Received: 31 October 2019 Revised: 10 November 2020 Accepted: 20 November 2020 DOI: 10.1002/joom.1130

RESEARCH ARTICLE

Commitment follows beliefs: A configurational perspective on operations managers' commitment to practice adoption

Maricela C. Arellano1 | Johannes Meuer2 | Torbjørn H. Netland3

1Department of Logistics and Operations Management, HEC Montréal, Montreal, Abstract Quebec, Canada Companies that seek to improve their operational performance by adopting 2Group for Sustainability and Technology, new practices often report disappointing adoption rates. The literature con- ETH Zurich, Zurich, Switzerland cerning practice adoption has tended to focus on efficacy and legitimacy 3Chair of Production and Operations drivers at the organizational level. However, there exists convincing evidence Management, ETH Zurich, Zurich, Switzerland that practice adoption largely depends on the commitment of those managers involved in the adoption of a given practice. Thus, we investigate what pro- Correspondence Maricela C. Arellano, Department of mpts operations managers to commit to practice adoption. We draw on the Logistics and Operations Management, theory of planned behavior to explore the cognitive foundations of 76 opera- HEC Montréal, 3000, Côte-Sainte- tions managers' commitment to new operational practices. Using fuzzy-set Catherine Road, Montreal, QC H3T 2A7, Canada. qualitative comparative analysis, we identify three belief configurations associ- Email: maricela-connie.arellano-caro@ ated with high levels of commitment—“the Follower,”“the Pragmatist,” and hec.ca “the Reformer.” We contribute a behavioral operations perspective to the liter-

Handling editor: Gopesh Anand ature on practice adoption by providing an individual-level and configura- tional view of managerial commitment to change.

KEYWORDS behavioral operations, beliefs, managerial commitment, operations managers, qualitative comparative analysis

1 | INTRODUCTION Historically, the majority of scholars have examined practice adoption at the organizational level, concerned In pursuit of competitiveness, firms adopt practices that with explaining why and under what conditions compa- they consider to be superior to, or more legitimate than, nies successfully adopt certain practices. They have other practices (Ketokivi & Schroeder, 2004; Leseure largely focused on two primary drivers of practice adop- et al., 2004; Voss, 2005). It is well established that the tion: efficacy and legitimacy (Leseure et al., 2004; commitment of managers is essential to the success of Voss, 2005). Efficacy drivers dominate when companies practice adoption (e.g., Done et al., 2011; Herold set out to adopt practices intended to increase their oper- et al., 2007; Kostova & Roth, 2002; Netland et al., 2015). ational performance, while legitimacy drivers dominate Ultimately, the practice adoption process is a result of when companies set out to adopt practices intended to the behaviors of individuals, which in turn are driven by increase their validity in the eyes of key stakeholders. their commitment to the practice (Kostova & However, the results at the organizational level provide Roth, 2002). However, only limited scientific research mixed evidence of the implications of efficacy and legiti- has studied the individual-level conditions that drive the macy drivers in practice adoptions (Huang et al., 2010; commitment of operations managers to a new opera- Ketokivi & Schroeder, 2004). Moreover, the focus of prior tional practice. research on the drivers of adoption at the organizational

J Oper Manag. 2020;1–26. wileyonlinelibrary.com/journal/joom © 2020 Association for Supply Chain Management, Inc. 1 2 ARELLANO ET AL. level has inadvertently led scholars to ignore the role 2 | DRIVERS OF PRACTICE played by individuals. In these studies, operations man- ADOPTION agers are usually portrayed as hyper-rational agents who reactively respond to the dominant pressures on them by 2.1 | Efficacy and legitimacy as adopting certain practices (Croson et al., 2013; Donohue organizational-level drivers of practice et al., 2020). adoption Nevertheless, several recent individual-level studies reject the assumption that managers mechanically imple- Both the efficacy and legitimacy perspective have domi- ment practices that are deemed to be effective or legiti- nated scholarly explanations of why and with what level mate. In this view, a few authors maintain that practice of success organizations adopt new operational practices adoption is not the result of legitimacy and efficacy fac- (Jacqueminet, 2020; Ketokivi & Schroeder, 2004; tors in their own right but rather of people's evaluations Leseure et al., 2004). According to the efficacy perspec- of these factors (e.g., Bendoly & Cotteleer, 2008; tive, adoption depends on the fit between the practice Braunscheidel et al., 2011; Jacqueminet, 2020; Rogers and a range of internal and external contextual factors. et al., 2007). These studies suggest that managers, Most studies on practice adoption employ this perspec- through complex cognitive processes, base their decisions tive, as exemplified by the following three highly cited and actions on their beliefs regarding the practice and the articles in the current journal.1 Flynn et al. (2010) study context (e.g., Croson et al., 2013; Eggers & Kaplan, 2013; how different configurations of supply chain integra- Helfat & Peteraf, 2015; Kennedy & Fiss, 2009; Nadkarni & tions affect performance. Zhu and Sarkis (2004) investi- Barr, 2008). Yet, it remains unclear how operations man- gate how lean and quality practices moderate the agers' beliefs influence their commitment to the adoption relationship between green supply chain management of operational practices. practices and organizational performance. Shah and To address this gap, we study the beliefs that under- Ward (2003) report how combinations of lean practice pin managerial commitment and, hence, shift the conver- bundles and contextual variables affect performance sation on practice adoption away from the organizational outcomes. These three studies indirectly assume the effi- level to the specific dynamics at the individual level. We cacy perspective, portraying firms as rent-seeking orga- argue that operations managers' beliefs regarding the nizations that would proactively adopt expectedly adoption of a new operational practice establish their effective practices. commitment to it. Our theoretical arguments are drawn In contrast, the legitimacy perspective portrays firms from Ajzen's (1985, 1991) theory of planned behavior, as responding passively and reactively to exogenous pres- which explains how the behaviors of individuals are sures in search of compliance (Burns & Wholey, 1993; driven by three types of salient beliefs—namely, behav- Kostova et al., 2008). The legitimacy perspective draws ioral beliefs, control beliefs, and normative beliefs. We on the neo-institutional theory and suggests that institu- maintain that these three types of beliefs collectively rep- tional pressures explain the behavior of firms—for exam- resent an individual's actual disposition and, therefore, ple, adopting certain practices (DiMaggio & interact in complex ways to shape individuals' commit- Powell, 1983). From this perspective, adopting practices ment. Because managers activate their salient beliefs makes firms “look good,” which helps them attain simultaneously rather than in isolation (Piderit, 2000; legitimacy. Straatmann et al., 2018), we adopt a configurational per- Certain studies have juxtaposed the efficacy and legit- spective and ask the following question: What configura- imacy perspectives in an effort to investigate whether tions of beliefs commit managers to a new operational they compete or complement each other in terms of driv- practice? ing the adoption of operational practices (e.g., Bansal & To answer this question, we survey 76 operations man- Roth, 2000; Barreto & Baden-Fuller, 2006; Huang agers who implemented an operational practice within et al., 2010; Rogers et al., 2007). A few studies even sug- their organization. We ask about their beliefs regarding gest that the efficacy and legitimacy drivers co-exist, practice adoption and analyze their responses using fuzzy- regardless of contextual variables such as timing and set qualitative comparative analysis (fsQCA) (Ragin, 2000), maturity (Kennedy & Fiss, 2009; Liu et al., 2010). Overall, a method for configurational analysis. In doing this, we researchers have provided evidence of both the efficacy contribute to the literature on practice adoption by provid- and legitimacy explanations of firms' decisions to adopt a ing a detailed account of how the interdependencies practice—either because a firm considers adoption to be between operations managers' efficacy- and legitimacy- necessary (the efficacy argument), expected (the legiti- related beliefs drive their commitment. macy argument), or both. ARELLANO ET AL. 3

2.2 | Individual-level drivers of practice understanding of individual-level mechanisms, they nei- adoption ther explain operations managers' cognitive interpreta- tions of efficacy and legitimacy drivers nor the cognitive Organizational-level studies have been criticized for underpinnings of high levels of managerial commitment. ignoring cognitive factors at the individual level that may provide further insights into why practice adoption occurs. For example, Kostova and Roth (2002: 229) reflect 3 | A CONFIGURATIONAL MODEL on how practice adoption “is not necessarily driven by OF COMMITMENT TO PRACTICE rational, efficiency-based decisions but can be better ADOPTION explained in terms of the interpretive [emphasis added] social processes through which employees build percep- To explain how the belief configurations of individual tions about the efficiency of a practice.” Scholars in the managers relate to their commitment to a given opera- field of behavioral operations increasingly endorse this tional practice, we develop a configurational model of argument, thereby highlighting additional research commitment to practice adoption. The model presented design limitations in numerous organizational-level stud- in Figure 1 proposes that the combinations of operations ies and promoting the study of individual cognition managers' beliefs concerning efficacy- and legitimacy- (e.g., Busse et al., 2017; Choo et al., 2015; Croson related pressures affect the managers' commitment to et al., 2013; Fahimnia et al., 2019). new operational practices. Despite the repeated call for more individual-level We draw on Ajzen's (1991) theory of planned behav- studies on practice adoption, evidence remains scarce. ior (Ajzen, 1991) and Armenakis et al.'s (2000) model of Only a few empirical articles in the field of operations the institutionalization of change to develop our model. management explain the adoption of practices on the We define commitment to a new practice as a psychologi- basis of a cognitive interpretation (e.g., Bendoly cal state or that indicates an individual's level of et al., 2008; Bendoly & Cotteleer, 2008; Braunscheidel attachment to that practice (Becker et al., 1996). Attach- et al., 2011; Rogers et al., 2007). Strategy scholars have ment comprises an individual's identification with the employed an individual-level perspective to study the practice and his/her involvement in its use. When there adoption of practices related to corporate social responsi- is a high level of attachment, managers interpret the bility, among other issues, thus suggesting that cognition practice as part of what they are and do (at work) and, (studied as attitudes, , opinions, emotions, or thus, fully support it. beliefs) and related behaviors of managers are key predic- The theory of planned behavior states that people tors of practice adoption. Appendix A provides an over- reflect on specific situations in an effort to “act in more view of representative individual-level studies on practice or less logical ways” (Ajzen, 2005: 29) by “tak[ing] adoption. Although these studies have contributed to our account of available information and implicitly or

Efficacy-related beliefs

Belief in valence Behavioral Belief in beliefs appropriateness Belief in discrepancy Operations Operations managers’ managers’ commitment to belief Control Belief in an operational configurations beliefs self-efficacy practice

