پاسخ های پژوهشی مدیران به چه معنی است و چه چیزی آنها را تحت تاثیر قرار می دهد؟ در مورد بازار گرایی و عملکرد شرکت

Using cross-sectional surveys to support proposed causal sequences in theoretical models is problematic, especially when the informants are managers and performance is a dependent variable. The results of such surveys reveal more about managers sensemaking processes than actual business processes, a rival causal theory that researchers tend to ignore. This problem leads to unsound conclusions and management advice. We illustrate the argument in terms of survey research concerning the relationship between market orientation and firm performance. We show how sensemaking theory can account for existing results including some that otherwise lack explanation. The issues raised challenge many accepted interpretations of research results and have far-reaching implications in terms of the meaning and role of survey research. In addition, they point to important areas for future research.

Much research and model building in marketing and management is in danger of leading research and management astray because it fails to consider alternative models that fit the data equally well but which have quite different theoretical and causal implications and because it fails to take into account the impact of management sensemaking on measurement biases and feedback effects. Much reported research tends to have the following characteristics. (1) Based on prior literature and insight researchers propose a model to explain a causally dependent variable such as firm performance. (2) The model comprises a set of independent or explanatory variables that are assumed to be directly or indirectly causally antecedent to the dependent variable. (3) The models assume unidirectional causality, which is indicated via one-way arrows in model diagrams. More complex structural equation models can have layers of dependent and independent variables but they generally retain a unidirectional causal structure. (4) Surveys of managers’ perceptions and recall are used to measure the variables. (5) Various methods are used to fit the proposed model to the data collected. If the results show that the proposed model is consistent with the data it is regarded as supported and the proposed causal sequences are explicitly or implicitly accepted. Two issues cause problems with this approach, which researchers tend to ignore. First, the methods used do not provide a test of the causal directions assumed in the model because they are, for the most part, based on cross-sectional, correlational data. As we all know and teach, correlation does not equal causation, only association. Yet many models are interpreted and described, once model fit is established, in causal terms using causally loaded words such as affects, impacts, enhances, determines, produces, and leads to conclusions, such as: “Overall, the results demonstrate that market orientation has a positive impact on organizational performance” (Kirca et al., 2005, p.37). Only rarely are attempts made to test alternative causal sequences and feedback and interaction effects. The second issue is that alternative theory based, causal sequences among variables are ignored. Some of our reviewers argued that researchers are entitled to use such causal interpretations because these stem from our theories and the model fit is not inconsistent with the theory. But the causal directions proposed are not confirmed or tested by the model fit and it is dangerous to assume otherwise. Researchers need to consider alternative models that fit their data equally well, including alternative causal sequences among the variables and the potential existence of spurious links due to omitted variables. The relevance and importance of these issues are clearly shown by research carried out by Henley et al. (2006), who systematically retested 79 structural equation models reported in top management journals. They find that 75% of the models have at least one equivalent model, i.e., models that have the same structure or number of parameters, but which have radically different theoretical and practical implications. We ignore alternative causal explanations of our correlation matrices at our peril. If we ignore alternative possible explanations and theories we are not conforming to the norms of evidence based science but are engaged in a process that is more akin to advocacy. The theories we use are not usually so robust that alternative causal sequences can be easily dismissed. In many cases rival causal sequences have been proposed already in the literature. For example, there is the everlasting debate about whether attitude change drives behaviour change or the reverse and, in studies of inter-firm relations, the direction of the causal links between variables such as commitment, trust, conflict, communication and power can be and are argued to occur both ways in different researchers’ models. In truth, the causal direction is probably two-way, not one way, because the state of a relation in one period affects its state in the next period and hence so all dimensions of relations are interconnected over time. Such feedback mechanisms over time imply different types of causal sequences to those usually included in reported