Metatheory in Consumer Behavior Research Rap Session Position Paper

Citation:

Gerald Zaltman, Reinhard Angelmar, and Christian Pinson (1971) ,"Metatheory in Consumer Behavior Research Rap Session Position Paper", in SV - Proceedings of the Second Annual Conference of the Association for Consumer Research, eds. David M. Gardner, College Park, MD : Association for Consumer Research, Pages: 476-498.

Proceedings of the Second Annual Conference of the Association for Consumer Research, 1971     Pages 476-498

METATHEORY IN CONSUMER BEHAVIOR RESEARCH

RAP SESSION POSITION PAPER

Gerald Zaltman, Northwestern University

Reinhard Angelmar, Northwestern University

Christian Pinson, Northwestern University

[The authors wish to express their appreciation to George Brooker for his critical reading of the manuscript and to Jean Zaltman for typing this manuscript under acute pressure and to the Educational Foundation of the Association of American Advertising Agencies for their partial support of Mr. Angelmar in his investigation of the metatheoretical aspects of social marketing.]

[Gerald Zaltman is Associate Professor of Behavioral Science, Director of Research, Graduate School of Management and Faculty Associate, the Center for the Interdisciplinary Study of Science and Technology, Northwestern University. Reinhard Angelmar and Christian Pinson are currently completing their doctoral requirements in the Marketing Department, Graduate School of Management, Northwestern University.]

There have been remarkable efforts made during the past few years toward the advancement of knowledge in the behavioral sciences. While important and impressive steps toward this goal have been recorded the actual accomplishment in the behavioral sciences has not been consistent with the magnitude of the effort. The applicability of this observation, of course, varies considerably among disciplines and among specialties within disciplines- One important reason underlying the failure of actual achievement to keep pace with effort expended is the seeming absence of use of explicit criteria in the methodological approach (National Science Council, 1969). Certainly consumer behavior as an interdisciplinary area of research is not exempt from this statement. This has very important implications for marketing research in consumer behavior. The market researcher needs a set of criteria to evaluate the material he tries to build upon and, equally important, he needs a set of criteria, i.e4, a thinking methodology, to guide his own theoretical and applied research and to ensure the greatest value for his efforts.

The area of metatheory promises to provide fruitful guidelines for using and conducting research in marketing. It is hoped that presenting a brief overview of the domain of metatheory will help sensitize researchers to some of the many and complex issues involved and thereby encourage a more critical and selective development, testing and application of behavioral science in marketing research.

METATHEORY AND CONSUMER BEHAVIOR

Broadly defined, metatheory is: The investigation, analysis, and description of the technology of: (1) theory building, (2) theory testing, and (3) theory utilization. This definition seems to hold over all sciences. Metatheory is not concerned with the context, i.e., substantive area, of scientific activity but rather with the conceptual procedures of science.

Recently many marketers have recognized the need for being concerned with metatheory. Bartels (1970) in his Marketing Theory and Metatheory deals with the subject although not in a strict behavioral context. He points out that "the term metatheory appears only once in marketing writings, in the preface of John Howard's Marketing Theory wherein he says merely that 'a metatheory of marketing . o . is needed."' More recent statements demonstrating the need and utility for metatheory work in the behavioral field have come from: (1) Kollat, Engel, and Blackwell (1970), particularly in their treatment of variable and construct problems, although they don't make explicit use of the term metatheory; (2) Zaltman (1970) in his mention of criteria for evaluating theories and his discussion of scientific inference in using behavioral theory in the study of consumer behavior; and (3) Hunt (1971) in his attempt to evaluate from a metatheoretical viewpoint an earlier expression of marketing by Bartels. Earlier writings which in some way touch upon one or more dimensions of metatheory without labeling it as such include Taylor (1965), Buzzell (1963), Baumol (1957), Schwartz (1963), Halbert (1965), Bliss (1964), Sheth (1967), and Bartels (1951). There is, however, no published writing which represents a systematic account of metatheory in consumer behavior. Perhaps the theory which comes closest to expressing a metatheoretical basis or orientation is the so-called Howard-Sheth (1969) theory of consumer behavior.

In the discussion to follow the authors outline some of the basics of metatheory relevant to consumer behavior. Obviously not all the various criteria, principles, and issues mentioned can be elaborated upon within the confines of this paper Only selected topics will be expanded in the discussion. The other elements are mentioned at the risk of the confusion and monotony usually attendant upon the simple citing of unelaborated notions, in order to give the reader a broader feeling for the scope and variety of considerations metatheory holds for marketing research. In all cases appropriate references are provided so that the challenged reader may pursue and at least partially satisfy his curiosity. Hopefully, too, in the process of doing this the reader will internalize metatheory guidelines as part of his own thinking methodology

Since metatheory is a theory whose subject-matter is some other theory it provides the opportunity to treat that theory in a more rational way. Accordingly, it is appropriate to analyze the components of theories, namely hypotheses and concepts, before turning to the notion of theory itself. However, a theory is a conceptual device that is useful for explaining, predicting, and controlling events. Therefore the discussion begins with these three functions of information, i.e-, explanation, prediction, and control- Following this will be a treatment of concepts. hypotheses, and finally theories.

EXPLANATION

Explanation, defined here as the ascription of causes to an event or type of event (Morgenbesser, 1968:117), plays a crucial role in the scientific process. What is most distinctive about the major goals of science is the "organization and classification of knowledge on the basis of explanatory principles (Nagel, 1961:4). In the area of consumer behavior, nearly all research is oriented toward the development of explanatory principles which attribute causes to instances of consumer behavior.

Scientific explanations must meet two systematic requirements. These have been labeled the requirement of explanatory relevance and the requirement of testability. Explanatory relevance means the account of some phenomenon provided by an explanation would constitute good grounds for expecting that the phenomenon would appear under the specified circumstances. Explanatory relevance is achieved when the explanatory information adduced affords good grounds for believing that the phenomenon to be explained did, or does, indeed occur. This condition must be met if we are to be entitled to say: 'That explains it -- the phenomenon in question was indeed expected under the circumstances"' (Hempel, 1966:48).

An explanation having no test implications is devoid of empirical content; no empirical findings could support it or disconfirm it and consequently it provides little or no ground for expecting a particular phenomenon -- it lacks what Hempel calls objective explanatory power- Thus the requirement of testability is that scientific explanations must be capable of empirical test- An explanation which meets the first requirement (empirical relevance) also meets the requirement of testability whereas the converse does not hold (Hempel, 1966:49)

Causality in Explanation

The definition of an explanation employed in this paper relies heavily on the concept of causation and a very brief exploration of the meaning of the term "cause" will be helpful at this point. For the researcher, the object of explanation should be to provide, with as high a degree of certainty as possible, information about what variables influence, produce, or affect other variables. It is necessary to determine in a given context what the marketing decision variables (causes) are so that procedures can be established for manipulating them. Only when hypotheses or theoretical statements are presented in this way can they be of maximum utility in deriving marketing implications

It will be useful to formulate a causal statement in a marketing context and use this statement to discuss the properties of causal laws Such a statement might be: Norms of reciprocity among consumers affect the outcome of personal selling situations. Stinchcombe (1968) describes this type of proposition as one in which one, in particular, of the variables within a broad class of phenomena explains a particular variable in another class of phenomena. The two variables are norms or feelings of reciprocity and the outcome of a selling effort. The connecting mechanism at the individual level might be that feelings of obligation (in turn explained by social exchange theory) develop within a consumer as he comes to perceive the salesman as investing in the selling situation resources valuable to the salesman. Because of an apparent opportunity cost incurred by the salesman (as perceived by the prospect) in his relationship with the consumer, the consumer will reciprocate by rewarding the salesman with a purchase. A number of things in the causal sentence which are properties of causal statements should be noted:

1. The statement assumes, for example, that high levels of reciprocity among consumers will be found in successful personal selling situations; this statement would probably be better stated as a probabilistic prediction.

