Empiricism, Paradigms and Data



Robert N. Bostrom

University of Kentucky


Abstract


            Rather than being based on an indefensible distinction between empiricists and interpretivists, this essay argues that such a distinction, although generally held by many persons, is actually not defensible on the basis of a supposed theory-data interaction. Covering law explanations are not part of common theory in communication research. Paradigmatic distinctions are typically thought of as being fundamental to the point of irresolvable, and when they are based on philosophical positions should be sharply questioned. Research is best performed when paradigms do not limit the activity or the thought of researchers.



Empiricism, Paradimgs, and Data


             It is extremely gratifying to know that someone has actually read something you have written, and even more pleasing to know that the reader has agreed with most of what you have said. Pavitt raises a number of important issues that deserve further consideration and I am happy to discuss them here. These issues include the nature of empiricism, the existence of paradigmatic differences, and most importantly, how these views affect what we take to be data statements in communication study.

            The Nature of Empiricism. In the introduction to my essay I noted that we have generally studied communication study from empirical and interpretive points of view. Prior to that I mentioned some fundamental distinctions drawn by philosophers, definitional issues arising from attempts to deal with concepts like reality, and some difficulties inherent in classification. Pavitt is troubled by my description of empirical study as that “traditionally ...associated with covering law explanations and [as following] traditional behavioristic agenda” (2004, p 6). The quotation might seem to indicate, and Pavitt apparently assumes, that I take a fairly conservative view of empiricism and of covering law explanations. Of course, the point of my essay was the exact opposite, and further, that attempts to justify paradigmatic differences by philosophical argument thus far have failed.

            It would help, however, to examine the quotation more closely. What I actually said was “Traditionally, empiricism has been associated with covering law explanations and has followed traditional behavioristic agendas” (Bostrom, 2003, p. 276). Note the past tense. Few in social science today believe that covering laws are the desired end product of our research. There are a number of legitimate alternatives. In 1992 Donohew and I discussed three basic models for research: inductive-statistical, deductive-statistical, and deductive-nomological (Bostrom & Donohew, 1992, pp. 114-116). In that discussion, we showed that only the deductive-nomological model has the characteristics of a traditional covering law and is more typical of chemistry or physics than social science. Donohew and I made it clear that most of the time communication research follows the inductive-statistical model, though occasionally deductive-statistical models appear. In the inductive-statistical model observers begin with the characteristics, or facts of a situation, proceed to empirically connected observations (explanans), and then go on to formulate a general theory which explains the observations (explanandum). Also, we used the term “law-like relationships” rather than “laws,” to further distance ourselves from early postivism (Carnap, 1953).

            These models, of course, are usually concerned with causal laws, often specifying a degree of probability, though that is not always primary (Fetzer, 1974). Having empirical characteristics is not enough to form an adequate explanatory base. According to Hunt (1976), "Pragmatically speaking, we may refer to causal explanations as those explanations that employ nonspurious theoretically supported, sequential laws in their explanans" (p. 51). Hunt goes on to list four criteria for acceptance of such laws: temporal sequentiality, associative variation, nonspurious association, and theoretical support. Such statements, of course, must also have empirical content, making them susceptible to measurement; possess nomic necessity, that is, have potential causality, be capable of integration into a system of compatible lawlike statements; and be intersubjectively certifiable.

            I think it is obvious that the models question cannot really be described by a simple division into empirical and interpretive, and when I used the word empirical, I was simply describing the bipolar state of affairs in communication theory, a situation that Pavitt attempted to remedy in proposing a third way. But the empirical label is worth a little more attention. Ancient skeptics were atheists, the most prominent of which was Sextus Empiricus (about 150 AD). Today, when one calls oneself an empiricist, some of the skepticism of the ancients still remains. When many of us first adopted the label, we were less concerned with the nature of empiricism and more concerned with distinguishing ourselves from the rhetoricians. On the face of it, an empiricist is someone who does not believe in metaphysical entities, such as truth, justice , and the American Way, and whose watchword is “I don’t believe it unless I see it.” Endnote This naive skepticism fails immediately, because we all believe in things we have not seen. I have never seen my liver, but I believe that I have one, however much bourbon has been poured through it. Emily Dickinson never saw the sea (even though Amherst is only about one hundred miles from the ocean) but felt that she was justified in believing in its existence. So any empiricist not only believes in the evidence of one’s senses, but also believes in descriptions of entities furnished by other persons, and, in addition, entities that could possibly be observed given sufficient effort and expenditure to do so. We can believe in Afghanistan, the planet Mars, and the AIDS virus.

