JOB FLEXIBILITY, WORK-NONWORK INTERRUPTIONS, AND IMPLICATIONS FOR WORK-NONWORK CONFLICT By Jessica Keeney A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Psychology 2012 ABSTRACT JOB FLEXIBILITY, WORK-NONWORK INTERRUPTIONS, AND IMPLICATIONS FOR WORK-NONWORK CONFLICT By Jessica Keeney Organizations offer employees flexibility in when and where they work to help them better manage their work and nonwork demands. However, research does not always demonstrate a negative relationship between job flexibility and work-life conflict. Boundary theory suggests that roles with flexible boundaries pose tradeoffs. One reason for inconsistent findings may be that job flexibility has potential costs to individuals that mask its benefits. In particular, a flexible job may open the door to increased frequency of interruptions between work and nonwork. In a study of 1,141 employees both flexibility in the timing and location of work were negatively related to work-to-nonwork conflict and unrelated to nonwork-to-work conflict. Importantly, flexibility in the location of work was associated with more frequent self-initiated work interruptions of nonwork and the results demonstrated that the negative relationship between flexibility in the location of work and work-to-nonwork conflict would be more strongly negative if it were not for these interruptions. Similarly, flexibility in both the timing and location of work were associated with more frequent nonwork interruptions of work, and the nonsignificant relationship between these predictors and nonwork-to-work conflict became significantly albeit modestly negative when controlling for interruptions. The present study largely supports the underlying proposition of boundary theory that there are tradeoffs inherent in flexible boundaries between roles. There were some unexpected findings, however. Differential results were found depending on the source of work interruptions of nonwork (i.e., self-initiated versus other-initiated), which is not predicted by boundary theory. Job flexibility was actually modestly negatively related to the frequency of colleagues interrupting one’s personal time. Despite some of the apparent challenges of job flexibility in the form of greater worknonwork interruptions, the results largely imply benefits of job flexibility for individuals, particularly for flexibility in the timing of work, which was more strongly negatively related than flexibility in the location of work to work-to-nonwork conflict. Importantly, both forms of job flexibility were positively associated with boundary control suggesting that giving employees more autonomy over the timing and location of their work has positive psychological effects overall. As we improve our understanding of how boundaries between work and nonwork affect individuals, organizations can play the role of educating their employees about the various tradeoffs of job flexibility. Employees and their managers have a joint responsibility to set expectations about the acceptable types and frequency of interruptions between work and nonwork, and how these should be dealt with when they arise. Maintaining an open dialogue should help ensure that the benefits of job flexibility outweigh the costs. ACKNOWLEDGMENTS First and foremost, thank you to my advisor, Dr. Ann Marie Ryan, for her support and guidance throughout graduate school and during the completion of this dissertation. I am grateful for having a wonderful committee consisting of Drs. Ellen Kossek, Neal Schmitt, and Daisy Chang. Thank you for providing your expertise. Thank you to the individuals who made it possible to collect the data for this dissertation, Rick and Tom. You displayed exceptional persistence to make it happen. I would also like to thank the American Psychological Association (APA) for awarding me an American Psychological Foundation (APF) grant to conduct this research. Last but certainly not least, thank you to my parents, sister, and husband Michael for their encouragement. iv TABLE OF CONTENTS LIST OF TABLES ......................................................................................................................... vi LIST OF FIGURES ...................................................................................................................... vii INTRODUCTION ...........................................................................................................................1 Purpose and Organization ................................................................................................................1 Theoretical Background ...................................................................................................................5 Role Theory .........................................................................................................................5 Boundary Theory .................................................................................................................7 Job Flexibility ................................................................................................................................13 The Upsides of Job Flexibility .......................................................................................................17 The Downsides of Job Flexibility ..................................................................................................19 Nonwork Interrupting Work ..............................................................................................19 Work Interrupting Nonwork ..............................................................................................21 Empirical Findings Linking Job Flexibility and Work-Nonwork Interruptions ................22 The Type of Job Flexibility Matters ..................................................................................24 Moderators of the Job Flexibility-Work-Nonwork Interruptions Relationship .................27 Work-Nonwork Interruptions and Conflict .......................................................................30 The Moderating Role of Preference for Segmentation ......................................................32 The Moderating Role of Boundary Control .......................................................................35 Inconsistent Mediation: The Downsides of Job Flexibility Mask the Upsides .............................36 Exploratory Questions ...................................................................................................................38 PILOT STUDY ..............................................................................................................................40 Method ...........................................................................................................................................40 Sample................................................................................................................................40 Procedure ...........................................................................................................................40 Measures ............................................................................................................................41 Results ............................................................................................................................................41 Job Flexibility and Job Permeability..................................................................................41 Work-Nonwork Interruptions ............................................................................................43 MAIN STUDY...............................................................................................................................45 Method ...........................................................................................................................................45 Sample................................................................................................................................45 Procedure ...........................................................................................................................45 Measures ............................................................................................................................46 Results ............................................................................................................................................50 Discriminant Validity of Scales .........................................................................................50 Descriptive Statistics and Scale Intercorrelations ..............................................................52 Hypothesis Testing.............................................................................................................52 v DISCUSSION ................................................................................................................................81 Work-to-Nonwork..........................................................................................................................81 Potential Explanations for the Unexpected Benefits of Job Flexibility .............................81 Nonwork-toWork ...........................................................................................................................83 Questioning Assumptions: Are Interruptions Undesirable? ..........................................................86 Flexibility in the Timing versus Location of Work .......................................................................87 Limitations .....................................................................................................................................88 Future Research Directions ............................................................................................................90 Practical Implications.....................................................................................................................91 APPENDICES ...............................................................................................................................94 Appendix A: Focal Employee Informed Consent and Survey.......................................................95 Appendix B: Significant Other Informed Consent and Survey ...................................................107 REFERENCES ............................................................................................................................119 vi LIST OF TABLES Table 1: Types of Job Flexibility ...................................................................................................15 Table 2: Job Flexibility and Job Permeability Factor Loadings from Pilot Study EFA ................42 Table 3: Work-Nonwork Interruptions Factor Loadings from Pilot Study EFA ...........................44 Table 4: Confirmatory Factor Analysis Results.............................................................................50 Table 5: Phi Matrix for Job Flexibility/Permeability Confirmatory Factor Analysis ....................51 Table 6: Phi Matrix for Interruptions/Conflict Confirmatory Factor Analysis..............................52 Table 7: Scale Descriptives and Intercorrelations .........................................................................53 Table 8: Confirmatory Factor Analysis Results for Significant Other Reports .............................71 Table 9: Intercorrelation between Self and Significant Other Reports ..........................................73 Table 10: Correlations with Significant Others’ Job Flexibility....................................................77 Table 11: Polynomial Regression Results for Employees’ and Significant Others’ Flexibility in the Timing of Work Predicting Employees’ Nonwork Interruptions of Work ..............................78 Table 12: Polynomial Regression Results for Employees’ and Significant Others’ Flexibility in the Location of Work Predicting Employees’ Nonwork Interruptions of Work ...........................78 Table 13: Polynomial Regression Results for Employees’ and Significant Others’ Flexibility in the Timing of Work Predicting Division of Household Labor ......................................................79 Table 14: Polynomial Regression Results for Employees’ and Significant Others’ Flexibility in the Location of Work Predicting Division of Household Labor....................................................80 vii LIST OF FIGURES Figure 1: Summary of Hypotheses ..................................................................................................4 Figure 2: Support for Hypothesized Relationships ........................................................................56 Figure 3: Significant Results of Hypothesis Tests .........................................................................57 Figure 4: Job Flexibility Attenuates the Positive Relationship between Workload and SelfInitiated Work Interruptions of Nonwork ......................................................................................64 Figure 5: Job Flexibility Attenuates the Positive Relationship between Workload and OtherInitiated Work Interruptions of Nonwork ......................................................................................65 viii INTRODUCTION Purpose and Organization Understanding how individuals manage the boundary between work and nonwork has important implications. Work-life conflict has been linked to job dissatisfaction, turnover