The Relativistic/Constructionist Perspective on Scientific Knowledge and Consumer Research

THE RELATIVISTIC/CONSTRUCTIONIST PERSPECTIVE ON SCIENTIFIC KNOWLEDGE AND CONSUMER RESEARCH

J. Paul Peter, University of Wisconsin

Jerry C. Olson, Penn State University

In a recent article Calder and Tybout (1987) promoted their conceptions of scientific method, philosophy of science, and the nature of scientific knowledge. According to them, there are three types of knowledge, each with its own methodology or philosophy of science. Everyday knowledge implies qualitative methodology; scientific knowledge implies sophisticated methodological falsificationist methodology; interpretive knowledge implies critical relativistic methodology.

Any attempt to label some research as science" or "scientific knowledge" and to withhold this label from other research needs careful scrutiny. This is because in Western cultures such labels have special meaning and are highly honorific; approaches to knowledge that are not scientific arc commonly denigrated in our culture.

The purpose of this paper is to critique the Calder and Tybout position. First, we review arguments that led to the demise of falsification as a viable philosophy of science as well as our own reasons for eschewing that approach. We do not exclude falsification studies from science, however, we reject falsification as a general philosophy of science and as the only approach that produces scientific knowledge. Second, we demonstrate that Calder and Tybout have misconstrued our relativistic/ constructionist approach leading to improper inferences about what it has to say about the nature of science. In the final section, we discuss several implication of our R/C perspective for consumer research.

WHAT'S WRONG WITH CALDER AND TYBOUT'S PERSPECTIVE?

Calder and Tybout offer two criteria that are designed to distinguish scientific knowledge from other types. Their criteria are (1) there are empirical attempts to refute the theory, and (2) the theory has performed better than available competitors. They argue that the tenets of sophisticated methodological falsification (Lakatos 1970; Leong 1985) provides the only methodology for scientific knowledge. Other forms of knowledge have different objectives, require their own methodologies, and do not produce scientific knowledge.

There are at least two major reasons why the conception of science sketched by them is unacceptable. We discuss them below in terms of the demarcation problem and problems with falsification. While they have been discussed previously in the consumer research literature, they bear repeating since they apparently have been ignored.

The Demarcation Problem

Professional philosophers of science have been unable to construct criteria by which science could be demarcated from non-science. A review of the history of debates over demarcation criteria reveal that they are typically used in polemical battles between rival camps, such as Popper's attempts to discredit Marx and Freud (see Laudan 1983).

It seems clear that Calder and Tybout are using a ploy similar to that of Popper. As practitioners and advocates for laboratory experimentation in the social psychology tradition, they have singled out their preferred approach as scientific and categorized qualitative and interpretive research as something less than science. However, the Calder and Tybout criteria fail to demarcate scientific knowledge from other forms on several grounds.

First, by excluding from science any theory that has not been empirically tested by researchers with falsification motivations, Calder and Tybout have eliminated almost all of what philosophers of science and practicing researchers would consider science. For example, in physics, the Work of Newton and Einstein would be classified as non-science since they were not concerned with falsifying their own theories but instead, championed them strongly. Most people recognize them as two of the greatest scientists in history.

In consumer research there are also many other types of work that would be eliminated from science if Calder and Tybout's criteria were accepted. For example, attempts to construct general models of consumer behavior, such as the work of Howard and Sheth, would be considered non-scientific activity. In fact, all attempts to create theories that do not involve empirical falsification activities would be considered non-scientific. Similarly, attempts to construct mathematical models to fit data would be eliminated under the Calder and Tybout constraints.

Second, few social science researchers seriously attempt to falsify theories. Rather, in most cases, they attempt to falsify the null hypothesis in order to support their favored hypothesis or they attempt to falsify someone else's theory in order to support their own or one that they champion. Greenwald et. al.(1986) have discussed at length the strong confirmation bias in social science and have pointed out that publication practices overwhelmingly favor supportive results. In fact, Greenwald et. al. argue that researchers who do not obtain supportive results are unlikely to even try to publish the work. In many cases, then, there is little chance that a "falsifying" result will be widely disseminated in a field and become part of scientific knowledge.

Third, we note that the Calder and Tybout criteria are not specified precisely enough to be applied. How many attempts to refute a theory must be made to conclude that a theory is falsified? In comparative "tests" of alternative theories, how much better must the empirical results be before the winner is identified? Who decides what constitutes a valid falsification test? What is to be concluded when some comparative results support one theory and some support the other?