models because much research in marketing and management is static not dynamic. For example, models of firm performance do not allow performance to switch roles and become an independent variable. Yet March and Sutton (1997) describe a number of potential mechanisms by which performance at an earlier time can affect future performance, including investment, motivation and learning effects. In this article the focus of attention is on the role and importance of one particular type of feedback mechanism, management sensemaking (Weick, 2000), and the way this affects the meaning and interpretation of survey based research. Sensemaking refers to the way managers process incoming information over time and make sense of their organization, its environment and its performance. We will show that sensemaking processes affect the way managers perceive their firm and the way they answer questions about it. This, in turn, leads to rival causal explanations that also fit the data and which can help explain otherwise unexplained results (March and Sutton, 1997). The kinds of problems we identify are of general importance and relevance because they apply to any model tested using cross-sectional survey data of people’s perceptions and recall. They are especially important when the data comes from surveys of managers and performance is a key dependent variable because such models are used as the basis for proposing normative theories advising managers and policy makers about how to improve manager and firm performance and behaviour. Models with different causal sequences that explain the observed data equally well lead to different types of management advice in terms of which things to focus on to achieve a firm’s objectives. We need to face up to these issues and work out ways to better deal with them, rather than hiding them and pretending they do not exist. The purpose of this paper is to call attention to these problems, and to illustrate them in terms of research concerning the link between market orientation and firm performance. We show how an alternative, sensemaking theory, that is equally if not more strongly established, can account for the same research results as well as others that are hitherto not well explained. The sensemaking interpretation of results has rather different implications in terms of the nature and degree of support for theories of marketing’s impact on firm performance. It also leads to a number of important issues to accept and deal with in future research. In short, we argue that performance drives managers’ perceptions of their firm and its environment, including its market orientation and this affects how they answer survey questions and what these answers mean. Although our focus is, for the purposes of illustration, on research concerning market orientation and performance, we would like to emphasise that the issues raised are of general relevance and importance. We chose this research area because these concepts are of central importance to the marketing discipline, because extensive research has been carried out based mainly on surveys of managers and because most marketing academics are familiar with the research. Our purpose is not to criticise this research per se, or to argue that this research area is more guilty than others in terms of the issues we raise. Far from it, it is, we repeat, used only to highlight and illustrate the issues that are of much more general importance. We believe in the concept of market orientation and that marketing can and does play an important role in driving firms’ performance. What we do not believe is that the results of cross-sectional surveys of managers’ perceptions and recall support claims of a causal link between market orientation and performance. It has nothing to do with the reliability and validity of the measures we use, though this is of course does matter, it is about the way managers answer survey questions and what affects their answers. We could have focused on other dimensions of firms that have been used to explain performance and firm behaviour, such as research on technology, entrepreneurial, relational or international orientations or research on various types of firm competences such as innovative, relational and network competences. We are not the first to raise these issues. Although we developed our ideas independently, we take comfort in finding James March and Robert Sutton’s important 1997 paper in Organization Science ( March and Sutton, 1997). This reassured us that the issues are real and important and yet ignored. It also helped us to further develop our ideas. The marketing literature has largely ignored the March and Sutton paper. The paper is organized as follows. Section 2 reviews MO-Performance (MO-P) research and methodology, revealing some of the inconsistencies and problems that have arisen. Section 3 describes the problems that threaten the validity of cross-sectional survey research, illustrating them with examples drawn from MO-P research. Section 4 describes a psychological sensemaking perspective that can account for existing MO-P results as well as some unexpected and otherwise unexplained findings. Finally, in Section 5, we discuss possible solutions to the problems identified and future research opportunities.

