Abstract
Previous research has conceptualized and modeled customer orientation (CO) in one of two ways: as a psychological phenomenon antecedent to critical job states (i.e., stress and engagement) or as frontline employee behaviors that are caused by these same job states. Building on meta-analytic data, this study finds greater support for the causal relationships implied by a psychological construal of the construct and reveals that CO influences frontline employees' job outcomes through its effects on stress and engagement. Moderation analyses also indicate that CO's influence on model variables is stronger when frontline employees' customer workloads increase and is weaker as the need for customer persuasion increases. These findings contradict widely held assumptions rooted in a behavioral view of CO—namely, that CO is a consequence of job states, a proximate determinant of job outcomes, and most beneficial when ample opportunity for customer engagement exists. Overall, the results support a broadened perspective that recognizes that CO improves job outcomes because it enhances frontline employees' psychological welfare in addition to being good for business. These findings suggest that managers should consider CO an important criterion in frontline employee decisions, recognize CO as beneficial when limited opportunity for customer engagement exists, and avoid efforts to curtail CO's costs at the frontline employee level.
Keywords
After reportedly trying to deal with an unruly passenger on a parked aircraft Monday, JetBlue Airways Corp. flight attendant Steven Slater used the plane's intercom system to curse out the passenger. He allegedly grabbed a beer from the beverage cart and deployed an emergency escape slide on the runway…. Anyone who has worked in customer service … can relate to Slater. On Tuesday, online message boards and social networks flooded with empathy for the man who gave new meaning to the term “jumping ship.”
Customer orientation, the manifestation of the marketing concept at the individual worker level (Saxe and Weitz 1982), has been the object of sustained interest for three decades. This interest has been fueled, in part, by the expectation that CO is a valuable resource that should positively influence important psychological (e.g., commitment) and job (e.g., performance) outcomes among frontline employees (e.g., Brown et al. 2002; Donavan, Brown, and Mowen 2004). However, some research has cast doubt on CO's influence on these outcomes, leaving valid questions about how CO operates to influence workers' job outcomes, when this influence is likely to be most pronounced, and its relevance to employee selection decisions and training efforts (e.g., Plouffe, Hulland, and Wachner 2009).
In terms of CO's relationship with critical psychological variables—including job stress (e.g., role ambiguity and role conflict) and job engagement (e.g., job satisfaction and organizational commitment)—researchers are strongly divided regarding the true direction of influence (Donavan, Brown, and Mowen 2004; Schwepker 2003). Specifically, scholars who conceptualize CO as a psychological (i.e., individual difference) variable have argued and found support for the proposition that CO decreases stress and enhances engagement among frontline employees (e.g., Harris, Mowen, and Brown 2005; Rod and Ashill 2010). In contrast, scholars who view CO as a set of worker behaviors have typically proposed and found support for the opposite causal order: Stress reduces and engagement increases frontline employees' CO (e.g., Bettencourt and Brown 2003; Kelley 1992).
Furthermore, inconsistent empirical evidence has led scholars to question the efficacy, or at least the universal efficacy, of CO as a determinant of frontline employee job outcomes (e.g., Franke and Park 2006; Homburg, Müller, and Klarmann 2011b). For example, CO has been shown to increase self-rated (e.g., Licata et al. 2003) and manager-rated (e.g., Grizzle et al. 2009) performance and decrease employees' propensity to leave the organization (e.g., Babakus, Yavas, and Ashill 2009). Nonetheless, an approximately equal number of studies have also concluded that CO has no effect on self-rated and manager-rated performance (e.g., Flaherty et al. 2009) or workers' intentions to quit their frontline jobs (e.g., Harris et al. 2006). These inconsistent findings may be explained by extant studies' focus on narrowly specified models in which CO is posited as a proximate determinant of job outcomes, particularly performance (Plouffe, Hulland, and Wachner 2009). While parsimonious, these models rarely consider the effects of moderators and even more infrequently account for the role of intervening variables or common antecedents, all of which may alter the observed pattern of results across studies (Crawford, LePine, and Rich 2010).
To help reconcile these divergent perspectives and inconsistent empirical evidence, we build on insights from the (firm-level) market orientation literature to conceptualize employee CO as a psychological phenomenon that is best understood as a work value. Consistent with this conceptualization, we employ job demands-resources (JD-R) theory (Bakker and Demerouti 2007; Demerouti et al. 2001) to propose that job stress and job engagement are related psychological processes that mediate CO's influence on frontline employees' job outcomes (performance and propensity to leave). We test this proposed causal structure using meta-analytic data and compare it with a rival model that represents the dominant view in the literature and is based on a behavioral understanding of CO. Furthermore, we draw on Saxe and Weitz (1982) to identify two previously unexamined moderators—employees' customer workload and use of persuasion in customer interactions—that may help explain inconsistencies in empirical findings across studies.
This research extends the literature in several meaningful ways. First, it represents an important first step toward reconciling divergent perspectives on the nature of employee CO and its role within broad construct networks. This contribution is realized by advancing a new conceptualization of employee CO that serves to bridge extant psychology-and behavior-based perspectives on the construct and by using robust data to empirically assess the relative merits of the causal structure implied by this new conceptualization.
Second, this study builds on JD-R theory to introduce and evaluate an integrative explanatory model that offers new insight regarding how CO operates to influence boundary spanners' job outcomes and when this influence is likely to be most pronounced. To do so, this study uses meta-analytic data to test structural models and moderating effects not previously investigated in published research to date. Specifically, this research investigates employee CO's relationship with job stress, job engagement, and job outcome variables within the same model; this approach represents a clear departure from extant research, which typically examines CO's (direct or unmediated) relationship with job states (i.e., stress and/or engagement) or job outcomes, but not both. Furthermore, by focusing on differences across studies, this research is able to assess the moderating role of two high-level job characteristics not considered in extant research.
Finally, unlike previous meta-analyses that focus exclusively on employee CO in sales contexts, this research represents the first meta-analytic investigation of the effects of CO among service workers as well as salespeople. As such, this study updates and extends previous research syntheses on CO (Franke and Park 2006; Jaramillo et al. 2007) and other related meta-analyses in the personal selling literature (e.g., Brown and Peterson 1993).
We organize the rest of this article as follows: We begin by advancing our conceptualization of the employee CO construct. Then, we offer an overview of the study's theoretical framework, in which we discuss the basic tenets of JD-R theory and define relevant constructs. Next, we develop hypotheses in support of the proposed JD-R model and introduce a rival model grounded in a behavioral understanding of employee CO. We then present the study methods and the results for the proposed and competing models. We conclude with a discussion of the findings' implications for theorists and managers.
The CO of Frontline Employees
Grounded in the marketing concept, the employee CO construct has enjoyed a rich history of empirical investigation since Saxe and Weitz's (1982) influential presentation of the selling orientation–customer orientation (SOCO) scale. Although scholars seem to be in general agreement regarding the core meaning of the construct (Schwepker 2003), empirical studies have usually conceptualized employee CO in one of two ways: (1) as a set of employee behaviors aimed at engendering customer satisfaction (hereinafter referred to as the “behavioral perspective”) or (2) as a psychological variable (e.g., mind-set, attitude, statelike individual difference, surface trait) that motivates employees to satisfy customers' needs (hereinafter referred to as the “psychological perspective”). In the marketing literature, the behavioral perspective's origins are in the work of Saxe and Weitz (1982, p. 343), who define CO as a behavioral phenomenon representing “the practice of the marketing concept at the level of the individual salesperson and customer.” The psychological perspective, in contrast, is largely grounded in Brown et al.'s (2002, p. 111) conceptualization of CO as “an employee's tendency or predisposition to meet customer needs in an on-the-job context.”
Despite employing comparable measures (variants and extensions of the SOCO scale), studies grounded in the behavioral and psychological perspectives suggest that CO plays markedly different roles within nomological nets that include relevant job state and job outcome variables. Specifically, research grounded in the behavioral perspective has predominantly argued that CO is a consequence of job stress (i.e., role ambiguity and role conflict; e.g., Flaherty, Dahlstrom, and Skinner 1999; Hoffman and Ingram 1991) and job engagement (i.e., job satisfaction and organizational commitment; e.g., Joshi and Randall 2001; O'Hara, Boles, and Johnston 1991) and a proximate determinant of job outcomes, such as manager- and self-rated performance (e.g., Cross et al. 2007; Plouffe, Hulland, and Wachner 2009). In contrast, studies proceeding from a psychological perspective typically posit that CO is antecedent to job stress and job engagement (e.g., Donovan, Brown, and Mowen 2004; Rod and Ashill 2010) and have more frequently considered the possibility that CO is a distal (or mediated) predictor of job outcomes, including employee performance and propensity to leave the organization (e.g., Babakus, Yavas, and Ashill 2009; Rod and Ashill 2010).