            There are limits, however. Emily Dickinson felt that because she knew how heather looked, and what a wave must be, she also could believe in a Supreme Being. Most of us would not be that liberal with our empiricism, and usually we set strict limits on what we would consider real in a practical sense. If we are very conservative on these limits, we might agree with Berkeley that gravity, as Newton described it, could be considered simply a convenient term that helps explain relationships among observations. Endnote

            A conservative empiricism may keep us from believing in mental telepathy, subliminal persuasion, and ghosts, but it also has its problems. Many perfectly sensible entities, such as attitude are hard to defend from an empirical point of view. If we cannot believe in gravity, we cannot believe in attitudes or black holes, and end up admitting that they are actually only intervening constructs. This position is the one I took in my persuasion book (Bostrom, 1983). Pavitt (1999), however, pointed out that scientific realism enables us to allow a measure of reality to these constructs, and I must admit, that when I first read his article I was completely convinced that his approach was the correct one. Endnote I must also point out, however, that today, with the advances we have seen in cognitive psychology and magnetic resonance scanning that a conservative empiricist might well be justified in believing that attitudes are real entities. In addition, Pinker’s (1997) description of how the mind works, together with Lakoff and Johnson’s (1999) description of the metaphorical nature of theory also are strong underpinnings for a more conservative empiricism than Pavitt might like.

            I once thought the label “empiricist” should be replaced with “objectivist” (Bostrom, 1983, Harrington & Bostrom, 1996) but you can see from my recent paper that I do not think much of objectivsm any more. My main defense about the point of view I took in my 1983 book is that it seemed like a good idea at the time. I do not mind being wrong sometimes because the only way you can be right all the time is to say nothing. Many of my students have enjoyed reminding me that in 1966 I thought cognitive dissonance was a useful way to describe motivation in argument (Bostrom, 1966). Endnote

            Paradigmatic Thinking. Another problem that Pavitt raises is my skepticism about the inevitability of paradigmatic thinking. Pavitt contends that paradigms are real and will not go away. They seem illogical to me, as well as potentially hurtful, and I am hard put to find real justification for them. It would seem useful to me to define more carefully what we might take to be a paradigmatic difference. It does not seem reasonable to classify every difference of opinion as paradigmatic. In politics, as Pavitt notes, examples are easy to find. Krugman (2003) notes that the new right wing does not just want to change the tax structure, but they basically do not believe in our system of government-–a real paradigmatic difference. I would think that paradgmatic distinctions in communication should be more than simple disagreements about definitions. Endnote

            This interpretation seems to be what is implied by the “set of beliefs” that form Kuhn’s (1970, 1977) definition of the term. In communication study, how drastic do the differences have to be before they qualify as paradigmatic? I think I can make a case that the differences between Berger and Sunnafrank are less than paradigmatic, as well as the differences between Larson and Sanders as opposed to Steinfatt and Infante. The Shimanoff paradigm is simply a theory-metatheory confusion rather than a real difference. A true paradigmatic difference, to me, would be something irresolvable.

            The way we look at nonverbal communication is ane excellent example. Motley, of course, feels that communication should be intentional, but if you believe in the unconscious mind, you will feel that not everything we do is planful, that is, may be a manifestation of an unconscious wish or desire. Many psychotherapists have a strong belief in the existence of an unconscious mind. Often this belief leads them to take undefensible positions. In an effort to show how the unconscious makes plans without our knowing it, Weiss (1990) studied patients’ reactions to the planfulness of a therapist’s comments (called insights) and examined how often a therapist generated pro-plan insights as opposed to anti-plan insights. Then Weiss compared the degree of planfulness in the therapists’ interventions to patients’ reaction on the Experiencing scale, which measures how well people are experiencing a particular thought. “People are vividly experiencing a thought or feeling if they articulate it clearly and focus on understanding its significance. Individuals who talk in vague terms and whose words seem detached from their feeling are deemed to be weakly experiencing a thought of a feeling”( Weiss, 1990, p. 105). In short, when the therapist was on topic, the patient’s responses were clearer and less detached. On the basis of this finding with three patients Weiss concluded that the unconscious mind is planful. A simpler explanation seems more desirable, but belief in the unconscious with all of its ids, egos, and superegos is firmly fixed in a diminishing number of psychotherapists.