2. Changes in the level of reciprocity will produce changes in the frequency of successful personal selling.

3. Successful selling efforts by salesmen do not produce feelings of reciprocity among consumers(this may at first glance seem contradictory but salesmen may only activate or stimulate this variable, not create it). (The question of reciprocal causation cannot be considered here due to space limitations.

4. For a change in reciprocity to produce a change in sales, there do not have to be changes in other variables

5. The variables involved in a given causal law may be of different classes, e.g., one variable may be dichotomous and the other continuous.

6. There can be contexts where the causal law does not apply- Presumably it would not apply to "order-taking" personal selling situations but would apply to "creative" personal selling situations.

7. Other variables such as changes in level of disposable income or advertising could cause a change in sales without invalidating the causal statement.

8. Host importantly, we do not know that a given change in sales is in fact caused by a change in reciprocity (either among a given group or by exposure of other groups to the sales effort). Even if we hold constant the effect of advertising, income change and other imaginable variables and still find variations in reciprocity to be associated with variations in sales we cannot conclude with absolute certainty that the causal statement is true. There is always the chance that a variable or set of variables (including measuring errors and problems in the research design) that we have not thought of has produced the change in sales. As Campbell notes, a fundamental limitation in the scientific process is that, "Not only are scientific truths logically unproven, they also lack certainty in any other sense -- inductive, empirical, scientific, or implication. Yet they are in some sense 'established.' The best of theories if not 'confirmed' are at least 'corroborated' . . . thus the only process available for establishing a scientific theory is one of 'eliminating plausible rival hypotheses' . . . this is the best we can do-" (Campbell, 1969:352, 354-55)

Types of Scientific Explanation

There are basically five types or models of scientific explanation: the nomological model, the deductive model, the probabilistic model, the functional or teleological model, and the genetic model (e.g. Nagel, 1961: 20-26; HemPel and Oppenheim, 1948: 135-174; Hempel, 1966: 49-69; Brown, 1963: 47A). However, the discussion here must necessarily be limited to the deductive and probabilistic models due to space considerations.

The Deductive Model. The deductive model, widely regarded as the ideal form of explanation is of the following form: If explanatory statements or sentences S1, S2 ... Sn contain one or more laws and if a phenomenon P can be deduced from S1, S2 ... Sn then S1, S2 ... Sn is a satisfactory conjunction of explaining sentences. Thus the explicandum, that which is explained, is a logical necessary consequence of the explanatory premises (the explicans). In deductive models, then, the premises state a sufficient condition for the truth of what is being explained. It is sometimes added that the explaining premises, i.e. S1, S2 ... Sn must contain or constitute a theory.

For example, in a consumer behavior context, the question of brand loyalty among a certain group of customers might be raised, i.e., Why, in a particular instance, is loyalty so durable a phenomenon? The explanatory premise or explanans in this case might be certain learning principles involving the selective perception, retention and reinforcement theories. Thus continued brand loyalty (P in terms of our earlier notation) is directly and logically deducible from a particular learning theory (S1, S2 ... Sn in our earlier notation). This does not exclude other explanatory premises from being introduced as supplementary or competing explanans.

Probabilistic Explanations. Probabilistic explanation can be contrasted with deductive explanation: With a deductive explanation,the explanatory premises would, if true, provide conclusive evidence for the conclusion, constituting a totally sufficient guarantee of the explanatory conclusion. With a probabilistic explanation, the explanatory premises do not provide a guarantee of the conclusion, but merely render it relatively likely ...'~ (Rescher, 1970: 37)

Social scientists usually encounter probabilistic explanations when the explanatory premises (e.g., reinforcement theory) contain a statistical assumption about some class of elements (e.g., consumers exposed to random reinforcement as opposed to routine reinforcement) while the explanandum (brand loyalty) refers to a given consumer in that class of consumers. Thus if we want to explain why a given consumer is brand loyal (assumed to be a dichotomous variable for purposes of illustration) we would point out that a known percent age of consumers exposed to random reinforcement will be brand loyal. Notice that this is not deducible.

Levels of Explanation

There are at least four levels of explanation in the behavioral sciences (Doby, 1969). They are presented in Table 1.

TABLE 1

LEVELS OF EXPLANATION

A good illustration of these four levels of explanation involving behavioral phenomena of relevance to marketing is presented by H. G. Barnett's (1953) theory of innovation as a basis for cultural change. Only elements of this theory, his notion of basic wants as necessary conditions for innovation, will be treated. At level one a phenomenon, an act R innovation, i.e., adoption, is observed; it is known -- not just assumed -- to have occurred or to be in the process of taking place.

At level two the phenomenon is observed to be of a certain nature, Q, in this instance, the purchase of an ultra modern architectural blueprint for a permanent home or possibly the actual purchase of such a home. Thus the nature of the event, Q, consists of a purchase (a particular behavior) of an object perceived as new with new being defined in terms of qualitative distinction rather than in terms of time. Q, then, is composed of three factors: (1) purchase behavior, (2) perceptual processes, and (3) an object having qualitative distinction from other objects in the same general class of objects. Q may have been produced by or be a result of central subliminal wants (a type of self want) and creative wants (a type of want that we relabel as autotelic wants). These two wants represent xl and x2 in Table 1. Central subliminal wants are those which relate to the individual's need for self-preservation and self-definition. They influence how we structure and organize our environment. Creative wants emphasize accomplishment with the process or act of being creative being at least and probably more important than the resultant innovation or objects. In general, wants of this nature result from dissatisfaction with the accepted way of doing things.

Level three is concerned with explaining how central subliminal wants (xl) and creative wants (x2) interact in manner yj to produce the adoption of the innovation in question. Explanation in this case takes the form of describing what y; is. In our example creative wants interact with central subliminal wants. The need to define oneself as unique, avant-gardish, etc. together with dissatisfaction of existing modes of architecture as means of achieving this self definition, lead to the adoption of radical architectural style. But in level three the emphasis is upon the manner of interaction. It could be explained that creative wants stimulate (the manner of interaction) central subliminal wants and that for reasons of congruence or cognitive consistency the central subliminal wants are expressed in creative ways, i.e., the individual establishes a self-definition of being an innovator. Being interested in doing innovative things and being dissatisfied with existing conditions brings about the idea that he is an innovator which becomes expressed in such behavior as the acquisition of a radically or at least significantly different home.

At level four explanation goes beyond the relationship of the x's to each other and attempts to account for reasons (wl, w2, etc.) factors x interact in manner yg. An explanation at this level has already been given. It was stated above that for reasons of cognitive consistency Factor xl, creative wants, cause Factor x2, central subliminal wants to express themselves in innovative, i.e., creative ways. The notion of cognitive consistency in this illustration constitutes the reason, w1.

Evaluating Explanations

There are four basic criteria for evaluation explanations. These are: scope, precision, power, and reliability- Each is discussed briefly below.

Scope. Scope refers to the range of events to which an explanation can be applied. It is difficult in the social sciences to achieve wider scope without introducing ambiguity.