Legitimacy-related beliefs

FIGURE 1 A Belief in configurational model of Normative coercive pressures operations managers' beliefs Belief in commitment to practice mimetic pressures adoption 4 ARELLANO ET AL. explicitly consider[ing] the implications of their actions” Discrepancy captures the belief that a change is neces- (Ajzen, 1985: 12). In our study, the behavior (i.e., action) sary because of a mismatch between current and desired of interest is the adoption of a new practice by an opera- states (Armenakis et al., 2007). Discrepancy depends on tions manager. According to the theory of planned the level of managerial attention paid to environmental behavior, the main cognitive precondition of behavior is cues (e.g., Eggers & Kaplan, 2013) and on managers' the to perform the behavior. Intention, in turn, problem-solving approaches (Cantor & Macdonald, 2009; is explained by the manager's attitude toward the behav- Choo et al., 2015). Certain managers prefer their current ior, subjective norm, and perceived behavioral control. processes and methods—as evidenced by statements like These aggregated dimensions comprise the manager's “this is how things are done here” (Loch, 2017: 599). salient beliefs2 regarding the practice adoption and Such complacency restrains the implementation of together define the manager's readiness for change changes intended to improve operations. A high level of (Armenakis et al., 2000). The theory of planned behavior risk aversion can also explain a low level of discrepancy distinguishes three types of salient beliefs (Ajzen, 1991, (Loch, 2017). In contrast, certain operations managers 2005): behavioral beliefs represent an individual's evalua- actively seek to identify gaps between the current state of tion of engaging in the behavior in question; control operations and a possible better state of operations. Thus, beliefs represent an individual's sense of ability to per- discrepancy mentally prepares managers for change and form the behavior; normative beliefs represent an indi- forms a basis for supportive change behaviors vidual's of the social pressure to perform a (Armenakis et al., 2000). given behavior. Appropriateness refers to the belief that a specific cor- rective action will remedy a problem (Armenakis et al., 2007). In this sense, appropriateness points to the 3.1 | Behavioral beliefs perceived fit between the new practice and the organiza- tion's strategy and context (Bansal & Roth, 2000; According to Ajzen (1991), behavioral beliefs arise from Braunscheidel et al., 2011; Cordano & Frieze, 2000), a an individual's concerns regarding the consequences of a key tenant in the efficacy perspective. Rather than certain action. Managers may, for example, assess the assessing the benefits for themselves, managers evaluate value of adopting a practice for either themselves the change based on its implications for their unit's per- (e.g., advance their career or provide greater job auton- formance (Jacqueminet & Durand, 2020; Stevens omy) or their organization (e.g., solve a problem or et al., 2005). Because appropriateness relates to the per- improve performance). The more an individual believes ception of the usefulness of a practice, this belief encour- in the efficacy of a certain practice, the more committed ages managers to adopt the practice (e.g., Bendoly that individual will be to adopting the practice. Thus, et al., 2008; Bendoly & Cotteleer, 2008). behavioral beliefs are closely associated with the efficacy perspective, as individuals primarily evaluate a given change in terms of the efficacy of the practice. Behavioral 3.2 | Control beliefs beliefs can take three forms: valence, discrepancy, and appropriateness (Armenakis et al., 2007). Control beliefs—also referred to as self-efficacy—refer to Valence describes how attractive the outcomes of a an individual's “perceived ease or difficulty of performing particular change are perceived to be and captures an a behavior” (Ajzen, 1991: 188). In the context of practice individual's evaluation of gains and losses (Bartunek adoption, an individual's self-efficacy captures his/her et al., 2006; Kennedy & Fiss, 2009). Several studies indi- own belief in his/her ability to undergo the change and cate that managers care not only about monetary execute the methods and procedures associated with the rewards but also about status, power, recognition, and new practice (Armenakis et al., 2007; Stevens career development (Balogun et al., 2017; Done et al., 2005). Thus, control beliefs offer another efficacy- et al., 2011). Individuals also assess the potential nega- related logic, as managers evaluate their efficacy and that tive effects of changes in their own jobs. For example, of their team in terms of successfully coping with practice employees may question the increased workload and adoption. responsibility brought about by a new practice The higher an individual's belief in his/her self-effi- (Bartunek et al., 2006). Facing the challenge of keeping cacy, the more he/she supports change (Jimmieson the daily business running, managers' levels of valence et al., 2008). In this regard, a high sense of control trans- strongly affect their support for the practice adoption— lates into a commitment to change because individuals for example, in terms of the help they offer to their set ambitious performance goals (Tarakci et al., 2018), employees (Balogun, 2003). feel a sense of control over the adoption process ARELLANO ET AL. 5

(Vardaman et al., 2012), and handle stressful situations a “taken-for-granted” way of working. The adoption of well (Herold et al., 2007). In contrast, individuals with a practices by imitation enables managers to reduce the low level of belief in their self-efficacy frequently lack search costs and to avoid the risk of being the first to commitment because they consider their actions to be explore an unknown territory (Barreto & Baden- inconsequential in terms of achieving a particular out- Fuller, 2006; Chandler & Hwang, 2015; come (Vardaman et al., 2012). Jacqueminet, 2020). However, the literature does not explain whether mimetic pressures form the basis for actual commitment to a new practice or simply translate 3.3 | Normative beliefs into short-term supportive behaviors.