Given the vastly different causal structures implied by these two perspectives, the issue of whether CO is behavioral or psychological seems to be fundamental to an adequate understanding of the phenomenon. Although in recent years, researchers have attempted to develop more finegrained, context-specific conceptualizations and measures of the CO construct (e.g., Homburg, Müller, and Klarmann 2011a, 2011b), the question of whether employee CO is behavioral or psychological seems to have gone unaddressed in the literature. In this study, we proceed from the viewpoint that employee CO is psychological. Although we recognize that a definitive answer to whether CO is behavioral or psychological is perhaps not possible, the following section briefly reviews insights from the firm-level market orientation literature that inform our view on this important issue. Our intent in doing so is to lay a proper conceptual foundation for the meta-analytic work that follows.
Employee CO as a Psychological Phenomenon
Although both constructs share a common origin in the marketing concept, research on employee-level CO and market orientation (or firm-level CO 1 ) has largely proceeded along parallel paths. However, their common grounding in the marketing concept implies that both constructs are a manifestation of the same underlying phenomenon (Kelley 1992; Saxe and Weitz 1982), albeit at different levels of analysis. As a consequence, the rich body of literature on market orientation has strong potential to inform our discussion on the nature of employee CO.
Consistent with Deshpandé, Farley, and Webster (1993), we use the terms “firm-level CO” and “market orientation” (Kohli and Jaworski 1990) interchangeably.
As is the case for employee CO, the literature on market orientation reveals dueling conceptualizations regarding the nature of the construct, with some authors defining it in behavioral terms and others calling it a fundamental value of a firm's culture (Homburg and Pflesser 2000). For example, in their seminal work on the organizational implementation of the marketing concept, Kohli and Jaworski (1990) conceptualize market orientation as a set of behaviors largely focused on generating, disseminating, and responding to information about customers' needs. Conversely, Deshpandé, Farley, and Webster (1993) challenge this behavior-driven view of market orientation, arguing that the core of the construct rests on the deeply rooted set of values and beliefs that reinforce and give rise to behaviors aimed at satisfying customers. Consistent with this perspective, several authors, including Narver and Slater (1995), Homburg and Pflesser (2000), Kennedy, Goolsby, and Arnould (2003), and Gebhardt, Carpenter, and Sherry (2006), subsequently define market orientation as a value or deep-rooted belief that places customers' interests first and serves to guide firm behaviors and organizational decision making. While differing opinions regarding the nature of market orientation still persist, the prevailing view is perhaps best captured in the following excerpt from Narver and Slater (1998, p. 235):
Can market orientation truly be merely activities with no connection to an underlying system of values of the organization?… Of course, one measures market orientation by measuring certain specific activities, but in doing so we are measuring the manifestations of an underlying belief system…. If a market orientation were simply a set of activities completely disassociated from the underlying belief system of an organization, then whatever an organization's culture, a market orientation could easily be implanted by the organization at any time. But such is not what one observes. In sum, logic, scholarly treatises … and empirical evidence strongly contradict the idea that market orientation is other than the manifestation of a culture.
Given a cultural understanding of the market orientation of the firm, it is clear that the organizational implementation of the marketing concept is not simply about executing a set of customer-driven behaviors. Rather, this view implies that to implement the marketing concept effectively, firms must transform their organizational culture (Gebhardt, Carpenter, and Sherry 2006; Kennedy, Goolsby, and Arnould 2003) and attract, select, and socialize employees so that they too share this fundamental cultural value of the firm (Farrell and Oczkowski 2009; Hartline, Maxham, and McKee 2000; Kelley 1992). From this vantage point, employee CO can thus be understood to be an important work value that directs frontline workers' on-the-job behaviors (Bardi and Schwartz 2003; Verplanken and Holland 2002) and determines their level of fit within market-oriented organizations and/or customer-oriented work roles (Donavan, Brown, and Mowen 2004; Grizzle et al. 2009). Consistent with this line of theorizing, we conceptualize employee CO in this study as a psychological variable that captures an individual (as opposed to cultural or collective) work value.
Employee CO as a Work Value
“Values” refer to relatively stable, broad goals that guide people's perceptions, attitudes, and behaviors across time and over different contexts (Allport 1961; Bardi, Calogero, and Mullen 2008; Brown and Treviño 2009; Rokeach 1973). In his influential work, Rokeach (1973) proposes that values can serve as a unifying concept across sciences involving social behavior because they are highly abstract and thus underlie more concrete phenomena such as attitudes, preferences, and intentions. As an aspect of a person's psychology, values vary in important ways from other individual difference variables. Unlike personality traits, values enable people to reconcile potential conflicts between competing goals or inclinations (Rokeach 1973). In contrast to attitudes, which refer to evaluative judgments that have the potential to influence behavior, values are devoid of judgment and, in general, operate as guiding principles in people's lives (Steenkamp and De Jong 2010).
Work values are more specific or narrower manifestations of human values that refer to generally enduring beliefs about the desirability of different aspects of work and work-related outcomes (Lyons, Higgins, and Duxbury 2010). Work values have been shown to influence employees' job attitudes (e.g., job satisfaction, organizational commitment; Locke 1976; Meyer, Irving, and Allen 1998) and decision making (Ravlin and Meglino 1987). The level of congruence or match between employees' work values and those of the organization (Edwards and Cable 2009) and supervisors (Meglino, Ravlin, and Adkins 1989) has also been shown to influence worker job attitudes and other managerially valued outcomes. Furthermore, consistent with Schneider's (1987) attraction–selection–attrition (ASA) model, organizational values have been found to influence people's job choice decisions (Judge and Bretz 1992), which—at the organizational level—translates into within-firm homogeneity in values (Giberson, Resick, and Dickson 2005). Finally, while some debate exists in the literature regarding the extent to which work values can be influenced through socialization, recent research suggests that certain forms of leadership can help induce workplace value congruence (Brown and Treviño 2009).
Against this backdrop and consistent with the widely held understanding that customer satisfaction is at the core of the marketing concept, and therefore the CO construct, we offer the following definition as a foundation for our meta-analytic review: Customer orientation is a work value that captures the extent to which employees' job perceptions, attitudes, and behaviors are guided by an enduring belief in the importance of customer satisfaction.
Before turning our attention to the study's theoretical framework, we note that extant operationalizations of employee CO are consistent with the proposed value-based conceptualization along two critical dimensions. First, dominant measures of employee CO (e.g., Brown et al. 2002; Saxe and Weitz 1982) are composed of items that prompt employees to indicate whether they tend to engage in behaviors that engender customer satisfaction and to hold favorable attitudes toward satisfying customers. That is, extant CO measures assess tendencies in worker behaviors and attitudes, as opposed to their behaviors and attitudes toward a focal customer in a given exchange episode. This approach to measuring CO is consistent with how values (Rohan 2000) and other individual difference variables whose influence is manifest across situations are assessed in the literature (e.g., Pervin and John 1997; see also Dona-van, Brown, and Mowen 2004, as it relates to this issue).
Second, from a content perspective, dominant CO scales focus on behaviors and attitudes that indirectly reveal whether employees believe customer satisfaction is an important on-the-job goal. Stated differently, extant CO measures assess the extent to which employees tend to engage in value-expressive behaviors (Bardi and Schwartz 2003) and hold value-expressive attitudes (Wilcox, Kim, and Sen 2009) that indicate an underlying concern for customers' welfare. The use of value-expressive behaviors and attitudes to measure work values is well established in the literature (e.g., Bardi, Calogero, and Mullen 2008; Ravlin and Meglino 1987) and is deemed preferable by scholars who question respondents' ability to accurately articulate or express their values when asked to do so directly (Rohan 2000).