A number of hypotheses and theories relevant to marketing and having broad scope can be cited briefly- The two-step flow hypothesis is a good example of an explanation of marketing-relevant communication behavior. Exchange theory as articulated by George Homans (1961) and Peter Blau (1964) is a theory of wide scope and relevant to marketing (Zaltman, 1970:30-32). E. T. Hall's (1961) theory of culture as communication is another explanation with extremely broad scope although difficult to test empirically.

Precision. "The precision of an explanation refers to the exactness with which the concepts used in explanation are related to empirical indicators, and the precision with which the rules of interaction of the variable in the system are stated" (Meehan, 1968:117). Note that there are two areas of precision referred to in this quote. The first concerns the relationship between a concept and its empirical indicator. The second concerns precision in the statements concerning the relationships among concepts.

With regard to precision in the first case there is always an unavoidable gap, a lack of precision, between a concept and its empirical operationalization.

In a very real sense no theoretically defined concepts can be directly translated into operations nor can theoretical propositions be tested empirically" (Blalock, 1968). In an article largely devoted to this problem of precision, Zaltman (1970:32) concludes: "Perhaps one of the greatest obstacles inhibiting the effective application of the behavioral sciences to marketing problems is that this very important quality of isomorphism (between theoretical and operational systems) can only be determined intuitively." Hard-headed researchers tend to disagree with this position (Bechtoldt, 1959), but the disagreement ultimately becomes one of two diverging basic philosophical positions, (Harre, 1967)

The second aspect of precision concerns precision in the stated relationships between concepts- Blalock (1969:18) argues vigorously for specifying relationships in the form of direct causal links stated in terms of covariations and temporal sequences for reasons of explication, testing, and measurement.

This last factor, measurement, is a key factor influencing precision. In the first case it affects the accuracy of the operationalized concept: the smaller the measurement error the more isomorphic the relationship between theoretical concepts and their empirical indicators. In the second place measurement affects the detection of such things as interaction effects which influence the interpretation of relationships among explanatory variables.

Power. Power refers to the degree of control over the environment an explanation provides. Power depends upon the precision of the description and explanation and upon the completeness of the variables. An explanation encompassing all, or many, relevant variables and providing linking statements, as discussed in the preceding section on precision, is considered more powerful than an explanation which involves few variables inarticulately expressed and relies heavily on the clause, "other things being equal."

Reliability. Reliability refers to the frequency with which factors not included in the explanation interrupt the situation the explanation concerns. It would be unusual of course to have complete reliability in marketing and the behavioral sciences in general. Reliability must be considered as a relative thing and not in absolute terms. In some ways reliability is related to degree of precision. We can say that a certain behavior is accounted for -- or explained by -- the life style of the actor. This statement is rather reliable given its vague all-inclusiveness but not very precise; certainly not as precise as breaking life style down into principal component parts and explaining behavior in terms of these parts.

PREDICTION

In the social sciences prediction refers to a stated expectation about a particular aspect of behavior or some particular set of circumstances which may be verified by subsequent observation.

Prediction is generally used in two ways. First, it is often used "for making deductions from known to unknown events within a conceptually static system" (Schuessler, 1968). An example is the use of regression analysis to estimate one variable from one or more concurrent variables- This is a frequent practice in using personality variables to "predict" buyer behavior or to use discriminant function analysis to predict adopter categories on the basis of psychosocial and demographic variables.

A second use of the term involves making assertions about future outcomes on the basis of recurring sequences of events. Sometimes, such statements are time bound, as in predicting for a given period. This is a form of prediction frequently labelled forecasting. Another form of prediction common to marketing is test marketing, which assumes that processes occurring in the test market will be "re-occurring not only in the test market but in comparable other areas, so that the recurring sequence of events in the test market adequately predict an outcome in the relative future in other contexts.

Prediction appears to be a special form of explanation, future-oriented (genetic) explanation, although we may have the paradox of being able to predict without explaining and explaining without being able to predict (Kaplan, 1964: 346-349; Bunge, 1967b; Rescher, 1970:30-36; Hempel and Oppenheim, 1948:138). However, "the logical structure of scientific (prediction) is . . . the same as that of explanation, namely, a deductive inference from laws and data' (Bunge, 1967b:69)

"The test then is to get (any explanation) to use the deductive technique of explanation on a future event; that is, to take the risk of being proved wrong by making a prediction. The argument involved in making a prediction is a slight modification of that involved in an explanation. In a prediction the conditions -- some antecedent event or events -- are the starting point, working through a law to the predicted event, whereas in an explanation the explanandum is the starting point, and only such conditions are involved as to make its following from the law plausible" (Caws, 1965:94).

Scientific prediction is forecasting with the help of scientific or technological theories and data. It is a statement or set of statements (1) whose premises are true, (2) which contains data statements which are true but refer to times no later than the present, and (3) which relates to the relative future. All scientific prediction involves some theory from which deductions are possible and some factual evidence relevant to the propositions of the theory.

A good example of scientific prediction in consumer behavior is the use of attitudes, or more strictly empirical measures of attitudes, to predict buyer behavior. The general principle involved is a psychological one concerning the striving for consistency between attitudes and behavior. The empirical data statements may be past or current information describing how the attitudes and buyer behavior are related. On the basis of the general theory and data statements conclusions are drawn -- deduced -- about some state in the relative future. Cognitive consistency models and market data have also been used to predict brand preferences and actual buyer behavior.

Levels of Prediction

As in the case of explanation, there are also different levels in prediction. These levels correspond to those shown in Table 1 except that they are future oriented. Consider an example of different levels of prediction derived from theoretical models discussed by Rogers (1969), Lerner (1958) and others concerning innovation in developing societies. The basic theory involves four interrelated concepts: literacy, cosmopoliteness, empathy, and innovativeness.

At level one the simple statement is that in the relative future more innovative social units (society, group, individual) will exist. The social unit is the phenomenon of concern.

The second level of prediction is that more innovative social units, Q, will exist as a result of increases in literacy, cosmopoliteness, and empathy. These three variables are the Factors xl, x2 etc , referred to in Figure 1. Q is the nature of the phenomenon.

At the third level of prediction we say that the three factors have interacted in the past, and/or are now interacting and/or will interact in the future in such a manner as to produce a change in innovativeness- The "in such a manner" is the focus of prediction at this level. For example, we might predict the following manner or type of interaction between literacy and empathy: higher degrees of mastery over symbols in written form (literacy), widen experiences through meaningful encounters with the print media, and unlock mental abilities, allowing the social unit to encompass these new experiences- This, in turn, increases the ability to project oneself into the role or situation of another. Such identification is a direct antecedent of being innovative (Barnett, 1953).

At level four, emphasis is on the reasons why (w;) literacy and empathy (and the other factor combinations, e.g., literacy-cosmopoliteness, etc.) interact in the manner (yg) just described and why empathy and the other variables may produce innovativeness directly independent of the impact of interaction effects. The reason why empathy affects innovativeness may be found in the concept of vicarious modelling. The reason for the literacy-empathy relationship may be found in contemporary theories of creativity. Thus, the prediction that Q is of a specified nature ultimately rests on the prediction that vicarious modelling, certain creative processes, and other reasons have been or will be in operation. The reasons just cited are themselves subject to analysis at each of the four levels of explanation and prediction.

Evaluating Predictions

Scope. The scope of a prediction refers to the range of events that it covers. The range of events may be viewed both longitudinally and latitudinally. If there is a long chain of events which must take place before a given phenomenon will occur and there are no other causal events capable of producing that phenomenon then a prediction of that phenomenon must encompass or include accurate prediction of the state of events in the chain. Scope of prediction may also be viewed laterally, i.e., in terms of the number of final events the predictive explanation covers.