intentions, depression, and poor physical health (Allen, Herst, Bruck, & Sutton, 2000). Organizations are increasingly offering employees flexibility in when and where they work to help them better manage their work and nonwork demands. According to a report issued by the Society for Human Resource Management in 2011, 53% of employers offered flextime while 63% offered telecommuting. Most human resource professionals believe that job flexibility will become even more prevalent over the next five years. Yet researchers have made surprisingly little headway in evaluating potential solutions to work-life conflict despite increased scholarly attention to the problem over the past several decades (Kossek, Baltes, & Matthews, 2011). Most notably, research is equivocal with regard to whether job flexibility reduces work-life conflict (Allard, Haas, & Hwang, 2007). As noted by Shockley and Allen (2007) studies have produced inconsistent estimates of the relationship between job flexibility and work-life conflict, even at the level of meta-analysis. Whereas Byron’s (2005) meta-analysis revealed a negative relationship between flexibility and both directions of work-family conflict (-.30 for work-to-family and -.17 for family-to-work), Mesmer-Magnus and Viswesvaran’s (2006) meta-analysis revealed no relationship with either direction of conflict. The above inconsistencies in findings have been attributed to various factors such as differences in the measurement of flexibility and the presence of moderators (Shockley & Allen, 2007). Another possibility, demonstrated by Schieman and Young (2010) is that there are 1 potential downsides to flexibility that mask its benefits. Boundary theory is a recently advanced framework applied to the work-nonwork interface (Ashforth, Kreiner, & Fugate, 2000; Clark, 2000; Nippert-Eng, 1996) that suggests flexible boundaries offer distinct advantages but are not without their own set of challenges when it comes to establishing a healthy work-life balance. Job flexibility can make balancing work and personal demands easier in a way, but can also contribute to a blurred boundary between work and personal life. To the extent that higher job flexibility makes managing the boundary between work and personal life more difficult in some respects, one would expect an overall negative relationship between job flexibility and work-life conflict to be attenuated. Therefore, the overall relationship between job flexibility and work-life conflict may vary from study to study depending on the relative extent to which the potential benefits and costs of job flexibility are realized. A complex portrait of job flexibility emerges when one considers that it can reduce worklife conflict, but at the same time can indirectly contribute to work-life conflict. Whereas job flexibility has primarily been researched for its beneficial effects, its potential costs have received less attention. The mechanisms by which job flexibility may contribute to higher worknonwork conflict are the main focus of this paper. It is proposed that individuals with greater latitude in when and where work is performed (i.e., high job flexibility) are more likely to experience interruptions between the work and nonwork roles, which may fuel perceptions of work-nonwork conflict. Thus, the purpose of this paper is to identify whether and how job flexibility has an indirect positive relationship with work-nonwork conflict (i.e., through worknonwork interruptions). This indirect positive relationship is expected to obscure the direct negative association between job flexibility and work-nonwork conflict (i.e., result in inconsistent mediation). That is, the relationship between job flexibility and work-nonwork 2 conflict would be even more negative if it were not for work-nonwork interruptions. Figure 1 provides an illustration of the major hypotheses. The proposed study is intended to make several contributions. First, as already mentioned, the potential downsides to job flexibility are under-researched. The presence of inconsistent mediation would at least partially explain the equivocal results with respect to the relationship between job flexibility and work-nonwork conflict. Second, job flexibility is a multifaceted construct (Hill et al., 2008), but the majority of previous research only examines one specific type of flexibility (i.e., either the location or timing of work). While conceptualizing job flexibility as an aggregate construct, this study also considers the possibility that its effects differ according to type of job flexibility. Third, the circumstances under which work-nonwork interruptions are most likely to materialize as a result of job flexibility are investigated. Workload and family responsibility are explored as moderating variables. The goal served by investigating these as moderators is to assess whether job flexibility poses the same challenges for everyone (i.e., work-nonwork interruptions) or whether a more nuanced approach is required to understand and manage job flexibility. Finally, work-nonwork interruptions may not constitute costs to all individuals. The proposed study builds on what we know about individual differences in preferences for work-nonwork segmentation and perceived boundary control. The paper is organized as follows. The first section provides a theoretical background for work-nonwork conflict to define terms and establish a general framework for predictions. Role theory and boundary theory are presented as the theoretical perspectives that underlie the proposed study. The concept of integration versus segmentation of the work and nonwork domains is introduced to help explain the merits and potential costs of role flexibility. The second section explains the rationale for focusing the present paper on job flexibility. The third 3 Figure 1 Summary of Hypotheses Preference for Segmenting Nonwork from Work Family + Responsibility - + Nonwork Interruptions of Work - + + + Nonwork-to-Work Conflict Job Flexibility Composite - Flexibility over Timing > Start/Quit Times > General Autonomy - Flexibility over Location - + + Boundary Control - + - Work Interruptions of Nonwork - Workload Work-to-Nonwork Conflict + + Preference for Segmenting Work from Nonwork + Note. Hypothesis 6 (not illustrated here) states that the relationship between job flexibility and nonwork interruptions of work is mediated by job permeability. 4 section acknowledges the positive and desirable aspects of job flexibility, whereas the fourth and most detailed section is devoted to exploring the downsides that job flexibility may pose for some individuals. The fifth and final section leads up to the inconsistent mediation hypothesis, which explains what we should expect to observe if job flexibility is indeed both beneficial and costly (i.e., in terms of the experience of work-nonwork conflict). Theoretical Background Role Theory Work-nonwork conflict has been primarily studied through the lens of role theory (Kahn et al., 1964), which proposes that individuals can experience opposing pressures that result from participation in different roles. Work and family are thought to be the dominant life roles for most individuals and have thus been the focus of most investigations (Friedman & Greenhaus, 2000). Because the work and family domains are often associated with different purposes and cultures (Clark, 2000), individuals are often forced to cope with competing demands and expectations. Work-family conflict occurs when “participation in the work (family) role is made more difficult by virtue of participation in the family (work) role” (Greenhaus & Beutell, 1985, p. 77). Conflict can occur in both directions, such that work interferes with family or family interferes with work. Recently, research has focused on work-life conflict (Fisher, Bulger, & Smith, 2009) to be more inclusive with respect to aspects of nonwork other than family (e.g., health, friendships, etc.). The term “work-nonwork” is adopted in this paper as a response to the criticism that “work” is actually a part of “life” (Kossek, Baltes, et al., 2011). This paper will refer to work-nonwork conflict, but will also make reference to work-family conflict to accurately describe past research findings. Because family is often the most salient identity of the nonwork domain, and because the vast majority of individuals have family, however it may be 5 defined, in some cases family is used as a specific example of nonwork. “Personal life” has also been used in the literature to describe nonwork and will be used interchangeably. Work-nonwork conflict can take three forms (Carlson, Kacmar, & Williams, 2000; Greenhaus & Beutell, 1985). Time-based conflict occurs when personal life takes up time that one would prefer to dedicate to work, or vice versa. This form of conflict can present itself as clashes between roles in the scheduling of activities or as excessive time dedicated to one role at the expense of another (Pleck, Staines, & Lang, 1980). Strain-based conflict occurs when stress (e.g., fatigue, irritability) in one’s personal life interferes with performance at work, or vice versa. Finally, behavior-based conflict refers to the situation where styles of behavior that are helpful in one’s personal life are counterproductive if applied to the work role, or vice versa. These three forms of conflict are positively interrelated, with a particularly strong association observed between time-based and strain-based conflict (r = .58 work-to-family, .76 family-towork; Carlson et al., 2000). In practice, studies typically average together items or subscales representing different forms of conflict in order to examine work-family conflict, or worknonwork conflict, at the aggregate level. This is the approach that will be taken in the present paper. Though role theory is the foundation for the definition of interrole conflict, it has not been a productive basis for deriving hypotheses (Grandey & Cropanzano, 1999). In fact, workfamily research has been criticized for being atheoretical (Kingston, 1988; Zedeck, 1992). There are actually multiple theoretical perspectives that have been used to explain the interface between work and nonwork, such as spillover, compensation, and segmentation (Edwards & Rothbard, 2000). What scholars appear to be lamenting is not the fact that there is insufficient theory guiding research, but instead the lack of a comprehensive, integrative theory for work- 6 nonwork conflict (Westman & Piotrkowski, 1999). A relatively recently proposed framework that may have the potential to bridge together existing perspectives is border theory (Clark, 2000) also known as boundary theory (Ashforth et al., 2000; Nippert-Eng, 1996). Boundary Theory Boundaries can be defined as the lines that demarcate different life domains (Ashforth et al., 2000). The idea that we are able to view life as compartmentalized into distinct domains, such as work and family, provides some evidence in and of itself in support of the notion that these boundaries exist (Nippert-Eng, 1996). While these boundaries are experienced phenomenologically as “mental fences,” they are reinforced and enacted temporally and physically (Ashforth et al.). Traditionally, companies give their employees specific hours and work spaces designated for work, whereas employees have their own times (traditionally evenings and weekends) and spaces (homes) for their personal lives. Employees who work in nontraditional arrangements (e.g., telecommuting) also manipulate time and space to reinforce boundaries, for example, by shutting the door to their home office during periods of peak work engagement. Boundary theory (Ashforth et al., 2000; Zerubavel, 1991) proposes that individuals actively manage the strength of boundaries around roles. This theory has been applied to the roles of work and nonwork, but could easily be applied to other roles (Clark, 2000; Nippert-Eng, 1996). Strong boundaries characterize segmentation, which exists when “the mental boundary between realms is clear and impregnable…with no conceptual overlap between realms and their contents, there is no physical or temporal overlap between them either” (Nippert-Eng, 1996, p. 568). Work and nonwork are treated as separate. Weak boundaries reflect the opposite end of the continuum, integration, which at its extreme is when “no distinction exists between what 7 belongs to home or work and when and where they are engaged” (Nippert-Eng, 1996, p. 567). In reality, integration and segmentation lie at opposite ends of a continuum with most individuals falling somewhere in between the two extremes (Ashforth et al., 2000; Nippert-Eng, 1996). There is a consensus among organizational scholars that neither integration nor segmentation is universally ideal (Ashforth et al., 2000; Clark, 2000). Rather, each is thought to have advantages and disadvantages. The primary merit of integration is that of less effortful role transitions. One can move from one role to another more easily—if one is accustomed to doing so, it becomes practiced. Furthermore, the environment of an integrator is typically set up to facilitate such transitions (e.g., work e-mail and personal e-mail merged). Such transitions make it possible to meet role demands as they arise. The disadvantages of integration are susceptibility to interruptions between roles and potential confusion about the appropriate role identity to enact (i.e., due to lack of symbolic markers and psychological compartmentalization). Segmentation, because it lies at the opposite end of the continuum, naturally has advantages and disadvantages reverse of integration (Ashforth et al., 2000). There is reduced likelihood of interruptions between roles, but the cost of high segmentation is that it is more effortful to change roles. To provide a concrete example, consider an attorney who relies heavily on physical documents and keeps them all in his/her office. If this attorney wanted to eat dinner at home and then access a file later in the evening, he/she would have to physically return to the office. The flipside of segmentation, perhaps because it is more difficult to transition between roles, is that roles tend to be highly psychologically compartmentalized, which can reduce the spillover of strain. Also, it is easier for segmenters to cue the appropriate identity at any given time (e.g., tough business negotiator versus forgiving spouse). 