Theoretical Framework
Overview
Conceptually and empirically, CO appears to belong within a broad nomological net that includes job stress, job engagement, and job outcome variables (e.g., Brown et al. 2002; Siguaw, Brown, and Widing 1994). However, in general, previous studies have investigated employee CO's relationship with each of these variables in isolation; as a consequence, theoretical insight into CO's likely role within a nomological net that includes constructs such as role ambiguity, employee satisfaction, and turnover intentions is limited. We fill this void in the literature by relying on JD-R theory to advance an integrative framework that posits a specific role for employee CO and relevant contextual factors within this broad construct network.
According to JD-R theory, while every job has its own set of specific factors associated with employee stress and engagement, these factors can be classified in two categories: job resources and job demands (e.g., Bakker, Van Veldhoven, and Xanthopoulou 2010; Crawford, LePine, and Rich 2010). Job demands are aspects of the job that are central to role fulfillment and require frontline employees to expend sustained physical and/or psychological effort, resulting in various physiological and/or psychological costs. Examples of common job demands include high workloads and emotionally demanding customer interactions (Bakker and Demerouti 2007). Job resources are aspects of the job and the person that enable frontline employees to achieve work goals, help in reducing or coping with job demands, and/or provide for personal growth. Examples of job resources include self-efficacy, job security, and supervisor support (Bakker and Demerouti 2007; Nahrgang, Morgeson, and Hoffmann 2011; Xanthopoulou et al. 2007).
In addition to proposing that job demands and job resources are working conditions common to every occupation, the JD-R model proposes that demands and resources interact to influence employee job outcomes through their effects on dual mediating processes (Bakker and Demerouti 2007). One process focuses on worker stress-strain effects, and the other involves employee engagement. The job stress aspects of the strain process result from frontline employees' exposure to uncertain or conflicting environmental stimuli (i.e., stressors) while performing their jobs (Chang, Rosen, and Levy 2009; LePine, Podsakoff, and LePine 2005; Schuler 1980). Job engagement focuses on the degree to which frontline employees are invested in their organizations and have positive attitudes toward their jobs (Crawford, LePine, and Rich 2010; Nahrgang, Morgeson, and Hoffmann 2011).
Collectively, the tenets of the JD-R model imply a causal structure in which frontline employees' levels of job stress and job engagement are the proximate determinants of worker outcomes, such as performance and turnover intentions. Frontline employees' stress and engagement levels are in turn a function of the resources available to employees and of the demands employees face on the job. Combinations of demands and resources operate to produce both health-impairing (i.e., stress) and job-enhancing (i.e., engagement) effects. Furthermore, JD-R theory suggests that demands and resources interact such that the beneficial impact of resources on stress and engagement is often enhanced when workers face demanding job conditions (Bakker and Demerouti 2007; Bakker, Van Veldhoven, and Xanthopoulou 2010).
Definitions of Model Constructs
The demands and resources that play a role in any given organizational environment depend on the characteristics of the job (Bakker, Demerouti, and Verbeke 2004). Consequently, the specific resources, demands, mediators, and outcomes considered in JD-R-based studies tend to vary considerably (e.g., Bakker and Demerouti 2007; Bakker, Van Veldhoven, and Xanthopoulou 2010; Nahrgang, Morgeson, and Hoffmann 2011). The remainder of this section identifies and defines the constructs that represent the five JD-R job factors in this study—resources, demands, stress, engagement, and outcomes. Figure 1, Panel A, provides a graphic representation of how these job factors (and the constructs they subsume) are expected to relate to one another based on JD-R theory.
Proposed and Rival Meta-Analytic Models
Job Resource
Frontline employee CO is this study's focal job resource. As a work value, CO can be categorized within the JD-R framework as a personal resource or aspect of the self (as opposed to organizational or work-group resource) that influences how individual employees conduct themselves and operate within their work environment (Xanthopoulou et al. 2007).
Job Demands
As we illustrate in Figure 1, Panel A, this study considers how two job demands, customer workload and persuasion use, interact with CO to predict job stress and engagement. We define “customer workload” as the extent to which frontline employees divide their attention among a relatively large number of customers during any work shift or day (Bakker, Van Veldhoven, and Xanthopoulou 2010). “Persuasion use” refers to the extent to which frontline employees use influence tactics to elicit responses from customers (e.g., bring a car in for service while under warranty) that are necessary for achieving job goals (McFarland, Challagalla, and Shervani 2006).
Job Stress
Role ambiguity and role conflict represent the job stress process in our model. Both facets have been shown to have important health and performance implications among boundary-spanning employees (e.g., Brown and Peterson 1993; Singh, Goolsby, and Rhoads 1994). “Role ambiguity” refers to the degree to which frontline employees are uncertain about what others expect from them in their roles, the best way to fulfill known expectations, and the consequences of role performance (Singh, Goolsby, and Rhoads 1994). In contrast, “role conflict” refers to the “degree of incongruity or incompatibility of expectations communicated to a role incumbent by role senders” (Michaels, Day, and Joachimsthaler 1987, p. 30), such as managers and customers.
Job Engagement
Consistent with Bakker, Van Veldhoven, and Xanthopoulou (2010) and Nahrgang, Morgeson, and Hoffmann (2011), job engagement is represented in our model by both employee satisfaction and organizational commitment. “Employee satisfaction” is defined as a person's positive emotional state resulting from an appraisal of his or her job experiences (Locke 1976). In contrast, “organizational commitment” refers to the strength of an employee's psychological bond with and level of psychological investment in his or her employing organization (Hunt, Chonko, and Wood 1985).
Job Outcomes
This study investigates two important job outcomes: propensity to leave and employee performance. “Propensity to leave” refers to the subjective likelihood that frontline workers will voluntarily leave their employing organizations within a relatively limited time frame (Fried et al. 2008). Finally, we define “job performance” as the extent to which an employee contributes to organizational effectiveness given the expectations associated with his or her work role (Treadway et al. 2005). Self-rated and manager-rated performance measures are evaluated separately within the empirical model.
Hypothesis Development
CO Decreases Job Stress
Frontline employees are susceptible to role stress for a multitude of reasons, including their direct, often improvised, boundary-spanning contact with customers (Behrman and Perreault 1984). Customers tend to have unique needs, which hampers organizations' ability to prescribe exactly what frontline employees are supposed to do when engaging with them (Agarwal 1999). This lack of a concrete role routine may heighten frontline employees' levels of perceived role ambiguity in terms of role expectations and methods for role fulfillment. Moreover, customers' unique needs give rise to requests that may run counter to frontline employees' understanding of their own role obligations, thus creating the opportunity for increased levels of perceived role conflict (Jones, Busch, and Dacin 2003).
Although boundary-spanning contact with customers gives rise to potentially stressful job demands (e.g., unique customer requests), these demands do not necessarily create role stress because stress is said to exist “in the eye of the beholder” (Kammeyer-Mueller, Judge, and Scott 2009, p. 179). This assertion is supported by the differential exposure hypothesis (Bolger and Zuckerman 1995), which states that personal characteristics or individual difference variables alter the way employees interpret or perceive their job environment (Treadway et al. 2005) and thus represent an important resource that aids in stress resistance (Hobfoll 1989, 2001). Moreover, research suggests that values are personal attributes that contribute to differential exposure (Bandura 1991) and consequently serve to alter employees' interpretation of their work experiences (Ravlin and Meglino 1987).
Consistent with this literature, we propose that CO represents an individual resource that helps alleviate role stress by altering frontline employees' appraisal of their job environment (Folkman et al. 1986). Specifically, CO is likely to reduce perceptions of role ambiguity because it offers frontline employees strong guidance regarding the purpose or objective (i.e., customer need satisfaction) of their job roles. Thus, even in the absence of clear role expectations, frontline employees will perceive less role ambiguity because their belief in the importance of customer satisfaction (i.e., internalized work value) will help define the role for them. Likewise, CO is likely to reduce role conflict because it brings customer–employee and manager–employee role expectations into greater alignment, such that frontline workers perceive requests to satisfy customers' unique needs as being more consistent with their job role (Jones, Busch, and Dacin 2003). Prior research offers empirical support for our expectation that CO will lower perceived role ambiguity (e.g., Mengüç 1996) and role conflict (e.g., Jones, Busch, and Dacin 2003) among boundary-spanning employees and also supports the general proposition that resources protect employees from the effects of job stress (Bakker, Demerouti, and Euwema 2005; Singh 2000). Building on the premise that work values contribute to differential exposure and related empirical evidence, we expect the following:
Increasing CO decreases job stress among frontline employees.