Precision. The precision of a prediction refers to the degree of isomorphism between the concepts involved in the prediction and their empirical indicators. In addition to the important quality of isomorphism, precision is also affected by the accuracy of the empirical indicators, i.e., the quality of measurements with which the prediction is made. This is a methodological problem of some concern. As with nonpredictive explanations, the precision of the statement describing the rules of interaction among system variables is also important. This is very relevant to the problem of scope. The greater the longitudinal scope of a prediction the greater the degree of precision required in stating the relationships among the X's and between X's and Q's. As Stinchcombe (196& 129) has observed, "the elegance and power of an explanation can only be as good as the causal connection among variables allows it to be." This is also a reminder that statements of relationships between variables should not only be precise but be causal as well.

Power. Power in precision is a function of the precision of the predictive statement and its completeness. The more precise and encompassing of variables a predictive explanation is, the more powerful it is. It is more powerful in the sense that it provides a greater opportunity to control, and to control more accurately, those variables amenable to control or external manipulation by the researcher or planner.

One problem affecting that aspect of power involving completeness is the issue of prediction as feedback: a stated prediction may affect its own fulfillment. This is largely relevant to social rather than physical contexts- A promotional campaign stressing the virtues of a new non-prescription drug may, through the power of suggestion, cause people to actually experience relief solely on the basis of the prediction that it would provide relief. Studies on placebo effects provide the scientific grounding for such a statement. However, the feedback could be negative if excessive unrealistic expectations were built in the consumer's mind which the product could not match in actual use.

Reliability. Reliability in prediction concerns the frequency with which factors not included in the predictive explanation interfere with the predicted phenomenon, i.e., cause it not to happen exactly as the explanation predicted. To some extent there can be a trade-off between reliability and precision. The less precise the prediction or, alternatively, the more vague it is, the greater the degree of fluctuation allowed in the phenomenon Q- It is up to the researcher to determine the final trade-off point.

CONTROL

Explanation serves a twofold function: (1) to satisfy the human need to anticipate events (prediction), and (2) to be able to control future events. It is this latter phenomenon, control, which is of concern here. Many scientists claim that control is the central factor in the scientific enterprise (Rychlak, 1968). Even so, the philosophy of science has relatively little to say about a metatheory of applied behavioral science. Yet, the control function of scientific knowledge is of such great importance in applied activities such as marketing that it is essential to investigate the various criteria and guidelines for exercising control and evaluating control efforts.

Definition of Control

Control is the systematic manipulation of some element related to or contained within a system so as to effect a change in one or more elements in that system. "A system is an entity which is composed of at least two elements and a relation that holds between each of its elements and at least one other element in the set" (Ackoff, 1971:662). Control over a particular event is achieved if the relations specified in the explanation may be manipulated; manipulation of relationships requires manipulation of variables. The definition of control also specifies that an external state, i.e. the environment of a system, may be a source of causal forces. The environment of a system is defined as "a set of elements and their relevant properties, which elements are not part of the system but a change in any of which can produce a change in the state of the system" (Ackoff, 1971:662-63)- Control over an event is achieved if endogenous and/or exogenous variables contained in an explanation are manipulated so that a desired result is obtained.

Levels of Understanding in Control

Following the structure of Table 1, four different levels of understanding in the control process can be defined. At the first level there is the simple identification of the criterion variable(s) and the noting that a certain condition can be wrought by the change agent. The second level involves the identification of manipulable causal factors (the x's) capable of bringing about the phenomenon Q. The third level represents a still higher level of knowledge about the control situation. It identifies the strategies and tactics (the y's) to be used to alter existing relational patterns among x's and between x's and the condition out of which Q is to emerge. The final level is the specification of the reasons (w's) why strategy y, etc. can affect the relations among x's and between x's and the criterion variables- This involves identifying facilitating factors in the system or the system's environment

At Level 1 the criterion variable is identified- In the illustration used here this variable will be product quality perception. It is ascertained that this is a manipulable variable, i.e., within the deliberate control of the change agent. The next level of understanding permits the marketer to say that the product will have a perceived quality Q as the result of his manipulating price and channel of distribution (Stafford and Enis, 1969:456-58; McConnell, 1968:300-03). At level three the marketer knows through adequate research that, for his product class and market segment, there is a certain price range within which there is a relatively high positive association between price and perceived product quality. Presumably, consumers reason that a high price is due to high workmanship and/or more durable and functional features, thus enhancing the objective quality of the product. There may also be an interaction between causal factors having an additional effect on perceived product quality. The imputed relationship between price and quality and the added causal impact of the interaction effect between price and store images (Stafford and Enis, 1969) would presumably lead to a strategy of relatively high price and distribution through outlets having favorable images. At a still higher level of understanding he would base this action on certain reasons (wl, w2 ... w3)such as the income level of the relevant market segment, the importance of quality, etc.

Translating Practical Problems into Theoretical Issues

Argyris (1970) argues for a deductive approach to control problems although he does not explicitly label hi s approach as such. The reader should recall that the basic deductive model is one in which an explanansis sought for an explanandum under certain necessary constraints. In parallel fashion Argyris sees merit in translating practical problems into theoretical issues. The client's problems become empirical illustrations of more general theories from which the empirical problem phenomena can be deduced.

By translating the client's problems into theoretical issues, the problems may be analyzed using a wealth of concepts and findings in other settings which may be generalizable to the specific situation of the client. Without having first translated the particular problems into theoretical issues, the new suggestions and insights provided for control strategies might otherwise be missed. Marketing has benefitted considerably from this process. For example, the translation of brand selection and loyalty problems into learning theory and small group theory issues has proved very fruitful and the recasting of salesman-prospect relations into an exchange theory issue has provided new guidelines for the recruitment, training, and assignment of salesmen.

The process involved in the iteration between practical problems and theoretical issues is shown in Figure 1. Given a marketing problem the first step is to translate it into a more abstract theoretical issue as defined by the current state of knowledge in the discipline(s) contributing information to the particular issue- Exploration of the theoretical issue at a theoretical level can contribute insight and marketing strategy clues directly to the marketing problem. The marketing problem is viewed simply as an explanandum, i.e., as an empirical manifestation of the theoretical issue. Additional insight and guidelines may be derived by examining the implications of a theory in its application in areas traditionally considered as nonmarketing settings.

FIGURE 1

ITERATION BETWEEN PROBLEMS AND THEORETICAL ISSUES

Categories of Evaluation of Control Efforts

Effort. Five categories of criteria for evaluating the success or failure of a control performance have been suggested in the literature (Suchman, 1967:6073). The first category concerns effort. This involves what was done and how well it was done and uses such criteria as the quantity and quality of activity occurring. Emphasis is on input rather than output- This is one of the easiest evaluative tasks. The number of dollars invested in advertising is easy to assess in detailed ways, e.g., by market segment, by media, etc. The approximate number of consumer exposures to advertising can also be known with a high degree of accuracy.