Normative beliefs refer to “the likelihood that important referent individuals or groups approve or disapprove of 3.4 | Belief configurations: From belief performing a given behavior” (Ajzen, 1991: 195) and cap- to commitment ture the desire to conform to the expectations of signifi- cant others. These beliefs correspond to the legitimacy Beliefs do not work in isolation. Instead, the theory of perspective in that they represent the legitimacy drivers planned behavior suggests that beliefs collectively repre- at the individual level. sent an individual's actual disposition (Ajzen, 1985, 1991, Concerning legitimacy, two important conceptual 2005). Originally, Ajzen's (2005: 118) theory argued that boundaries are worth noting. First, we must define legiti- the relative importance of the aggregated dimensions of macy in our context. Most studies perceive legitimacy to beliefs varies in different situations: “In some instances, be “a physical property” in the sense that it is “a charac- only one or two of the factors are needed to explain the teristic to be gained, increased, and lost” (Suddaby intention, while in others, all three factors are important et al., 2017: 453). In contrast, we view legitimacy as “a determinants. In addition, the relative weights of the perception” that “uses the metaphor of property as taste, three factors may vary from one person to another, or assessment, or judgment” (Suddaby et al., 2017: 463). Sec- from one population to another.” The literature on cogni- ond, we must clarify the boundaries within which indi- tion, attention, and strategic renewal also highlights the viduals seek legitimacy. In the context of multinational complex associations among the , perceptions, and firms, the “important others” for operations managers evaluations that managers hold and their impact on both are arguably senior managers at the headquarters and strategic choices and organizational change (Eggers & peers at sister facilities (Jacqueminet, 2020; Kostova Kaplan, 2013; Nadkarni & Barr, 2008). Thus, the means et al., 2008). Accordingly, operations managers' observa- through which different types of beliefs either foster or tions of the headquarters' and prestigious sister facilities' suppress individuals' behaviors are sophisticated. How- actions and communications provide the basis for their ever, most studies that apply the theory of planned legitimacy-related beliefs. behavior oversimplify the model assumptions of complex Further, two types of legitimacy pressures explain the associations due to methodological constraints. search for legitimization: coercive and mimetic pressures3 Given the limited scholarly understanding of how (DiMaggio & Powell, 1983). Coercive pressures stem from individuals construct the joint effects of the efficacy and individuals' beliefs regarding what others expect from legitimacy motives that explain commitment (Kennedy & them. These expectations foster individuals' commitment Fiss, 2009; Leseure et al., 2004; Rogers et al., 2007), a con- due to their desire to build or protect an image of confor- figurational approach can provide valuable new insights. mity (Bansal & Roth, 2000; Stevens et al., 2005). In orga- This perspective offers a structured and comprehensive nizations, the outcomes of coercive pressures are assessment of causally complex mechanisms that are cap- unknown, as they can trigger both compliant and rebel- tured by outlining the three tenets of causal complexity: lious behaviors (Balogun et al., 2017; Leseure conjunctural causations, equifinality, and asymmetric et al., 2004). For example, managers in subunits that are causality (Fiss, 2007; Ragin, 2008). dependent on resources provided by their headquarters First, conjunctural causation refers to the that may accept a corporate mandate to adopt a new practice combinations of multiple causal conditions may produce (Kostova & Roth, 2002), while managers who perceive an outcome of interest. The notion of conjunctural causa- the power of their headquarters to be limited, may choose tion is congruent with the assumption of the theory of to ignore such a mandate. planned behavior that the mere activation of certain Mimetic pressures occur when managers adopt a beliefs may prove sufficient to explain intention in cer- practice driven by their perception of the positive impacts tain situations (Ajzen, 1991) as well as with the literature of the practices adopted by others or when the practice is on managerial cognition that attests to the complex 6 ARELLANO ET AL. managerial mental structures that support strategic function of an explanatory condition may differ across decision-making (Helfat & Peteraf, 2015). Second, equi- configurations). The analytical logic underpinning QCA finality conveys the idea that multiple combinations can considers cases (i.e., rows in a data set) as configurations satisfactorily explain the same outcome. Although studies of conditions and argues that by comparing cases system- on belief configurations at the individual level remain atically and in a stepwise manner, researchers can elimi- uncommon, three notable exceptions (Bansal & nate any irrelevant and redundant conditions that are Roth, 2000; Jacqueminet & Durand, 2020; Straatmann unrelated to the outcome of interest. Thus, QCA follows et al., 2018) present different configurations that are Mill's (2011) logic of paired comparisons, arguing that based on combinations of underlying beliefs that lead to when two cases exhibit the same outcome and only differ the same outcome (i.e., adoption-supportive behaviors). in terms of one causal condition, the distinguishing con- Third, asymmetric causation implies that both the pres- dition is irrelevant in explaining the outcome. ence and the absence of certain attributes may be con- Between 1987 and 2000, QCA was only possible when nected to the outcome. In relation to the theory of using so-called crisp sets—that is, dichotomous condi- planned behavior, researchers have provided certain evi- tions that define cases as being either fully in (1) or fully dence that supports the notion of asymmetric causation out (0) of a set. In 2000, Ragin (2000) introduced fsQCA, between beliefs and behaviors (Bansal & Roth, 2000; a variant of QCA that enables researchers to include con- Straatmann et al., 2018). For example, Piderit (2000) ditions that range from 0 to 1. Between 2007 and 2011, argues that individuals may simultaneously experience several articles laid the conceptual, methodological, and positive and negative cognitive orientations toward prac- empirical foundations for the application of QCA in the tice adoption. management literature (e.g., Fiss, 2007, 2011). Due to the additional opportunities that QCA now offers to researchers, there is an increased number of studies that 4 | RESEARCH METHOD employ QCA in the management literature. QCA involves five steps. First, the outcome is defined. To explore the complex associations between managers' Second, the explanatory conditions are selected. Third, belief configurations and high levels of commitment to both the outcome and explanatory conditions are cali- operational practices, we employ a method capable of brated. Calibration, which is the process of transforming handling the configurational perspective—fsQCA. We variables into set membership scores, assigns a degree of obtained our data by surveying operations managers membership of every case in a pre-defined set. Fourth, working for multinational companies. We targeted opera- the calibrated data are transferred to a truth table that tions managers employed by global firms because they lists all the logically possible combinations of both absent are not only in a good position to evaluate the operations and present conditions. A truth table contains 2k rows of their own facility but also serve as boundary spanners (where k equals the number of conditions), with each between the top management and shop-floor employees row representing a logical configuration. Fifth, the Bool- as well as between their facility and sister facilities. Their ean minimization process is employed to systematically boundary position exposes these managers to both effi- compare the configurations and identify the most parsi- cacy and institutional pressures related to the adoption of monious configurations that explain the outcome. new operational practices at their facilities. Finally, the research draws upon the insights provided by the QCA results and additional descriptive information through post-QCA case studies or descriptive statistics to 4.1 | Fuzzy-set qualitative comparative explain the results. The following subsections first analysis describe our data and then offer a stepwise explanation of our analysis. Charles Ragin (1987) introduced QCA intending to iden- tify configurations of explanatory conditions associated with the presence of an outcome. QCA draws on set the- 4.2 | Data ory and Boolean algebra to systematically analyze data (Ragin, 2008; Schneider & Wagemann, 2013). This Given the exploratory nature of our study as well as the method enables researchers to analyze complex causal cognitive elements included in our theoretical model, we associations characterized by conjunctural causation designed a survey-based study involving single respon- (i.e., multiple conditions in conjunction explain an out- dents at the individual level and purposefully selected come), equifinality (various configurations may lead to operations managers with substantive experience with the same outcome), and asymmetry (i.e., the contributing the adoption of operational practices. To select the ARELLANO ET AL. 7 respondents, we defined three criteria. The respondents practice adoptions mandated by their headquarters (26%) had to (1) hold managerial responsibilities; (2) work in a and those adoptions that were voluntary site-initiated facility with operational responsibilities that belongs to a (74%). Table 1 summarizes the demographic characteris- site network comprising at least three facilities in differ- tics of the sample. ent geographical regions, and (3) have recently partici- pated in the adoption of a new operational practice at their facility. We applied a snowball-sampling approach, 4.2.2 | Common method variance which involved contacting several operations managers in our network and asking them to share the survey As our causal conditions and the outcome variable are all among their professional networks. self-reported measures, we followed the approach of We developed a questionnaire to capture the different Podsakoff et al. (2012). We applied various procedures to beliefs held by operations managers and their commit- mitigate the risk of common method variance. Before ment to adopt an operational practice (see items in data collection, we designed the survey in a way that Appendix B). In order to ensure that the respondents reduces possible sources of method bias. After data shared an understanding of what a “practice” was and to trigger reflective processes concerning the process of implementation, at the beginning of the questionnaire, TABLE 1 Sample demographics (N = 76) we asked them to name a recent process of practice adop- tion in which they had participated and to which they N Percentage would refer in their answers. We also explained that “A Respondent years with the company ‘practice’ is a solution, a method, a procedure, a project, <1 year 6 8.20 or an improvement effort (depending on the vocabulary 1–5 years 29 39.70 used within the company) regarding the management of 6–10 years 18 24.70 the operations of the facility. Adopting a practice involves a change in the way of working and can manifest in dif- Over 10 years 20 27.40 ferent forms.” Further, we provided a few examples of Respondent age well-known practices (e.g., lean, total quality systems, Six Less than 25 years old 0 0.00 Sigma) and indicated that a practice may also have a pro- 25–35 years old 27 37.00 prietary name (e.g., the Coca-Cola Quality System). To 36–45 years old 31 42.50 increase the construct validity, we discussed the format Over 45 years of age 15 20.50 and content of the questionnaire with three operations Facility role management scholars and then pre-tested it with six operations managers. Level 4: Leader/hub of knowledge 45 60.00 Level 3: Supply chain management 16 21.30 Level 2: Production improvement 6 8.00 4.2.1 | Sample demographics Level 1: Production only 8 10.70 Facility location From a total of 106 initial responses, we derived 76 usable South America 32 47.40 responses from operations managers working in facilities run by various multinational firms. Most respondents North America 31 39.50 (60%) worked in facilities that played a lead operational Rest of the world 10 9.20 role within their firm, 21% of them worked in facilities Practice content type with principal responsibility for managing the supply Operational excellence/lean 44 57.89 chain, while 19% of them worked in facilities only Information and technology 23 30.26 responsible for production. Approximately half the facili- management ties were located in South America (47.4%), with the Certifications 5 6.58 second-largest group being located in North America Product/process redesign 4 5.26 (39.5%) and the remainder (9.2%) in Europe or Asia. The Practice adoption type respondents had adopted several types of operational practices, including operational excellence or lean Mandated by headquarters 20 26.30 (57.9%), information and technology management sys- Voluntarily initiated by the facility 56 73.70 tems (30.3%), certifications (6.6%), and product or process Notes: The N (number) of a few subgroups does not add up to 76 due to redesign (5.2%). The respondents also reported on missing values. We report only valid percentages. 8 ARELLANO ET AL. collection, we conducted two post-hoc statistical tests— Given the limited quantitative research available on these the Harman's single-factor test and the confirmatory fac- types of institutional pressures, we adapted the scales tor analysis (CFA) marker variable procedure. Appendix used in prior studies concerning the role of institutional C.1 includes a detailed explanation of our procedures. and external pressures in practice adoption (Kostova & Taken together, the results suggest that the presence of Roth, 2002; Liu et al., 2010). Our measures are based on common method bias is not a concern in our model. an exploratory factor analysis of 11 items that relate to managers' perceptions of the pressures originating from both the headquarters and the sister facilities (see Appen- 4.3 | Analyzing belief configurations by dix C.2). The reliability of the resulting scales was found means of fsQCA to be high for coercive pressures (α = 0.77) and mimetic pressures (α = 0.89). 4.3.1 | Step 1: Defining the outcome Thereafter, we tested the measurement model by including all the latent factors and their corresponding Our outcome measure is an operations manager's level of items. This CFA, which we conducted in AMOS 25, evalu- commitment to a practice that is new to the plant. The ates the convergent and discriminant validity of the vari- practice entails a new way of working in the eyes of the ables. Given our small sample size (N = 76), the number manager at the outset of the adoption. We used Kostova of factors, the number of indicators per factor, and the and Roth's (2002) validated scale of commitment to qual- magnitude of the factor loadings required for the CFA ity practices after adjusting the scale to our research con- (Wolf et al., 2013), we used the item-parceling approach text (e.g., because they studied a specific quality process, to estimate the model. This approach combines items we replaced “quality process” with “this practice”). into subsets (or parcels), which are subsequently treated We measured the construct reliability using as indicators of the latent construct. Each parcel is con- Cronbach's alpha (α) and established the high reliability structed as the average of the measured items (see of our adjusted commitment scale (α = 0.90). In addition, Appendix C.3). The item-parceling approach reduces the we also followed Kostova and Roth's (2002) procedure for number of estimated parameters and also tends to assessing nomological validity. We examined the correla- decrease the measurement error (Bagozzi & tions between our commitment measure and a measure Edwards, 1998). of the respondents' psychological ownership of the prac- The validity and the reliability of the measurement tice (an example item is “I feel that this is my practice”), model are supported by the CFA results: χ2/df = 1.276, which also reflects an individual's approval of the prac- CFI = 0.94, TLI = 0.93, RMSEA = 0.06 (90% CI: 0.03– tice in question. The measures were found to be signifi- 0.08), and SRMR = 0.079. For small sample sizes, a com- cantly correlated (r = 0.61, p < 0.01), which provided bination of CFI >0.95 and SRMR <0.08 results in a low additional support for the validity of our outcome. sum of Types I and II errors (Hu & Bentler, 1999). Appendix B summarizes the outcome and explanatory conditions for the efficacy and legitimacy beliefs, includ- 4.3.2 | Step 2: Selecting the explanatory ing their composite reliability (CR) and average variance conditions extracted (AVE) estimates. The convergent validity of each construct is evidenced by each manifest item load- To operationalize the efficacy-related beliefs, we adopted ing significantly (all the loadings >0.50, p < 0.001; four measures from Armenakis et al.'s (2007) validated val1 > 0.50, p < 0.01). To further test the discriminant scales of the beliefs of organizational change recipients.4 validity, we used the heterotrait-monotrait (HTMT) tech- Valence captures how attractive a manager perceives the nique (Henseler et al., 2014) with a cutoff value of 0.85. outcome of a given change to be (α = 0.60). Discrepancy The HTMT technique has been shown to achieve the best measures the manager's belief in the need for change balance between high detection and low arbitrary viola- because he/she perceives there to be a mismatch between tion rates (Voorhees et al., 2015). As indicated in Appen- the current and desired states (α = 0.71). Appropriateness dix C.4, no warnings were issued. captures the manager's belief that a specific action is appropriate for solving the problem at hand (α = 0.80). Finally, self-efficacy measures the manager's perception of 4.3.3 | Step 3: Calibrating the outcome her ability to adopt the practice (α = 0.82). and conditions To operationalize the legitimacy-related beliefs, we developed two scales for perceived coercive pressures and As fsQCA analyzes set intersections, all the measures perceived mimetic pressures (DiMaggio & Powell, 1983). must be transformed into set membership scores. A ARELLANO ET AL. 9 fuzzy-set membership score ranges from 0 to 1, with latest version of the fs/QCA software (version 3.0; three defined anchor points: the full membership, the full Ragin & Davey, 2017) and added a constant of 0.001 to non-membership, and the crossover point (Ragin, 2008). all 0.5 values to avoid the configurations being mechanis- In large-sample QCA studies in which researchers are tically excluded by the software. We named the sets of usually less familiar with the individual cases as well as conditions in our study in a manner that reflected our in studies in which the meaning of a measure makes it calibration approach (e.g., the set of cases with “above- difficult to calibrate sets based on substantive and theo- average belief in valence”). Table 2 presents the descrip- retical knowledge (e.g., cognition), researchers have occa- tive statistics and correlation matrix associated with our sionally opted to directly calibrate the sets by drawing on measures. the distribution of the measures (Fiss, 2011; Meuer et al., 2015). To calibrate our measures, we first considered the 4.3.4 | Step 4: Analyzing the truth table possible approaches to calibration. We evaluated them in light of the possibility of drawing on external anchors The truth table, which lists all the logically possible com- (substantive or theoretical), the nature of our constructs binations of the six explanatory conditions, has 26 =64 and measurements, and the skewness of the distribution rows (an abbreviated truth table is provided in Appendix within our sample. We opted to use the sample distribu- C.6). We found empirical evidence for 35 configurations tion to identify the cut-off points (see Appendix C.5 for a (i.e., managers in our data set exhibited 35 out of the pos- detailed description of our assessment process). More sible 64 belief configurations). However, we found no evi- concretely, we calibrated the 75th percentile to set the dence for 29 configurations, which resulted in a situation full membership, the 25th percentile to set the full non- known as “limited diversity,” in which not every logically membership, and the mean to set the crossover point.5 possible combination of conditions has an empirical Further, we used the calibration function available in the representation.

TABLE 2 Descriptive statistics

Non- calibrateda Calibrated measuresb Correlations of non-calibrated measuresc

Neither fully in nor Fully fully Fully Measures Mean SD out (0) out (0.5) in(1)1234567 Explanatory measures: Beliefs Efficacy-related beliefs 1. Valence 3.76 0.63 ≤3.50 3.76 ≥4.25 0.62 2. Discrepancy 4.40 0.42 ≤4.00 4.40 ≥4.75 0.35** 0.67 3. Appropriateness 4.16 0.56 ≤4.00 4.16 ≥4.60 0.36** 0.61** 0.77 4. Self-efficacy 4.15 0.53 ≤4.00 4.15 ≥4.40 0.40** 0.50** 0.67** 0.80 Legitimacy-related beliefs 5. Coercive 4.21 0.57 ≤3.81 4.21 ≥4.75 0.07 0.22 0.17 0.20 0.73 pressures 6. Mimetic 3.69 0.81 ≤3.25 3.69 ≥4.19 0.04 0.08 0.24* 0.19 0.50** 0.86 pressures Outcome measure 7. Commitment 4.24 0.54 ≤4.00 4.24 ≥4.67 0.45** 0.60** 0.60** 0.61** 0.21 0.17 0.88 8. Marker variable 3.19 1.29 N.A. N.A. N.A. 0.09 −0.08 −0.09 0.04 0.01 0.14 0.09

Note: *Pearson's correlations statistically significant at the 0.05 level (two-tailed); ** correlations statistically significant at the 0.01 level (two-tailed). aAll the measures were collected using a five-point Likert-type scale. bThe 75th percentile and 25th percentile are the anchors for full membership and full non-membership, respectively. The mean is the crossover point. cSquare root of the AVE in the diagonal. 10 ARELLANO ET AL.