            Talking with someone who believes in the unconscious is frustrating. I believe that I am overweight because I love to eat. A Freudian would tell me that I eat too much because I unconsciously have a will to fail, a manifestation of the death wish. When I deny such a wish, I am told that, of course, I do not know about it because it is in my unconscious mind, and my denials are yet another manifestation of the death wish. These beliefs are harmless unless they are being told to me at the rate of $150 per half hour session under the pretense that I have mental imbalance.             Because paradigms rely on axiomatic beliefs, it is common to support them with basic philosophical positions, A good example is found in some of the justifications for qualitative research. Hamilton (1994), for example, uses Kant’s phenomenology to indicate that perception is more than seeing, which, Hamilton asserts, “opens the door to subjectivism, idealism, perspectivism, and relativism” (1994, p. 63). This logic is bizarre. Kant’s interpretation is more easily explained by invoking linguistic functions rather than than some mysterious phenomenon. Pavitt has difficulty in seeing why phenomenology is considered fundamentally different from cognitive psychology, and the answer is not rational, but paradigmatic.

            Some of Pavitt’s examples of current paradigms are worth examining. The difference between Skinner and Chomsky was primarily methodological, Skinner basing his contentions on experimental observation, and Chomsky basing his on logic. When Pinker provided hard data for Chomsky’s position in The Language Instinct (1994), most Skinnerians recanted.

            Some of the other distinctions raised by Pavitt are definitional. The intentionality issue is one such–what shall we define as communication? The silly statement that “You cannot NOT communicate” is one such, and for someone to take that position might well be paradigmatic. I feel that a truly paradigmatic issue is one that cannot be solved, one that is only resolved by taking up sides. Is this a constructive situation? I think not.

            My argument is that paradigms do indeed exist-–but a more meaningful question is: should they exist? A distinction between qualitatve and quantitative is currently given paradigmatic status. I am not as convinced of the inevitability of this distinction. I think I have demonstrated that the distinction cannot be maintained on philosophical grounds. In addition, I feel strongly that all researchers should be involved in both quantitative and qualitative processes. It is interesting to me to note that the two qualitative researchers that I know best, Anderson and Lindloff, are both highly competent in statistical methods. Those who have seen my research methods book (Bostrom, 1998) will know that I feel that qualitative and quantitative research are essential parts of the same process. In that book I took the position that one should do qualitative research first, and then proceed to what I called confirmatory research in the search for more general statements.             There are some excellent exemplars that might cause us to revise what we think are paradigmatic differences between qualitative and quantitative research. My favorite example is detailed in the book When Prophecy Fails (1956). In the early 1950's, in a large American city, a prophetess (Mrs. Keech) began telling persons that she was in touch with a group of supernatural beings, which she called the Guardians. These Guardians were travellers in space and time, told Mrs. Keech that the world was going to end soon, and that a few persons would be saved, primarily Mrs. Keech and her associates. When the prophecy failed, Mrs. Keech and her associates, rather than admitting that they were duped and disbanding the group, instead redoubled their efforts to recruit new members and insisted that minor errors had contributed to the failure of the prophecy. The complete account of the movement and its subsequent effects were told in Festinger, Riecken, and Schacter (1956). Their method of study was simple. They joined the group. Here is their description of the infiltration process:

In our very first contact with the central members of the group, their secrecy and their general attitude toward nonbelievers made it clear that a study of the group could not be conducted openly. Our basic problem then was obtaining entree for a sufficient number of observers to provide the needed coverage of members' activities, and keeping at an absolute minimum any influence which these observers might have on the beliefs and actions of members of the group. ... Our initial contact was Mrs. Keech, whom one of the authors telephoned shortly after the newspaper story about her appeared in late September. He told her his name and said he had called to see if he might talk with her about some of the things she had told the reporter, especially the matter of the predicted flood and flying saucers. (Festinger, Riecken, & Schacter (1956), pp. 234-235).