Performance. Performance criteria relate to measuring the results of efforts rather than the effort itself. Such questions as the following are asked: What changes occurred? Were these the intended changes? Was it of the desired magnitude? Did the advertising create positive images? Did it reach the intended audience? What was the purchase response rate? Given that many programs involve a hierarchy of objectives, performance criteria can be applied at each level of objective

Adequacy. The adequacy of a performance given the total need is another important criterion. A promotional campaign intended to precipitate trial of a product can hardly be adjudged adequate if it only succeeds in stimulating interest. It is less adequate still if it only stimulates awareness and is least adequate when it only reminds consumers of the product's existence. Both exposure and impact must be considered as essential elements of adequacy. Bigman (1961:113) notes: We must distinguish between effectiveness and impact. By the latter term I mean the strength of the influence upon exposed individuals. A program or activity may have considerable impact, affecting markedly the thoughts and actions of those it touches, it will be necessarily judged ineffective if it is so designed that this impact is confined to a small fraction of the group it is intended to reach and influence. "

Efficiency. The next criterion is efficiency. It is "concerned with the evaluation of alternative paths or methods in terms of costs ... In a sense. it represents a ratio between effort and performance -- output divided by input (Suchman, 1967 : 64) . This is one of the central features of operations research. Operations research concepts and techniques have been applied to marketing quite successfully and need not be elaborated upon here

Process. Finally, we have the criterion of process. This involves the analysis of the means whereby a program achieves whatever effects it may have. It calls for an overview and analysis of the impact of particular sequences in the control program. It is concerned, in other words, with the overall program and with the interaction among parts of the program or elements in the system. Are particular components interacting in such a way as to produce dysfunctional effects? Are there bottlenecks in the process?

CONCEPTS

The concepts used in consumer behavior research can be ordered according to their closeness to the realm of observation. Howard and Sheth explicitly use this dimension to classify their concepts into hypothetical constructs and intervening variables (Howard and Sheth, 1969; Sheth, 1967; MacCorquodale and Meehl, 1948). These two types of concepts can be complemented by the type of concepts which serve as definiens of the intervening variables, i.e. observational concepts. Hence we have: (1) observational concepts, (2) intervening variables, and (3) hypothetical constructs (Hempel, 1952; Carnap, 1956; Hesse, 1967) . We can distinguish two subtypes of concept-type three, namely isolated hypothetical constructs and theoretical concepts. Isolated hypothetical constructs are constructs that are somewhat removed from the observational realm or plane, i.e., they have acquired some "surplus meaning" and are not exhaustible by and translatable into intervening variables, nor are they included in a "nomological network" (Cronbach and Meehl, 1955) as are theoretical concepts. They are relatively "isolated" (Sheth, 1967) and not explicitly related to other concepts of the same kind. The distinction between these three (four) types of concepts has several implications. Two important dimensions, namely the type of definition and the criteria imposed on each type will be dealt with here.

Types of Definitions

Observational concepts are defined by ostensive definition. Ostensive definitions are object-term associations (rather than term-term associations). An object-term association relates a term to an object, while a term-term association is a definitional relationship between terms. Intervening variables are defined by nominal definition in terms of observational concepts, i.e. by operational definitions. Isolated hypothetical constructs do not receive definition but "illustrations in use," i.e. their meaning is clarified by the various contexts in which they are being used (Caws, 1965). The theoretical concepts, finally, are defined by setting forth the relationships these concepts form with other concepts of the same kind. The latter procedure is called "theoretical definition" (Hempel, 1952; Caws, 1965).

This nice distinction between concepts is blurred by the fact that the same term may be used for expressing concepts that are at different observational levels. As an example, consider the multiple use of the term "attitude" by Howard and Sheth. Two answers are possible to the question "What is attitude?" The first answer is: "Attitude is the numerated response to a set of bi-polar scales." This is an operational definition of "attitude" and hence, attitude is construed as an intervening variable. The second possible answer is:

"'Attitude' is a function of 'choice criteria' and 'brand comprehension,' and it feeds into 'intention'." Evidently this does not tell us how to measure attitude. Rather, it tells us how attitude is related to other concepts (Howard and Sheth, 1969) .

Even more disturbing than the multi-level definition of concepts is the fact that, particularly at the level of intervening variables, different operational definitions bear the same name. Examples of concepts displaying such disorderliness are brand loyalty (Engel, Kollat, Blackwell, 1968), attitudes (Summers, 1970) and innovators (Robertson, 1971 ; Rogers, 19622. One possible remedy to this situation is to call for "standard definitions' (Kollat, Engel, Blackwell, 1970). Another alternative is increased methodological research that preserves the necessary variety of operational measures while investigating their common substantive basis (Webb, Campbell, Schwartz, Sechrest, 1966: Campbell, 1969).

Types of Criteria for Concepts

Now let us look at the various criteria that arise with respect to each type of concept. The principal criteria that are applicable to observational terms are: (1) determinacy, i.e., is their use well determined for every user of the language? and (2) uniformity of usage, i.e., are the conditions of usage the same for all users? (Hempel, 1952; Mandler and Kassen, 1959) . As to the intervening variables, the criterion of reliability is most applicable (Bohrnstedt, 1970). This criterion is intimately related to the criterion for observational concepts. Together, they serve to guarantee the inter-subjective character of science (Popper, 1959).

These criteria, however, tell us nothing about whether our concepts are true in some sense, or useful with respect to explanation, prediction, and control, the three major purposes of our scientific endeavor- "Beer" may be a highly reliable concept, but its value for explaining, predicting, and controlling consumer behavior would appear to be minimal. Criteria that do deal with the relevance of concepts for the three mentioned purposes are called criteria for concept validity

The first such criterion applies to the intervening variables, and it can be called "practical validity" (Campbell, 1960), or "criterion-oriented validity" (Cronbach and Meehl, 1955). Here one's primary interest does not concern the concept in question but some other concept -- the criterion -- which one wants to predict. The concept's ability to predict the criterion is called its practical validity. Indices of practical validity include correlation and regression coefficients, and R2. The next type of concept is the trait concept. ["Trait" corresponds to "isolated hypothetical construct" as introduced previously- We decided to use the term trait at this point because most of the relevant methodological literature available is in psychology and uses this term. We do not, however, want the reader to associate the substantive connotations just the methodological.] To it the criterion of trait-validity is applicable (Campbell, 1960). Trait validity consists of convergent and discriminant validity (Campbell and Fiske, 1959; Campbell, 1960) . One has to show that the concept can be observed in more than one situation (using more than one method of measurement) and that it can be meaningfully differentiated from other, similar concepts. As an example of the lack of convergent validity, consider the concept of innovativeness. Attempts to demonstrate the existence of "general" innovators, i.e. consumers having the characteristic of innovativeness in varying contexts, have apparently failed. An area of consumer research in which discriminant validation would be desirable are the various consumer typologies that have been proposed.

Another criterion applies to the hypothetical constructs- This is the criterion of construct validity (Cronbach and Meehl, 1955; Bechtoldt, 1959; Campbell, 1960; Bohnstedt, 1970). Strictly speaking, it is not a criterion for a concept but for an inference. Construct validity is required whenever we want to establish a link between a hypothetical construct and its operational definition. To go back to our example of the two definitions of attitude, one theoretical and the other operational, we need a statement saying "attitude' measures 'attitude'." Otherwise, our theoretical concept may lack the crucial criterion of empirical significance (Carnap, 1956). For if theoretical concepts were only related to other theoretical concepts, neither of which were linked to intervening variables by means of correspondence rules, the system would serve no purpose whatsoever with respect to explanation, prediction, and control. The existence of a correspondence rule is a sufficient but not necessary condition for the empirical significance of theoretical concepts. It is possible to measure a concept indirectly, i.e. relate it to theoretical concepts which do possess correspondence rules. An example of this is indirect methods of attitude measurement. To illustrate, take the measurement of attitudes through biases on an information test (Kidder and Campbell, 1970)- In this case the two theoretical concepts are "attitudes" and 'beliefs," and they are related by an hypothesis claiming that the direction and strength of an attitude toward a social object have systematic effects on the beliefs concerning the object. The theoretical concept "belief system" is now linked directly to an information test. And the distortions on the information test then give us an indirect measure of the direction and strength of the underlying attitude.