4.3.5 | Step 5: Boolean minimization several important background factors between the entire sample (all cases in our analysis) and each of the The truth table also provides information regarding the raw resulting belief configurations. We used a manager's consistency, the proportional reduction in inconsistency membership score to allocate cases to the belief configu- (PRI), and the frequency of each configuration. This infor- rations then conducted the one-sample t-tests to assess mation is used to select the configurations for the subse- whether the configuration in focus is specific to a back- quent Boolean minimization. The raw consistency can be ground factor. The use of one-sample t-tests ensures the of as the proportion of cases in a truth table row comparability of our findings across different belief con- that displays the outcome.6 An alternative measure, the figurations because we compare each configuration PRI, indicates the degree to which a configuration is not against the “average” committed manager (for an exam- simultaneously sufficient for the occurrence and non-occur- ple of a similar application, see Meuer et al., 2015). rence of the outcome—guarding against the reporting of According to the theory of planned behavior, the fun- false positive configurations. For our analysis, we followed damental information regarding the determinants of a recent research (Greckhamer, 2016; Schneider & given behavior is contained within the three beliefs of the Wagemann, 2013) and examined the truth table of our data; model, for contextual variables “will be related to behav- we chose to include only those configurations that exhibited ior if, and only if, they influence the beliefs that underlie a raw consistency threshold of at least 0.79 and a PRI the behavior's attitudinal or normative determinants” threshold of at least 0.67. The frequency threshold specifies (Ajzen, 1985: 14). Nonetheless, Ajzen (1991) indicates the minimum number of cases required for a configuration that researchers can include factors relevant to the to be considered important in terms of explaining an out- behavior under investigation. Based on Ajzen's recom- come of interest. In accordance with comparable studies, mendations, the literature concerning knowledge transfer we used a frequency threshold of one case (e.g., Meuer and adoption, and the global context in which managers et al., 2015; Reimann et al., 2017), which ensured a rec- work, we included a set of background variables on three ommended inclusion rate of above 80% (Ragin, 2008). Over- levels: individual level, organizational unit (i.e., facility) all, 32 cases were categorized as configurations that level, and network level. exceeded both the consistency and frequency thresholds. Appendix C.7 provides detailed information on the By minimizing the configurations, QCA enables items that we used to measure the background variables. researchers to formulate assumptions, or directional expecta- At the individual level, the two demographic variables tions, in order to generate three types of solutions: conserva- were manager seniority (1: <1 year, 2: 2–5years, 3: tive, intermediate, and parsimonious solutions (Schneider & 6–10 years, and 4: >10 years) and manager age Wagemann, 2013). First, the conservative solution only (1: <25 years, 2: 25–35 years, 3: 36–45 years, and 4: includes the truth table rows sufficient for the outcome. Sec- >45 years). We also accounted for the managers' negative ond, the intermediate solution only includes logical remain- emotions at the onset of the implementation (an example ders if they are in line with the directional expectations item is “This whole adoption made me feel kind of specified by the researcher. By formulating directional upset”), which is not uncommon in a situation of change expectations, a researcher expresses informed assumptions (Oreg, 2003). Finally, to account for everyday attitudes regarding unobserved configurations (i.e., logical reminders) toward change, we included managerial proactivity, which and, thus, engages in a form of counterfactual analysis is “the extent to which an individual takes self-directed (Schneider & Wagemann, 2013). Based on our literature action to anticipate or initiate change in the work system review, we expect that the presence of efficacy-related beliefs or work roles” (Griffin et al., 2007: 329) measured with a contributes to operation managers' commitment and that nine-item construct (5-point Likert scale). legitimacy-related beliefs may be either present or absent. At the organizational level, we considered that man- Third, the parsimonious solution minimizes the sets of causal agers' beliefs might be influenced by their position as conditions across the entire truth table (i.e., both observed managers within intra-firm networks. In this regard, we and unobserved configurations, irrespective of how tenable included two variables. First, we considered the role of the configurations are) to identify the configurations with the organizational unit within the network, which we the lowest number of conditions. measured on a scale ranging from 1 (fulfillment of basic tasks to “get the job done”) to 4 (a center of excellence that actively participates in the development of strategic 4.4 | Tests of variable influence capabilities). Second, we considered the importance of the organizational unit's location (5-point Likert scale) To contextualize the fsQCA results, we conducted one- regarding access to low-cost resources, proximity to mar- sample t-tests to compare the differences in the means of ket, and access to knowledge and skills. ARELLANO ET AL. 11

At the network level, we selected two factors consid- 5.1 | The Follower ering that the intra-firm network exposes managers to a range of information and knowledge, which can influ- We label the first belief configuration presented in ence their beliefs. The first factor is the network size that Table 3 “the Follower” due to the importance of its two refers to the number of facilities with the same main legitimacy-related beliefs. The Follower's core condi- responsibility (1: <5, 2: 5–10, 3: 11–20, 4: 21–30, and 5: tions include above-average beliefs in self-efficacy, >30). The second factor is the scope of the variable net- appropriateness, and perceived coercive pressures. work that differentiates none (1), domestic (2), regional These beliefs are complemented by an above-average (3), and global (4) networks. perception of strong mimetic pressures as a peripheral condition. This suggests that the perceived coercive pressures function together with other beliefs to activate 5 | RESULTS managerial commitment. Followers' high levels of self- efficacy (i.e., their perceived ability to adopt the prac- As is typical for QCA studies, we ran a necessity analysis tice) make “following” an efficient strategy when the before conducting the sufficiency analysis. The necessity practice appears to fit well with the objectives of their analysis revealed that no efficacy- or legitimacy-related organizational unit (Choo et al., 2015; Jacqueminet, belief is by itself necessary for managers to show high levels 2020). This implies that when Followers commit to and of commitment.7 Table 3 reports the results of the suffi- then implement a practice, they do so to boost their self- ciency analysis regarding the presence of above-average image, which also motivates them to engage in further managerial commitment to a new operational practice.8 improvement (Tarakci et al., 2018). Moreover, the The leftmost column of Table 3 lists the explanatory above-average levels of self-efficacy appear to reduce the measures. The remaining columns summarize the belief feeling of coercion suggesting that managers may inter- configurations associated with above-average levels of com- pret coercive pressures as a source of empowerment and mitment. Table 3 also lists the consistency, raw coverage, corporate support for the new practice. In addition, and unique coverage scores for each configuration. We only motivated by others' successful adoptions (perceived report those configurations that have a unique coverage mimetic pressures), Followers reporting an increased (i.e., the share of cases that uniquely take a given path to self-efficacy belief seem to experience a reduction of the outcome) of above 0.01. The overall solution consis- their overall fear and stress (Vardaman et al., 2012), tencyofourresultsis0.90,whichissubstantiallyabovepre- which, in turn, stimulates their commitment viously reported coefficients of consistency. Furthermore, (Armenakis et al., 2000: 650). the overall solution coverage stands at 0.72, which is in line Considering the contextual variables, we focus on find- with other studies using fsQCA (e.g., Fiss, 2011). ings with statistical significance at the 1 and 5% levels in We identify three belief configurations for high levels Table 4. For the Follower, this is the case for skill access of commitment that differ with regard to their combina- from the facility's location advantage (compare, 0.39***), tions of core conditions (large circles in Table 3). We the manager's proactivity at the company level (0.44**), label these the Follower, the Pragmatist, and the and the manager's intention at the onset of the implemen- Reformer. For two belief configurations—the Pragmatist tation (0.44***). First, a facility established to access skills and the Reformer—our results identify two and three must be proficient in absorbing lessons from others. Thus, neutral permutations, respectively, which represent com- when Followers perceive high appropriateness and self- binations of peripheral conditions that surround and efficacy, they do not hesitate to adopt new practices. Sec- reinforce the core features of the first-order belief config- ond, we expect Followers' strong belief in appropriateness urations9 (Fiss, 2011). The results of the tests of variable to be connected to the managers' generally proactive orien- influence are presented in Tables 4 and 5. tation toward following company policies. Third, Fol- Taken together, Tables 4 and 5 indicate that one or a lowers exhibit a more positive intention than the average few contextual variables do not primarily drive our manager in the sample, which may help them to interpret results. In Table 4, the t-tests indicate that only a few var- the perceived coercive pressures as a meaningful encour- iables' means differ significantly between a specific belief agement when it comes to implementing effective prac- configuration and the full sample mean. We return to the tices that strengthen the legitimacy of their facility. nuances in Table 4 in our presentation of the resulting Therefore, contrary to studies that report legitimacy pres- configurations. In Table 5, the cases appear to be ran- sures to reduce the level of managerial commitment domly distributed among the configurations in terms of (Balogun et al., 2017; Kostova & Roth, 2002; the facility location, the type of practice, the decision- Prajogo, 2011), we find that Followers are highly commit- maker for the adoption, and the origin of the practice. ted despite (or because of) legitimacy pressures. 12

TABLE 3 Configurations for achieving above-average managerial commitment to a new operational practice

Notes: A full circle indicates that the presence of an explanatory measure is important in terms of explaining commitment. A crossed-out circle represents the same but in relation to absence. Large circles represent core ARELLANO conditions, which are central to explaining managers' commitment. Small circles represent peripheral conditions, which contribute to commitment. Blank spaces represent “don't care,” which indicates that a certain condition may be present or absent (i.e., it is irrelevant) when explaining commitment. The table reports all cases with a membership score >0.50. Because a case can be a member of more than one configuration, the overall number of cases may exceed the number of cases reported in the truth table. TAL ET . ARELLANO ET AL. 13

TABLE 4 Differences in the means across the three belief configurations

1. The 2. The Descriptive statistics Follower Pragmatist 3. The Reformer

2a. 2b. 3a. 3b. 3c. Variables Mean SD Min./max. One sample t-tests of the belief configurations Contextual factors Network size 2.75 1.58 1.00/5.00 −0.20 2.25a 0.25a 0.05 0.00 −0.25 Network scope 3.34 0.82 2.00/4.00 0.10 0.66a 0.66a −0.08 −0.13 −0.74 Facility role in the network 3.63 1.30 1.00/5.00 0.31 0.37a 0.37a 0.11 0.16 0.17 Facility location advantage: Access 3.42 1.13 1.00/5.00 0.14 0.58a −1.42a 0.32 0.15 0.58 to low-cost resources Facility location advantage: 3.77 0.91 1.00/5.00 0.23 0.23a −0.77a 0.18 0.23 0.83** Proximity to market Facility location advantage: Access 4.11 0.90 1.00/5.00 0.39*** 0.89a 0.89a 0.15 0.25 −0.11 to knowledge/skills Managers' characteristics Manager's years with the firm 2.71 0.96 1.00/4.00 0.02 −0.71a 0.29a 0.03 −0.14 −0.31 Manager's age 2.84 0.75 2.00/4.00 −0.17 −0.84a −0.84a −0.16 −0.34** −0.04 Manager's proactivity at the 3.88 0.71 1.33/5.00 0.08 0.12a −0.21a 0.24 0.23 0.39 individual level Manager's proactivity at the facility 3.88 0.73 2.00/5.00 0.22 1.12a 0.12a 0.17 0.27 0.59* level Manager's proactivity at the 3.81 0.81 1.67/5.00 0.44** 1.19a 0.19a 0.34* 0.43** 0.52 company level Negative affection at the onset of the 2.65 0.94 1.00/4.67 −0.41* 2.02a 1.02a −0.22 −0.38 0.08 implementation Intention at the onset of the 4.40 0.70 1.75/5.00 0.44*** 0.35a 0.60a 0.45*** 0.48*** 0.45** implementation

Notes: * p < 0.1; ** p < 0.05; *** p < 0.01. at-tests are not possible because configurations 1a and 1b have one case each. We report the arithmetic difference of the means. In Appendix C.10, we present the results of additional one-sample t-tests based on only the first-order belief configurations (1. The Follower, 2. The Pragmatist, and 3. The Reformer).