Cross-sectional survey research has many strengths and obviously plays an important part in research in marketing, management and other disciplines (Rindfleisch et al., 2008). That this type of research is inappropriate is not the argument here, Instead, such surveys face a number of problems, especially when they are based on managers’ perceptions and recall and when performance is a dependent variable. Furthermore, these problems have been neglected in marketing and management research but they need attention, otherwise our interpretations of survey results can misdirect research efforts and lead to misleading advice to managers and policy-makers. As Das (2003) points out, “the inaccuracies of manager perceptions should be taken as a given,” and it is more important “to see where the inaccuracies arise from and what their consequences are in practice” (p.23). Managers engage in a constant process of trying to understand and make sense of what they know and believe about themselves and their firm and to reconcile this with their ongoing experience of firm behaviour and performance in a particular environment – what Weick (2000) describes as sensemaking. The process of sensemaking does not result in an accurate and unbiased view of the firm and its environment but one that is governed by various psychological principles related to the way managers seek and process information about their firms, the causal attributions they tend to make, and their desire for cognitive consistency (Ailawadi et al., 2004). Sensemaking theory suggests an alternative causal ordering of variables to that usually proposed. In the case of MO-Performance research it is a theory that is consistent with existing results and is even capable of explaining some results that are otherwise difficult to explain. The relevance and significance of sensemaking biases go way beyond the study of MO and performance, important though this is, because such biases force us to reassess how academics use and interpret the results of any survey of management perceptions and recall. The impact of such sensemaking biases is an important issue because explaining variations in firm performance has been the subject of many proposed models and research studies in marketing and management and most of these are tested using cross-sectional surveys of managers. In these studies firm performance is linked to various types of factors including firm orientations, resources and other characteristics and to environmental characteristics. The resource-based theory of the firm (Barney, 1991) and resource-advantage theory (Hunt and Morgan, 1996) are examples of general theories of this form. Examples of research on specific factors affecting firm performance include, technology orientation (Gatignon and Xuereb, 1997), production orientation (Pelham, 2000), learning orientation (Barker and Sinkula, 1999), entrepreneurial orientation (Matsnuno et al., 2002) and strategy type (Matsuno and Mentzer, 2000). In addition to research on firm performance, management perceptions and recall are used in a variety of other types of research. In each case management sensemaking processes offer another potential theoretical explanation for the research results and the causal links among variables in the model. One example is the study of organizational characteristics and how they are interrelated, including studies of organizational culture, innovation, information flows, growth and decision-making. A manager’s perceptions and recall of one type of organizational variable is likely to influence and be influenced by their perceptions of other types of variables thus introducing potential bias and other causal directions linking variables. Another example is the study of inter-organizational and buyer–seller relations that examine the effects of relation dimensions such as power, conflict, trust and commitment on each other and on other characteristics of relations such as longevity, cooperativeness, growth, performance and innovation. For this research managers from only one side of the relation tend to be interviewed and asked about both sides of the relationship. This results in additional sensemaking problems for the respondent as they use various types of information to determine the way a counterpart sees the relation. Other types of survey research potentially affected by management sensemaking include studies of firm strategic behaviour, supply chain management, customer relationship management, organizational buying behaviour and so on. But, no matter what the specific focus of the research, the potential impact of management sensemaking on the results cannot be avoided and should not be ignored. Given the problems and limits of cross-sectional survey-based research using management perceptions and recall, why do researchers persist with this type of research and why is it published? We are all aware that cross sectional surveys cannot determine causality and that research based on management perceptions and recall is subject to potential bias – this is not new. March and Sutton (1997) suggest an answer: “properties of the research context, rather than individual ignorance or journal incompetence, may be primary contributors to this curiosity” (p.702). There are barriers to carrying out the kind of longitudinal and detailed studies required to overcome the problems identified, including short term research funding, the need for publications for career advancement and what professional journals demand. The managerially focused environment in business schools also encourages academics to develop and promote the types of theories that managers want (Gray and Wilkinson, 2007), i.e., ways of predicting and controlling firm performance, even if the research support is equivocal. This leads to research that is believed and accepted more because it fits with researchers’ management orientation, not because of its research standards and attention to causal inferences. We could even see this as a reflection of researcher sensemaking, where the research culture serves to limit the attention given to rival hypotheses such as those resulting from management sensemaking.