File spoon-archives/bhaskar.archive/bhaskar_2003/bhaskar.0304, message 16

Date: Thu, 10 Apr 2003 11:12:11 -0400
Subject: Re: BHA: Research Methods Texts

Hi Marsh,

I'm also teaching a course on research methods, using Babbie's "The 
Practice of Social Research."  I am not using a CR based book, as you are, 
but am trying to complement Babbie's presentation with lectures on CR 
alternatives.  (I have to say right away, however, that my interpretation 
of CR has been formed by many years of struggling with Lonergan and 
Polanyi, only recently supplemented by my reading of Bhaskar and his 
commentators and followers.)  Babbie does not really think of himself as a 
"positivist," and has some very good things to say about different 
paradigms and theories.  But there remains a strong positivist bias in his 
book, especially when he is discussing quantitative research designs and 
methods.  He does not really want to assent to the reality of generative 
mechanisms which are neither empirical nor actual.  It is ironic, in a way, 
because so much of the quantitative research examples he gives involve 
measurements of dispositions of individuals, but in his explicit 
ontological statements he denies the reality of the very things he devotes 
so much time to measuring.  I find this same kind of ontological 
ambivalence in all of the mainstream methods texts I have any familiarity with.

In addition, there is what I regard as a fundamental misrepresentation of 
statistics.  I believe that the fundamental heuristic anticipation in 
social scientific statistical analyses is of "conditional 
probabilities."  Measurements are events, and the relative probabilities of 
certain values or ranges of values on the dependent variable are 
conditioned by the frequencies of measurement events of one or more 
independent variables.  Conditional probabilities are illustrated visually 
in contingency tables, but they are also expressed in measures of 
association and regressions (these are not "measurments" in the same sense 
as measurements of variables, but different ways of expressing relations 
between or among two or more measurements made on the same cases).

The misrepresentation comes both from lack of clarity about the degree to 
which statistical analyses are about conditional probabilities, and from 
the way in which the slogan "correlation is not causation" triggers 
attempts turn statistical relations into causal relationships.  The image 
of causation behind these attempts is still the old mechanistic "billiard 
ball" world view, even though it is disguised in rather sophisticated 
fashion.  When an event is understood to be the cause -- the necessary and 
sufficient condition -- of a subsequent event, not notion of causality is, 
in Aristotelian terms, efficient causality.  But the dispositions of 
individuals can best be understood as formal causes of the behaviors we are 
trying to explain.  In CR terms, these dispositions are in the realm of the 
real, even though they are neither empirical or actual.  In Aristotelian 
terms, they are potentialities for the kind of behavior they inform -- the 
function as programs or templates for behavior, as generative mechanisms.

The problems with mainstream methods texts, I believe, are not so much in 
techniques of data collection (I prefer "data construction") or in 
techniques of data analysis.  The come from too narrow an ontology, and a 
consequent restriction in conceptualization and theory construction.

Sorry for the length of this response, but there is a long history of 
arguments that goes into the current state of methods texts, a history of 
which I know only bits and pieces.


Dick M

At 05:31 PM 04/09/2003 -0400, you wrote:
>I'm teaching a course on research methods and using two textbooks. One,
>_Social Research Methods_ by Lawrence Neuman is fairly standard although far
>less dogmatic than most that I've seen. The other, _Explaining Society_ by
>Danermark, et al. is explicitly based in CR. At one point we covered the
>logic of research. Neuman discusses inductive and deductive reasoning;
>Danermark discuses both of these as well as abductive and retroductive
>reasoning. One of the students asked what I thought was a very good
>question. Why doesn't Neuman cover these other modes of reasoning as well? I
>explained that Neuman is updating a text (now in its 5th edition) that
>originally didn't have to deal with CR and there's a certain path-dependence
>for the textbook that makes incorporating such concepts now very difficult.
>In effect, the entire text would have to be reorganized.
>I'm not entirely satisfied with this answer. Why do you think this stuff
>hasn't become more common in methods textbooks? Do you know of any that do a
>better job with this?
>         Marsh Feldman
>      --- from list ---

     --- from list ---


Driftline Main Page


Display software: ArchTracker © Malgosia Askanas, 2000-2005