5.2 | The Pragmatist The absence of an above-average belief in discrepancy suggests that Pragmatists do not pioneer change but tend The Pragmatist predominantly exhibits efficacy-related to see the actual ways of working as satisfactory beliefs. This configuration is characterized by the presence (Loch, 2017). The absence of urgency encourages a utili- of above-average beliefs in valence and appropriateness tarian approach to supporting the change in which the and the absence of an above-average belief in discrepancy. manager does a cold calculation of threats and opportuni- The Pragmatist commits to a new practice by prioritizing ties (Bartunek et al., 2006; Kennedy & Fiss, 2009). The the benefits of the change and does so intending to secure belief configuration capturing the Pragmatists has two both facility gains—appropriateness—and personal neutral permutations that we label the Legitimacy- gains—valence—(e.g., Bansal & Roth, 2000; Bartunek seeking Pragmatist when the perceived legitimacy drivers et al., 2006; Prajogo, 2011). Pragmatists may, for example, are above average (solution 2a) and the Entrepreneurial consider that recognition will result from the increased Pragmatist when they are below average (solution 2b). performance they envisage following the adoption of a Because only two cases uniquely correspond to the new practice (Done et al., 2011). In this sense, this mana- Pragmatist, we seek to identify differences in means that gerial profile most closely resembles the traditional view of are larger than one standard deviation from the full sam- a rational, goal-seeking manager, who maximizes effi- ple when cautiously interpreting the variable influence ciency by ensuring the technical fitness of the practice tests in Table 4. In our study, the Legitimacy-seeking (Jacqueminet & Durand, 2020). Pragmatist works in a company with a large network size 14 ARELLANO ET AL.

TABLE 5 Distribution of cases The Follower The Pragmatist The Reformer across the contextual variables Variables 12a2b3a3b3c Facility location North America 9 – 1 6 6 1 South America 4 1 – 10 6 4 Rest of the world 2 – – 3 2 – Type of practice Operational Excellence/lean 12 – – 12 9 1 Information technology 3 – – 4 3 1 Product/process redesign 2 1 1 3 2 1 Certifications – – – 1 1 1 Decision-maker for the adoption The manager's facility 13 1 – 12 11 3 Headquarters 4 – 1 8 4 2 Origin of the practice The manager's facility 2 – – 2 2 – Sister facility 4 – 1 10 5 3 Headquarters 11 1 – 8 8 2

Notes: This table lists the distribution of cases across the variables listed in the Variables column.

(2.25 difference in the mean from the full sample) and is The Reformer suggests a path toward commit- notably younger than the average manager (−0.84). This ment to practice adoption that has hitherto been left Pragmatist displays a relatively higher level of proactivity largely unexplored in the practice adoption literature. at the facility (1.12) and company (1.19) levels compared This configuration challenges prior assumptions in to that at the individual level and exhibits a negative that managers do not (only) focus on the benefits of affection at the onset of the practice implementation the practice adoption to the organization or to them- (2.02). Altogether, these insights point toward managers selves. A strong belief in self-efficacy is associated with little experience who seek legitimacy in a relatively with the freedom to experiment, and it heightens a large network and who are possibly worried about having manager's self-confidence to innovate (Tarakci to comply with pressures from superiors and peers et al., 2018). Thus, it appears that the Reformer's related to practice adoption. On the other hand, the desire to improve operations—discrepancy—prompts Entrepreneurial Pragmatist shows a negative pre- them to view the adoption of a new practice as an implementation affection toward the practice (1.02), is opportunity to learn via iterative experimentation responsible for a facility in a high-cost country (−1.42), and problem-solving (Cantor & Macdonald, 2009; and is relatively young (−0.84). These characteristics sug- Choo et al., 2015). gest that the Entrepreneurial Pragmatist becomes com- We uncover three neutral permutations of the mitted to a new practice only under conditions of low Reformer. The belief in the necessity of change and desire legitimacy pressures. A possible explanation is that their for differentiation go hand-in-hand in each permutation. lack of experience and high-cost location urge them to The Company-oriented Reformer (configuration 3a) is focus on their facility's performance and to adopt unique motivated by the opportunity to support the facility— practices. appropriateness. The Self-oriented Legitimacy-seeking Reformer (configuration 3b) seeks a change that enables taking advantage of personal opportunities—valence— 5.3 | The Reformer and would also readily adopt practices in an effort to please the headquarters—perceived coercive pressures. The third belief configuration, the Reformer, is character- The Self-oriented Entrepreneurial Reformer (configura- ized by an above-average belief in discrepancy and an tion 3c) also seeks personal returns—valence—but above-average belief in self-efficacy. Reformers believe ignores or does not perceive peer pressure—below-aver- that change is necessary and, further, that they possess age perceived mimetic pressures—thereby indicating his/ the skills and knowledge required to implement the prac- her disposition to favor practice differentiation over tice. Thus, they generally appear to be “open to change.” replication. ARELLANO ET AL. 15

Considering the analysis of the contextual variables 6 | DISCUSSION included in Table 4, we again use the cut-offs of the 1 and 5% confidence intervals. All the three permutations of the Most prior research on practice adoption has considered Reformer (Self-oriented, Legitimacy-seeking, and Entre- operations managers to be “technical middleware” who preneurial) report a higher intention to adopt the practice are responsible for the successful adoption of operational than the entire sample (cf. 0.45***, 0.48***, and 0.45** in practices (e.g., Ketokivi & Schroeder, 2004; Rogers Table 4, respectively), which was also the case for the Fol- et al., 2007). The efficacy and legitimacy perspectives por- lower. This indicates that our belief measures also capture tray managers as hyper-rational actors or as passive and intention, which is a central construct in the theory of obedient agents, respectively. A vast literature on planned behavior (Ajzen, 1991). Besides, the Self-oriented organizational-level practice adoption emphasizes the Legitimacy-seeking Reformer in our sample is relatively commitment of managers as a key success factor younger than the sample mean (−0.34***), thus suggesting (e.g., Done et al., 2011; Kostova & Roth, 2002; Netland that such a manager is more susceptible to conforming to et al., 2015), yet cognition at the individual level has company-level rules and programs—perceived coercive rarely been studied (Bendoly et al., 2008; Loch, 2017). We pressures. Indeed, managers who exhibit this belief config- challenge the assumptions of “the ideal profile” of a com- uration also appear to be more proactive at the company mitted operations manager by acknowledging the possi- level (0.43**), thereby indicating their eagerness to please. bility of cognitive heterogeneity. Compared to prior Finally, the Self-oriented Entrepreneurial Reformer literature that has focused on independent effects, our reports that the main advantage of the facility's location emphasis on complex causality enabled the identification lies in its proximity to the markets (0.83**), which perhaps of different belief configurations that lead to a high man- causes such a manager to value a standard operational agerial commitment to a new practice. We discuss our practice low and to seek alternative local practices. This is contributions and our study's implications for research in line with the absence of strong mimetic pressures in this and practice in the following sections. profile.

6.1 | Rethinking the role of the efficacy 5.4 | Robustness tests and legitimacy perspectives

We conducted a comprehensive set of robustness tests Our study contributes to the unresolved debate surround- to assess how sensitive our results were toward ing efficacy versus legitimacy drivers (Barreto & Baden- changes in the model parameters and alternative Fuller, 2006; Kennedy & Fiss, 2009; Ketokivi & approaches to calibration. First, we tested the robust- Schroeder, 2004; Leseure et al., 2004). The first contribu- ness of our results in relation to a more lenient model tion relates to the omnipresent role of efficacy drivers in (consistency cutoff = 0.75, Appendix C.11, Test 1) and relation to practice adoption. We add important details to two more conservative models (frequency cutoff = 2; the overall understanding of how efficacy drivers affect consistency cutoff = 0.8; Appendix C.11, Tests 2 and managerial commitment. The literature on practice adop- 3). The results reveal that all the configurations are tion assumes that operations managers select and imple- stable in the lenient model. The two conservative tests ment operational practices almost entirely based on their indicate that the Pragmatist is highly stable toward efficacy-related beliefs—the beliefs in the improvements changes in the model parameters, the Follower is in operational performance that the practice adoption robust toward a shift in the frequency cutoff but sensi- would bring. Our results indicate that this dominant view tive to more conservative consistency cutoffs, while the offers a fair approximation of some managers' motiva- Reformer is robust to consistency but sensitive to tions but does not accurately represent a commitment to frequency. practice adoption in general. Second, we conducted seven robustness tests using On the one hand, the efficacy perspective has merit alternative calibration approaches (see Appendix C.11, because efficacy-related beliefs are omnipresent in all the Tests 4–10). The results indicate rather good stability of configurations. We found a generally stronger influence the Reformer across all the models and demonstrate the of efficacy factors (core conditions) when compared with specific sensitivities of the Follower and the Pragmatist. legitimacy factors on operations managers' commitment Overall, the robustness tests suggest that our results to a new practice (even after discounting the fact that remain substantively unchanged, thereby corroborating there are more efficacy-related beliefs). Overall, the three the cognitive configurational perspective developed in identified configurations suggest different levels of ratio- this study. nality. Most closely corroborating the efficacy perspective 16 ARELLANO ET AL. on practice adoption, the Pragmatist and the Reformer & Baden-Fuller, 2006; Rogers et al., 2007). Only one con- account for approximately two-thirds of the operations figuration—the Company-oriented Reformer—emerges managers included in our study. Between the two, the exclusively from efficacy-related beliefs. The other five Pragmatist—who reports a strong belief in belief configurations encompass both efficacy- and appropriateness—most closely resembles the hyper- legitimacy-related beliefs. If we consider only the first- rational actor who focuses on the importance of a prac- order belief configurations, the Follower combines the tice (Jacqueminet & Durand, 2020). presence of the two types of beliefs. Relatedly, although On the other hand, the efficacy perspective fails to the most popular premise concerning the theory of grasp the different forms of rationality that drive the final planned behavior is that different behaviors may depend commitment to practice adoption. Although the Fol- on different combinations of beliefs, our findings also lower, the Pragmatist, and peripherally also the reveal that the same behavior may rely on different combi- Company-oriented Reformer rationally evaluate an oper- nations of beliefs. Furthermore, the absence of conditions ational practice concerning its potential improvement is also important (e.g., discrepancy for the Pragmatist), effect for the facility (appropriateness), other beliefs also which supports the importance of adopting a behavioral contribute to their commitment. For example, the Prag- perspective enabling a careful exploration of the presence matist also evaluates the personal benefits that are likely or absence of beliefs. to be obtained—valence—while the Follower and the Taken together, our findings suggest that researchers Company-oriented Reformer pay attention to how well must not consider efficacy- and legitimacy-related argu- they believe they can implement the practice—self-effi- ments in isolation, not because the arguments inherent cacy. Furthermore, our results suggest that the Follower to both perspectives are wrong but rather because the and the Company-oriented Reformer appear to prioritize various interdependencies that exist among the two per- collective benefits (i.e., they are committed when it is spectives are ultimately more important in terms of good for the company) over individual benefits, thereby explaining commitment to a new practice than the inde- hinting at a relationship between belief structures and pendent links between them. The tests of variable influ- organizational identity (Jimmieson et al., 2008; Tarakci ence suggest that this coexistence holds irrespective of et al., 2018). background factors. Hence, our study corroborates that Our second contribution relates to the role of legiti- commitment results from operations managers' own macy drivers. Our results provide limited support for the interpretations, that is, their salient beliefs, rather than legitimacy perspective. Most notably, the Follower, who organizational-level variables per se (Eggers & has coercive pressures at his/her core, most clearly Kaplan, 2013; Helfat & Peteraf, 2015). resembles the manager portrayed by the legitimacy per- spective. The Legitimacy-seeking Pragmatist and the Self- oriented Legitimacy-seeking Reformer also exhibit legiti- 6.2 | Specifying the role of self-efficacy macy-related beliefs, albeit only as peripheral conditions. in the commitment to practice adoption Yet, none of the belief configurations that exhibit the presence of a certain level of legitimacy-related factors Our fourth contribution is to provide new insights into the explain managers' commitment without the inclusion of role of self-efficacy in the belief configurations that are rel- efficacy-related beliefs. Thus, we note that legitimacy evant to the commitment to a new practice. This explana- pressures are occasionally of some importance, although tory condition appears in four out of the six configurations not in their own right. for achieving above-average managerial commitment to a Our third contribution relates to the different roles of new operational practice. The prior literature concerning each of the beliefs to managerial commitment. Although operational practices has largely neglected the role played all beliefs are present in at least one of the belief configu- by self-efficacy in practice adoption, although a few excep- rations, their role differs from one configuration to tions have provided initial evidence that self-efficacy another. Contrary to previous studies asserting that com- relates positively to managers' adoption intention. For mitment is triggered by the additive strength of the example, Straatmann et al. (2018) identify it as a condition beliefs, our results show that every belief performs a dif- for change intention, Jimmieson et al. (2008) as a periph- ferent function. In this regard, our findings concerning eral condition, while Cordano and Frieze (2000) note a both the efficacy and legitimacy perspectives on man- negative effect on the intention to adopt practices in the agers' commitment to practice adoption strongly chal- short term. Our findings indicate that an above-average lenge the implicit assumption in the literature regarding belief in self-efficacy is of importance for both the the adoption of either an efficacy perspective or alegiti- Reformer (largely efficacy-dominated) and the Follower macy perspective on practice implementation (e.g., Barreto (efficacy- and legitimacy-dominated). ARELLANO ET AL. 17