CO Increases Job Engagement
Consistent with the well-established finding that values are an important determinant of human motivation (Verplanken and Holland 2002), the JD-R model suggests that CO should heighten employees' work-related motivation and, as a consequence, their job engagement levels. Specifically, JD-R theory proposes that resources such as CO can be intrinsically motivating because they help satisfy employees' basic needs (e.g., need to make a difference, need to help others) or extrinsically motivating because they are instrumental to the achievement of work goals (Bakker and Demerouti 2007). Using this same logic, two recent meta-analyses grounded in JD-R theory (Crawford, LePine, and Rich 2010; Nahrgang, Morgeson, and Hoffmann 2011) provide ample evidence in support of the desirable effects of different types of resources (e.g., knowledge, autonomy, supervisor support, feedback) on worker engagement, represented in these particular studies by measures of job satisfaction and organizational commitment.
For most employees, CO is likely to serve as a potent extrinsic motivator because it focuses their job effort on customer need satisfaction, a critical ingredient for success or superior performance in boundary-spanning roles. In addition, CO may act as an intrinsic motivator because it enables workers to succeed in helping others through their jobs, a basic need common among those who select certain types of frontline jobs (e.g., social work, health care). Thus, by facilitating the fulfillment of workers' own needs and those of their customers, CO improves employees' fit with the demands imposed on and values supplied by frontline job environments (Edwards 1996; Edwards and Cable 2009; Hobfoll 2001). This improved fit not only increases job satisfaction but also leads to enhanced employee commitment to the organization because the work is more fulfilling (Xanthopoulou et al. 2007). Building on similar arguments, Donavan, Brown, and Mowen (2004) and Farrell and Oczkowski (2009) posit and find that CO increases worker satisfaction and organizational commitment among those employed in frontline jobs. Thus, we propose the following:
Increasing CO increases job engagement among frontline employees.
Job Stress Has Undesirable Effects on Job Engagement and Job Outcomes
The JD-R model and related stress theories (e.g., conservation of resources theory; Hobfoll 1989) suggest that role stress will have a detrimental effect on employees' job engagement and job outcomes for two related reasons (Bakker, Demerouti, and Verbeke 2004). First, job stress is a proximate determinant of job strains, including anxiety, tension, and exhaustion (Jex 1998). Strain results in negative job attitudes (e.g., decreased satisfaction) because workers attribute the strain to the job itself (Schaubroeck, Cotton, and Jennings 1989). Strain also decreases employees' sense of obligation to and desire to remain with the organization because the strain reflects poorly on the value conferred by the employee–organization exchange relationship (Chang, Rosen, and Levy 2009; Cropanzano, Rupp, and Byrne 2003). Moreover, strains are physically and emotionally taxing and thus deplete workers' energy levels (Crawford, LePine, and Rich 2010). Depleted employees lack the mental and physical energy needed to fulfill role obligations effectively, which ultimately translates into poor performance (Nahrgang, Morgeson, and Hoffmann 2011). Consequently, to the extent that role stress produces job strain, it will decrease worker engagement levels and performance and will also increase workers' propensity to leave the organization.
The second reason for expecting negative effects of role ambiguity and conflict is that both are types of hindrance-related stress (Cavanaugh et al. 2000). Unlike challenge stress, which results from job conditions that provide workers with an opportunity for personal growth and thus may increase employee engagement, hindrance stress stems from work conditions that thwart goal attainment (LePine, Podsakoff, and LePine 2005). Beyond affecting performance, hindrance stress also discourages employees from investing further in the employment relationship (Crawford, LePine, and Rich 2010) and leads them to actively withdraw from the organization in an attempt to protect remaining resources from the demands imposed by the unfavorable work environment (Hobfoll 2001; Schaufeli, Bakker, and Van Rhenen 2009). As hindrance stresses, role ambiguity and role conflict are likely to have undesirable effects on employee engagement and job outcomes.
The literature provides ample empirical evidence to support the proposed relationships. Several meta-analyses, such as Jackson and Schuler (1985) and Brown and Peterson (1993), provide evidence consistent with the proposition that role stress decreases frontline employees' level of job engagement (i.e., employee satisfaction and organizational commitment). Meta-analytic evidence also offers broad support for the deleterious effect of role stress on performance (e.g., Gilboa et al. 2008; Tubre and Collins 2000) and for the positive relationship between role stress and employees' propensity to leave the firm (Brown and Peterson 1993; Mor Barak, Nissly, and Levin 2001). Thus, we propose the following:
Increasing job stress decreases job engagement among frontline employees.
Increasing job stress has undesirable consequences on frontline employees' job outcomes.
Job Engagement Improves Job Outcomes
Engaged employees focus their physical, cognitive, and emotional energies on goal attainment (Nahrgang, Morgeson, and Hoffmann 2011) and consequently perform better than their less engaged counterparts (Bakker and Demerouti 2007). Rich, LePine, and Crawford (2010) suggest that each of these energies contributes uniquely to employee performance. Specifically, they posit that physical energies facilitate the performance of the behavioral routines necessary for the fulfillment of role obligations. Cognitive energies fuel employee performance by providing for a sharper focus and improved attentiveness to the details that are relevant to effective role execution. Emotional energies contribute to enhanced employee performance by helping employees meet the emotional demands of their role through a more complete and authentic performance. These effects of physical, cognitive, and emotional energies explain why engaged employees experience heightened levels of performance. Extant empirical evidence supports the idea that engaged frontline employees perform better, as researchers have shown that both job satisfaction (e.g., Franke and Park 2006) and organizational commitment (e.g., Meyer et al. 2002) influence performance among boundary-spanning employees.
Furthermore, multiple applications of the JD-R model show that worker engagement decreases employees' propensity to leave the organization (e.g., Bakker, Demerouti, and Schaufeli 2003; Schaufeli and Bakker 2004). Schaufeli, Bakker, and Van Rhenen (2009) argue and find support for the notion that absenteeism occurs because workers want to withdraw from aversive or undesirable work conditions. Building on this same logic, we propose that engaged employees find their work conditions to be favorable or desirable and thus are unlikely to express an intention to leave their employing organization. Meta-analytic work among frontline employees supports this expectation, as our pair of engagement constructs (job satisfaction and organizational commitment) have been found to decrease employees' propensity to leave their organizations (e.g., Mor Barak, Nissly, and Levin 2001; Tett and Meyer 1993). Considering the expected effects of job engagement on frontline employees' performance and their propensity to leave the organization, we propose the following:
Increasing job engagement improves frontline employees' job outcomes.
Job Stress and Job Engagement Mediate CO's Effects on Frontline Employees' Job Outcomes
As mentioned previously, a basic premise of JD-R theory is that the influence of job demands and job resources on job outcomes is mediated by dual intervening processes, one involving stress and the other engagement. This fundamental aspect of the theory has received support in several empirical studies across a variety of different contexts (e.g., Bakker, Demerouti, and Verbeke 2004; Nahrgang, Morgeson, and Hoffmann 2011). In our current application of JD-R theory, this aspect of the model implies that the effects of frontline employee CO on job outcomes are mediated by job stress and job engagement. Thus, we expect that CO—a psychological resource that guides frontline employees' job perceptions, attitudes, and behaviors—will result in improved performance and decreased turnover intentions among frontline employees because (1) it will help reduce the amount of role stress employees perceive in their work environment and (2) it will increase employees' levels of engagement or investment in their jobs. This expectation is formally captured in the following hypothesis:
The influence of CO on job outcomes is mediated by (a) job stress and (b) job engagement.
Moderating Effects of Job Demands
Given inconclusive evidence regarding CO's performance effects (e.g., Franke and Park 2006), recent research has reflected a renewed focus on understanding when employee CO is more likely to produce outcomes valued by managers (Homburg, Müller, and Klarmann 2011a). Interest in this issue can be traced back to Saxe and Weitz (1982), who emphasize that CO's performance benefits are likely to be situation specific. Studies that attempt to identify factors that constrain or enhance CO's purportedly beneficial effects are limited (Homburg, Müller, and Klarmann 2011b) and suggest that salesperson characteristics (Stock and Hoyer 2005; Wachner, Plouffe, and Gregoire 2009), aspects of the work environment (Donavan, Brown, and Mowen 2004; Grizzle et al. 2009), and customer and product characteristics (Homburg, Müller, and Klarmann 2011b) may all play a role in moderating CO's influence.