8 There are two characteristics of role boundaries that determine the likely extent of integration versus segmentation. Flexibility is defined as “the extent to which the physical time and location markers, such as working hours and workplace, may be changed” (Hall & Richter, 1988, p. 215). Thus, flexibility corresponds to the temporal and physical aspects of a role’s boundary—whether it can be enacted in various settings and at various times (Ashforth et al., 2000). An example of role flexibility is provided by professors, who can perform at least a portion of their jobs (e.g., writing articles, grading papers) at any point or place in the day. An example of an inflexible work role is a nurse who must work during specific hours in the hospital. The second aspect of boundaries that determines strength, or level of integration, is referred to as permeability, or “the degree to which a role allows one to be physically located in the role’s domain but psychologically and/or behaviorally involved in another role” (Ashforth et al., p. 474). An example of a permeable work role is a secretary who is permitted to use the internet and phone at work for personal reasons (e.g., making doctor’s appointment). An example of impermeability is a call center employee who is not allowed to accept personal phone calls at work. According to boundary theory, flexibility and permeability weaken domain boundaries and thus promote integration, whereas inflexibility and impermeability contribute to strong boundaries and segmentation (Ashforth et al., 2000). Ashforth et al. (2000, p. 475) purport that roles characterized by inflexible and impermeable boundaries also tend to have significant contrast in role identities, defined as “socially constructed definitions of self-in-role (this is who a role occupant is).” For example, a person’s work identity may emphasize self-reliance and tenaciousness, whereas his or her family identity is defined by nurturance and compromise. Role contrast and boundary characteristics are reciprocally related. A person with high role contrast is likely to adopt impermeable and 9 inflexible boundaries in order to preserve the essence of each identity. Impermeable and inflexible boundaries allow role identities to evolve independently thus encouraging high contrast. Because Ashforth et al. (2000) notes that work and nonwork identities are typically high in identity contrast, it will not be part of the present investigation. Some researchers have used a configuration of role characteristics and boundary characteristics to describe where a person falls along the integration-segmentation continuum. For instance, Ashforth et al. (2000, p. 475) suggested that “the concepts of flexibility, permeability, and [identity] contrast jointly define a given pair of roles as segmented or integrated.” Similarly, Bulger, Matthews, and Hoffman (2007) performed a cluster analysis of boundary flexibility and permeability ratings and described resulting profiles according to what level of the integration-segmentation continuum those individuals represent (e.g., a cluster with high flexibility and permeability was deemed to represent high integration). However, other researchers have distinguished between ability to integrate versus actual integration (Matthews, Barnes-Farrell, & Bulger, 2010). Flexibility and permeability reflect a hypothetical or perceived capacity to integrate roles (i.e., weaken boundaries), but a high level of these boundary characteristics does not mean that a person will necessarily integrate roles. For example, a freelance writer may have the ability to intersperse work and personal activities, but may choose to keep them separate. The actual level of integration (i.e., boundary weakness), consistent with the definition offered earlier, is reflected by whether a role is enacted in various settings and at various times, or while physically located in another role. This is the position taken in the present paper—that role flexibility and permeability enable but are not equal to integration. Boundary strength can be critiqued as a broad concept that lacks the level of concreteness required for operationalization. In this paper, it is argued that boundary strength can only be 10 studied empirically by its various indicators. One can study the subjective sense of integration, or the global perception that one’s work life and home life are integrated, as did Desrochers, Hilton, and Larwood (2005). These authors mostly dealt with the issue of separate versus merged work and family identities (e.g., whether a clear boundary exists between one’s career and one’s role as a parent). One can also study boundary strength by measuring role referencing, the frequency that an individual acknowledges one role while in an alternative role through artifacts (e.g., work mementos at home, family photos at work) and conversations (Clark, 2002; OlsonBuchanan & Boswell, 2006). Several authors have studied inter-domain transitions, or the psychological and behavioral movement between the work and nonwork domains (Desrochers et al., 2005; Matthews et al., 2010). Previous measures of inter-domain transitions have included behaviors such as stopping what one is doing at work to attend to a family responsibility, attending to family responsibilities during a lunch break, and leaving work early to meet a family obligation. The intent of this paper is to focus explicitly on those inter-domain transitions that signify punctuation of one role by another role, referred to as work-nonwork interruptions. Worknonwork interruptions are defined as the blending of work and nonwork activities in time and space. To position work-nonwork interruptions within the theoretical nomenclature of boundary theory, the construct is most closely aligned with the concept of permeability. Recall that permeability refers to whether one can be physically located in one role but behaviorally or psychologically preoccupied by another role (Ashforth et al., 2000; Hall & Richter, 1988). A work-nonwork interruption is the actual permeation, “the psychological shift that occurs when a person is physically in one domain and becomes mentally concerned with the other” (Hall & Richter, 1988, p. 215). Hall and Richter described how work-nonwork interruptions can be 11 cognitive (thinking about one domain while in another) but also behavioral (actually performing activities related to the other domain). Work-nonwork interruptions can be self-initiated or otherinitiated. They do not necessarily take place during a pre-specified break (Jett & George, 2003)—rather, they may occur at unplanned times as in the case of unexpected contact from a member of another role, or at natural breaking points depending on the natural rhythms of a task, personal fatigue, etc. One issue surrounding boundary strength deserves clarification. Do work and nonwork, simply lie on a continuum of segmentation to integration, or should work-to-nonwork segmentation be described separately from work-to-nonwork integration? This is essentially the issue of whether strength pertains to one boundary between a pair of roles, or whether each role has its own boundary strength. Pleck (1977, p. 423) suggested that work and nonwork can be “asymmetrically permeable.” Based on qualitative research conducted at Best Buy, Ammons (2009) developed a typology of boundary strength. Two types implied symmetry in boundary strength. There were employees who kept work and nonwork separate from one another and, conversely, there were those employees labeled “Holistic” who had few if any boundaries separating the work and personal domains. The former group focused on their jobs with few personal breaks or activities, and then left all work-related thoughts and items at the door at the end of each workday and workweek. The latter group included an employee who would talk with family and friends throughout the day, exercise during work time, and take work home in the evenings and weekends. However, two types of employees emerged that reflected asymmetries according to direction. There was the traditionally ideal worker, who reluctantly or eagerly integrated work into nonwork (e.g., responded to work requests at home) but kept their personal 12 life contained. Then there was the employee who freely integrated their personal lives into work (e.g., completed family tasks while working at home) but kept work contained. As suggested by Ammon’s research and as other studies have confirmed (e.g., Bulger et al., 2007; Hecht & Allen, 2009; Kossek, Ruderman, Braddy, & Hannum, 2012) work and nonwork do appear to have their own levels of boundary strength. Thus there are two directions of integration (i.e., work-to-nonwork and nonwork-to-work) that are considered orthogonal to one another. Correspondingly, work-nonwork interruptions will be studied as bidirectional. Job Flexibility Past research has focused on the boundary characteristics of both the work and nonwork roles (e.g., Hecht & Allen, 2009; Matthews et al., 2010). Each role has its own boundary with an associated level of flexibility and permeability. Individuals vary in the extent to which their work allows flexibility with regard to location and timing, but they also vary in the extent to which their personal responsibilities can be shifted in time and place. Flexibility in the nonwork domain can depend a great deal on the aspect of nonwork to which one is referring. For example, one can exercise midday or in the evening and thus that part of nonwork may be flexible. Other aspects of nonwork may be less flexible; for instance, one cannot change when a child becomes hungry and needs dinner. Although the boundary characteristics of both work and nonwork are important, the present study will focus only on the boundary characteristics of the work role. The rationale for this narrow focus is that job characteristics are more easily influenced by organizations. Moreover, the purpose of the proposed study is to better understand the implications of job flexibility for work-nonwork balance. This is not to say that all characteristics of the nonwork role will be ignored in this study—family responsibility, a likely correlate of nonwork flexibility and permeability, is expected to play a role in how job flexibility is utilized. 13 Although work role flexibility in the abstract sense is theorized to promote integration, in practice there are several forms of job flexibility that may foster integration to varying degrees. Organizations formally offer a variety of flexible work arrangements, such as flextime, compressed workweeks, part-time work, job sharing, and telecommuting. Some arrangements (e.g., job sharing, reduced workload) are intended to reduce the amount of time worked. These types of arrangements do not embody role flexibility as defined in boundary theory and are assumed not to facilitate integration. Of most relevance to the present paper are those practices that increase flexibility in the timing or location of work. These are the types of job flexibility examined in this study (see Table 1 for examples of representative policies and measurement). An example of a practice designed to increase flexibility over the location of work is telecommuting, defined as “a mode of work in which employees perform all or part of their work outside the employing organization’s physical boundaries, operating and communicating mainly through information technology” (Thatcher and Zhu, 2006, p. 1078). An example of a formal policy designed to provide flexibility over the timing of work is flextime. Under this arrangement, employees are free to vary their starting and quitting times, typically around a set of core hours during which they must be present at work. Flextime provides limited flexibility when compared to more liberal policies, such as “results only work environment” (ROWE) where work can be conducted at anytime an employee chooses (Ammons, 2009; Kelly, Moen, & Tranby, 2011). This suggests the timing of work can be further subdivided into flextime versus more general autonomy over work hours. In addition to formal use of flexibility policies, employees may informally practice flexibility if their supervisors allow it or if they have access to work materials outside of core business hours (Kossek, Lautsch, & Eaton, 2006). Importantly, the focus of this paper is on 14 Table 1 Types of Job Flexibility Type Sub-type Timing General autonomy over work hours Start/end time Location n/a Representative Formal Policies Flextime in which employees are expected to work a certain number of hours per week, but can work whatever