The above-average commitment exhibited when man- contexts of scheduling, forecasting, inventory control, agers perceive themselves as skillful suggests that control risk management, and new product development beliefs act through two mechanisms. First, self-efficacy (Fahimnia et al., 2019). However, in the context of prac- appears to buffer legitimacy pressures (in the Follower tice adoption, this is a tremendously difficult task due to case) and to foster both experiential learning and entre- the limited understanding of the cognitive underpinnings preneurial spirit (in the Reformer case). The only belief of key actors. Our study helps to disentangle the complex configuration in which self-efficacy does not matter is the interdependence of the beliefs that influence the commit- Pragmatist. In this particular case, it appears that above- ment, and hence behavior, of operations managers. average beliefs in valence and appropriateness are suffi- Our results suggest that corporate managers who seek cient to drive high commitment levels to adoption. to encourage practice adoption must be aware of the dif- ferent possible means to prompt operations managers to commit to practice adoption—although these means do 6.3 | Rediscovering the configurational not share the same triggers. Being sensitive to different nature of the theory of planned behavior belief configurations and adjusting the experienced pres- sures or motivational mechanisms accordingly must allow We also contribute to the research that draws on the the- corporate managers to increase commitment and, in turn, ory of planned behavior in operations management. By the success rate associated with practice adoption. outlining a configurational perspective on the theory of First, Followers commit when there are strong exter- planned behavior, we focus attention on the theory's fun- nal pressures from superiors and peers, adequate damental assumption of the interdependency of beliefs. resources to support them during the change, and when This perspective enabled us to achieve a more comprehen- they feel that they can master that change. Practical tac- sive understanding of the cognitive underpinnings of tics to motivate commitment of Followers include formal behavior in general and, more specifically, to explore the communication (e.g., leaders' speeches and newsletters) combinations of behavioral, normative, and control beliefs that conveys the appropriateness of new practices, inter- associated with high levels of managerial commitment. nal best-practice , and references to other We complement this configurational perspective on successful improvement efforts both within and outside the theory of planned behavior through a corresponding the organization. Because Followers have a strong inten- methodological approach—namely, fsQCA—to account tion to adopt the practice and are usually supportive of for equifinality, conjunctural causation, and asymmetric company improvements, such activities can help to grow causation. Prior studies have primarily drawn on statistical momentum. Second, Pragmatists require individual frameworks that make it challenging to model the possi- incentives and tangible proof of the fit between the prac- bility that different belief configurations may explain com- tice and the needs of their units. They may be young and mitment. This study is among the first to apply fsQCA to emotionally affected by the adoption, so effective tactics the individual-level investigation of planned behaviors in to prompt Pragmatists to commit to practice adoption the context of practice adoption (for a notable exception, include communicating the message of practice fit and Straatmann et al., 2018). Moreover, in contrast to prior implementing clear incentives for practice adoption. studies that use the theory to predict behavior—generally Third, Reformers commit when they sense continuous by testing the relationships among the aggregated dimen- learning opportunities that trigger both their belief in dis- sions of antecedents (attitudes, subjective norms, and per- crepancy and a sense of high capability to embrace the ceived behavioral control) and individuals' or change. Effective tactics to strengthen operations man- behaviors (Cordano & Frieze, 2000; Jimmieson et al., 2008; agers' commitment to adopt the practice include provid- Straatmann et al., 2018)—we draw on the theory to ing continuous support for their efforts and avoiding explain commitment. We do so by including the non- punishing errors. For example, Armenakis et al. (2000) aggregated specific types of beliefs (behavioral, normative, reference a global awareness program in which and control beliefs). employees were sent to visit other organizations that had implemented the practice in question to ascertain the level of (introducing discrepancy) and to 6.4 | Implications for practice boost both ambition and confidence (i.e., self-efficacy). Our study also suggests which profiles to look for The behavioral operations management literature has when assigning managers to specific (large-scale) change previously focused on how to make operations managers programs. Our results reveal that a “single best operations behave in the “right manner” through incentives or fram- manager” for leading practice-adoption processes does not ing or on how to control individual decision biases in the exist. If corporate managers prefer a top-down approach to 18 ARELLANO ET AL. practice adoption, they may favor the selection of Fol- Another set of limitations concerns the use of fsQCA. lowers, who are motivated by pressures stemming from First, because our study is among the first to use fsQCA at superiors and peers rather than by the potential benefits the cognitive level, we encountered some challenges in for themselves. Followers are less inclined to question the analyzing cognitive measures using QCA. For example, the utility of a new practice partially because of their desire to standard fsQCA literature asks the use of “substantive and maintain their unit's location advantage. However, if a theoretical” knowledge during calibration, but the aggre- firm prefers bottom-up change, it would be worth consid- gate measures used in our study (validated, externally ering the selection of Reformers. The three neutral permu- valid, and multi-item measures) make this difficult. We tations of Reformers suggest that they have a strong desire addressed this limitation by transparently documenting the to change and improve (belief in discrepancy). Finally, if decision-making process that guided our approach to cali- corporate managers care most about cost reduction or rev- bration and evaluating the robustness of our results using enue growth, they must consider choosing Pragmatists. several robustness tests. Second, despite several prior Pragmatists are committed when they believe in the value attempts to apply QCA to longitudinal data, QCA remains and the fit of a given practice. They are not limited by self- unsuitable for analyzing temporal dynamics. While we efficacy beliefs and their commitment to new practices identify three belief configurations, managers may shift does not depend on coercive and mimetic pressures. from one profile to another dependingonthepracticetobe Considering the implementation of the adopted or the specific stage of the adoption process. abovementioned advice, the challenges for corporate Finally, it is unclear if and to what extent QCA studies managers are twofold. First, to project or determine the suffer from endogeneity. Endogeneity might lead to false operations managers' commitment levels, they require a conclusions due to the overlooking of factors that may tool to map the different belief configurations among impact how belief configurations influence managers' com- their managers. Although such a tool is not readily avail- mitment. In our study, we consider the risk of endogeneity able, a starting point could be the survey we employed in to be low due to the comprehensiveness of the theory of the present study. Second, we discussed how different planned behavior, the corresponding validated measures tactics trigger different salient beliefs and motivate man- of beliefs, and the various procedures for addressing issues agers with varying configurations of belief. Thus, deter- of common method variance and construct validity. mining the best tactics to use with managers with We note several promising paths for future research. different belief configurations is challenging. In the Scholars could directly investigate the cognitive mecha- absence of a profiling tool, all tactics can be used in the nisms underlying the identified configurations. The belief form of overtreatment, although this could prove to be an configurations may be partially formed by the expensive strategy. We leave these interesting aspects unobserved heterogeneity of managerial cognitive capa- open to be addressed in future research. bilities, such as attention, perception, and problem- solving (e.g., Cantor & Macdonald, 2009; Choo et al., 2015; Eggers & Kaplan, 2013; Nadkarni & 6.5 | Limitations and future research Barr, 2008), or by managerial cognitive biases like fram- ing (Bartunek et al., 2006). Future studies could also One set of limitations of our study is related to the sam- examine changes in managers' cognition or behavior dur- ple and design. We collected retrospective data through a ing the adoption process as well as the consequences for survey, thereby restricting the possibility of assessing the commitment and adoption, namely the evolution of depth and the breadth of practice adoption. Another limi- belief configurations and the effects of corporate and tation is that our single-respondent survey may raise the employee initiatives at various times during the adoption question if the commitment of one manager, let alone to process. To study the temporal stability of belief configu- lead a successful practice adoption, is sufficient to drive rations, future studies could develop new methodological the commitment at the facility level. Moreover, although approaches that enable the analysis of patterns of config- our results did not show the predominance of any prac- urational change. Furthermore, as Ajzen (1991) suggests, tice type or facility location in the different belief configu- the theory may be extended to include alternative or rations, we cannot completely rule out confounding additional beliefs and constructs. This could be done by effects of practice types (e.g., technology vs. people orien- treating the new contextual variables as additional causal tation, strategic scope, and content), the industry context, conditions or conducting fsQCA for different samples. As or other specific intra-firm dynamics. We also note that cultural differences and personality traits may be impor- our results only discovered two cases of the Pragmatists, tant antecedents of belief configurations, future studies which is a low number for drawing robust conclusions could corroborate these aspects' impacts on the salience for the tests of variable influence. of behavioral, control, and normative beliefs. Also, the ARELLANO ET AL. 19 validity of the valence and discrepancy scales could be efficacy perspective illustrates the dominant position of this type put to the test by conducting a mixed methods study that of study in the literature. involve a wider variety of practices to identify the most 2 Ajzen (1991) notes that people can hold many beliefs regarding a salient beliefs for operations managers working in multi- certain behavior. However, at a given moment, they can only attend — — national corporations. Finally, scholars could compare to a small number of what he calls salient beliefs. Only salient beliefs represent essential determinants of a person's commitment the beliefs of several managers within the same facility to and behavior. This is an important nuance because a manager's provide insights into the antecedents of collective belief belief regarding, for example, “the adoption of 5S in my factory next structures. May” may differ from that manager's belief in 5S in general. 3 In our context, DiMaggio and Powell's 1983) third type of institu- tional isomorphism—“normative pressures” stemming from “pro- 7 | CONCLUSION fessionalization”—is of less relevance because the managers share the same profession. Implicit norms are transmitted by colleagues In this article, we investigated how operations managers' with similar professions working in sister organizations. The beliefs concerning adopting a given practice relate to interpersonal network within an organization exerts the strongest their actual commitment to that practice. We selected a influence on individuals (Armenakis et al., 2007). We follow the approach of Burns and Wholey (1993) in that mimetic pressures micro-foundations perspective to reveal the cognitive account for normative pressures. Thus, we model how normative underpinnings of managers' commitment. We contribute beliefs in the context of practice adoption may be affected by the to the literature on practice adoption by providing an expectations of both internal and external actors. individual-level and configurational view of managerial 4 Armenakis et al. (2007) include a fifth measure, namely principal commitment to change by building on the theory of support, which captures the pressures from superiors within facil- planned behavior and using a configurational model and ities. We replaced this measure with items concerning the legiti- method (fsQCA). The Follower, the Pragmatist, and the macy drivers of practice adoption to better reflect the pressures in Reformer represent three different profiles of highly com- the intra-organizational context. mitted operations managers who rely on unique combi- 5 We used the mean rather than the median as a crossover points nations of beliefs regarding practice adoption. Our study because of the skewed distribution of some of our measures (for calls for a reappraisal of how best to assess and manage most measures, the median and mean are nearly identical and our operations managers' commitment to practice adoption. robustness tests indicate not substantive differences in our results). 6 For fsQCA, please refer to the detailed definition and exact for- ACKNOWLEDGMENTS mula for calculating consistency of a fuzzy condition of Ragin (2008: 44-68) or Schneider and Wagemann (2013: 126). The authors are very grateful to the review team for their 7 valuable and constructive comments which have substan- A necessity analysis identifies individual conditions that are neces- sary for the outcome to occur. Conditions are considered necessary tively helped to improve the earlier versions of this arti- if they are always present when the relevant outcome is present cle. We also thank the participants of the KLU Research (Ragin, 2008). In empirical studies, a consistency value above 0.90 Seminar 2018, the Academy of Management Annual is considered an indication of necessity. Our analysis reveals that Meeting 2018, and the International QCA Workshop no present or absent explanatory condition is necessary for man- 2018 at which earlier drafts of this paper were presented. agers to show high levels of commitment (see Appendix C.8). We thank Silvia Ponce and Louis Hébert for their helpful 8 We also run an fsQCA for the negated outcome (below-average advice during the survey development and data collection managerial commitment to a new practice). Note that the negated as well as Andrea Furlan and Christian Rupietta for their outcome does not test the opposite of commitment (i.e., refusal to insightful feedback on earlier versions of this manuscript. commit) but rather the absence of high commitment. Thus, the results presented in Appendix C.9 do not aid significantly in answering our research question. ORCID 9 The core and peripheral conditions correspond to the difference Maricela C. Arellano https://orcid.org/0000-0003-4473- between the parsimonious and intermediate solutions. Core con- 2732 ditions appear in both solutions, although peripheral conditions Johannes Meuer https://orcid.org/0000-0003-3443-6761 are eliminated in the parsimonious solution and so only appear in Torbjørn H. Netland https://orcid.org/0000-0001-7382- the intermediate solution. Thus, core conditions are more strongly 1051 associated with the outcome.