In spite of the authors' insistence that they were only listeners and did not contribute to the central beliefs of the group, it is clear from their description that they had pretended to have psychic experiences in order to ingratiate themselves with the group, and that on other occasions they might have reinforced the beliefs of the group by pretending to agree with them.

The study of this group led one of the researchers (Festinger) to formulate a theory called cognitive dissonance (Festinger, 1956), which generated much interest and provoked a good deal of further research. The research effort, however, was high productive, in that many other better theories were generated.

            Another excellent example of qualitative research as a starting point is the initial exploration of the concept of credibility by Berlo, Lemert, and Mertz (1969). These researchers asked respondents to furnish adjectives that they used to describe message sources, both believable and unbelieveable. The researchers intruded in the process only by matching adjectives into bipolar pairs (such as clean-dirty or smart-dumb). Then they subjected the hundreds of adjective pairs to factor analysis. The credibility scales that resulted were an extremely productive tool in the analysis of persuasive messages sources.

            Qualitative research needs no philosophical justification in that its generation of information can stand on its own. But would it not be more interesting to follow up on a qualitative investigation with more general investigations? For example, Ruud (1995) studied the activities in a regional symphony orchestra. Primary participants in this orchestra were divided into the categories of musicians, directors, and management. These categories seem fairly simple and straightforward. But an in-depth analysis of the discourse of the three groups, both within groups and between groups, showed significant differences in the use of communication. A more general investigation of this phenomenon would not only be useful for arts groups, but also for organizational communication. Obviously, this argument is old, raising as it does, the issue of programmatic research. Sometimes one can combine qualitative and quantitative in one study as Nancy Harrington and I did when we studied compulsive talkers. We began with case studies and interviews and then proceeded to more generalizable statements (Bostrom & Harrington, 1998).

            When Reardon and Rogers (1998) tried to explain how mass communication and interpersonal communication came to be so strongly divided (another paradigm), they attributed this illogical formulation to the nature of the current academy in which we live and work. This attribution led me to speculate in my footnote, which reads “These paradigmatic divisions have probably arisen because of the basic nature of the university culture and the demands of academic life, and have no sensible counterpart outside of the academy.”

            The Nature of Data. Pavitt’s third main point centers on the nature of data statements, describing a procedure of transcribing speaking turns based on the interpretations of the coder and the subsequent categorizations that result. Pavitt notes that this is an objective procedure in which bias keeps building through every step in the operation. Once again, the problem here is oversimplification. The categorizations that result from coding can be of many types, some of which may suffer bias, and some that may not. Basic types of categorizations are single stimulus (points on a scale), stimulus comparison (greater or smaller, stronger or weaker), stimulus similarity (nearer or farther from a given exemplar), and stimulus preferences (likes or dislikes). Mixing these categories invites disaster. Jacoby (1991) suggests that the problems inherent in the kinds of categorization can be solved by using dimensional analysis. Virtually any stimulus can be viewed from a number of dimensions, and Jacoby insists that context and the researchers’ intent are crucial. He goes on to show how physical and conceputal dimensions are analogous and can be utilized in the construction of data statements (pp. 28-31).

            An excellent example of the use of dimensional analysis with observations of interpersonal interactions is found in Goldsmith and Baxter (1996). These researchers combined dimensional analysis and clustering to form a model of interpersonal interaction. This kind of data construction is, admittedly more effortful, but certainly solves the problem Pavitt poses, and shows that even potentially biased data can be objectified in systematic ways. If one followed Goldsmith and Baxter’s example, the systematic bias that Pavitt refers to will appear in one of the dimensions, making corrections easy. If the bias is not systematic, then it will be part of the error variance, and, of course, if it gets too large the data set will be meaningless.