hours they choose Sample Measurement of Perceived Flexibility Clark (2002) from “Work Border Flexibility” scale: - I am able to arrive and depart from work when I want. - I am free to work the hours that are best for my schedule. - I could easily take a day off of work if I wanted to. Kossek et al. (2006) from “psychological job control” scale: - To what extent does your job permit you to decide about WHEN the work is done? - I do not have control over when I work (reverse). Flextime in which employees Voydanoff (2005) “Schedule may vary starting and ending Flexibility”: times within limits. Examples: - Dummy variable coded 1 for those who reported being able to choose - Employee establishes their starting and quitting times at starting and ending times but work within some range of hours keeps same schedule each day - Employee must be present Schieman & Young (2010) “Schedule during specified core hours Flexibility”: but may adjust arrival and - Who usually decides when you start departure times each day and finish work each day at your main - Employees may take a job? Is it someone else, or can you longer than scheduled break decide within certain limits, or are you than usual if they make up the entirely free to decide when you start time by starting work earlier and finish work? or staying later Telecommuting – employees Kossek et al. (2006) from regularly work at home during “psychological job control” scale: part or all of a work schedule - I have the freedom to work wherever is best for me—either at home or at work. - To what extent does your job permit you to decide on your own about WHERE the work is done? 15 employee perceptions of job flexibility, regardless of whether it is formal or informal, because this is what determines actual employee behavior and psychological experiences. Another important point is that flexibility in the timing and location of work are likely to correlate. For instance, telecommuters who are primarily identified by their flexibility in location may also have greater flexibility over their schedule than office-bound employees. Similarly, remote access to work through technology allows a person to work anywhere and anytime. Thus, the types of job flexibility outlined in Table 1 can be considered together as a composite. Throughout this introduction the term job flexibility will refer to a composite of perceived flexibility in the location of work and in the timing of work, the latter which includes both flexibility over starting/quitting times and autonomy over working hours more generally. This dissertation proceeds from the viewpoint that it is possible and desirable to distinguish between job flexibility and job permeability. Recall that permeability refers to whether one can be physically present in a role but psychologically or behaviorally involved in another role. Flexible jobs are likely to be permeable. However, one can imagine a job where one must work certain hours within the physical confines of an office, but still has the autonomy to attend to personal concerns throughout the day. Likewise, a person may be able to work at home but due to the nature of their job needs to be completely absorbed in their work. Thus, job flexibility and permeability are not one and the same. Job permeability is facilitated by job flexibility, but also depends on factors like the culture of the organization and nature of the job. Despite the conceptual distinction, several measures of job flexibility have confounded flexibility with permeability. For example, the Clark (2000) measure of job flexibility contains one item that reads “My employer allows me to carry out non-work projects during spare time at work.” 16 The possibility to remain at work but engage in personal business (i.e., permeability) will be considered separately from the ability to choose work hours and location (i.e., job flexibility). The Upsides of Job Flexibility Though the purpose of this paper is to further understanding of the potential challenges of job flexibility, by no means is the intent to refute or discount the benefits of a flexible job. Many employees desire more flexibility in the work role, and with good reason. Job flexibility can reduce work-nonwork conflict through several means. Job flexibility enables employees to schedule work during times and places that work best for them. Work tasks can be scheduled so that they do not interfere with family responsibilities and vice versa (Piotrkowski, 1979). Job flexibility can even increase the quality of time spent in both roles, as explained by Hill, Erickson, Holmes, and Ferris (2010, p. 356): “The highest quality work hours are not always between 8 a.m. and 5 p.m. The best strategic ideas may come to one at 5 a.m. or at 11 p.m. Likewise, the highest quality family time may occur during the regular work day. For example, the best time to hear about school may be right after children come home from school. Putting one’s time to its best use, regardless of the hour of the day, may lead to greater work-life harmony and less conflict.” Job flexibility may be particularly useful for working parents with young children who often pose unpredictable demands (Hill et al., 2010; Poppleton, Briner, & Keifer, 2008). Imagine the common scenario in which a child is sick and needs to leave school or daycare. In this situation, an inflexible job can be stressful because the parent must take a vacation day or find someone to provide alternative childcare. Thus this person may experience family-to-work conflict because he/she is worrying about how to resolve the situation which negatively impacts 17 job performance. This person probably also perceives work-to-family conflict because his/her job is preventing him/her from fulfilling the parent role. A flexible job, on the other hand might allow an employee to take some time off from work to deal with the situation and perhaps even continue working, which could reduce perceptions of conflict (Hill et al., 2010). Other practical benefits have been cited specifically for telecommuting. Employees working from home appreciate the time saved by not having to commute into an office (Ammons & Markham, 2004; Hilbrecht, Shaw, Johnson, & Andrey, 2008; Musson & Marsh, 2008). Time can be reallocated to paid employment, caregiving, or housework. Also, work interruptions by colleagues can be a frequent occurrence in offices that can negatively impact productivity (O’Connell, 2008). Although working at home can open the door to interruptions between work and nonwork, it should on the other hand reduce the frequency of face-to-face interruptions within the work role, which may allow employees to focus more intently on one work task at a time (Ammons & Markham, 2004; Hilbrecht et al., 2008). Finally, there are some psychological benefits of a flexible job apart from the practical ones. First, the belief that one has personal control over demanding situations can reduce stress even if that control is not exercised (Ganster & Fusilier, 1989). Ammons (2009) found that with the rollout of a ROWE initiative at Best Buy, employees perceived a big change just knowing that they could integrate their family and personal lives if needed. Christensen and Staines (1990) suggested that satisfaction with job flexibility is partly due to perceptions of enhanced autonomy, independent of how much that flexibility actually resolves practical dilemmas between work and nonwork. Second, organizational offerings of job flexibility can play a symbolic role, enhancing perceptions of support (Kossek, Lautsch, & Eaton, 2005). Perceptions of how much supervisors or organizations care about their employees’ work-life balance can 18 influence attitudes in a positive manner, reducing perceptions of work-nonwork conflict (Allen, 2001; Kossek, Pichler, Bodner, & Hammer, 2011). The multitude of the benefits described above leads to the hypothesis that, overall, a negative relationship may be observed between job flexibility and both directions of worknonwork conflict. H1: Job flexibility is negatively related to (a) work-to-nonwork conflict and (b) nonworkto-work conflict. The Downsides of Job Flexibility Despite the oft cited benefits of job flexibility, numerous organizational and family scholars have noted the possibility that flexible job arrangements may have the unintended consequence of facilitating interruptions between work and nonwork (e.g., Kossek, Baltes, et al., 2011; Mirchandani, 1998; Rau & Hyland, 2002). Poppleton et al. (2008) commented with regard to their qualitative diary study of employees with high flexibility: “…in an organization where employees had extensive control over time, time was a central theme of almost all of the worknon-work conflicts.” Employees in their study could take time out of their work day to attend to personal needs, but not without guilt and other consequences. Interrupting personal time to meet work demands was also an issue. Professionals with flexibility grapple with just how available they should make themselves to their employers outside of work hours. The process by which job flexibility enables interruptions in both directions is considered next. Nonwork Interrupting Work Given sufficient autonomy over when and where work is performed, one can deal with personal issues as they arise during hours traditionally reserved for work. Family-initiated intrusions are more likely to occur if family members are under the impression that an employee 19 can make up the work later (e.g., in the evening). This is especially the case if an employee’s partner has a relatively less flexible job that does not permit him/her to deal with family emergencies (Rafnsdottir & Heijstra, in press). Flexible schedules may encourage employees to take on even more family responsibilities than they may have if they previously lacked flexibility (Hammer, Neal, Newsom, Brockwood, & Colton, 2005). Having a flexible schedule opens up the possibility of scheduling work around family activities, but inevitably some family demands will clash with time parameters set by clients or colleagues in the form of interruptions (Standen, Daniels, & Lamond, 1999). Likewise, being able to work at home may lead an employee to believe that he/she has more time available for domestic responsibilities, but may end up extending those responsibilities into work time. Several qualitative studies have documented the capacity for telecommuters to multitask work and family duties, for example, by throwing in a load of laundry while waiting for data to process, or unpacking groceries while on a conference call (Ahrentzen, 1990; Hilbrecht et al., 2008; Musson & Marsh, 2008). Additionally, the proximity of family when working at home makes interruptions more likely. If children are present, they may demand immediate attention or need to be driven places. Teleworkers have cited instances of relatives showing up at their homes unannounced while they were working, expecting that they have the free time for a visit (Ahrentzen, 1990). Technology may also play a role in nonwork interruptions of work. The same technologies that enable job flexibility (e.g., cell phone, laptop) are readily available for personal use (Sadler, Robertson, Kan, & Hagen, 2006). Having the internet at one’s fingertips provides a convenient means to take care of personal business (e.g., paying bills electronically) and a way for family members to keep in touch throughout the work day (D’Abate, 2005). Some 20 individuals may find a constant connection comforting whereas others might find it distracting. Whether individuals appraise interruptions as interference should depend on a number of factors, including personal preferences and perceived control, which will be discussed later. Work Interrupting Nonwork When an individual has the ability to work flexible hours and to work from home, work may end up intruding on time or space normally reserved for personal activities. Hall and Richter (1988, p. 219) opined, “The greater flexibility permitted by arrangements such as flextime and flexplace is often at the expense of the greater permeability of the home domain: Work is literally moving into the home.” Colleagues and clients may not be aware of the appropriate time for contact, and employees may feel pressure to respond to unexpected requests immediately even if they are in the middle of an activity with their family. Whereas the use of mobile technology may decrease the emphasis on face-time, it may ironically increase an employee’s sensed obligation to be available around the clock, increasing their susceptibility to otherinitiated intrusions (see work by Towers, Duxbury, Higgins, & Thomas, 2006 on technology as “time thieves” and “space invaders”). In addition to workplace norms that may promote contact outside of traditional business hours (Fenner & Renn, 2010), employees may reason that they owe it to their employer in return for the flexibility that they are provided. Employees who work at home have physical reminders of unfinished work that may prompt them to disengage from personal leisure. The flexibility to communicate anywhere or anytime via laptop or Blackberry encourages the habit of regularly taking breaks from personal activities to check in with work. Individuals may find it difficult to set limits for themselves regarding their work activity after hours. Furthermore, many employees who work in a traditional corporate setting perceive the ability to work after hours as useful—a means to avoid 21 interruptions from