ENDNOTES 1 These are the three most highly cited articles over the last 3 years REFERENCES in the Journal of Operations Management (source: Wiley/ Ajzen, I. (1985). From intentions to actions: A theory of planned CrossRef, May 2, 2020). The fact that all three study practice behavior. In J. Kuhl & J. Beckmann (Eds.), Action control: From adoption at the organizational level and assume the validity of the cognition to behavior (pp. 11–39). Springer. 20 ARELLANO ET AL.

Ajzen, I. (1991). The theory of planned behavior. Organizational Cantor, D. E., & Macdonald, J. R. (2009). Decision-making in the Behavior and Human Decision Processes, 50(2), 179–211. supply chain: Examining problem solving approaches and Ajzen, I. (2005). Attitudes, personality, and behavior. Open Univer- information availability. Journal of Operations Management, 27 sity Press. (3), 220–232. Armenakis, A. A., Bernerth, J. B., Pitts, J. P., & Walker, H. J. (2007). Chandler, D., & Hwang, H. (2015). Learning from learning theory: Organizational change recipients' beliefs scale: Development of A model of organizational adoption strategies at the micro- an assessment instrument. Journal of Applied Behavioral Sci- foundations of institutional theory. Journal of Management, 41 ence, 43(4), 481–505. (5), 1446–1476. Armenakis, A. A., Harris, S. G., & Feild, H. S. (2000). Paradigms in Choo, A.S., Nag, R., Xia, Y., (2015). The role of executive problem organizational change: Change agent and change target per- solving in knowledge accumulation and manufacturing spectives. In R. T. Golembiewski (Ed.), Handbook of organiza- improvements. Journal of Operations Management, 36(1), tional behavior, revised and expanded. Routledge. 63–74. Bagozzi, R. P., & Edwards, J. R. (1998). A general approach for rep- Cordano, M., & Frieze, I. H. (2000). Pollution reduction preferences resenting constructs in organizational research. Organizational of U.S. environmental managers: Applying Ajzen's theory of Research Methods, 1(1), 45–87. planned behavior. Academy of Management Journal, 43(4), Balogun, J. (2003). From blaming the middle to harnessing its 627–641. potential: Creating change intermediaries. British Journal of Croson, R., Schultz, K., Siemsen, E., & Yeo, M. L. (2013). Behavioral Management, 14(1), 69–83. operations: The state of the field. Journal of Operations Man- Balogun, J., Fahy, K., & Vaara, E. (2017). The interplay between agement, 31(1–2), 1–5. HQ legitimation and subsidiary legitimacy judgments in HQ DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: relocation: A social psychological approach. Journal of Interna- Institutional isomorphism and collective rationality in orga- tional Business Studies, 50(2), 223–249. nizational fields. American Sociological Review, 48(2), Bansal, P., & Roth, K. (2000). Why companies go green: A model of 147–160. ecological responsiveness. Academy of Management Journal, 43 Done, A., Voss, C., & Rytter, N. G. (2011). Best practice interven- (4), 717–736. tions: Short-term impact and long-term outcomes. Journal of Barreto, I., & Baden-Fuller, C. (2006). To conform or to perform? Operations Management, 29(5), 500–513. Mimetic behaviour, legitimacy-based groups and performance Donohue, K., Özer, Ö., & Zheng, Y. (2020). Behavioral operations: consequences. Journal of Management Studies, 43(7), Past, present, and future. Manufacturing & Service Operations 1559–1581. Management, 22(1), 191–202. Bartunek, J. M., Rousseau, D. M., Rudolph, J. W., & DePalma, J. A. Eggers, J. P., & Kaplan, S. (2013). Cognition and capabilities: A (2006). On the receiving end: Sensemaking, emotion, and multi-level perspective. Academy of Management Annals, 7(1), assessments of an organizational change initiated by others. 295–340. Journal of Applied Behavioral Science, 42(2), 182–206. Fahimnia, B., Pournader, M., Siemsen, E., Bendoly, E., & Wang, C. Becker, T. E., Billings, R. S., Eveleth, D. M., & Gilbert, N. L. (2019). Behavioral operations and supply chain management— (1996). Foci and bases of employee commitment: Implications A review and literature mapping. Decision Sciences, 50(6), for job performance. Academy of Management Journal, 39(2), 1127–1183. 464–482. Fiss, P. C. (2007). A set-theoretic approach to organizational config- Bendoly, E., Bachrach, D. G., & Powell, B. (2008). The role of opera- urations. Academy of Management Review, 32(4), 1180–1198. tional interdependence and supervisory experience on manage- Fiss, P. C. (2011). Building better causal theories: A fuzzy set ment assessments of resource planning systems. Production approach to typologies in organizational research. Academy of and Operations Management, 17(1), 93–106. Management Journal, 54(2), 393–420. Bendoly, E., & Cotteleer, M. J. (2008). Understanding behavioral Flynn, B. B., Huo, B., & Zhao, X. (2010). The impact of supply sources of process variation following enterprise system chain integration on performance: A contingency and config- deployment. Journal of Operations Management, 26(1), uration approach. Journal of Operations Management, 28(1), 23–44. 58–71. Braunscheidel, M. J., Hamister, J. W., Suresh, N. C., & Star, H. Greckhamer, T. (2016). CEO compensation in relation to worker (2011). An institutional theory perspective on Six Sigma adop- compensation across countries: The configurational impact of tion. International Journal of Operations & Production Manage- country-level . Strategic Management Journal, 37(4), ment, 31(4), 423–451. 793–815. Burns, L. R., & Wholey, D. R. (1993). Adoption and abandonment Griffin, M. A., Neal, A., & Parker, S. K. (2007). A new model of of matrix management programs: Effects of organizational work role performance: Positive behavior in uncertain and characteristics and inteorganizational networks. Academy of interdependent contexts. Academy of Management Journal, 50 Management Journal, 36(1), 106–138. (2), 327–347. Busse, C., Regelmann, A., Chithambaram, H., & Wagner, S. M. Helfat, C. E., & Peteraf, M. A. (2015). Managerial cognitive capabili- (2017). Managerial perceptions of energy in logistics: An inte- ties and the microfoundations of dynamic capabilities. Strategic gration of the theory of planned behavior and stakeholder the- Management Journal, 36(6), 831–850. ory. International Journal of Physical Distribution & Logistics Henseler, J., Ringle, C. M., & Sarstedt, M. (2014). A new criterion Management, 47(6), 447–471. for assessing discriminant validity in variance-based structural ARELLANO ET AL. 21