Saxe and Weitz (1982) argue that CO is more likely to have an effect on workers' job outcomes when the exchange context confers employees an opportunity for meaningful, continued interaction with customers. They also identify several job characteristics that may constrain this opportunity, including (1) jobs that do not provide employees with sufficient time to interact with their customers individually and (2) jobs in which employees are expected to pressure or influence customers to achieve desired goals. Consistent with this line of reasoning, this study investigates whether customer workload (which limits employee–customer interaction times) and persuasion use (which limits the amount of time employees can devote to understanding customer needs) moderate CO's effects on frontline employees' job outcomes.
The proposition that job demands and resources interact to predict job stress and engagement has received significant support in the JD-R literature (Bakker, Van Veldhoven, and Xanthopoulou 2010). According to JD-R theory, resources are most valuable when employees face demanding job conditions (Bakker and Demerouti 2007) because such conditions activate relevant employee resources as part of the coping process (Hobfoll 2001; Xanthopoulou et al. 2007). Consequently, the stress-reducing effects of a resource are accentuated in situations in which the resource can serve as a buffer from the potentially deleterious effects of high levels of job demands (Bakker, Demerouti, and Verbeke 2004). Similarly, the engagement-enhancing benefits of a resource increase in situations in which the high levels of job demands induce employees to make full use of relevant resources to achieve extrinsically and intrinsically motivated job outcomes (Bakker, Van Veldhoven, and Xanthopoulou 2010).
While JD-R theory predicts that the effects of resources are accentuated in demanding job conditions, it also recognizes that how particular demand–resource combinations interact depends on the specific nature of the demands and resources in consideration (Bakker and Demerouti 2007). In a frontline environment, high customer workloads are likely to be physically, emotionally, and mentally demanding (Demerouti et al. 2001; Lee and Ashforth 1996; Nahrgang, Morgeson, and Hoffmann 2011). Thus, as customer workloads increase, employees draw increasingly on their CO resource as a way of remaining effective on the job. Customer orientation is a particularly useful resource in this context because it focuses frontline employees' energies on the critical objective of customer need satisfaction, despite being constantly taxed by the simultaneous requests and varying needs of numerous customers. Stated differently, CO enhances workers' ability to meet the demands associated with high customer workload jobs, and consequently, its beneficial effects are magnified under such conditions (Kristof-Brown, Zimmerman, and Johnson 2005). This proposition is consistent with Donavan, Brown, and Mowen (2004), who show that CO's effects on workers' level of engagement are strengthened as the intensity of customer contact increases. Consequently, we expect high customer workloads to activate the CO resource, such that
(a) The negative influence of CO on job stress is stronger (weaker) when customer workloads are high (low), and (b) the positive influence of CO on job engagement is stronger (weaker) when customer workloads are high (low).
Jobs that require high levels of persuasion typically revolve around the goal of eliciting a desired response from customers (e.g., immediate purchase). This demanding goal requires that employees draw on their resource pools to remain healthy and engaged (Bakker and Demerouti 2007). However, CO is not a useful resource for helping frontline employees meet the demands of a job focused on customer persuasion. Indeed, jobs that involve high levels of persuasion use are the antithesis of CO (Saxe and Weitz 1982) and represent a strong threat to the CO resource. Such threats encourage resource conservation (i.e., withdrawal) among employees (Hobfoll 1989, 2001) and consequently are likely to limit CO's beneficial effects on stress and engagement. Furthermore, such threats are also likely to temper CO's desirable consequences because they signal a poor fit between workers' values and the environmental supplies available to fulfill those values (Edwards 1996). These arguments are consistent with the findings of Grizzle et al. (2009), who demonstrate that CO's performance effects are inhibited in work climates that do not support such an orientation. Thus, we expect that high-persuasion jobs will limit or offset the benefits that CO confers to frontline employees, such that
(a) The negative influence of CO on job stress is weaker (stronger) when persuasion use is high (low), and (b) the positive influence of CO on job engagement is weaker (stronger) when persuasion use is high (low).
Other Model Relationships
This study focuses on direct, mediated, and moderated effects of a single job resource, CO. This section explains other relationships not developed as specific hypotheses but specified in the empirical models. First, we allow role ambiguity to covary with role conflict and self-rated performance to covary with manager-rated performance. This approach recognizes that the constructs are related and may have unmodeled antecedents in common, without the need to propose a specific causal relationship between them.
Second, as in much prior research (e.g., Brown and Peterson 1993; Meyer et al. 2002; Tett and Meyer 1993), we specify job satisfaction as causing organizational commitment. Conceptually, this relationship is based on the idea that while both commitment and satisfaction are important job attitudes, satisfaction is causally antecedent to commitment because it is “more specific, less stable, and more rapidly formed” than commitment (MacKenzie, Podsakoff, and Ahearne 1998, p. 90). Moreover, research also suggests that when employees find a job that satisfies their needs, they invest more or embed themselves further in the organization in an attempt to ensure that they can continue to experience need satisfaction, thus enhancing organizational commitment (Greguras and Diefendorff 2009). Consequently, we expect to find a positive relationship between job satisfaction and organizational commitment.
Finally, we model self- and manager-rated performance as predictors of frontline employees' propensity to leave the organization. The literature offers differing perspectives on how performance affects employee turnover. As Nyberg (2010) explains, one tradition argues that top-performing employees are more likely to stay because they want to remain with an employer that values their performance. Others argue that top-performing employees are more likely to leave because they are more attractive to other potential employers. While the nature of the relationship continues to be debated, past meta-analytic evidence suggests that performance and turnover are related and that the relationship is likely negative (Griffeth, Hom, and Gaertner 2000). Consistent with this evidence, we expect to find a negative relationship between ratings of employee performance and turnover intentions.
A Rival Model Grounded in a Behavioral Understanding of Employee CO
In a previous section, we identify two perspectives regarding the nature of employee CO: the psychological and behavioral perspectives. Moreover, we note that each perspective implies substantially different causal roles for CO within broad construct networks. The preceding JD-R-based explanatory model is consistent with our proposition that CO is a psychological phenomenon best understood as a work value. To assess the relative merits of this perspective and the causal relationships it implies, we specify and test a competing model grounded in a behavioral construal of the CO construct.
The competing model—which represents the dominant viewpoint in the literature and is depicted graphically in Figure 1, Panel B—is consistent with the centrality ascribed to the CO phenomenon in the marketing literature and with the proposition that CO refers to a set of behaviors that are largely determinant (and thus a proximate predictor) of frontline employee performance (Plouffe, Hulland, and Wachner 2009). Furthermore, this model posits that CO serves as a mediator of the effects of job stress and job engagement on employee job outcomes. Consistent with this proposition, scholars who conceptualize employee CO as a set of behaviors argue that job stress decreases CO because it disrupts or interferes with frontline employees' strong focus on customer-need satisfaction (e.g., Hoffman and Ingram 1991; Kelley 1992). Moreover, proponents of this perspective posit that job engagement increases CO because frontline employees who are invested in their roles will place a premium on customer-need satisfaction as a way of helping their organizations further their goals (e.g., Huang and Dastmalchian 2006; O'Hara, Boles, and Johnston 1991). In summary, the competing model suggests that CO refers to a behavioral routine that is a proximate predictor of frontline employee job outcomes and that the extent to which workers engage in this behavioral routine is influenced by two key aspects of the work environment—namely, job stress and job engagement.
Method
Meta-Analytic Procedures
Multiple criteria guided the selection of studies used in the meta-analysis. The studies had to present individual-level data on employees whose jobs primarily involved either sales or service to external customers (e.g., excluding information technology support staff, purchasing agents, general samples of marketing managers). To limit possible cultural differences in the effects of CO, workers' primary language had to be English. To ensure this research synthesis adds clarity to the understanding of the CO construct within the marketing discipline and is based on samples of adequate quality, studies providing correlations had to be published in marketing or closely related business publications and had to employ comparable measures of the CO construct (i.e., variants and extensions of the SOCO scale). Finally, the results had to enable the coding of correlations between one or more pairs of variables examined in the study.
