Prelims

Focusing on Case Outcomes Rather than Variable Relationships

Arch G. Woodside, Coastal Carolina University, Conway, USA

This 2019 volume in the Advances in Business Marketing & Purchasing (ABMP) book series focuses on theoretical, empirical, and practical issues of trust. The papers in this ABMP volume contribute to the literature of business-to-business (B-to-B) relationship marketing by advancing knowledge, insights, and tools for understanding trust in B-to-B relationships and for learning how high trust versus distrust impact decisions. The papers in this volume embrace the theoretical stance that deep understanding of trust follows from examining the antecedents, mechanism, and outcomes of trust in specific contexts. Context research relating to trust includes cross-industries examination of advanced, emerging, and developing markets, culture, types of offerings, duration, and the stages of relationships (as well as additional dimensions) as antecedence conditions that two, three, or more persons engaging in communicating, acting, and assessing interpersonal and inter-organizational relationships. The benefits to the reader include a new appreciation of contextual influences and the mechanism of trust’s impacts on decisions affecting two or more individuals/organizations in relationship marketing.

Brown, Crosno, and Tong (2019) define trust as the belief that one’s channel partner can be relied on to fulfill its obligations and to behave in a benevolent manner. Rousseau, Sitkin, Burt, and Camerer (1998, p. 395) define “Trust is a psychological state comprising the intention to accept vulnerability based upon positive expectations of the intentions or behavior of another.” Morgan and Hunt (1994, p. 23) “conceptualize [high] trust as existing when one party has confidence in an exchange partner’s reliability and integrity.” This definition parallels that of Moorman, Deshpandé, and Zaltman (1993, p. 82) as “Trust is defined as a willingness to rely on an exchange partner in whom one has confidence.” Morgan and Hunt (1994) emphasize that both definitions draw on Rotter’s (1967, p. 651) classic view that trust is “a generalized expectancy held by an individual that the word of another […] can be relied on.”

The intention of this forward to the volume is to indicate a number of inherent weaknesses in the current dominate logic in constructing and testing theory relating to trust in marketing relationships as well as how to overcome these weaknesses. The inherent weaknesses in research on trust in relationship marketing occurs in almost all the theoretical and empirical studies in the discipline and includes the most cited study (i.e., Morgan & Hunt, 1994) and the more recent study by Brown et al. (2019). The inherent weaknesses arise because of applying symmetric theory and empirical tests relating to trust. For example, here are a few of the 12 hypotheses (H _i ) proposed by Morgan and Hunt (1994).

If Morgan and Hunt (1994) had proposed that the data in their study included different cases showing high trust and having both a negative and positive relationship with high scores among cases for the second variable in each of these four hypotheses, they would have found that a cross-tabulation of cases segmented into quintiles for each variable would support this perspective. Proposing and demonstrating symmetric directional relationships – as these four hypotheses propose – offers a shallow view of the causes and/or consequences of high trust. Moving away from symmetric theory construction and analysis is necessary to advance useful and accurate theory and empirical research of trust in relationship marketing.

The same observation holds for the study of antecedents, mechanism, and outcomes of distrust. An asymmetric stance is necessary because the causes, mechanisms, and consequences of distrust are likely to be dramatically different from those of high trust. Consequently, cases in a dataset will be observable where high distrust is associated with cases having high and low shared values – as well as cases with high and low scores with each of the other three variables in H5, H6, and H10, respectively, in Morgan and Hunt’s (1994) study. The issue of whether or not an overall relationship between X (e.g., trust as a dependent or independent variable) and Y (e.g., shared values, relationship commitment, or any other variable) is positive, negative, or close to zero has no importance. Beyond the conclusion made by the American Statistical Association (2016) that, “By itself, a p-value does not provide a good measure of evidence regarding a model or hypothesis” (Wasserstein & Lazar, 2016, p. 132) and the general conclusion that null hypothesis significance tests (NHST) is a corrupt research practice (Hubbard, 2016), the more relevant issue is when, not if, high (or low) trust is an ingredient in one or more complex antecedent configurations that consistently indicate cases having a specific outcome condition of interest. One should Ask, for what circumstances do cases with high trust contribute to cases resulting in high relationship commitment? Also, one should ask, for what circumstances do cases with high trust contribute to cases resulting in low relationship commitment? These two questions can and should be asked for distrust in the process of constructing and testing hypotheses. Thus, a “four-corner analysis” (Woodside, Nagy, & Megehee, 2018) is possible to theorize and test empirically.