PROPOSITIONS

Propositions establish relations between concepts. We can order them along various dimensions First of all, the relation between them may be an empirical or a non-empirical relation (Rozeboom, 1956). The latter is simplified by nominal definitions of intervening variables. The propositions, "Attitude is the response to a set of bipolar scales," or "brand loyalty is the proportion of a household's product purchases devoted to the most frequently purchased brand" are nonempirical relations. Another potential instance of such relations may be the correspondence rules. "Attitude1" is "Attitude2" for example may conceivably be construed to be a nonempirical relation between two symbols. This, however, is not a generally accepted position (see Hesse, 1967; Hesse, 1970).

The other types of propositions are empirical propositions, i.e., subject to testing and falsification. Examples of these propositions are: "innovators are more cosmopolite than non-innovators," "attitudes influence beliefs," and "attitudes are measured by attitude tests." Note that the three preceding propositions, while being all of an empirical character, have differing observational status. And this dimension, i-e., the degree of observability of the constituent concepts of propositions, can be used as a dimension for ordering the empirical propositions. First of all, there are those propositions whose constituent concepts are observational concepts or intervening variables. Secondly, there are propositions of a mixed character, namely the correspondence rules. They contain both observational and non-observational concepts. Finally, there are propositions containing only non-observational concepts.

A third dimension of propositions, in addition to the two already mentioned (empirical status, observational status) can be discussed, namely the generality of a proposition. Here, we can distinguish between propositions whose universe of discourse is an infinite set and those which open to a finite set. The former are called "universal statements" (Popper, 1959). Universal propositions make assertions that refer to an unlimited number of cases. For example, the proposition: "Innovators are more cosmopolite than non-innovators," if construed as a universal statement, refers to all innovators of the past, the present, and the future, and regardless of their geographical location. In contrast to this proposition, "A is an Innovator," or "The innovators in our sample are cosmopolite' are singular statements. Their universe of discourse is a finite set, and even statements like the latter one are only conjunctions of singular statements- A fourth dimension, finally, distinguishes between all-statements and existential statements (Popper, 1959). All-statements are of the form: "All innovators are cosmopolite," while existential statements are of the form: "There are innovators who are cosmopolite."

Let us now use the preceding classification to characterize the empirical propositions of consumer behavior research. We want to call hypotheses only those propositions containing theoretical concepts. Empirical generalizations are propositions containing only observational concepts or intervening variables. Correspondence rules, finally, contain both theoretical concepts and observational concepts or intervening variables. All three kinds of propositions, in addition, are universal propositions. That is to say, they make a(n) (explicit or implicit) claim for universal spatio-temporal applicability. Thus, empirical generalizations concerning the cosmopoliteness of innovators are, a priori, assumed to be applicable anywhere and at any time -- until proven otherwise. The same holds true for hypotheses and correspondence rules.

Testability and Confirmation

An important characteristic of hypotheses, correspondence rules, and empirical generalizations is their testability. Each of these types of propositions is testable when it is possible to derive from them implications for the form "if conditions C are realized, then outcome B will occur," (Hempel, 1966). It is not necessary that conditions C be realized or technologically realizable at the time when the proposition is propounded. (For example, the proposition "Consumers inhabiting the moon are more innovative than consumers inhabiting the earth" is testable but 'technologically' not realizable at this point in time.) While the propositions to be tested are universal statements, test implications are singular statements- Testability is not much of a problem for empirical generalizations. Since they contain only observational concepts or intervening variables, they are immediately testable with the help of a singular statement which provides the "initial conditions" (Popper, 1959). Hence, the statement All innovators are cosmopolite needs only the singular statement "A is an innovator in order to imply the test-implication "A is cosmopolite," also a singular statement. The testing of hypotheses is somewhat more complicated- Here we need correspondence rules in addition to the initial conditions. For example, if we want to test the hypothesis, "The greater the congruence of the self with the brand image, the more positive is the attitude toward the brand" we need the following two correspondence rules: "instrument X is a measure of congruence," "instrument Y is a measure of brand attitude." Only then can we, together with initial conditions such as "the congruence is high" or "the congruence is low," derive test implications. If our hypothesis has been stated in quantitative form, i-e. in form of an equation, for example, it will yield an infinite number of test implications such as "If congruence is .9 (= condition C), then attitude will be .9 (= outcome B)," or "If congruence is .35, then attitude will be .41," etc.

Another important characteristic of hypotheses, empirical generalizations, and correspondence rules is their confirmability. At the outset, we must say that there is a fundamental difference in the confirmatory aspects of those propositions construed as all-statements and those which make only existential claims. All-propositions can never be conclusively confirmed (or verified), while existential propositions can never be disconfirmed (or falsified) (Popper, 1959; Hempel, 1967). To illustrate the former claim, consider again the proposition that all innovators are cosmopolite. Since this proposition refers to an infinite set, we will never be able to find out whether the unexamined cases (still an infinite number) conform to our proposition. As far as the impossibility of falsifying existential claims is concerned, a similar reasoning applies. For example, if a study designed to test the proposition that "there are opinion leaders within industrial firms ..." (Martilla, 1971) failed to discover opinion leaders, one could argue that within the remaining cases (infinitely many) some of these surely will be opinion leaders. It seems that universal existential propositions are in a good position, quite in contrast to universal all propositions. The latter can never be confirmed. To disconfirm them, however, it suffices to find one simple negative instance.

THEORIES

Theories are sets of propositions. They can be likened to networks. The concepts are represented by the knots, while the threads connecting the latter correspond, in part, to definitions, empirical generalizations, correspondence rules, and hypotheses- This network floats above the plane of observation and is anchored to it by the observational concepts (Hempel, 1957; Quine, 1953; Hesse, 1970). It i s usually required that the network contain non-observational concepts and hypotheses in order to qualify as a theory (Nagel, 1961; Hesse, 1967). The network analogy implies a certain systemiticity of theories. This systemiticity is given by the relation of deducibility between parts of the network. Hypotheses, together with correspondence rules, serve as premises for the deduction of empirical generalization- Thus, the relation between hypotheses and empirical generalization is somewhat analogous to that of the latter with the data or facts (Hesse, 1967). As an example, consider the following hypothesis: "When a response is followed by a reward, the probability of its recurrence increases-" To this, we can add the correspondence rules: "brand purchase is a response," 'brand satisfaction is a reward,," "the probability of recurrence of brand purchase is measured by brand loyalty. ' From these three correspondence rules plus the hypothesis, we can deduce the empirical generalization "the brand loyalty increases with an increase in brand satisfaction." From this proposition we can now deduce test implications, i.e. singular statements that refer to directly observable facts, as shown in the previous section.