We identified relevant studies in multiple ways. We searched multiple electronic databases, such as ABI/Inform, Google Scholar, and ISI Web of Knowledge (reference citations). Keywords used included the focal variables plus related terms, such as “role clarity” for “role ambiguity” and “turnover intentions” for “propensity to leave.” We examined citations in the studies we found and in previous related meta-analyses in marketing (e.g., Franke and Park 2006) and other fields (e.g., Mor Barak, Nissly, and Levin 2001). We manually searched all available issues of related specialty-area journals such as Journal of Personal Selling & Sales Management and Journal of Service Research. Using studies found from 1979 through July 2011, this search process provided correlations from 323 samples reported in 291 publications based on data provided by 99,641 frontline workers (for a listing of the source studies included in the meta-analysis, see the Web Appendix at www.marketingpower.com/jm_webappendix).
One experienced analyst coded the study effects, discussing judgment calls with other authors as necessary. In addition, we employed high-inference coding procedures (Cooper and Hedges 1994) to develop scores for our study's two moderators, customer workload and persuasion use. Specifically, using agreed-on definitions and information provided in each study about the type of frontline role respondents performed, multiple coders independently rated each sample on five-point scales to assess the extent to which workers' jobs required that they (1) divide their attention among a relatively large number of customers during any work shift or day and (2) use persuasion to accomplish job goals. This approach to coding meta-analytic moderators is well established in the literature and has been shown to yield construct scores that possess convergent and discriminant validity (for related references, see Denson, Spanovic, and Miller 2009). Five coders provided ratings of the samples' customer workload, and six rated persuasion use, producing highly reliable scores for the two variables (Cronbach's a = .95 and .91, respectively).
We used random-effects models to calculate the mean correlations. This approach allows generalizations to a population of potential studies, provides more realistic estimates of average effect sizes, and indicates the variability in true effect sizes across studies (Raudenbush 2009). As Hunter and Schmidt (2004) suggest, we analyzed the correlation coefficients without first applying Fisher's r-to-z transformation. In addition, consistent with Franke and Park (2006), we used meta-regression procedures to test the proposed moderation effects.
Structural Model Analysis Strategy
The meta-analytic process produced a correlation matrix for the study variables (see Table 1) that we used to test the proposed and rival models using structural equation modeling (SEM) in Mplus 6.0. For all model constructs, we set error terms equal to 1 minus the mean reliability value obtained in the meta-analysis (ranging from .77 to .87). We used the harmonic mean of the correlations' total sample sizes (n = 5336) as the sample size for model estimation purposes (Viswesvaran and Ones 1995). Finally, we tested mediation hypotheses using the procedures Iacobucci, Saldanha, and Deng (2007) outline for evaluating mediation in SEM.
Meta-Analytic Correlations for the Study Samples
p < .05.
p < .01.
Results
Test of the Proposed Main Effects Model
Overall Model Fit
The proposed meta-analytic model provides a good fit to the data (χ2 = 152.0, d.f. = 3, p < .05; comparative fit index [CFI] = .98; standardized root mean squared residual [SRMR] = .02). However, the model residuals, modification indexes, and p-values supported adding two direct paths (from CO to self-rated performance and propensity to leave) and deleting three nonsignificant (p > .05) paths from the model. The resulting model provides a better fit to the data than the original model (χ2 = 7.1, d.f. = 4, p > .10; CFI = 1.00; SRMR = .00). Figure 2 provides a graphic representation of this final meta-analytic model, including standardized direct effects and R-square values.
Structural Path Estimates for the Best-Fitting Proposed Model
Test of Direct Effect Relationships
The test of the structural model supports a majority of the JD-R-based direct effect hypotheses. Specifically, the results support the first three hypotheses: CO decreases job stress (H1), CO increases job engagement (H2), and job stress decreases job engagement (H3). H4, which proposes that job stress has undesirable effects on job outcomes, is partially supported: (1) Role ambiguity decreases manager- and self-rated performance and exerts a direct negative effect but positive total effect on propensity to leave, and (2) role conflict increases (rather than decreases) self-rated performance, increases propensity to leave, and is unrelated to manager-rated performance. The results offer support for H5, which argues that job engagement has desirable effects on job outcomes including self- and manager-rated performance and propensity to leave. Finally, CO has a positive effect on self-rated performance and decreases workers' propensity to leave; employee satisfaction increases organizational commitment; and self-and manager-rated performance are unrelated to workers' propensity to leave the organization.
Mediation Hypothesis
Consistent with JD-R theory, H6 proposes that job stress and job engagement mediate CO's effects on employee job outcomes. The results, which we summarize in Table 2, Panel A, provide strong support for this expectation. That is, job stress and job engagement partially mediate CO's effects on self-rated performance and propensity to leave and fully mediate CO's effects on manager-rated performance. Furthermore, all the stress (role ambiguity and role conflict) and engagement (satisfaction and commitment) variables play a role in this mediation process; significant indirect (or mediation) effects occur in all 12 of the potential mediation paths tested (3 job outcomes x 4 intervening variables). Finally, the ratio of CO's indirect to total effects on the three job outcome variables ranges from 50% to 100%, suggesting that the intervening variables account for at least half, if not all, of CO's effects on the three outcome variables (Iacobucci, Saldanha, and Deng 2007).
Results of Mediation Analyses
Notes: We performed tests of mediation using the procedures outlined in Iacobucci et al. (2007). The ratio of indirect to total effects accounts for competing indirect effects present in the models. We assessed statistical significance (p < .05) of the mediation (indirect) effects using the Sobel test.
Effects of Study Quality
To assess whether the preceding structural model results are influenced by the quality of the studies that provided meta-analytic correlations, we estimated a second model using correlations obtained from studies published only in “high-quality” journals. High-quality correlations were identified by compiling the five year Social Sciences Citation Index score for all journals that contributed samples (nonindexed journals were assigned a score of 0) and then excluding correlations provided by journals with Social Sciences Citation Index scores that fell below the median. This stringent quality standard resulted in the retention of 155 samples that reported data on 60,482 frontline workers. The structural model results based on this high-quality data set were remarkably comparable to those obtained using the full set of meta-analytic correlations. Specifically, in addition to nearly identical fit statistics, the magnitude, direction, and significance of the path estimates based on the high-quality data set were consistent with those obtained for the model estimated using all available correlations. To illustrate, we found CO's total effect on manager-rated performance, self-rated performance, and propensity to leave using the high-quality (vs. all available) samples to be .09 (.10), .41 (.34) and -.29 (-.33), respectively. Overall, these results suggest that the observed pattern of effects is highly stable and robust to the impact of study quality.
Test of the Moderation Hypotheses
The results for the moderation hypotheses (see Table 3) provide support for five of the eight proposed effects, with partial support for H7a, no support for H8a, and full support for H7b and H8b. Specifically, customer workload and persuasion use moderate the influence of CO on employee satisfaction and commitment but not the influence of CO on role conflict. In addition, customer workload moderates CO's effect on role ambiguity. All the significant results are in the expected direction: (1) The negative relationship between CO and role ambiguity strengthens (i.e., becomes more negative) as customer workloads increase, and (2) the positive relationship between CO and job engagement (i.e., satisfaction and commitment) strengthens as customer workloads increase and weakens as persuasion use increases. Furthermore, an exploratory analysis reveals that customer workload and persuasion use moderate the relationship between CO and self-rated performance, in a manner consistent with the effects observed for the job engagement variables. 2
To assess whether the impact of the moderators is the same across sales versus service contexts, we also performed a series of meta-regression analyses in which we used the moderator, a sales-service dummy variable, and their product term (e.g., dummy x persuasion use) as predictors of the meta-analytic correlations. These analyses revealed that in general, the effects of the moderators are the same across sales versus service contexts; the only exception occurs in the case of the CO–role ambiguity relationship, for which the analyses reveal that the moderating role of customer workload is stronger in service than sales contexts.
Results of the Moderation Hypotheses
p < .05.
p < .01.