In sum, we concur with Laudan (1983) that "For these and a host of other reasons familiar in the philosophical literature, neither verificationism nor falsificationism offers much promise of drawing useful distinctions between the scientific and the nonscientific" (p. 26). Without this demarcation, Calder and Tybout's judgement of their own approach as scientific and the other approaches as non-scientific is fully discredited. We now turn to the second problem with their perspective, the technical inadequacies of sophisticated methodological falsification.

Problems with Falsification

It should be clear that falsification in all of its forms has been rejected as an adequate philosophy of science. Although there are a variety of problems with the approach, there is one of central importance to this discussion. Namely, that while-Calder and Tybout recognize the error of asserting that theories can be shown to be true, they apparently accept the idea that theories can be falsified empirically. However, Duhem (1953), Laudan (1983) and others have explained why it is impossible to do so. In addition, falsification does not give a good account of the history of science.

What Theory Do Data "Test"?

A central problem with the concept of falsification is that much more is involved in an empirical study than simply the substantive hypotheses under investigation. There are a whole set of assumptions about initial conditions, auxillary hypotheses, and the validity of measuring instruments, sampling, and data analysis procedures. No study can totally unravel these and "test" only the substantive hypothesis of interest. Thus, any attempt to conclude that a theory has been refuted can easily be deflected by suggesting that something else in the maze of assumptions and premises caused the "falsifying" result (Laudan 1977; Anderson 1983).

In addition, imaginative researchers can make ad hoc modifications to a theory to dispute inconsistent results or argue that any unsupportive results demonstrate certain boundary conditions to the theory's applicability, but do not refute the theory. Finally, they may produce another set of empirical results that again supports their favored theory and attempt to persuade the research community that their theory remains valid.

In sum, theories can not be falsified by empirical data alone. Thus, it is not surprizing that we know of no instances in our field (or others) where advocates for a theory simply gave up and accepted that their theory was totally incorrect because someone produced empirical results inconsistent with the theory. In fact, Mitroff (1973) found in his research on the NASA moon scientists that the most respected scientists believed strongly in their theories and ignored data that did not support them.

What Does History Show About Falsification?

Falsification has long been rejected as providing an adequate account of the history of science. This is because there are many examples of well-accepted theories for which there is abundant empirical evidence that they are incorrect (see Anderson 1983).

In consumer research, it is clear that many theories survive and prosper despite a lack of empirical support. Familiar examples include need theory, personality theory, and theories of the attitude-behavior relationship. In addition, given the weak statistical criteria used to judge the adequacy of empirical results and the failure to reject theories with little supporting empirical evidence, it appears that falsification does not provide an adequate account of consumer research.

In sum, the only reasonable conclusion is that falsification does not provide a good description of how science is performed. Apparently, it is also not a good normative model since the greatest scientists in history did not follow falsification tenets.

WHAT IS THE RELATIVISTIC/CONSTRUCTIONIST PERSPECTIVE?

Calder and Tybout have misconstrued our R/C perspective leading to inappropriate conclusions and implications about what relativism has to say about science. We will discuss Calder and Tybout's misconceptions in terms of the fallability, concensus, and methodology arguments.

The Fallability Argument

Calder and Tybout assert that

Both Anderson (1983; 1986) and Peter and Olson (1983) have argued that what is called scientific knowledge is the product of relativistic methodology. Basically they contend that because falsification depends on data and concensus about data, which may be fallible, this methodology is untenable (p. 138).

This statement in no way represents our view, and we are puzzled that anyone could interpret our writings in this way. There are at least two problems with the Calder and Tybout reconstruction of our position.

First, our major concerns with falsification are described earlier in this paper; the issue of the "fallibility of data" is not part of these arguments. Strictly speaking, in fact, to assert that data "may be fallible" suggests the possibility of perfect data that is not fallible. Such a view is inconsistent with relativism.

Second, we note that Calder and Tybout's concern with "fallibility" suggests their positivistic assumptions about the possibility of ultimate criteria and objective standards for judging the truth or falsity of data and/or consensus about data. However, from our R/C perspective, consensus involves subjective decisions by a group of scientists to accept a theory and its support or not to do so; whether the concensus is true or not is a non-question from our perspective.

The Consensus Argument

Calder and Tybout assert that

We believe that relativism retreats unnecessarily to the position that scientific knowledge is based not on empirical testing but on social agreement. The observation that concensus about data may be erroneous does not imply that it is preferable to avoid the use of data and instead rely on consensus per se. We need not give up the primacy of empirical data in confronting theory and relegate scientists to theory peddlers (p. 138).