colleagues and enhance productivity (Fenner & Renn, 2010). By engaging in ‘time slicing’ people can use spare minutes they have at home or in other third places (e.g., vacation locales) to get work done (Middleton, 2007). Empirical Findings Linking Job Flexibility and Work-Nonwork Interruptions Research testing boundary theory is relatively nascent, but results are generally supportive of the preceding theoretical arguments. Interpretations are somewhat strained by measurement issues. Clark (2002) developed two measures that from a content validity perspective would be considered nonwork interruptions of work and work interruptions of nonwork (e.g., family contacting a person at work, taking care of work-related business while at home). Note that Clark labeled these measures work and home “permeability” but this label is not used here to retain distinction between permeability and interruptions. Clark (2002) also developed a measure of job flexibility. Though it did have some construct contamination (i.e., one item reflecting job permeability), it is best considered a measure of general autonomy over work hours based on the majority of item content. In support of the idea that job flexibility enables work-nonwork integration, Clark (2002) observed a positive association of general autonomy over work hours with nonwork interruptions of work (r = .40) and with work interruptions of nonwork (r = .33). Matthews and Barnes-Farrell (2010) validated a measure labeled “work flexibilityability” adapted from Clark (2002) that captures the ability to change work hours (i.e., general autonomy over work hours) but specifically for the purpose of meeting family needs. For example, an item reads, “I am able to arrive and depart work when I want in order to meet my family and my personal life responsibilities.” Note the difference from a measure of general autonomy over work schedule, in which the ability to work in various times and places is not tied 22 to the purpose for doing so. That is, autonomy over work schedule enables a person to work during the evening or on vacation, but this behavior may be motivated by a demanding job instead of family responsibilities. By constraining autonomy over work hours to be specific to meeting family responsibilities, one would expect the construct to be more highly related to nonwork interruptions of work than to work interruptions of nonwork. This is precisely what is observed. Matthew and Barnes-Farrell (2010) found that their measure of work flexibility was correlated .36 with nonwork interrupting work and .14 with work interrupting nonwork (measures by Clark, 2002). Similar relationships between the same variables were observed in a study performed by Bulger et al. (2007). Also, Matthews et al. (2010) found the same measure of autonomy over work hours (i.e., specific to meeting family responsibilities) was positively correlated with work-to-family transitions (a concept similar to family interrupting work) but not with family-to-work transitions. At least two studies have investigated the relationship between working from home and work-nonwork integration. Desrochers et al. (2005) found that business professors who spent more hours working from home engaged in more frequent transitions between work and household-related tasks throughout the day. Kossek et al. (2006) found that the volume of telework performed was positively related to a work-family boundary management strategy favoring integration. The measure captured a blend of attempts, preferences, and actual behavior engaged in for the purpose of keeping work and family relatively separate versus integrated. Although the use of mobile technology has not been studied directly in relation to work-nonwork interruptions, it has been associated with work-to-family conflict suggesting that the ability to work anytime, anywhere may contribute to such interruptions (Batt & Valcour, 2003). Building on the literature linking flexibility with work-nonwork integration, it is expected that: 23 H2: Job flexibility is positively related to nonwork interruptions of work. H3: Job flexibility is positively related to work interruptions of nonwork. The Type of Job Flexibility Matters Ashforth et al., (2000, p. 481) proposed that “the greater the role integration, the greater the potential for confusion regarding which role identity to enact and for undesired interruptions.” However, as mentioned previously different forms of job flexibility may promote integration to varying degrees. Simply put, the type of flexibility matters. There is reason to believe that permitting employees flexibility in the location of work is more influential in promoting integration than flexibility in the timing of work. When an employee works at home, the physical environment contains cues for multiple roles, priming identities for work and family simultaneously (Shockley & Allen, 2007). Working at home, whether through formal telecommuting or through mobile technology after hours, reduces the traditional methods of coordination and control employed by organizations over their employees, which tend to keep the work and nonwork realms separate (Thatcher and Zhu, 2006). Thatcher and Zhu argue that working at home reduces social interaction with colleagues and supervisors, weakens the influence of organizational culture, and increases the likelihood of coming into contact with members of other life domains. According to social identity theory, a person’s active identity depends not only on subjective importance but also on situational relevance: “the surroundings of the focal individual influence the extent to which a certain identity is perceived as socially appropriate” (Thatcher & Zhu, 2006, p. 1077). For employees performing work in the context of their home or other personal spaces, the social constraints that normally keep work and nonwork separate (i.e., norms for being focused on work) are relaxed. 24 Standen et al. (1999) identified some of the characteristics of telecommuting that weaken the work-nonwork boundary, some of which coincide with those mentioned above: (1) work and home life have a greater opportunity for interaction because they are closer in space and time, (2) the lack of a commute decreases opportunity for role decoupling, (3) the absence of norms about time of attendance, breaks, nonwork phone calls, family visits, etc., (4) the work environment contains reminders of nonwork roles, and (5) unfinished work has a visible presence at home. Even if it is conceptually possible, is it practically feasible to separately measure flexibility in time and place and to investigate their independent effects? Some researchers have argued that alternative work arrangements, such as flextime and telework, should always be lumped together in research since they are not mutually exclusive (Berman, 1997). It is true that flexibility in location and timing of work is likely to co-occur—for instance, telecommuters may have greater control over their schedule, and mobility via technology is likely to affect both the timing and location of work. But they are not perfectly correlated and have been operationalized as separate variables (Shockley & Allen, 2007). For example, one can telecommute with fixed hours and one can have flexible hours without working at home. To the extent that their effects on work-nonwork interruptions can be disentangled, the results may yield useful information. Efforts to prevent undesired interruptions between work and nonwork can be better directed if one can determine whether they are prompted more strongly by the comingling of time or space. H4: Flexibility in the location of work is more strongly related than flexibility in the timing of work to (a) nonwork interruptions of work and (b) work interruptions of nonwork. Furthermore, some types of flexibility in the timing of work may enable work-nonwork integration more than others. Several scholars have suggested that providing employees with 25 choice over their starting and ending times for their workday (i.e., flextime) is actually a relatively segmenting policy. For example, Rothbard, Phillips, and Dumas (2005, p. 245) stated that flextime “requires the employee to distinguish between when they want to work and when they want to spend time in the nonwork domain…This temporal restructuring is intended to remove overlap in work and nonwork activities in a way that reinforces boundaries and reduces nonwork intrusions on work life and work intrusion on nonwork life.” Rau and Hyland (2002) also characterized flextime as an arrangement that reinforces impermeable boundaries. Though these authors offered only indirect empirical support for their assertions, their arguments make sense. If an employee’s temporal flexibility is limited to autonomy over start/end times, then he/she may choose to come into work earlier or later in the day depending on personal obligations but otherwise focus intently on work during business hours and entirely on family and leisure outside of business hours. In contrast, consider an employee who is required to work a fixed number of hours per week but has complete control over when those hours are; in effect, he/she can run personal errands in the middle of work and can work late into the evening. Thus, one might expect differential relationships between flexibility in the timing of work and worknonwork interruptions, depending on whether one is referring to flextime versus general autonomy over work hours. H5: Autonomy with respect to starting/ending times is less strongly related than general autonomy over work hours to (a) nonwork interruptions of work and (b) work interruptions of nonwork. Finally, some organizational researchers have incorporated job permeability into the measurement of job flexibility as mentioned earlier (Clark, 2002; Matthews et al., 2010). For example, the Matthews et al. (2010) flexibility scale contains the item: “While at work, I can 26 stop what I am doing to meet responsibilities related to my family and personal life.” Operationalizing job flexibility in this manner should make it so that the construct is more strongly related to nonwork interruptions of work – one should be able to predict the occurrence of interruptions much better if the perceived ability to engage in such interruptions is specifically measured. Instead of confounding two conceptually distinct constructs, the present study departs from previous research by conceptualizing and measuring job permeability as distinct from job flexibility. To complicate matters, while they are separate constructs they are not presumed to operate independently. Bulger et al. (2007) suggested that by relaxing constraints around when and where work is performed, job flexibility likely contributes to job permeability. Furthermore, job permeability is a far more proximal construct to nonwork interruptions of work. Therefore, it is predicted that job permeability transmits the effects of job flexibility on nonwork interruptions of work. Note that for simplicity sake, the proposed mediation effect is not included in Figure 1. H6: The positive relationship between job flexibility and nonwork interruptions of work is mediated by job permeability. Moderators of the Job Flexibility-Work-Nonwork Interruptions Relationship Whereas job flexibility influences the ability for a person to integrate their work and nonwork roles, there are additional factors that should influence whether they actually do integrate. One such factor is motivation. Matthews and colleagues have developed the construct of flexibility willingness to describe the motivation of individuals to engage in inter-domain transitions (Bulger et al., 2007; Matthews & Barnes-Farrell, 2010; Matthews et al., 2010). In several studies, they have found willingness to transition out of work into family is positively related to actual work-to-family transitions (e.g., leaving work early), and willingness to 27 transition out of family into work is positively related to actual family-to-work transitions (e.g., going into work on the weekend). An additional factor that should influence whether one engages in work-nonwork integration is one’s level of role demands, which corresponds more closely to need rather than motivation. Interrupting the work role to accommodate family responsibilities or the nonwork role to accommodate work is a way to cope with especially heavy role demands (Matthews & Winkel, 2011). When time resources within a role are perceived as insufficient to deal with role demands, an individual may have to “borrow” resources from another role. A person may willingly engage in work-nonwork interruptions for this purpose, or he/she may be a relatively unwilling participant. Work-nonwork interruptions can be initiated by other role members (e.g., supervisors, spouses), who are referred to in role theory as “role senders.” Clark (2000) refers to these individuals as “border keepers” because they have influence in defining role boundaries. In support of the hypothesis that role demands have a main effect on interruptions, Cardenas, Major, and Bernas (2004) found that work overload was significantly positively related to the perceived frequency of work distractions at home. Additionally, Matthews and Winkel (2011) found a positive relationship between work role overload and family-to-work transitions, as well as a positive relationship between family role overload and work-to-family transitions. In addition to the main effects of workload and family