# Marketing Education Review

theoretical concept. However, some of their measures appear to he predictions, such as "I expect my doctor to talk clearly...". Parasuraman, Zeithaml, and Berry (1988) define expectations as what consumers feel should be provided, which is congruent with their operational definition in which respondents were instructed to "show the extent to which you think firms offering ... services should possess the features described."

In marketing, practices on writing propositions and hypotheses vary considerably. Frequently, the type of state- ment that has been termed a "proposition" in this article is labeled a "hypothesis" in marketing articles, and the type of statement illustrated previously as a "testable hypothesis" is not stated. Rather, the testable hypothesis is given implicitly by describing the operational definitions of the independent and dependent variables. As an example. Brown and Swartz (1989) give the following statement, which they term a hypothesis: "The level of positive client evaluation of the professional service is inversely related to gap 1." The oper- ational definitions of the variables were given so it would be easy for a reader of Brown and Swartz (1989) to write out the testable hypothesis. The explicit statement of both a proposition and its testable hypothesis would add to the length of an article and is probably unnecessary. The value of writing a testable hypothesis is that it can assist researchers in checking their work for theory-setting and proposition-testable hypothesis congruency and help a researcher in conducting a critical evaluation of the litera- ture on the topic of interest.

Bridge Laws

Statements that tie the empirical setting to the theoretical statement and propositions are termed bridge laws and serve to close the gap between the theoretical level statement, with its theoretical definitions, and the setting-specific phe- nomenon, as expressed in a research hypothesis (Hunt 1991). Bridge laws are also termed guiding hypotheses or assumptions.

The setting and propositional concept to operational mea- sure step involves a series of bridge laws. Because the logi- cal test of an operational definition is homology with the propositional concept, it is appropriate to start with the propositional level definition of the concept and compare it with the operational definition (Dubin 1978). We use indus- trial salesperson job satisfaction with material rewards as an illustration. If material rewards were defined as resources with exchange value, then such items as salary, commission, bonuses, incentive awards such as an travel, and use of a company car would be included (see Table 1). A bridge law would be that the pay component of INDSALES was with- in the domain of the propositional level concept. The assumption would be reasonable, because by definition pay is a resource with exchange value. "Pay" would not be a complete measure of the concept and if additional measures, such as job satisfaction with incentive awards, could be obtained, other hypotheses could be formulated relating sat- isfaction with incentives to job search.

In summary, the definition of the concept in the theoreti- cal statement should be homologous with the definition in the proposition and the two definitions should be tied together by bridge laws. Next, the propositional level con-

cept and operational definition should be homologous. If a gap is found, an opportunity for an empirical study may be suggested. The next step in establishing theory to hypothe- sis consistency is to ensure that the form of the relationship between the concepts in the theoretical statement, proposi- tion, and hypothesis is consistent.

Consistency of Operational Linkage and Statistical Tests

The operational linkage specifies how the concepts in the theoretical statement are related (Hage 1972). The example posits that as extrinsic job satisfaction (X) decreases, seek- ing another job is likely to increase (Y). The operational linkage is negative and direct: As X decreases, Y increases. The same linkage should be found in the propositions and hypothesis.

At the hypothesis level, a statistical test that is homolo- gous with the operational linkage is needed, and measure- ment of the concepts becomes an issue. The INDSALES scale has been treated as an interval level measure (Churchill, Ford, and Walker 1974) and if the job search measure was also an interval measure, then a statistical test of association for interval measures such as regression would be appropriate.

A limitation of many studies in marketing is that the the- oretical statement specifies a direct relationship between variables—for example: "As X increases, Y increases"—and a linear model is used. Not all direct relationships are nec- essarily linear; as an example, a curvilinear relationship is direct. If a relationship was curvilinear but only linear regression was employed, empirical support for a hypothe- sis may have gone unrecognized.

How the variables are posited to be related theoretically should be congruent with the operational linkage, the rela- tionships created by the measurement process. Teas (1993) believes that Parasuraman, Zeithaml, and Berry (1988) had problems in that area. Essentially, Parasuraman, Zeithaml, and Berry (1988) specified that as Performance (P) minus Expectation (E) increases, service quality (SQ) increases. Teas (1993) noted that expectations were defined by Parasuraman, Zeithaml, and Berry as similar to ideal stan- dards. If so, the classical ideal point attitudinal model would be implied. However, in the ideal point model, performance above the ideal point could decrease SQ, so SQ would not always increase as P-E increases. Teas (1993) solved the problem by modeling SQ as an ideal point where SQ increased as performance came closer to the ideal point.

Domain Consistency

The domain of a theoretical statement is a specification of that part of the natural world in which the statement is expected to apply and the setting in which the hypotheses will be tested should be consistent with the domain. Dubin (1978) has presented an analytical approach to domain spec- ification by noting that both a theoretical model's concepts and linkages must be within their respective boundaries. He defines boundaries as that part of the theoretical and thus empirical world in which predictions can be made on the basis of the model. As an example, some boundaries were implicit in the sales force turnover example. Recall that the model of seeking another job in the same occupation was