equation modeling. Journal of the Academy of Marketing Sci- Netland, T.H., Schloetzer, J.D., Ferdows, K., (2015). Implementing ence, 43(1), 115–135. corporate lean programs: The effect of management control Herold, D. M., Fedor, D. B., & Caldwell, S. D. (2007). Beyond practices. Journal of Operations Management, 36(0), 90–102 change management: A multilevel investigation of contextual Oreg, S. (2003). Resistance to change: Developing an individual dif- and personal influences on employees' commitment to change. ferences measure. Journal of Applied , 88(4), Journal of Applied Psychology, 92(4), 942–951. 680–693. Hu, L.t., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in Piderit, S. K. (2000). Rethinking resistance and recognizing ambiva- covariance structure analysis: Conventional criteria versus new lence: A multidimensional view of attitudes toward an organi- alternatives. Structural Equation Modeling: A Multidisciplinary zational change. Academy of Management Review, 25(4), Journal, 6(1), 1–55. 783–794. Huang, X., Gattiker, T. F., & Schroeder, R. G. (2010). Do competi- Podsakoff, P.M., MacKenzie, S.B., Podsakoff, N.P., (2012). Sources tive priorities drive adoption of electronic commerce applica- of method bias in social science research and recommendations tions? Testing the contingency and institutional views. Journal on how to control it. Annual Review of Psychology, 63(539-569 of Supply Chain Management, 46(3), 57–69. Prajogo, D. I. (2011). The roles of firms' motives in affecting the out- Jacqueminet, A. (2020). Practice implementation within a multi- comes of ISO 9000 adoption. International Journal of Opera- divisional firm: The role of institutional pressures and value tions & Production Management, 31(1), 78–100. consistency. Organization Science, 31, 182–199. Ragin, C. (1987). The comparative method: Moving beyond qualita- Jacqueminet, A., & Durand, R. (2020). Ups and downs: The role of tive and quantitative strategies. University of California Press. legitimacy judgment cues in practice implementation. Academy Ragin, C. (2000). Fuzzy-set social science. University of Chicago of Management Journal, 63(5), 1485–1507. Press. Jimmieson, N. L., Peach, M., & White, K. M. (2008). Utilizing the Ragin, C. C. (2008). Redesigning social inquiry: Fuzzy sets and theory of planned behavior to inform change management. beyond. University of Chicago Press. Journal of Applied Behavioral Science, 44(2), 237–262. Ragin, C. C., & Davey, S. (2017). fs/QCA [Computer Programme], Kennedy, M. T., & Fiss, P. C. (2009). Institutionalization, framing, version 3.0 ed. University of California. and diffusion: The logic of TQM adoption and implementation Reimann, F., Kosmol, T., & Kaufmann, L. (2017). Responses to decisions among U.S. hospitals. Academy of Management Jour- supplier-induced disruptions: A fuzzy-set analysis. Journal of nal, 52(5), 897–918. Supply Chain Management, 53(4), 37–66. Ketokivi, M., & Schroeder, R. G. (2004). Strategic, structural contin- Rogers, K. W., Purdy, L., Safayeni, F., & Duimering, P. R. (2007). A gency and institutional explanations in the adoption of innova- supplier development program: Rational process or institu- tive manufacturing practices. Journal of Operations tional image construction? Journal of Operations Management, Management, 22(1), 63–89. 25(2), 556–572. Kostova, T., & Roth, K. (2002). Adoption of an organizational prac- Schneider, C. Q., & Wagemann, C. (2013). Set-theoretic methods for tice by subsidiaries of multinational corporations: Institutional the social sciences: A guide to qualitative comparative analysis. and relational effects. Academy of Management Journal, 45(1), Cambridge University Press. 215–233. Shah, R., & Ward, P. T. (2003). Lean manufacturing: Context, prac- Kostova, T., Roth, K., & Dacin, M. T. (2008). Institutional theory in tice bundles, and performance. Journal of Operations Manage- the study of multinational corporations: A critique and new ment, 21(2), 129–149. directions. Academy of Management Review, 33(4), 994–1006. Stevens, J. M., Kevin Steensma, H., Harrison, D. A., & Leseure, M. J., Bauer, J., Birdi, K., Neely, A., & Denyer, D. (2004). Cochran, P. L. (2005). Symbolic or substantive document? The Adoption of promising practices: A systematic review of the evi- influence of ethics codes on financial executives' decisions. dence. International Journal of Management Reviews, 5(3–4), Strategic Management Journal, 26(2), 181–195. 169–190. Straatmann,T.,Rothenhöfer,L.M.,Meier,A.,&Mueller,K. Liu,H.,Ke,W.,Wei,K.K.,Gu,J.,&Chen,H.(2010).Theroleof (2018). A configurational perspective on the theory of institutional pressures and organizational in the planned behaviour to understand employees' change- firm's intention to adopt internet-enabled supply chain man- supportive intentions. Journal of Applied Psychology, 67(1), agement systems. Journal of Operations Management, 28(5), 91–135. 372–384. Suddaby, R., Bitektine, A., & Haack, P. (2017). Legitimacy. Acad- Loch, C. H. (2017). Creativity and risk taking aren't rational: Behav- emy of Management Annals, 11(1), 451–478. ioral operations in MOT. Production and Operations Manage- Tarakci, M., Ates¸, N. Y., Floyd, S. W., Ahn, Y., & Wooldridge, B. ment, 26(4), 591–604. (2018). Performance feedback and middle managers' divergent Meuer, J., Rupietta, C., & Backes-Gellner, U. (2015). Layers of strategic behavior: The roles of social comparisons and organi- co-existing innovation systems. Research Policy, 44(4), zational identification. Strategic Management Journal, 39(4), 888–910. 1139–1162. Mill, J. S. (2011). A system of logic, ratiocinative and inductive: Being Vardaman, J. M., Amis, J. M., Dyson, B. P., Wright, P. M., & Van de a connected view of the principle of evidence, and the methods of Graaff Randolph, R. (2012). Interpreting change as controllable: scientific investigation. Cambridge University. The role of network centrality and self-efficacy. Human Rela- Nadkarni, S., & Barr, P. S. (2008). Environmental context, manage- tions, 65(7), 835–859. rial cognition, and strategic action: An integrated view. Strate- Voorhees, C. M., Brady, M. K., Calantone, R., & Ramirez, E. (2015). gic Management Journal, 29(13), 1395–1427. Discriminant validity testing in marketing: An analysis, causes 22 ARELLANO ET AL.

for concern, and proposed remedies. Journal of the Academy of SUPPORTING INFORMATION Marketing Science, 44(1), 119–134. Additional supporting information may be found online Voss, C. A. (2005). Paradigms of manufacturing strategy re-visited. in the Supporting Information section at the end of this International Journal of Operations & Production Management, article. 25(12), 1223–1227. Wolf, E. J., Harrington, K. M., Clark, S. L., & Miller, M. W. (2013). Sample size requirements for structural equation models: An evaluation of power, bias, and solution propriety. Educational How to cite this article: Arellano MC, Meuer J, and Psychological Measurement, 76(6), 913–934. Netland TH. Commitment follows beliefs: A Zhu, Q., & Sarkis, J. (2004). Relationships between operational configurational perspective on operations practices and performance among early adopters of green sup- managers' commitment to practice adoption. ply chain management practices in Chinese manufacturing J Oper Manag. 2020;1–26. https://doi.org/10.1002/ enterprises. Journal of Operations Management, 22(3), joom.1130 265–289. ARELLANO APPENDIX A: REPRESENTATIVE STUDIES OF PRACTICE ADOPTION AT THE INDIVIDUAL LEVEL TAL ET

Configurational . Author(s) Domain Theory Data and method Individual determinants Outcome(s) (I/O)a perspective Braunscheidel Operations Institutional theory Case study (seven Coercive pressures from (O) Adoption of Six Sigma No et al. (2011) management manufacturers) customers, mimetic responses, and normative pressures from a professional environment. Climate for implementation (rewards, support, and superiors' expectations) and fit innovation— personal values. Rogers et al. (2007) Operations Institutional theory Mixed methods (survey Institutional image (O) Adoption of a supplier No management data from 85 to 90 construction. development program suppliers, interviews with Internal motivation 14 suppliers) (pressure to improve performance). Bendoly and Behavioral Valence-instrumentality- Case study comparisons Perceived task (technology (I/O) Large-scale ERP No Cotteleer (2008) operations expectancy framework and controlled mismatch) and perceived adoption: short- and management experiments ease of deviation from long-term reactions the mandated procedure. Cordano and Management Theory of planned behavior Survey data (295 U.S. Pollution prevention (I) Behavioral preference No Frieze (2000) environmental managers) attitudes, perceptions of for reducing pollution at norms for environmental the facility regulation, perceived behavioral control, and past source reduction activity of the facility. Stevens Strategic Stakeholder management Survey data (302 senior Pressures from market and (I) Use of ethics code in No et al. (2005) management theory and the theory of financial executives) non-market stakeholders, decision making planned behavior perceived benefits, and training in ethics code principles. Bansal and Management Institutional theory, Case study (53 firms in the Competitiveness, (O) Adoption of ecological Yesb Roth (2000) economic theory, and U.K. and Japan) legitimation, and response initiatives literature on individual ecological responsibility,

values in addition to contextual 23 (Continues) 24 Configurational Author(s) Domain Theory Data and method Individual determinants Outcome(s) (I/O)a perspective factors (field cohesion, issue salience, and individual concern). Jacqueminet (2020) Management Institutional theory Logistic regressions with Influence of corporate and (O) Adoption of corporate No survey data from two branch headquarters, social responsibility measurement times implementation by practices (101/100 units of a peersc, and value multinational: 314/270 consistency. managers) Straatmann Psychology Theory of planned behavior Linear regression, latent Change-related attitude, (I) Intentions to support a Yes et al. (2018) profile analysis, and change-related perceived major change process fsQCA with data from behavioral control, and affecting the two measurement times change-related subjective organizational structure, (1589/1524 employees of norms. working processes, and a German city council) organizational culture Jacqueminet and Management Institutional theory (as fsQCA with survey data Industry-level (O) Adoption of three Yes Durand (2020) perception) from two measurement endorsementc, country- corporate social times (409 managers in level endorsementc, responsibility practices 65 units of a parent firm multinational enterprise) authorizationc, strategic importance of the practice, and value compatibility.

Notes: a I: individual level, O: organizational level; b These authors use the Discussion section of their article to extend their empirically grounded model based on their observations of combinations of motivations capable of explaining CSR adoption; c Measured using secondary data. ARELLANO TAL ET . ARELLANO ET AL. 25

APPENDIX B: DEFINITIONS AND ITEMS ASSOCIATED WITH THE OUTCOME AND EXPLANATORY CONDITIONS

Item parcel Item code Constructs and items Standard loading Commitment to the new operational practice (α = 0.897; CR = 0.91; AVE = 0.78) Indicate your current level of agreement or disagreement with the following statements Scale: strongly disagree = 1 to strongly agree = 5 p_comm1 comm1 I am willing to put in a great deal of effort, going beyond 0.84 what is normally expected, to continue helping with the advancement of this practice comm2 I talk up the practice to my friends as a great way of improving business p_comm2 comm3 I find that my values and the values promoted by this 0.92 practice are very similar comm6 I really care about this practice and its future p_comm3 comm4 This practice really inspires the very best in me in terms 0.88 of involvement in my work comm5 I am extremely glad that I was involved in this practice- adoption process Efficacy-related beliefs Indicate your level of agreement or disagreement with the following statements regarding your and impressions at the onset of the implementation: I believed that… Scale: strongly disagree = 1 to strongly agree = 5 Valence (α = 0.600; CR = 0.65; AVE = 0.38) p_val1 val1 The change resulting from the adoption of this practice 0.53 would benefit me p_val2 val2 If this practice was part of my job, I would experience 0.59 more self-fulfillment val3 I would earn higher pay from my job after the adoption of this practice p_val3 val4 The changes in my job assignments, as derived from the 0.73 adoption of this practice, would increase my feeling of accomplishment Discrepancy (α = 0.711; CR = 0.71; AVE = 0.45) p_disc1 disc1 We needed to change the way we did certain things at 0.79 this facility disc4 A change was needed to improve our operations p_disc2 disc2 We needed to improve the way we operated at this facility 0.63 p_disc3 disc3 We needed to improve our effectiveness by changing our 0.56 operations Appropriateness (α = 0.800; CR = 0.82; AVE = 0.60) p_app1 app1 The change resulting from the adoption of this practice 0.84 would have a favorable impact on our operations app5 The adoption of this practice was proving to be the best for our situation p_app2 app2 The adoption of this practice would improve the 0.68 performance of our facility app3 The practice that we were implementing was correct for our situation (Continues) 26 ARELLANO ET AL.

Item parcel Item code Constructs and items Standard loading p_app3 app4 The change to be brought about through this practice, I 0.79 realized that it was appropriate for our facility Self-efficacy (α = 0.824; CR = 0.84; AVE = 0.64) p_self1 selfe1 I had the ability to implement the practice 0.82 selfe5 We had the capability to successfully implement the practice p_self2 selfe3 I was capable of successfully performing my job duties 0.88 once the adoption began selfe4 We could successfully implement this practice p_self3 selfe2 I was sure I would successfully implement this practice 0.70 and integrate it into my job Legitimacy-related beliefs From your perspective, at the onset of the implementation, what were the expectations within your company regarding the adoption of the practice? Scale: strongly disagree = 1 to strongly agree = 5. The option “I do not know” was made available. Coercive pressures (α = 0.769; CR = 0.77; AVE = 0.54) p_coe1 coerc1 This company had high expectations regarding the 0.90 implementation of this practice at all its facilities coerc6 Prestigious sister facilities expected us to implement this practice p_coe2 coerc2 Our company expected us to adopt this practice 0.61 p_coe3 coerc4 Our main corporate centers (e.g., divisional head, regional 0.66 head) believed that we should implement this practice Mimetic pressures (α = 0.891; CR = 0.90; AVE = 0.75) p_mim1 mimet1 Most of our sister facilities have implemented this 0.94 practice mimet3 The sister facilities that implemented this practice have benefited greatly p_mim2 mimet4 Most of our sister facilities with similar characteristics 0.84 (e.g., they produce the same products/services, are located in the same region, belong to the same business division, use the same technology) have implemented this practice p_mim3 mimet5 The sister facilities that implemented this practice 0.81 improved their operational performance