Though widely practiced, the reporting of correlations and standardized partial regression coefficients (i.e., betas) in meta-analyses and accompanying significance tests (p-values) represent scant substance. Such research focuses on reporting the relative sizes of effect of each variable along with whether or not the effect size is different from zero. In reality, an influence for a condition can occur or be irrelevant no matter whether the correlation is in a two-variable relationship or the b-coefficient in a multiple regression analysis. The issue of substance lies inside the study of causal configurations of complex antecedent conditions that indicate a simple or complex outcome condition of interest. Again, telling that trust has a correlation with relationship commitment equal to 0.70 (p < 0.001) is not very informative given that high trust alone does not indicate consistency that high relationship commitment has occurred. Also, a researcher can expect – and should no longer ignore – anomaly cases that occur even when relationships empirically indicate a high effect size with high statistical significance. Discretizing by segmenting cases by quintiles or deciles and cross-tabulating almost always results in about 10 percent of the cases being classifiable as anomalies – even with the correlation between the two variables indicates a high effect size. While dichotomizing is never a good idea (cf. Cohen, 1983), discretizing by quintiles and building screens (McCampbell, 1998) or building screens by writing “fuzzy” statements via calibrating using 100 membership points for conditions (Ragin, 2008) rather than using continuous variables and symmetric tests is nearly always a better idea. Asking and answering the question – is there a positive relationship between trust and shared values – is bad science practice, along with asking similar symmetric directional questions for other variables. Good science practice includes asking under what circumstances do cases with high (low) trust indicate cases having high (low) shared values. Theory construction and empirical testing on trust in relationship marketing needs to shift from asking and answering variable relationship questions to asking and answering asymmetric case outcome questions.

The figure illustrates an example application of asymmetric case-based outcome theorizing and testing about trust in B-to-B relationship marketing. This study’s context is regarding managing relationships and purchase decisions by supermarkets buying committees and their suppliers of manufactured frequently purchased consumer brands (MFPBs). init is noted from the Figure that neither high trust nor the negation of high trust alone is sufficient for indicating accept or reject on a new product that a manufacturer brings to the supermarket buying committee. Actions/decisions in B-to-B relationship marketing depend upon configurations of complex antecedent conditions. Rather than demonstrating the rather obvious positive relationship between trust and acceptance, the empirical model in the figure does ask whether high trust leads to rejection for some cases. If yes, what are the circumstances when high trust indicates rejection? Also, the empirical model in the figure demonstrates that distrust leads to acceptance in some cases. If yes, this seeming anomaly begs the question – under what circumstances does distrust and accept occur?

“An anomaly is a fact that doesn’t fit received wisdom […] an anomaly marks an opportunity to learn something very valuable. In science, anomalies are the frontier, where the action is” (Rumelt, 2011, pp. 247–248). Most studies in behavioral sciences and the subdisciplines of business/management (e.g., accounting, finance, marketing, organizational behavior, and strategy) ignore anomalies in their testing of directionality of relationships (i.e., increases in X associates with increases in Y). These studies also fail to examine specific outcomes (e.g., firms with top-quintile profitability) and the antecedents to these outcomes – they focus on reporting precision in the directionality of relationships (e.g., p < 0.05) rather than constructing algorithms (i.e., screens) that accurately and consistently predict the occurrence of a given outcome.

This discussion supports two conclusions. First, the study of symmetrical variable-directional relationships and the reporting of small-to-large effect sizes with null hypothesis significance tests (NHST, e.g., r ² = 0.64, p < 0.001 for trust and acceptance) offers meager substance (for additional supporting details for this conclusion see Armstrong (2012) and Ziliak & McCloskey, 2008, 2009). This conclusion also applies to meta-analyses summarizing statistical effect sizes. Second, nowadays, researchers have tools available to enable them to shift from bad to good science practices by moving away from the study of symmetric variable-directional relationships and the use of NHST to study asymmetric screens (i.e., heuristics) to indicate specific outcomes consistently via somewhat precise outcome tests (SPOT).

Opens in a new window.

An Ethnographic Decision Process Model of Supermarket Committee Buying Decisions about a Manufacturer’s (M’s) New Product Offering.

The seventh paper elaborates on this asymmetric theory construction and empirical testing perspective for consistent outcome forecasting using screens and SPOT in great detail. The bottomline suggestion is to enjoy reading this volume in the ABMP series and consider embracing the perspective of shifting from the now pervasive perspective of theory construction and testing of symmetric directional variable relationships via NHST and effect sizes to asymmetric outcome configurational screens via somewhat precise outcome testing (SPOT) (Woodside, 2018).