereby to recognize how Mila had openly and yet with personal dignity been redirected without noes. All this from an “n of 1”as social methodologists might say about my data set. In many journal reviews I have received, especially early on when I was reporting a significant finding like the lesson Mila just taught me, a primary ground for rejection of my research reports was that I was merely anecdotal. Yes I was, yes I am.
I keep on learning from and sharing essentially anecdotal data because whether the stories are mine or others’, from the personal to the historical, stories have yielded far and away the most reliable (in statisticians’ terms merely “nominal”) empirical data I have in my storehouse of knowledge of how in the real world to get what I want and avoid what I don’t want from efforts at social control. Time and again, I use stories to guide me to look elsewhere, as to significance of other numbers (e.g., at a simple level, comparing when I told Mila no to what happened in other encounters with her I had witnessed). I keep finding that stories informants tell about their own experiences are more reliable and significant than stories social scientists tell about “them” (as in my own case, about prisoners and ritual abuse survivors). The lingering lawyer in me deeply respects my preference for personal over hearsay evidence. Feminists have taught me that in some social scientific circles, this preference is called “the narrative method.” Whatever people call it, that’s my research method--how I truly learn what is socially significant.
Back to the fundamental limit of all human knowledge: yeses and noes. When we try to reproduce language, as in the computer software I am using, language is fundamentally binary, a 1 or a 0, a yes or a no. At root, all our communication are binary, quantitative.
Decades ago, a grad student told me a story of his encounter with a noted quantitative sociologist, who presented a best-fitting path model of something or other. My acquaintance pointed to a particular path in the diagram, and asked, “What is the social significance of [a particular statistically significant path in the model)?” The presenter was baffled; after all, the path was significant at the .05 level.
I have noticed that self-identified quantitative criminologists share the struggle that any dedicated would-be social scientist including me faces, to reach findings that are socially and politically significant. I don’t have a clue as to what most reported research data mean to the informants, let alone figuring out what their data mean to me. I think I know what Mila meant to tell me. She wasn’t lying or trying to put on airs. By contrast, I don’t know for instance why I should accept at face value whether anyone who does or does not self-report victimizing or report being victimizated ought to be trusted, let alone understood.
A further problem I have with standard social research multivariate analyses is about claims of knowing that since a relationship is not statistically significant, the relationship makes no difference, let alone generalizing from group results to expectations of individuals. Statistically, the probability of any sample, let alone single case, having no effect on or being like what shows up in the best-fitting statistical model, IS ZERO (called “the fallacy of affirming the null hypothesis” at the beginning of chapter 1 of my graduate statistics textbook). Statistics only attach probabilities to what is NOT, NOT to what IS. After my late friend, co-instructor, and mentor Leslie T. Wilkins won a British Royal Statistical Society award for his work predicting which detained juvenile offenders would land back in the system after release, Les pointed out to me and countless others that prediction research is statistically bankrupt. (So much for profiling, period.)
As a criminologist in particular I’m brought back to a basic issue of yes or no in my field: whether to take each report of whether crime or criminality has occurred, or more generally of whether social control has been gained or lost, at face value. I can’t get past that methodological issue. Neither could people in 1986 in the Oslo criminology seminar I was invited in on, when we discussed a self-report crime survey, in this case of students in an introductory psychology class. An echo went round the room - How unscientific: the researchers didn’t know the students; how then could the researchers possibly know what the students know by students’ questionnaire responses?
I’ll spare you a separate “false debates” post on the topic of scientific vs. unscientific learning by declaring here and now that that I wish we self-styled social researchers would give up trying to prove who the better, truer, more elegant scientist is, and instead just trade notes on how and what plainly works and doesn’t in our social lives. False debates like that on quantitative vs. qualitative inquiry are beside any socially significant point of research, of socially meaningful learning. Transcending false debates is in my view where peacemaking discourse begins. I expect to follow up this essay with others on false debates that distract from accomplishing accountable, responsible, well informed management of our social problems. Love and peace—hal

No comments:

Post a Comment