responsibility, however, role demands could change the dynamics of the job flexibility-integration relationship. Job flexibility is an enabler of integration, but whether it actually leads to work interrupting nonwork or vice versa depends on a person’s level of role demands. Specifically, workload is hypothesized to operate in an amplifying manner for the relationship between job flexibility and work interruptions of nonwork. Family responsibility is hypothesized to amplify the relationship 28 between job flexibility and nonwork interruptions of work. Thus, the main effects and the moderating effects of role demands are hypothesized as follows: H7: (a) Workload is positively related to work interruptions of nonwork. (b) Job flexibility is more strongly related to work interruptions of nonwork when workload is high. H8: (a) Family responsibility is positively related to nonwork interruptions of work. (b) Job flexibility is more strongly related to nonwork interruptions of work when family responsibility is high. Role demands can also function as a limiting factor that prevents a person from interrupting a role. Being overwhelmed with work constrains one’s ability to suspend work for the sake of nonwork interruptions. If employees do not perceive that they have enough time to complete their work, then they probably will want to conserve their resources as much as possible. Consistent with this notion, Matthews and Barnes-Farrell (2010) found that people who worked more hours feel less able to make transitions out of the work domain to meet family responsibilities. Conversely, heavy family demands constrain a person’s ability to attend to work during personal time or in their home. Matthews and Barnes-Farrell (2010) found that employees who have more children report less ability to transition out of the family domain to meet work responsibilities (Matthews & Barnes-Farrell, 2010). Similarly, Desrochers et al. (2005) found that professors with young children reported less subjective integration of the work and family roles. They speculated that having the responsibility for infants makes the parent boundary less negotiable and therefore less apt to be interrupted by work. Importantly, in addition to the direct negative relationship expected between role demands and interruptions made to that role, there is proposed to be an interaction with job 29 flexibility. The positive relationship between job flexibility and work interruptions of nonwork should be attenuated by family responsibility. Even if employees could extend their work into times and places traditionally reserved for nonwork, their family lives could serve as a constraint thus reducing the likelihood that they would engage in work interruptions of nonwork. Similarly, the positive relationship between job flexibility and nonwork interruptions of work is expected to be attenuated by workload. Even if a person could suspend work temporarily for personal reasons, he/she would probably not be able to do so given a heavy workload. H9: (a) Workload is negatively related to nonwork interruptions of work. (b) Job flexibility is less strongly related to nonwork interruptions of work when workload is high. H10: (a) Family responsibility is negatively related to work interruptions of nonwork. (b) Job flexibility is less strongly related to work interruptions of nonwork when family responsibility is high. Work-Nonwork Interruptions and Conflict The purpose of this dissertation is to examine whether and how job flexibility might indirectly contribute to perceptions of work-life conflict through work-nonwork interruptions. Clark (2000) proposed that when two domains that are dissimilar from one another have weak borders, such as the case when work and nonwork are allowed to interrupt one another, conflict will ensue. The work and home domains typically have different cultures, fulfill different purposes, and are associated with distinct sets of goals. Weakening the boundary between the two creates a situation where work interruptions of personal life might be construed as detracting from one’s personal well-being, and where personal interruptions of work life might be construed as detracting from professional success. 30 There are three mechanisms proposed to account for the relationship between worknonwork interruptions and work-nonwork conflict. First, from a conservation of resources perspective (Hobfoll, 1989), individuals who transition their time and energy frequently out of a domain will perceive a resource loss (Matthews et al., 2010). This situation creates stress for them and they may attribute the source of their potentially insufficient resources to complete their work to their family life (or their family responsibilities to their work). In line with this reasoning, Cardenas et al. (2004) found that work distractions at home were negatively correlated with job satisfaction while family distractions at work were negatively correlated with family satisfaction. Second, there are transaction costs of switching between tasks (Kossek et al., 2005). Research has found that the impact of interruptions is not just the time lost to attend to the new task, but the time that it takes to reabsorb oneself in the original task (O’Connell, 2008). Worknonwork interruptions may inhibit successful role performance by preventing a state of flow and concentration. Indeed, Cardenas et al. (2004) observed that work distractions at home were negatively related to self-reported family performance. Third, by performing and thinking about one role while being physically located in another, strain generated in one domain is more likely to spill over into the other domain. Ilies, Wilson, and Wagner (2009) found that employees who perceived their work and family roles as highly integrated displayed stronger spillover of mood from work to home. The less clear the boundary is between work and home, the more difficulty people will have cognitively distancing themselves from work issues and concerns when they are trying to relax (Zijlstra & Sonnentag, 2006). Without fixed work hours or without “leaving the office,” it can feel as though work is never finished (Ahrentzen 1990). As a result work-related rumination may occur and hinder 31 psychological detachment. Likewise, if employees intersperse personal activities throughout their work day, their minds may easily drift back while working toward whatever nonwork issue they were dealing with last, interfering with their ability to concentrate on work. Evidence largely supports the hypothesized relationship between work-nonwork interruptions and conflict. One study found that the frequency of job contacts at home is positively related to work-to-family conflict (Voydanoff, 2005). Another study found that family intrusions into work, measured via daily diaries, were positively related to perceptions of family interfering with work, though the same study failed to find a significant correlation between work intrusions into family and work interference with family (Williams & Alliger, 1994). Finally, the frequency with which one transitions out of work to meet family responsibilities has been positively linked to family-to-work conflict (Matthews et al., 2010). The reverse was also true—transitions out of the family role to meet work demands were positively related to work-tofamily conflict. Based on these findings, it is predicted that: H11: Work interruptions of nonwork are positively related to work-to-nonwork conflict. H12: Nonwork interruptions of work are positively related to nonwork-to-work conflict. The Moderating Role of Preference for Segmentation The extent to which interruptions are seen as problematic may depend on idiosyncratic boundary management preferences. Individuals are presumed to have preferences regarding the extent to which their work and nonwork lives are separate versus integrated (Ashforth et al., 2000). A preference for segmentation reflects one approach to work-life balance focused on preventing cross-role contamination, but that is not the only approach. At the opposite end of the spectrum are individuals who prefer to integrate—their approach to balance is to deal with personal issues on work time and with work issues at home, so that neither overwhelms the other 32 (Hudson, Christensen, & Kellogg, 2002). More recently authors have clarified that preferences can differ according to the direction of segmentation (Powell & Greenhaus, 2010), such that an individual can prefer to strongly segment work from family (i.e., preserve the “purity” of family time and the family role identity) but prefer to integrate family to work (i.e., attend to family tasks while at work and invoke the family role identity while in the work role). Though individuals strive to attain their desired level of segmentation between work and nonwork, it can be challenging. If one has the ability to work in various places and times, for example, it can be tempting to become more attached to work than what is personally ideal. In the short run, one’s daily decisions are guided not by how many hours one wants to work, but by real considerations such as finishing tasks by deadlines. In the long run the accumulation of these decisions may result in an outcome that does not reflect one’s preferred balance between work and nonwork (van Echtelt, Glebbeek, & Lindenberg, 2006). Employees can also genuinely believe that tending to work outside of traditional times and places will help ease job demands; unfortunately, the work demands reduced by these practices may be offset by new demands that arise as a function of the employee’s availability (Middleton, 2007). There is a tension between wanting to avoid interruptions and appreciating their usefulness (Hudson et al., 2002). For example, talking with family members throughout the day may pose distractions, but also provides a way to stay informed and feel connected. Because of the above tensions as well as situational constraints on how much job flexibility an individual is afforded, preferences will not always align with one’s actual level of work-nonwork integration. This gives rise to various degrees of misfit that have implications for well-being and perceptions of work-nonwork conflict. A small literature has dealt with the fit between work-family segmentation supplies and preferences (Chen, Powell, & Greenhaus, 2009; 33 Edwards & Rothbard, 1999; Kreiner, 2006; Rothbard et al., 2005). These studies have focused on segmentation of work from personal life. Findings show that individuals can experience discomfort if their degree of segmentation exceeds their preference (Kreiner, 2006; Rothbard et al., 2005). However, this discomfort is less severe than that which occurs when the environment does not supply enough segmentation (Edwards & Rothbard, 1999; Kreiner, 2006; Rothbard et al., 2005). Thus there appears to be a general advantage of segmenting work from personal life in terms of higher well-being and less work interference with family, which is consistent with the hypothesized interruptions-conflict relationship. What these findings suggest is that individuals who prefer segmentation of work from nonwork should be especially likely to view work interruptions of nonwork as undesirable. Therefore, it is expected that: H13: The positive relationship between work interruptions of nonwork and work-tononwork conflict is stronger at higher levels of preference for segmenting work from nonwork. Only two studies were located that examined preferences for segmentation of nonwork from work. Bulger, Hoffman, and Matthews (2011) found that a preference for segmenting personal life from work exacerbated the relationship between work-to-personal life transitions and health symptoms. They did not find a significant moderating effect for the outcome of personal life-to-work conflict. Methot and LePine (2009) found that individuals who prefer to segment family from work were less likely to initiate a hypothetical workplace romance (an integrating behavior). This finding reinforces the idea that some people may be uncomfortable bringing nonwork elements into the work domain. Thus, a parallel prediction for the moderating role of segmentation preferences is made for the nonwork-to-work direction. 