Notes: All predictors were mean centered to facilitate interpretation of the results, α = random effects intercept. B = random effects regression coefficient. A significant B indicates statistical support for the moderating effect of customer workload or persuasion use on correlations between CO and other constructs.
Test of the Rival Model
The results reveal that the proposed competing model (with CO as full mediator of the effects of job stress and job engagement on job outcomes) offers a poor fit to the data (χ2 = 2198.4, d.f. = 12, p < .05; CFI = .76; SRMR = .12). After adding 11 direct effects from job stress and job engagement to the job outcome variables and eliminating two paths that became nonsignificant (p > .05) with the addition of the direct effects, the resulting competing model offers a good fit to the data (χ2 = 1.2, d.f. = 3, p < .05; CFI = 1.00; SRMR = .00). The structural parameter estimates for the best-fitting rival model, which Figure 3 graphically summarizes, indicate that role ambiguity decreases and satisfaction and organizational commitment increase employee CO. The results also reveal a direct positive effect but negative total effect of role conflict on CO (p < .01). In addition, the results suggest that CO increases self-rated performance (p < .01), is weakly related to workers' propensity to leave (p > .05), and is unrelated to manager-rated performance (p > .10).
Structural Path Estimates for the Best-Fitting Competing Model
Thus, these results suggest that CO does not mediate the effects of job stress and job engagement on manager-rated performance. Moreover, as Table 2, Panel B, indicates, of a total of 12 possible mediation paths (3 job outcomes x 4 independent variables), mediation only occurs two-thirds of the time. Importantly, mediated effects account for only a small fraction of the total effect of the stress and engagement variables. Specifically, the average ratio of the indirect to total effects (Iacobucci, Saldanha, and Deng 2007) for the competing model is 9%, which is substantially less than the average ratio for the JD-R-based meta-analytic model (78%). Collectively, this pattern of results provides greater support for the role of CO suggested by the psychological (JD-R-based) perspective than that implied by the behavioral perspective that currently dominates the literature.
Discussion
Prior research has conceptualized and modeled CO in one of two ways: as a psychological phenomenon that is antecedent to stress and engagement and (perhaps) a distal predictor of job outcomes or as frontline employee behaviors that are a consequence of stress and engagement and a proximate determinant of job outcomes. This research represents an important first step toward reconciling these two perspectives. Specifically, this study conceptualizes employee CO as a work value: an aspect of workers' psychology that guides their on-the-job perceptions, attitudes, and behaviors. This conceptualization bridges the psychological and behavioral perspectives on CO by allowing for the possibility that customer-oriented behaviors are the observable manifestation of an underlying psychological difference (Bardi, Calogero, and Mullen 2008). Importantly, the construal of employee CO as a work value is consistent with the development of the marketing concept at the organizational level (i.e., the market orientation of the firm; e.g., Homburg and Pflesser 2000) and thus also serves to build common ground between two important research streams that have proceeded along largely parallel paths despite their common origin (e.g., Grizzle et al. 2009).
Consistent with this new conceptualization and JD-R theory, the results of this meta-analytic study indicate that CO is antecedent to job stress and job engagement and that these variables represent part of the mediational mechanism through which CO influences frontline employees' job outcomes. That is, CO is a psychological resource that leads to desirable job outcomes because it helps shape employees' perceptions of and attitudes toward their jobs. This result largely supports the causal role for CO suggested by the psychological perspective on the construct and is thus contrary to arguments in the literature that certain aspects of the job (e.g., role stress, job satisfaction) may operate to decrease or increase employees' CO. However, it is important to note that our conceptualization of the CO construct as a work value implies that—while difficult to change—employee CO may be susceptible to the influence of certain aspects of the job environment such as training, socialization, and charismatic leadership (Brown and Treviño 2009).
The finding that CO decreases stress perceptions and enhances worker engagement supports the idea that CO affects frontline employee job outcomes by enhancing workers' fit with the demands of and values supplied by frontline jobs (Edwards 1996). This line of reasoning offers a sharp contrast to the bulk of the literature that presumes that CO improves workers' job outcomes because it represents the right way to do business or elicits favorable customer responses (e.g., Wachner, Plouffe, and Gregoire 2009). Consequently, recommendations in the literature to reduce or control workers' CO levels (e.g., Homburg, Müller, and Klarmann 2011a) may have undesired consequences—not because doing so directly leads to poor customer outcomes but because it signals a lack of fit between the work and what employees attracted to frontline jobs value (Donavan, Brown, and Mowen 2004), which eventually leads to increased stress, decreased engagement, and poor job outcomes (Edwards and Cable 2009).
Beyond disagreement regarding CO's nature and its role within broad construct networks, recent research has also called into question the relevance of CO to outcomes valued by managers (e.g., Franke and Park 2006; Homburg, Müller, and Klarmann 2011b; Plouffe, Hulland, and Wachner 2009). The results of this study indicate that CO has meaningful effects on managerially valued outcomes and that the strength of these effects depends on important contextual factors. Beyond its influence on self-rated performance (total effect = .34) and manager-rated performance (total effect = .10), this study demonstrates that CO decreases workers' propensity to leave through effects largely mediated by job stress and job engagement (total effect = -.33). This is an important finding, considering that turnover is a perennial problem among frontline workers (e.g., Rust et al. 1996) with substantial cost, performance, and customer loyalty implications (e.g., Palmatier, Scheer, and Steenkamp 2007). This finding underscores that the people recruited may be as important as what a firm does to the recruits after they have been hired; that is, hiring customer-oriented frontline workers will lead to longer employee tenures, thereby increasing the likelihood that those who become top performers do so because they stay at the organization long enough to be shaped by it.
An important aspect of this research effort is that it offers insight into how high-level job characteristics work to constrain or enhance CO's effects. A common assumption in the literature, rooted in a behavioral understanding of the construct, is that for CO to influence job outcomes, workers must have ample opportunity to interact with customers and thus learn about their needs and preferences (e.g., Saxe and Weitz 1982). As JD-R theory predicts and our results demonstrate, CO can be even more beneficial when opportunity for meaningful interaction is limited, as long as the constrained opportunity for interaction is not due to a direct threat to workers' CO. Consistent with this argument, we find that CO's influence on job stress, job engagement, and self-rated performance becomes stronger as customer workloads increase and weaker as employees' use of persuasion to achieve job goals increases. High customer workloads activate the CO resource as a mechanism for helping workers focus and remain effective in physically and emotionally demanding frontline roles. In contrast, jobs that involve high levels of persuasion represent a direct threat to the CO resource because pressuring customers is in direct opposition to a focus on customer need satisfaction. More broadly, in viewing CO as a personal resource within the JD-R framework, it becomes apparent that CO may be most beneficial when frontline job conditions are most demanding (Bakker and Demerouti 2007). This expectation again underscores the notion that CO's desirable effects on workers' job outcomes are attributable, in part, to the psychological benefits it confers.
Inconsistent empirical evidence has contributed to skepticism regarding CO's influence on important job outcomes (e.g., Franke and Park 2006; Plouffe, Hulland, and Wachner 2009). The study results support our expectation that these inconsistent findings may be due to a combination of model specification decisions and unaccounted-for moderation effects. For example, contextual factors may account for differences in the results reported by Brown et al. (2002) and Plouffe, Hulland, and Wachner (2009) regarding CO's influence on self-rated performance. Brown et al. investigate CO's effects in high-customer-workload, low-persuasion jobs, whereas Plouffe, Hulland, and Wachner assess CO's effects among employees in low-customer-workload, high-persuasion jobs. Despite comparable models, the studies arrived at different conclusions primarily because the zero-order correlations were dramatically different: Brown et al. find zero-order correlations to be almost three times as large as those in Plouffe, Hulland, and Wachner, and thus they find a significant direct effect of CO on performance, whereas Plouffe, Hulland, and Wachner do not. As another example, Cross et al. (2007) and Jaramillo et al. (2009) study CO's role among samples of predominantly business-to-business sales representatives. Despite relatively large and similar zero-order correlations between CO and self-rated performance, the studies arrived at different conclusions about CO's effects on performance because Jaramillo et al. (2009) control for the influence of more proximate predictors of performance (including some aspects of engagement and stress) and Cross et al. (2007) do not.