Calder and Tybout are in error in arguing that relativists eschew empirical data and empirical research. Relativists do not believe that data are worthless and should not be constructed. However, researchers with an R/C perspective do view empirical data differently than do positivistic falsificationists. For example, in line with modem philosophical positions, relativists recognize that data are theory laden. In other words, data are created and interpreted by scientists in terms of a variety of theories and theoretical perspectives. It is impossible to collect data that are "theory neutral", since at least some implicit theory is needed to create measures and attach a meaning to them.

Relativists accept empirical data as an important part of science and recognize that empirical results can play an important role in generating social consensus about theories. However, it should be clear that relativistic researchers are much less impressed with empirical data and its role in science than are Calder and Tybout. Part of this skepticism occurs because we know that well-trained researchers can construct empirical data and results to Support or refute almost any theory without violating "accepted standards" of research in a field. In fact, we have pointed out that the creation of empirical data and empirical results is a process that is controlled by the researcher (Olson, 1987; Peter, 1984). For example, if an initial laboratory experiment does not produce desired effects, a researcher can continue to do "pilot" studies (in the name of "pretesting" or "calibrating measures") modifying the manipulations and methods until the desired results are created. Given this view of data, it should be no wonder that we are more conservative in our evaluations of empirical results than are researchers who purport to be falsificationists.

The Methodology Argument

A final error made by Calder and Tybout was to categorize relativism as a research methodology. The relativistic/constructionist perspective is not a methodology. Rather, it is a set of metatheoretical presuppositions about the nature of reality and human capabilities for creating knowledge which, taken collectively, constitute a general philosophy of science. We believe that our perspective has a number of important implications for scientific practice, but it is not a methodology.

We suspect that Calder and Tybout charweri/c relativism as a methodology to avoid having to deal with the serious philosophical criticisms it levels at their preferred approach. The point of their paper is that relativism may be OK for non-scientific interpretivists, but has no relevance or value for scientists whose only valid approach is sophisticated methodological falsification. By characterizing relativism as a second-class methodology that does not produce scientific knowledge, Calder and Tybout try to dismiss it and the criticisms it has of their preferred approach.

Implications of the R/C Perspective for Consumer Research

To this point we have demonstrated that the Calder and Tybout conception of science is untenable and that they have misconstrued our R/C perspective. In this final section we offer several implications of our perspective for practicing consumer researchers. These include methodological pluralism, evaluative pluralism, and cognitive and behavioral changes in research practices.

Methodological Pluralism

From an R/C perspective, no single approach (such as sophisticated methodological falsification) can be universally applied to guarantee scientific knowledge. Rather, the relativist recognizes many approaches to creating scientific knowledge, including qualitative, quantitative, and interpretive approaches and that no single method is always better than the others for constructing scientific knowledge. Every method can be useful, can provide convincing evidence, and can create meaningful scientific knowledge in particular situations and contexts. Every method, including falsification, entails construction and interpretation by scientists. Restricting consumer research to a single method like falsification is likely to stifle the creativity of researchers in the field since it encourages them to discard any ideas they have for which they cannot readily design a "falsification experiment." For example, we find no reason to view Holbrook and Grayson's (1986) work as something less than science nor do we believe that a falsification experiment would have been appropriate for their purposes. Thus, we believe consumer researchers should welcome and respect research of many types, including qualitative and interpretive work, and not prejudge these as less scientific than traditional empirical studies.

Evaluative Pluralism

From our R/C perspective, no single criterion or set of criteria can be universally applied to judge theories, research results, or scientific knowledge. However, this position does not mean that R/C researchers have no standards for judging scientific knowledge.

In previous work we have subsumed a number of criteria under the general category of "usefulness." Usefulness is a relative concept. It involves many types of subjective evaluations by scientists that are contingent on or relative to a host of factors. Scientists may judge a theory to be useful for many reasons including (1) it describes the workings of a phenomenon in a way consistent with prior beliefs and/or behaviors; (2) it makes them feel that they better understand a phenomena; (3) it makes predictions, novel or otherwise, that are supported by empirical data; (4) it offers a parsimonious account of a complicated phenomenon; (5) it is aesthetically appealing or mathematically elegant; (6) it offers a potential solution to a problem deemed important by the research community or society; (7) it identifies a variety of new, interesting questions that can be researched; (8) it appears to helps people get along in the world; (9) it appears to increase the quantity and/or quality of human life; (10) it appears to lead to technological changes deemed good for society or an organization.

We believe that these and many other forms of usefulness are currently used to evaluate scientific theories and research. Which criteria are or should be applied depends on the particular context and the relevant goals and values of individual researchers and research communities. In consumer research, for example, we believe that different individuals and groups vary in the emphasis they place on criteria such as managerial relevance, mathematical sophistication, theoretical significance, and simply, whether the theory and research tell an interesting story.