34 H14: The positive relationship between nonwork interruptions of work and nonwork-towork conflict is stronger at higher levels of preference for segmenting nonwork from work. The Moderating Role of Boundary Control Whether or not work-nonwork interruptions are viewed as instances of work-nonwork conflict may depend on individual preferences, but perhaps more fundamentally also on a person’s sense of control over these interruptions. Although job flexibility is intended to increase employee’s control over managing work and nonwork demands, there are environmental influences that can constrict a person’s autonomy. Work-nonwork interruptions are not always self-initiated. Other-initiated interruptions seem particularly likely to inspire a sense of “helplessness to invasion” (Desrochers et al., 2005). In this case, work-nonwork interruptions are stressors because other people are determining where an individual allocates his/her resources. Alternatively, individuals may lack skills in self-regulation that allows them to control their work-nonwork interruptions. Taking breaks from work to attend to personal matters, for example, can be helpful for restoring energy, but can also turn into distractions. In that case work-nonwork interruptions play the role of stressors because which role an individual should attend to at any given moment may not be immediately clear to a person (i.e, a form of interrole ambiguity, which has not been studied). A sense of personal control is central to several theories about the stress process. In the demands-control model (Karasek, 1979) employee control (discretion, authority, decision latitude, etc.) buffers the impact of work stressors on strain. Similarly whether a potential stressor leads to strain depends on perceived control in the transactional model of stress and coping (Lazarus & Folkman, 1984). A qualitative study at IBM revealed that control over work- 35 life flexibility was important to employees. Even employees who clearly favored work-nonwork integration preferred to choose when to fit activities in rather than being involuntary interrupted (Hudson et al. 2002). Kossek and Lautsch (2008) contend that a sense of control over managing the extent of work/nonwork integration is more important than whether one actually integrates. Individuals have to weigh the tradeoffs of integration and segmentation, decide for themselves what strategy will work best for them, and actively self-monitor their behavior. Qualitative interviews by Kossek and Lautsch revealed that some participants felt trapped in a particular style of boundary management. A recent empirical investigation placed individuals into clusters based on their level of work-nonwork interruptions, work and family identity centrality, and boundary control (Kossek et al., 2012). Those clusters with higher boundary control tended to have more positive work and family outcomes, regardless of their profile of work interruptions of nonwork and nonwork interruptions of work. Thus, in the present study boundary control is expected to play a buffering role in the relationship between work-nonwork interruptions and work-nonwork conflict. H15 (a): The positive relationship between nonwork interruptions of work and nonworkto-work conflict is weaker at higher levels of boundary control. H15(b): The positive relationship between work interruptions of nonwork and work-tononwork conflict is weaker at higher levels of boundary control. Inconsistent Mediation: The Downsides of Job Flexibility Mask the Upsides Though the potential downsides of job flexibility are the main focus of the proposed study, the upsides cannot and should not be ignored if we are to fully understand the implications of job flexibility for work-life balance. Boundary theory suggests that work-nonwork integration is associated with various tradeoffs. How these tradeoffs combine to determine level of work- 36 nonwork conflict has not received much attention in the literature. Earlier it was hypothesized that job flexibility has an overall negative relationship with work-nonwork conflict because of its benefits. The positive relationship between job flexibility and work-nonwork interruptions, however, may be masking some of the benefit of job flexibility, because of the potential influence of work-nonwork interruptions on perceptions of conflict. This would suggest that the relationship between job flexibility and work-nonwork conflict could be more strongly negative if it were not for work-nonwork interruptions. Schieman and Young (2010) were the first to suggest that certain aspects of worknonwork integration may suppress the benefits of job flexibility. They found that the relationship between control over starting/ending times of work and work-to-family conflict became more strongly negative after controlling for work-family multitasking. The latter was measured with one item about frequency of trying to work on job tasks and home tasks at the same time while at home. The present study seeks to extend their findings by broadening the conceptualization and measurement of work-nonwork integration. Work-nonwork interruptions may result in multitasking, but they can also result in abandoning one task in favor of the other. Both types of integration can result in work-nonwork conflict. The present study also seeks to examine additional forms of job flexibility (i.e., more general autonomy over work hours and flexibility in location of work) and both directions of work-nonwork conflict. Inconsistent mediation is conceptually distinct from other third variable effects, such as consistent mediation and confounding (MacKinnon, Krull, & Lockwood, 2000). In consistent mediation, there is a direct and an indirect effect that are of the same sign, either positive or negative. Including the mediating variable in a regression reduces the magnitude of the relationship between the predictor and criterion, because it is presumed to be carrying at least 37 some of the predictor’s influence. With inconsistent mediation, there is also a direct and indirect effect but they are opposite in sign. That is, the predictor may have an overall negative relationship with a criterion, but an indirect positive relationship with that same criterion. When an inconsistent mediator is included in the regression, the magnitude of the relationship between the predictor and criterion increases (e.g., becomes more negative). Finally, there is a confounding effect when the third variable is not considered to be an “intermediate” variable but simply happens to be related to both the predictor and the criterion, artificially inflating or obscuring the relationship between the two. It is important to note that there can be evidence for work-nonwork interruptions acting as an inconsistent mediator even if job flexibility does not have an overall negative relationship with work-nonwork conflict. The negative direct path and positive indirect path could cancel each other out resulting in an overall nonsignificant relationship (Judd & Kenny, 1981). Thus, while the statistical calculation of indirect effects is equivalent in consistent mediation and inconsistent mediation, for the latter the requirement that the predictor and criterion be significantly related is relaxed (MacKinnon et al., 2000). The pattern of coefficients that indicates inconsistent mediation is (a) the estimate of the total effect is closer to zero than the direct effect, and (b) the third variable and direct effects are opposite in sign. H16(a): Work interruptions of nonwork is an inconsistent mediator of the relationship between job flexibility and work-to-nonwork conflict. H16(b): Nonwork interruptions of work is an inconsistent mediator of the relationship between job flexibility and nonwork-to-work conflict. Exploratory Questions 38 This study is primarily concerned with processes occurring within one person, to gain an in-depth understanding of how different aspects of work and nonwork come together for that person. However, much could be learned from studying the interpersonal relationships in which these processes are embedded. A person’s spouse or partner is especially likely to affect how he or she balances the demands of work and home. Poelmans (2001, p. LIT44) contended that work-life conflict “is basically a dynamic process between two individuals that are mutually independent and adaptive.” Some researchers have conjectured that it is the job flexibility of a person relative to his or her spouse that matters most in determining who interrupts work to take care of family responsibilities (Rafnsdottir & Heijstra, 2011; Shockley, 2010). The relative job flexibility of dual earner couples may also influence the division of household labor, which has implications for work-nonwork conflict. As will be described shortly, this study entails collecting responses from significant others to improve the methodological rigor of the study. Though the main purpose for doing so is to reduce common method bias (i.e., obtain ratings from another person’s perspective), there is an opportunity to ask some additional questions. Thus, I investigated the following exploratory research questions for those respondents with an employed significant other. 1. How does an employee’s job flexibility relative to the job flexibility of a significant other relate to the focal employee’s nonwork interruptions of work? 2. How does an employee’s job flexibility relative to the job flexibility of a significant other relate to the division of household labor? 39 PILOT STUDY Established scales were not available to adequately measure several of the constructs. First, previous studies of job flexibility have typically focused on one type of flexibility (i.e., location or time) or used less than three items to measure each type of flexibility (e.g., Schieman & Young, 2010; Shockley & Allen, 2007). Second, job permeability has often been confounded in existing scales as mentioned previously. Third, while there is a validated scale measuring work-nonwork interruptions, it does not distinguish between self- and other-initiated interruptions. Thus, a pilot study was conducted to develop new scales to measure these constructs, using items from existing scales as a starting point and supplementing where needed to ensure coverage of the construct space. Method Sample Participants were 287 members of the Professional Managers Association (PMA), a national association representing managers, management officials, and non-bargaining unit employees in the federal government. Respondents were 64.5% female with a mean age of 52. The racial composition was: 63.4% Caucasian, 12.9% African American, 8.4% Hispanic, and 2% Asian. The majority were married (69%) and less than half had children at home (44.5%). Respondents worked in diverse occupations (e.g., computer scientist, instructional designer, revenue officer) on average 42 hours per week. The highest level of education achieved was distributed as follows: college degree (50%), high school diploma (21%), graduate degree (16%), and technical school (4%). Procedure 40 An invitation to participate was sent to the PMA membership by the organization’s Executive Director via e-mail, which contained a link to the online survey. Of the approximately 1,500 members on the PMA listserv, 287 responded for a response rate of 19%. Measures Items to measure job flexibility, job permeability, and work-nonwork interruptions were derived from existing scales, which were supplemented with new items to ensure adequate coverage of the dimensions of interest (e.g., timing and location of work, self-initiated and otherinitiated interruptions). The items and their sources are provided in Tables 2 and 3. The first two measures had response scales of 1 = Not at All to 5 = Very Great Extent, whereas the latter measure had a frequency response scale of 1 = Never to 5 = Very Often. Results Job Flexibility and Job Permeability Exploratory factor analyses were conducted using principal axis factor, varimax rotation, and an extraction of factors based on a review of the scree plot. For the analysis of job flexibility and job permeability items, a three-factor solution was retained (see Table 2). The content of the factors corresponded to flexibility in the timing of work, flexibility in the location of work, and job permeability. Based on the results, the decision was made to drop two job flexibility items (1 and 4) and two job permeability items (11 and 12) from the final scales due to relatively lower loadings (less than .50), non-significant loadings, or cross-loadings. The results did not support flexibility over start/quit times as a distinct construct from flexibility in the timing of work more generally. The reliability of the scales formed as a result of the factor analyses were α = .88, .93, and .77 for flexibility in the timing of work, flexibility in the location of work, and job permeability, respectively. 41 Table 2 Job Flexibility and Job Permeability Factor Loadings from Pilot Study EFA Factor # Source Item Keep? a 3 A I have control over when I work. 5 B 2 B 6 C 1 A 4 10 B F 8 A 7 A 9 15 F F 13 B 14 D 12 C 11 E I am able to arrive and depart from work when I want. I am free to work the hours that are best for my schedule. I have choice in determining when I begin and end each workday. My job permits me to decide about when the work is done. I could easily take a day off of work if I wanted to. I have flexibility in selecting the location of where I work. I have the freedom to work wherever is best for me—either at home or at work. My job permits me to decide on my own where the work is done. I can work from almost anywhere. I could conduct personal errands during business hours if necessary. My employer allows me to carry out non-work projects during spare time at work. While at work, I can stop what I am doing to meet responsibilities related to my family and personal life. I have choice in deciding whether I can send or receive personal email while I work. I am expected to limit the number of times I make or receive personal phone calls while I work. 1 2 3 Yes .842 .181 .001 Yes .768 .168 .122 Yes .766 .158 .019 Yes .759 .259 045 No .493 .290 .120 No Yes .476 .283 .269 .883 .192 .116 Yes .322 .880 .122 Yes .293 .823 .022 Yes Yes .197 .127 .760 .119 .017 .749 Yes .021 .036 .734 Yes .234 .212 .662 No .053 .067 .383 No .018 .047 -.153 Notes. Item sources were: (A) Kossek et al. (2006), (B) Clark (2002), (C) Batt and Valcour (2003), (D) Matthews and Barnes-Farrell (2010), (E) Thomas and Ganster (1995), and (F) new a item based on literature review (Hill et al., 2010; Hecht & Allen, 2009). Item was revised from original to eliminate the need for reverse coding. 