This study adds to the increasing body of literature (e.g., Crawford, LePine, and Rich 2010; Nahrgang, Morgeson, and Hoffmann 2011) that offers support for the basic tenets of the JD-R theoretical framework. Thus, it extends JD-R theory by testing its predictions in a new context and evaluating resource–demand combinations substantially different from those considered in previous research. In addition, this effort indicates that the JD-R framework offers a useful, integrative structure that can help guide further research on CO and, perhaps more important, affords a different understanding of the CO phenomenon. Specifically, by recasting CO as a personal resource that enhances boundary spanners' ability to cope with the demands of their roles, JD-R theory offers a plausible explanation for why CO is likely to be a proximate antecedent to important psychological outcomes and a more distal predictor of employees' job outcomes.
Finally, the study reveals a somewhat unexpected relationship between two variables considered in prior meta-analytic work: We find that role conflict increases workers' self-rated performance. Such an effect has been proposed in the past on the basis of the expectation that workers give themselves credit for successfully navigating jobs that pull them in different directions (e.g., Behrman and Perreault 1984). This result is similar to a recent meta-analysis on the role stress–performance relationship (Gilboa et al. 2008), which indicates a negative correlation between role conflict and job performance but a positive correlation when other effects are partialed out.
Managerial Implications
Would JetBlue's Steven Slater have fared better if he were more customer oriented? Our research implies the answer to this question is yes and offers several related suggestions to marketing and sales managers. First, contrary to the findings of a previous meta-analysis (Franke and Park 2006), we find that CO influences outcomes that managers value, including manager-rated performance and employees' propensity to remain at or leave the organization. Moreover, we find that this happens because CO leads workers to experience less job stress and to be more engaged on the job as well, both of which are outcomes that are important to managers in their own right (e.g., Crawford, LePine, and Rich 2010). Our research thus indicates that resources invested to attract, select, and retain customer-oriented frontline sales and service employees may offer firms substantial returns in the form of improved employee welfare, heightened engagement, better performance, and longer job tenures. Consistent with this conclusion, we urge practitioners to make CO a criterion in their employee selection, retention, and compensation processes.
Second, the study's findings suggest that while CO is beneficial across a wide variety of jobs, its effects are more strongly pronounced in certain situations, and thus context should factor into managers' decisions when evaluating potential investments in employee CO. Contrary to the assumption that CO should offer the most benefits in jobs in which employees have ample opportunity to interact with customers and therefore learn about their needs (e.g., relational settings; Homburg, Müller, and Klarmann 2011b; Saxe and Weitz 1982), our research indicates that CO may be most beneficial when jobs confer employees with little opportunity to interact meaningfully with individual customers. For example, CO offers greater benefits to workers employed in fast-food restaurants, retail sales, and call centers (high-customer-workload jobs) than to those employed in pharmaceutical sales or industrial services companies (low-customer-workload jobs). We also find evidence that jobs that require workers to use persuasion to achieve job goals inhibit CO's benefits; bank tellers and providers of business-to-business contract services (low-persuasion jobs) thus benefit more from CO than telemarketers and insurance sales representatives. This finding is significant because it reveals that the purpose of the persuasion effort is not critical; rather, it is the need to persuade that accounts for the effect.
Third, our study indicates that employee CO may not be very responsive to some aspects of the job that can be influenced by managers. Specifically, our research reveals that employees' stress perceptions and engagement levels are more likely to be a consequence than a cause of frontline workers' CO. This finding implies that managers should not assume that their efforts to provide for a positive work climate, in which job stress is limited and job engagement is enhanced, will result in meaningful improvements in employee CO levels. This also suggests that CO's importance as a selection, retention, and compensation criterion may be greater than originally thought. It is important to note, however, that our study does not rule out the possibility that efforts to achieve value congruence (Brown and Treviño 2009), by training and socializing employees to adopt the cultural values of the firm, are beneficial (Hart-line, Maxham, and McKee 2000).
Finally, we caution managers against explicit attempts to curtail workers' CO levels as a mechanism for achieving costs savings, as some prior research has advocated (e.g., Homburg. Müller, and Klarmann 2011a). Beyond any potential effects on customer satisfaction, such attempts may be particularly harmful because CO helps shape frontline workers' perceptions of and attitudes toward their jobs. In other words, efforts to constrain workers' performance of customer-oriented behaviors (the tangible manifestation of their CO) are likely to result in heightened job stress and reduced engagement, which will have detrimental effects not only on workers' performance but also on their desire to remain employed in the organization. Initiatives aimed at optimizing workforce CO levels should thus be carried out at the firm level (rather than the employee level) and should be implemented through recruitment, selection, and retention programs.
Limitations and Further Research
As is common in SEM-based meta-analytic work, our study only includes constructs that resulted in a full correlation matrix for model estimation purposes. As a consequence of this limitation, we were unable to include any customer response variables (e.g., customer satisfaction) in our study. This presents an important opportunity for further research because insights may be gleaned from studies that simultaneously consider CO's employee-focused (as was done in this study) and customer-focused consequences. For example, such research might consider whether CO's effects on customer outcomes (e.g., customer satisfaction) are mediated through job stress and job engagement, as is suggested by both the emotional contagion hypothesis (e.g., Stock and Hoyer 2005) and particularly low correlations between self-rated CO and customer satisfaction noted in prior research (e.g., Homburg, Müller, and Klarmann 2011a).
While our research suggests that CO is a relevant criterion for personnel decisions, including frontline worker selection, retention, and compensation, the literature offers little guidance as to how CO can be used to inform these decisions. Thus, research is needed to address implementation issues within this domain. Moreover, our results suggest that workers' CO may not be highly responsive to certain aspects of the work environment. Thus, research is needed to ascertain whether training and socialization can be employed to enhance workers' CO. More broadly, our conceptualization of CO as a work value suggests that attraction–selection–attrition (ASA) processes are likely to result in CO homogeneity across firms (Giberson, Resick and Dickson 2005); thus, to understand how managers can best enhance their workforce CO levels, studies are needed that offer insight into the relative impact of ASA processes versus training and socialization efforts on employee CO. Furthermore, even though our meta-analytic findings underscore the importance of recruiting, longitudinal and/or experimental research is needed to achieve more reliable conclusions about what happens to employee CO after recruits are hired. For example, growth modeling could be used to explore, in a more fine-grained manner, whether aspects of the job have any meaningful, lasting impact on workers' CO.
This study employs high-inference coding procedures to develop the construct scores used to test the moderating effects of customer workload and persuasion use. While high-inference coding procedures offer rich and valid theoretical insights beyond those provided by meta-analytic source articles (Cooper and Hedges 1994; Denson, Spanovic, and Miller 2009), the resulting construct scores represent external raters' assessment of the job situation rather than respondents' views of their own psychological climate. Thus, a logical extension of this research would be to determine whether the moderating effects identified in this study hold when perceived job demand (i.e., psychological climate) measures are employed.
Finally, this study proceeds from the perspective that employee CO is a work value, a type of individual difference variable. As we noted in our exposition on the CO construct, this perspective is consistent with the development of the marketing concept at the firm level and with extant operational measures of the construct. While our study is grounded in this perspective, we recognize that, as a work value, employee CO will logically influence worker behaviors toward customers (as demonstrated by Grizzle et al. 2009) and that it can also influence related attitudes that affect employees' subsequent behaviors toward customers (as Stock and Hoyer 2005 argue). Thus, as Homburg and Pflesser (2000) did for the market orientation construct, research is needed that offers insight into the rich chain of effects through which CO, as a work value, influences the behaviors frontline employees exhibit in dyadic (i.e., one-to-one) exchanges with customers. We note that, within any such chain of effects, employee CO would represent the relatively enduring, psychological element that manifests itself across situations and is distinct from and causally antecedent to the episodic behaviors frontline employees perform when interacting with a focal customer.
Concluding Remarks
As a foundation for future research efforts, the results from more than 300 samples of boundary-spanning employees indicate that CO influences job outcomes for those in sales and service roles. This influence seems to be mediated by dual psychological processes, one involving job stress and the other involving job engagement, and to be stronger in jobs in which customer workloads are high and the use of persuasion to accomplish job goals is low. Thus, the results of this study underscore the merits of a psychological perspective on the CO phenomenon that recognizes that CO leads to improved job outcomes, not simply because it is a better way to do business but because it confers employees with important psychological benefits as well.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