In sum, from an R/C perspective, theories, research results and scientific knowledge can be and should be judged on a variety of context- dependent criteria. Contrary to Calder and Tybout, we do not believe that adopting a relativistic approach to evaluating theories has or will have a negative impact on the field nor will it lead to anarchy in the sense that research reports will be accepted uncritically.

Cognitive and Behavioral Changes

Finally, we think it is important for consumer researchers to make a number of changes in the way they think about, conduct, and report their research. For one thing, it is quite clear that much more consideration should be given to constructing interesting theories and creating insightful questions rather than "testing" theories.

We also believe that it would be useful for consumer researchers to develop a less pretentious language for reporting research results than the outdated one used by positivists and falsificationists. For example, empirical data and research do not and cannot "test" hypotheses or theories. Rather, they typically provide demonstrations of the researcher's predilections or skill at post hoc rationalization. In fact, we agree with McGuire (1973) who referred to laboratory experimenters as "stage managers." Words such as "confirm," "disconfirm," "prove," and "disprove" should have no place in modem consumer research. Stripped of such pretentious language, traditional empirical studies could be seen more clearly to have no magical powers which make them universally or scientifically superior to qualitative and interpretive efforts.

SUMMARY AND CONCLUSIONS

This paper investigated Calder and Tybout's conception of scientific knowledge and their misinterpretation of our relativistic/constructionist perspective. Their conception of scientific knowledge was found to be untenable and long rejected by philosophers and historians of science. Their interpretation of our perspective and its implications for science were demonstrated to be in error.

We continue to believe that relativism is currently the best available philosophy of science for consumer researchers. We also continue to believe that our analysis of science as marketing is a useful conception. We view successful scientists as skilled manufacturers and marketers of knowledge rather than "theory peddlers," as suggested by Calder and Tybout. Finally, we consider the diversity of interests and research approaches currently accepted in consumer research to be a welcome sign of growing maturity for the field.

REFERENCES

Anderson, Paul F. (1983), "Marketing, Scientific Progress, and Scientific Method,: Journal of Marketing, 47 (Fall), pp. 18-31.

Anderson, Paul F. (1986), "On Method in Consumer Research: A Critical Relativist Perspective," Journal of Consumer Research, 13 (September), pp. 155-173.

Calder, Bobby J. and Alice M. Tybout (1987). "What Consumer Research Is..." Journal of Consumer Research, 14 (June), pp. 136-140.

Duhem, Pierre (1953), "Physical Theory and Experiment," in Readings in Philosophy of Science, Herbert Feigl and May Brodbeck, eds., New York: Appleton-Century-Crofts, pp. 235-252.

Greenwald, Anthony G., Anthony R. Pratkanis, Michael R. Leippe, and Michael H. Baumgardner (1986), "Under What Conditions Does Theory Obstruct Research Progress?" Psychological Review, 93, No. 2, pp. 216-229.

Holbrook, Morris B. and Mark W. Grayson (1986), "The Semiology of Cinematic Consumption: Symbolic Consumer Behavior in Out of Africa," Journal of Consumer Research, 13 (December), 374-381.

Lakatos, Imre (1970), "Falsification and the Methodology of Science Research Programs," in Criticisms and the Growth of Knowledge, Imre Lakatos and Alan Musgrave, eds., London: Cambridge University Press, pp. 91-196.

Lauden, Larry (1977), Progress and Its Problems, Berkely CA: University of California Press.

Lauden, Larry (1983), "The Demise of the Demarcation Problem," in The Demarcation Between Science and Pseudo-Science, Vol. 2, Rachael Lauden, ed., Blacksburg, VA: Center for the Study of Science in Society, Virginia Polytechnic Institute.

Leong, Siew Meng (1985), "Metatheory and Metamethodology in Marketing: A Lakatosian Reconstruction," Journal of Marketing, 49 (Fall), pp. 23-40.

McGuire, William J. (1973), "The Yin and Yang of Progress in Social Psychology: Seven Koan," Journal of Personality and Social Psychology, 26 (June), 446-456.

Mitroff, Ian I. (1974), "Norms and Counter-Norms in a Select Group of Apollo Moon Scientists: A Case Study of the Ambivalence of Scientists," American Sociological Review, 39 (August), 579-595.

Olson, Jerry C. (1987), "Constructing Scientific Meanings," in Marketing Theory, Russell W. Belk, et al., eds., Chicago: American Marketing Association, pp. 384-388.

Peter, J. Paul (1984), "On Ignoring a Research Education," in Distinguished Essays in Marketing Theory, S. Brown and R. Fiske, eds., New York: Wiley, pp. 324-327.