Confirmation of Theories

We have previously discussed the confirmability of propositions and come to some clearcut conclusions concerning their verifiability or falsifiability. In reality, the situation is much messier than that, primarily because of the "fallacy of affirming the consequent." The example of innovators which was used previously to demonstrate the impossibility of confirming universal propositions was relatively harmless, mainly because we were stating only a correlation rather than making a causal claim- Now let us give an example involving a hypothesis with causal ambitions. Suppose we want to test the hypothesis that "messages that are congruent with the values of the audience lead to positive attitude change-" We can derive and test the implication "peer oriented messages will lead to positive attitude change with adolescents- ' The correspondence rules required to deduce this implication are: "Peer oriented messages are congruent with the values of adolescents," and "Attitude questionnaire X is a valid measure of attitudes at Time t and t+l." Now suppose that we obtain positive results Naturally, we would be inclined to say that our hypothesis was confirmed. Suppose, however, that there were an empirical generalization of the following sort: "The administration of attitude questionnaires at Time t leads to an increase in scores on the same questionnaire at time t+l." This proposition says that, in effect, the test results can be attributed to either hypothesis- Therefore, unless we control for the plausible rival hypothesis of test effects we cannot hope to validly test our value-congruence hypothesis (see CamPbell, 1969)

The preceding example showed that it is exceedingly difficult, if not impossible, to confirm a theory; for every seemingly confirming instance has a confirming effect not only on the theory of interest, but also on a large number of alternative theories which would have predicted essentially the same result. This raises the problem of how we can choose among competing theories. One of the choice-criteria which we may use is the evidential strength of theories. The problem of the evidential strength of a theory can be viewed as a relation between a body of actual or potential evidence, formulated in observation statements, and a theory (Hempel, 1967) The criterion of evidential strength, however, most often does not help us to decide among competing theories, because if they have some validity at all they usually come to similar predictions (McGuire, 1969). This dilemma has been thought to be remediable by means of crucial tests." In order to perform a crucial test one has to find a situation for which the two competing theories predict conflicting outcomes, i.e. one needs a test condition C for which "the first hypothesis (theory) yields the test implication 'If C then E1,' and the second hypothesis (theory) yields 'If C then E2,' where E1 and E2 are mutually exclusive outcomes- Performance of the appropriate test will then presumably refute one of the hypotheses (theories) and suggest the other." (Hempel, 1966:25-26) (term in parentheses added). The power of crucial tests, however, is somewhat diminished by the fact that strictly speaking it tells us only that one or more of the premises of the apparently refuted theory are wrong. But it does not tell us which one(s). Therefore, if we want to retain the hypothetical framework of the refuted theory we only need to claim that one or several of the correspondence rules were false, or even that a singular statement specifying the initial conditions was not accurate. Such a reasoning may be justified in a case where, for example, we have extended the theory to some previously ignored context for which we had to develop new correspondence rules that lack construct validation- In other situations, such an ad hoc procedure "rescues the theory from refutation only at the price of destroying, or at least lowering, its scientific status" (Popper, 1963, 1965:37)Generally speaking, the evidential strength of theories is not a decisive criterion for choosing among theories. The criterion of simplicity has been proposed as a complement if not an alternative to the criterion of evidential strength. It seems intuitively obvious that we will choose the simpler among competing theories having approximately equal evidential strength. Difficulties arise, however, when one tries to state clear criteria of simplicity (Hempel, 1966). This and many other problems still remain to be solved in the area of metatheory.

CONCLUSION

Above all else, the chief virtue of exploring the domain of metatheory is the resulting creation of a more constructively critical mental set for viewing scientific activity. Metatheory "... helps in raising fundamental scientific and philosophic questions, it helps asking them in the right ways, it discloses conceptual sickness and prescribes treatments for it, and it widens the horizon of research" (Bunge, 1959:26) These values can accrue to both the scientist researcher and the practitioner who relies upon the researcher as well as his own inductive and deductive cognitive processes- Applying metatheory to marketing and Particularly consumer behavior research produces a number of benefits whose end results can only mean increased sophistication and accuracy in marketing thought and practice. It is contended here, as suggested by Bunge, that the imposition of a metatheory thinking methodology on marketing research and practice will: (1) help detect, correct, and systematize previously unrealized inconsistencies in working assumptions; (2) help minimize such confusions as mistaking predictability and correlation for causation; (3) sensitize marketing men to the need to be more conceptually self exacting; e.g., to make explicit basic hypotheses assumed by cognitive propositions; (4) help sensitize researchers to the serious problems in attempting to measure quantitatively concepts which are inadequately elaborated. Examples of concepts which are sometimes poorly articulated are innovation, creativity, and cosmopoliteness. This problem is especially likely to occur since advances in quantitative skills often occur more rapidly than advances in qualitative knowledge; (5) keep the researcher and practitioner problem oriented, i.e., not allow methodology to replace or be confused with theory as is sometimes the case in the use of factor analysis where there is no discussion of relationships among variables in a cluster; and (6) stimulate the search for logical and rational accountings of phenomena made observable by advancements in methodological procedures.

It has been the authors' goal to leave the reader with an awareness of and sensitivity to a thinking methodology labeled metatheory involving the investigation, analysis, and description of various aspects of theory construction, the components of theory, the overall theory itself, and the use of theories- The desired consequence is a more constructively critical evaluation and more sophisticated work in the behavioral science aspects of marketing.

REFERENCES

Ackoff, R. L. Towards a system of systems concepts. Management Science, 1971, 17, 661-671.

Argyris, C. Interaction theory and method: a behavioral science view- Reading: Addison-Wesley, 1970.

Barnett, H. G. Innovation: the basis of cultural change. New York: McGraw Hill, 1953.

Bartels, R. Can marketing be a science? Journal of Marketing, 1951, 15, 319-328.

Bartels, R. The general theory of marketing- Journal of Marketing, 1968, 32, 29-33.

Bartels, R. Marketing theorY and metatheory. Homewood: Richard D. Irwin, 1970.

Baumol, N. J. On the role of marketing theory- Journal of Marketing, 1957, 21, 413-418.

Bechtoldt, H. P. Construct validity: a critique. American Psychologist, 1959, 14, 619-629.

Blalock, H. M. Jr. The measurement problem: a gap between the languages of theory and research. In H. M. Blalock, Jr. and A. B. Blalock (Eds.), Methodology In social research. New York: McGraw-Hill, 1968.

Blau, D. M. Exchange and power in social life. New York: John Wiley, 1964.

Bliss, D. How we can know more about marketing. In R. Cox, W. Alderson and S. J. Shapiro (Eds.), Theory In marketing. Homewood: Richard D. Irwin-, 1964.

Bigman, S. K. Evaluating the effectiveness of religious programs. Review of Religious Research, 1961, 2.

Bohrnstedt, G. W. Reliability and validity assessment in attitude measurement. In G. F. Summers (Ed.), Attitude measurement Chicago: Rand McNally, 1970.

Brown, R. Explanation In social science. Chicago: Aldine, 1963

Bunge, M. Metascientific queries. Springfield: Charles C. Thomas, 1959.

Bunge, M. Scientific research I: the search for system. Berlin: Springer, 1967.

Bunge, M. Scientific research II: the search for truth. Berlin: Springer, 1967b.

Buzzell, R. D. Is marketing a science? Harvard Business Review, 1963, 14.

Campbell, D. T. and Fiske, D. W. Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 1959, 56, 81-105.

Campbell, D. T. Recommendations for APA test standards regarding construct, trait, or discriminant validity. American Psychologist, 1960, 15, 546-553.

Campbell, D. T. Prospective: artifact and control- In R. Rosenthal and R. L. Rosnow (Eds.), Artifact In behavioral research- Academic Press, 1969.

Carnap, R. The methodological character of theoretical concepts. Minnesota Studies In the PhilosoPhy of Science 1, 1956.

Caws, P. The philosophy of science. Princeton, New Jersey: D. Van Nostrand Company, 1965.