            I appreciate the opportunity to explore some of these points in greater depth. We often seek philosophical justification for our points of view, and part of this search is inherent in an emerging discipline like communication. We need to be careful, however, not to turn serious examination of the nature of theory into endless philosobabble. The truth is that qualitative research can and should stand on its own without such elaborate apology. The converse is also true. Research does not deserve special attention just because it claims to be qualitative. It would be useful for us all to concentrate on the content of our research, which, I assume, is still focused on how to help persons communicate more effectively.

   




References

 

Berkely, G. (1710/1957). A treatise concerning the principles of human knowledge. Indianapolis, IN: Bobbs-Merrill.

Berlo, D. Lemmert, J., & Mertz R. (1969). Dimensions for evaluating the acceptability of message sources. Public Opinion Quarterly, 33, 563-576.

Bostrom, R. N. (1966). Motivation and argument. In Perspectives in argumentation, ed. Gerald R. Miller and Thomas Nilsen (Scott, Foresman, and Company,), pp. 110-128.

Bostrom, R. N. (1983). Persuasion. Englewood Cliffs, NJ: Prentice-Hall.

Bostrom, R. N. (2003). Theories, data, and communication research. Communication Monographs, 70, 275-294.

Bostrom, R. N., & Donohew, R. L. (1992). The case for empiricism: Clarifying fundamental issues in communication theory. Communication Monographs, 59, 109-128.

Bostrom, R. N. & Harrington, N. G. (1998). An exploratory study of compulsive talkers. Communication Edcuation, 43, 78-91..

Carnap, R. (1953). Testability and meaning. In H. Feigl and M. Brodbeck (Eds.) Readings in the philosophy of science (pp. 47-52). New York: Appleton-Century-Crofts.

Cohen, B. (1985). Revolution in science. Cambridge, MA: Harvard University Press.

Festinger. L. (1956). A theory of cognitive dissonance. New York: Harper and Row.

Festinger, L., Riecken, H. W. & Schacter, S. (1956). When prophecy fails: A social and psychological study of a modern group that predicted the destruction of the world. New York: Harper and Row.

Fetzer, J. (1974). Statistical explanation. In K. Shaffner & R. Cohen (Eds.). Boston studies in the philosophy of science (pp. 71-93). Boston, MA: Philosophy of Science Association.

Goldsmith, D. J., & Baxter, L. A. (1996). Constituting relationships in talk: A taxonomy of speech events in social and personal relationships. Human Communication Research, 23, 87-114.

Hamilton, D. (1994). Traditions, preferences, and postures in applied qualitative research. In N. K. Denzin and Y. S. Lincoln (Eds.) Handbook of qualitative research. Thousand Oaks, CA: Sage.

Harrington, N. & Bostrom, R. (1996). Objectivism as the basic context for theory and research in communication. In J. L. Owen (Ed.) Context and communication behavior. Reno, NV: Context Press, (pp. 118-143).

Hunt, S. (1976). Marketing theory: Conceptual foundations for research in marketing. Columbus, OH: Grid.

Jacoby, W. G. (1991). Data theory and dimensional analysis. Newbury Park, CA: Sage.

Krugman, P. (2003). The great unravelling: Losing our way in the twentieth century. New York: W. W. Norton.

Kuhn, T. S. (1970). The structure of scientific revolutions. (2nd ed.) Chicago: University of Chicago Press.

Kuhn, T. S. (1977). The essential tension: selected studies in scientific tradition and change. Chicago: University of Chicago Press.

Lakoff, G.. & Johnson, M. (1999). Philosophy in the flesh. New York: Basic Books.

Pavitt, C. (1999).The third way: Scientific realism and communication theory. Communication Theory, 9, 162-188.

Pinker, S. (1994). The language instinct. New York: W. W. Norton.

Pinker, S. (1997). How the mind works. New York: W. W. Norton.

Reardon, K. K., & Rogers, E. M. (1988). Interpersonal versus mass media communication: A false dichotomy. Human Communication Research, 15, 284-303.

Ruud, G. (1995). The symbolic construction of organizational identityies and community in a regional symphony. Communication Studies, 46 , 201-222.

Weiss, J. (1990, March). Unconscious mental functioning. Scientific American, 262 (3),103-109.


Notes: