تاثیر وضعیت خرید در ساختار مرکز و مشارکت خرید : یک متا آنالیز انتخاب از تحقیقات رفتار خرید سازمانی

Researchers investigating various aspects of organizational buying behavior often have reported mixed, sometimes contradictory results. Recently, there have been attempts to consolidate work in this area [cf. Bunn, M.D. 1993. Taxonomy of buying decision approaches. Journal of Marketing; 57:38–56 (January); Johnston, W.J., Lewin, J.E. Organizational buying behavior: toward an integrative framework. Journal of Business Research 1996; 35 (January): 1–15]. However, to date no one has attempted a meta-analytical integration of this research stream. To fill this gap, the authors conduct a meta-analysis of the relationships between the nature of the purchase situation and buying center structure and buying center involvement. While results indicate that cumulative findings in some cases are robust, in other cases they are not. Subsequent moderator analysis indicates that study design characteristics account for significant variation in findings across studies.

Over the past decades, researchers have studied the processes and behaviors used by organizations in their purchasing activities. Marketers, in particular, have encouraged these investigations as an aid to better understand, serve, and retain their organizational customers. As a result, scores of theoretical and empirical articles have examined the constructs associated with organizational buying behavior. Additionally, like most business activities, the environment surrounding organizational purchasing is dynamic—evolving with the emergence of new techniques and technologies (Cannon and Perrault, 1999). For example, in recent years new processes have been introduced to aid organizations in their procurement efforts. These aids include (1) Web catalogs offering product specification, price, and availability, (2) Internet-based ordering and tracking systems, (3) electronic data interchange to facilitate inventory control, credit approval, invoicing and receivables, and (4) direct communication and relationship management tools. These new activities and aids provide opportunities for new research in organizational purchasing. However, before conducting new research a meta-analysis of the existing literature could prove useful in several ways. For example, by quantitatively summarizing the accumulated body of knowledge systematic sources of differences can be identified (Farley and Lehmann, 1986), new work in the field can be more effectively designed (Farley et al., 1998), and certain unresolved questions may be either clarified or settled (Brown and Peterson, 1993). One of these unresolved questions is elucidated by McQuiston (1989) who points out that researchers have had mixed success examining participation and influence (involvement) in organizational purchasing. Additionally, McCabe (1987) observes that the literature continues to support apparently conflicting views of the relation between buying center structure and the nature of the purchase situation. Summarizing these issues, Ghingold and Wilson (1998) argue that generalizable conclusions regarding the dynamic nature of buying center structure and its implications remain cloudy at best. To fill this void and to address some of the inconsistencies currently existing in the literature, we conduct a select meta-analytical integration of organizational buying behavior research. More specifically, we examine the strength, significance, and generality of relationships between the nature of the purchase situation and (a) buying center structure and (b) buying center involvement (see Fig. 1). We also examine several study design characteristics that may moderate the relationships between these antecedent and outcome variables.

One benefit of meta-analysis is its ability to identify and/or clarify unresolved questions in an area of study. To that end, our study contributes in the following ways. (1) As previously discussed, over the years the buyclass schema has received mixed conceptual and empirical support, resulting in a number of critics. However, contrary to these criticisms, our results provide general support for buyclass as a predictor of buying center size (r′=.47), lateral involvement (r′=.26), vertical involvement (r′=.26), and relative influence (r′=.29)—all significant at p≤.05. At the same time, however, we also find support for the proposition that buyclass alone may be insufficient for understanding important distinctions in buying center structure and involvement. Instead, our findings point to purchase importance, and to a lesser extent purchase uncertainty and complexity, as equally compelling explanatory variables. (2) The results of our meta-analysis refute allegations that prior studies examining the effects of purchase situation on buying center involvement are inconclusive. Rather, overall our results indicate that findings surrounding the relationships between the nature of the purchase situation and lateral involvement, vertical involvement and relative influence in buying centers are generally robust across studies. (3) In contrast, our findings support the proposition that, in sum, prior works examining the effect of the purchase situation on buying center formalization and centralization are inconclusive. Within this meta-analysis, these variables generally exhibit the weakest correlations, with all relationships found to be either non- or only marginally significant. These findings, coupled with the concerns expressed by earlier researchers, point to the need for new, carefully designed research to investigate these important relationships. A second benefit of meta-analysis is the ability to potentially identify systematic sources of differences across studies within an examined body of knowledge. When such systematic sources are found, future work in the field can be more effectively designed to either eliminate or control for these potential biasing influences. In this regard, our study makes the following contributions. (1) In sum, our findings indicate the seven examined study design characteristics likely bias effect sizes systematically (regression r-squares ranging from .42 to .87), thus potentially creating confounds with explanatory variables. These findings are reflective of the concerns expressed by several other researchers (cf. Lynch, 1999) over the external validity of study results where background factors are assumed irrelevant. (2) Ofparticular interest are the following select findings related to the moderator analysis. (a) The inconsistency across studies examining the effect of purchase situation on buying center formalization appear to be caused, at least in part, by the number of firms represented in the studies and the method of data collection used. Our findings indicate that in studies examining three or more firms the strength of the relationship between purchase situation and buying center centralization were negatively impacted. One possible explanation is that buying center centralization is dependent on firm characteristics, and thus is difficult to generalize across firms. Similarly, data gathering via mail survey also weakened the relationship between these two constructs. One implication of this finding is the need for the use of multi-methods of data collection in future studies designed to measure the relationship between these two constructs. (b) With regard to inconsistent relationships between purchase situation and buying center formalization, the size of the sample and the type of respondent used, along with the number of industries included in the study all seem to significantly influence between study results. More specifically, the relationship between these two constructs is negatively influenced in studies with larger sample sizes and/or limited to purchasing respondents only, and positively influenced in studies where a larger number of industries are included. We believe that (1) the inconsistencies across findings related to the influence of the purchase situation on buying center centralization and formalization, combined with (2) the robust results of our moderator analysis, provide the most worthwhile directions for future research. First, our findings indicate that prior attempts to examine the relationship between the nature of the purchase situation and buying center structure are inconclusive. This may be due, in part, to divergent study design characteristics—including some of the characteristics examined here, as well as potential others. One way to test this in future research would be to specify/model expected background variables and then test for the asserted pattern of interactions. Another approach would be for researchers to test the robustness of findings by systematically varying one or more of the domains of their study through replication. An alternative explanation of our moderator results may lie in the inappropriate nature of prior assumptions. For example, prior studies assume a linear relationship between purchase situation and buying center structure. It is possible, however, that this relationship is nonlinear. Where, for example, levels of formalization and/or centralization increase as levels of purchase importance, complexity, and uncertainty increase to some critical level. Thereafter, levels of formalization and/or centralization begin to decrease as levels of purchase importance, complexity, and uncertainty increase further. Future research might examine these potential relationships.