Cronbach, L. J. and Meehl, P. E. Construct validity in psychological tests. Psychological Bulletin, 1955, 52.

Doby, J. T. Logic and levels of scientific explanation. In E. F. Borgatta (Ed.), Sociological methodology, 1969. San Francisco: Jossey-Gass, 1969.

Engel, J. F., Kollat, D. T., and Blackwell, R. D. Consumer behavior. New York: Holt, Rinehart and Winston, 1968.

Halbert, M. The meaning and sources of marketing theorY. New York: McGraw Hill, 1965.

Hall, E. T. The silent language. New York: Fawsett, 1961.

Harre, R. History of philosophy of science. In P. Edwards (Ed.) Encyclopedia of Philosophy. New York: Macmillan and The Free Press, 1967.

Hempel, C. G. and Oppenheim, P. Studies in the logic of explanation. Philosophy of Science, 1948, 15, 135-175.

Hempel, C. G. Fundamentals of concept formation in empirical science, volume 2, number 7, in International encyclopedia of Unified science. University of Chicago Press, 1952

Hempel, C. G. PhilosoPhy of natural science. Englewood Cliffs: Prentice Hall, 1966.

Hempel, C. G. Confirmation: qualitative aspects. In P Edwards (Ed.), Encyclopedia of philosoPhY. New York: Macmillan and The Free Press, 1967.

Hesse, M.Laws and theories. In P- Edwards (Ed.), Encyclopedia of philosophy. New York: Macmillan and The Free Press, 1967.

Hesse, M. Is there an independent observation language? In D. G. Colodny (Ed.), The nature and function of scientific theories- University of Pittsburgh Press, 1970.

Homans, G E. Social behavior: its elementary forms- New York: Harcourt, Brace, 1961.

Howard, J. A. Marketing theory- Boston: Allyn and Bacon, 1965.

Howard, J A. and Sheth, J. N The theory of buyer behavior- New York: John Wiley, 1969.

Hunt, S. D. The morphology of theory and the general theory of marketing. Journal of Marketing, 1971, 35, 65-68.

Kaplan, A. The conduct of inquire: methodology for behavioral science. San Francisco: Chandler, 1964.

Kidder, L. H. & Campbell, D. T. The indirect testing of social attitudes. In G. I. Summers (Ed.), Attitude measurement- Chicago: Rand McNally, 1970.

Kollat, D. T., Engel, J. F. & Blackwell, R. D. Current problems in consumer behavior research- Journal of Marketing Research, 1970, 7, 327-332.

Lerner, D. The passing of traditional society. New York: The Free Press, 1958.

MacCorquodale, K. & Meehl, P. E. On a distinction between hypothetical constructs and intervening variables. Psychological Review, 1948, 55, 95-107.

McConnel, J. D. The price-quality relationship in an experimental setting. Journal of Marketing Research, 1968, 5, 300-303.

McGuire, W. J. The nature of attitudes and attitude change. In G. Lindzey & E. Aronson (Eds.), Handbook of social Psychology Reading: Addison-Wesley, 1969

Mandler, G., & Kessen, W. The language of psychology. New York: John Wiley, 1959.

Martilla, J. A. Word-of-mouth communication in the industrial adoption process. Journal of Marketing Research, 1971, 8, 173-178.

Meehan, E. J. Explanation in social science: a system paradigm. Homewood: The Dorsey Press, 1968.

Morgenbesser, S Scientific explanation. In D. L. Sills (Ed.), International encyclopedia of the social sciences. Cromwell Collier and Macmillan, 1968.

Nagel E. The structure of science. New York: Harcourt, Brace, 1961.

National Research Council. The behavioral and social sciences- Prentice-Hall, 1969.

Popper, K R The logic of scientific discovery. New York: Science Editions, 1961.

Popper, K. R. Conjectures and refutations: the growth of scientific knowledge. New York: Harper & Row, 1965.

Quine, W. V. O. From a logical point of view. Cambridge: Harvard University Press, 1953.

Rescher, N. Scientific explanation. New York: The Free Press, 1970.

Robertson, T. S. Innovative behavior and communication. New York: Holt, Rinehart and Winston, 1971.

Rogers, E. M. Diffusion of innovations. New York: The Free Press, 1962.

Rogers, E. M. Modernization among peasants: the impact of communication. New York: Holt, Rinehart and Winston, 1969.

Rozeboom, W. W. Mediation variables in scientific theories. Psychological Review, 1956, 63, 249-264.

Rychlak, J. F. A philosophy of science for personality theory. Boston: Houghton Mifflin, 1968.

Schuessler, R. Prediction. In D. L. Sills (Ed.), International encyclopedia of the social sciences. Cromwell Collier and Macmillan, 1968.

Schwartz, G. Development of marketing theory. Cincinnati: South-Western, 1963.

Sheth, J. N. A review of buyer behavior. Management Science, 1967, 13, B718-B756.

Stafford, J. E. & Enis, B. M. The price-quality relationship: an extension. Journal of Marketing Research, 1969, 6, 456-458.

Stinchcombe, A. L. Constructing social theories. New York: Harcourt, Brace, 1968.

Suchman, E. A. Evaluative research. New York: Russel Sage, 1967.

Summers, G. F. (Ed.), Attitude measurement. Chicago: Rand, McNally, 1970.

Taylor, W. J. "Is marketing a science?" revisited. Journal of Marketing, 1965. 29, 49-53.

Webb, E. J., Campbell, D. T., Schwartz, R. D. & Sechrest, L. Unobtrusive measures: nonreactive research in the social sciences. Chicago: Rand, McNally, 1966.

Zaltman, G. Marketing inference in the behavioral sciences. Journal of Marketing. 1970, 34, 27-32.

 

RAP SESSION SUMMARY

ACR AS A RESEARCH STANDARDS BODY

Standards for Whom?

Although there was much discussion about the existing process for evaluating journal and conference papers, it was generally agreed that there was a way of providing a review of the scholarly work done by academics. However, there is a large amount of consumer behavior research which is done by groups whose professional interest is not in behavioral research but rather in influencing policy. It is this research which is more often picked up, misquoted, and misused by media and by policy makers. Consequently, it is this group for which standards seem to be most necessary at this time.

Methods of Relating Standards to the Task:

Four approaches to standard setting used by other groups were noted. The first is the ad hoc technical committee which reviews a given piece of research, and gives guidance on its adequacy for the decisions that called forth the research. The second is the recognition of researchers for the excellence of their research. The third approach is the establishment of some code or check list which deals with general research procedures. The fourth is a certifying body which tends to focus on the education of researchers and implies a certain amount of public damage done if ill prepared researchers are allowed to operate.

The conclusion of the group was fairly clear, supporting some variation of alternative one, the ad hoc technical committee, as being the only promising route for the present. The enforcement of any scheme is always a tricky matter. However, making available technical aid seems to be a safe move since its impact is only on those who want to use it.

Education - A Next Step:

Although it might be expected from a group primarily made up of educators, it seems that the ACR might well consider encouraging its members to make themselves available to local groups attempting to do consumer behavior research. Such involvement, although time consuming and potentially very frustrating, is appropriate to the ACR's informal organization and is at the most critical level.

It was suggested that we might explore the relationship between the ACR and such groups as the American Council on Consumer Interests, the Advertising Research Foundation, the American Marketing Association, and the like.

It was agreed that the topic of standards, particularly in the area noted, warranted some attention at subsequent meetings if we are to make good on our objective to stimulate research that focuses on a better understanding of consumer behavior.

----------------------------------------