42 Work-Nonwork Interruptions A preliminary factor analysis of all work-nonwork interruptions items suggested a fourfactor solution. However, an inspection of the factor matrix revealed that two factors were composed of only two items each relating to monitoring and responding to communications (i.e., from work at home and vice-versa). These items did not correlate highly with any of the other items and, as such, appeared to be measuring a different construct. A decision was made to proceed with the factor analysis excluding these four items. In the final analysis a three-factor solution was extracted with factors that appeared to represent self-initiated work-to-nonwork interruptions, other-initiated work-to-nonwork interruptions, and nonwork-to-work interruptions (see Table 3). Three items (2, 8, and 17) were identified to be dropped from the final scale due to relatively lower loadings (at or below .50) or substantial cross-loadings. Additionally, two items (3 and 13) related to thinking about personal matters while at work and thinking about work during personal time, respectively, were flagged for deletion. They were among the items with lower loadings and further consideration revealed that they overlapped conceptually with items from established scales of strain-based work-life conflict. The reliability of the scales formed as a result of the factor analyses were α = .90, .89, and .76 for self-initiated work-tononwork interruptions, other-initiated work-to-nonwork interruptions, and nonwork-to-work interruptions, respectively. 43 Table 3 Work-Nonwork Interruptions Factor Loadings from Pilot Study EFA Factor # Source Item 10 A 11 9 12 A A A 13 16 A B 14 B 15 B 17 A 18 C I take care of work issues during my personal time away from work. I work during my vacations. I bring work home. I bring work materials with me when I attend personal or family activities. I think about work during my personal time. My personal life is interrupted by unexpected contact (e.g., emails, texts, and phone calls) from a colleague. My colleagues make unexpected requests of me during my leisure time. I stop whatever personal activity I am engaged in because a colleague requires my attention. I allow work to interrupt me when I spend time with my family or friends. I receive calls from co-workers or my supervisor while at home. I stop what I am working on because a family member or friend requires my attention. I take care of personal or family needs during work. I allow family/friends to interrupt me when I am working. My work is brought to a halt due to unexpected contact (e.g., emails, texts, and phone calls) from a family member or friend. I think about my family, friends, or personal a interests while working. My family members or friends make unexpected requests of me while I am working. I receive calls from family members while at work. When I work from home, I handle personal or family responsibilities during work. 5 B 1 A 7 B 6 B 3 A 4 8 B C 2 A Keep ? Yes 1 2 3 .849 .322 .067 Yes Yes Yes .768 .765 .683 .393 .197 .436 .042 .071 .111 No Yes .563 .360 .286 .829 .193 .113 Yes .366 .794 .066 Yes .276 .777 .232 No .550 .663 .175 Yes .302 .559 .015 Yes .031 .100 .667 Yes .163 -.116 .663 Yes .064 .110 .659 Yes -.019 .075 .580 No .185 .100 .558 Yes -.015 .231 .552 No No .016 .281 .072 .498 -.255 .465 Notes. Item sources were: (A) Kossek et al. (2012) – revised to omit frequency in item stem, (B) a new item, and (C) Matthews et al. (2010). Item was revised from original to eliminate the need for reverse coding. 44 MAIN STUDY Method Sample Participants were employees at the headquarters of a large manufacturing organization. The final sample of 1,141 individuals whose responses were used for data analysis was 75.5% male. The majority was married or partnered (85.5%) and a large percentage also had a spouse or partner who also worked (58.3%). Over two-thirds had children living at home (71.4%). The age distribution of respondents was as follows: 33 or younger (2.3%), 34 to 46 (42.4%), 47 to 64 (53.8%), and 65 and older (.4%). Respondents worked 51 hours per week on average. The percentage of respondents according to job level was: individual contributor (15.7%), manager – supervise individual contributors only (60.2%), manager of other managers (18.8%), and director/executive/and above (4.3%). The sample of 196 significant others was 18% male. About two-thirds of significant others were employed (67.9%), more frequently in a full-time capacity (45.9%) than part-time (21.9%). Those who worked full-time worked 45 hours per week on average, and those who worked part-time worked 19 hours per week on average. About four-fifths of participating significant others reported having children at home (79.6%). The age distribution of significant others was as follows: 33 or younger (3.6%), 34 to 46 (49.0%), 47 to 64 (44.9%), and 65 and older (<.1%). Procedures A random sample of 2500 employees was invited to participate. A senior-level Human Resources (HR) manager sent advance notice of the survey initiative to HR leads of the participating skill teams (Finance, Product Development, Manufacturing, Human Resources, IT, 45 Marketing Sales & Service, Purchasing, and Staff/Other) followed by an e-mail to the employees informing them about the survey. The actual survey link was sent to employees in an email from the primary investigator, followed by a reminder e-mail a week and a half later. Responses were received from 1,189 employees constituting a response rate of 47.6%, which falls between the company’s typical response rate for limited audience, single purpose surveys (10 to 30%) and the historical response rate for the annual company-wide survey (70%). Some individuals left the survey early and, thus, did not provide responses to a sizable portion of the items. The responses of 48 individuals who completed less than 70% of the survey were excluded from data analysis. This specific cutoff was chosen because it meant that individuals did not complete any of the work-nonwork interruptions items. Employees were given the option to volunteer the e-mail address of a significant other with whom they live. Using the contact information provided, 508 significant others were invited to complete a survey of which 229 responded (45% response rate). To match responses and keep the study anonymous, a coding system was used whereby employees and their significant others each provided the first two letters of their street names as well as the months of both their birthdays. Using these codes, 196 of the 229 significant other responses could be matched to the employee dataset. The failure to match all cases was due to a substantial number of employees (89) not providing the code in the requested format, an issue which was prevented from occurring in the significant other survey. All participating significant others were entered into a drawing for a $100 gift certificate to an online retailer. Measures All measures with items in the order of presentation are contained in Appendix A. 46 Job Flexibility. Two dimensions of job flexibility were assessed using the scales developed in the pilot study. Flexibility in the timing of work was measured with 4 items (α = .92) and flexibility in the location of work was measured with 4 items (α = .89). Additionally, four supplemental items were included to measure specific types of schedule flexibility. Respondents were asked to check the types that applied to their job (e.g., “You may take a longer than scheduled break if you make up the time by starting work earlier or working later”). Job Permeability. Job permeability was assessed using the scale developed in the pilot study. The 3-item scale had acceptable reliability (α =.77). Work-Nonwork Interruptions. Three dimensions of work-nonwork interruptions were assessed using the scales developed in the pilot study. The wording of one item was revised to improve construct validity (i.e., “I receive calls from my co-workers or my supervisor while at home” was modified to “I receive calls from my co-workers or my supervisor during the evening or weekend.”). Self-initiated work-to-nonwork interruptions and other-initiated work-to-nonwork interruptions were measured with 4 items each (α = .87 and .92, respectively). Nonwork-to-work interruptions were measured with 5 items (α =.80). Workload. Workload was measured using an abbreviated version of the scale by Caplan, Cobb, French, Harrison, and Pinneau (1975). This four-item scale has been used in several studies (Cardenas et al., 2004; Greenhaus, Parasuraman, Granrose, Rabinowitz, & Beutell, 1989; Parasuraman, Greenhaus, & Granrose, 1992). A sample item is “My job requires me to work very hard.” Reliability of the scale was acceptable (α =.84). Participants also reported number of hours worked as an alternative means of measuring workload that more closely parallels the measure of family responsibility. 47 Family Responsibility. A three-item scale by Konrad (2003) was used to measure family responsibility. Respondents were asked to indicate the average hours per week they spend on housework (e.g., cooking, cleaning, laundry), child care and/or parent care, and household maintenance (e.g., lawn care, painting, household repairs). A fourth item was added that referred to “support of spouse or partner, including providing emotional support, attending his/her work or social functions, entertaining his/her friends or coworkers.” The number of hours reported for each of the four categories was summed. This scale is formative and therefore high internal consistency is not expected. Work-Nonwork Conflict. Work-nonwork conflict was measured using the scale developed by Fisher et al. (2009). Five items assessed work-to-nonwork conflict (α = .90), and six items assessed nonwork-to-work conflict (α = .85). A sample item is “My personal life suffers because of my work.” Preference for Segmentation. Kreiner’s (2006) four-item scale was used to measure preference for segmenting work from nonwork (α = .86). A sample item is “I prefer to keep work life at work.” Preference for segmenting nonwork from work was measured using Kreiner’s scale adapted by Methot and LePine (2009) to measure the opposite direction of preference (α = .88). Boundary Control. A three-item measure developed by Kossek et al. (2012) was used to measure boundary control (α = .91). A sample item is “I control whether I combine my work and personal life activities throughout the day.” Telecommuting. Two items were included in the survey to explore whether policy use makes a difference beyond perceptions of actual flexibility. Although not directly related to the study hypotheses, this information was thought useful to the company sponsoring the research. One item asked respondents if they had a formal arrangement with up with their employer to 48 work from home. The second item asked for an estimate of how many days, in increments of .5, the respondent worked at home during standard business hours. Informal Practice of Job Flexibility. Two questions were also included to understand what types of informal flexibility employees were practicing. One item asked respondents to estimate the number of hours per week they typically spend working at home during the evenings or weekends. The second item asked respondents to indicate the type of flexible schedule they practiced if any. The company did not have a formal program for schedule flexibility. A sample response option was “You establish your starting and ending times but maintain the same schedule each day.” Demographics. Participants completed questions related to sex, age, job level, number of hours worked, marital status, employment of spouse or partner, and number of children living at home. Significant Other Survey. Participating significant others answered questions about the focal employee on measures of job flexibility, job permeability, work-nonwork interruptions, and work-nonwork conflict. Questions were adapted to reference the third person (see items provided in Appendix B). Self reports and other reports of these constructs were expected to be imperfectly but substantially correlated. Previous research suggests that self reports and spousal reports of work-to-family conflict, for instance, do not perfectly agree (e.g., r = .49, Grandey, Cordeiro, & Crouter, 2005). To address the exploratory research questions, significant others also received questions about their own job flexibility and level of family responsibility (i.e., the same scales that were administered to the focal employees). Division of household labor was calculated by dividing the focal employee’s hours by the total hours summed across partners (Sawyer, 2009). 49 Results Discriminant Validity of Scales A series of confirmatory factor analyses were conducted to examine the disctinctiveness of the scales. In each model, items were specified to load on only one factor and factors were free to correlate. Maximum likelihood estimation was used to estimate parameters. Models with fit indices at .08 or below for SRMR and .95 and above for CFI or .06 or below for RMSEA are generally considered to display good fit (Hu & Bentler, 1999). The results for these analyses are presented in Table 4. Table 4 Confirmatory Factor Analysis Results 2 Fit indices Constructs Job Flexibility and Job Permeability Work-Nonwork Conflict Work-Nonwork Interruptions Work-Nonwork Interruptions and WorkNonwork Conflict 2 χ -difference test 2 Models 1-Factor 2-Factor χ 2373.54** 1752.58** df 44 43 SRMR .11 .08 CFI .85 .89 RMSEA .25 .22 Δχ -620.96** df -1 3-Factor 286.54** 41 .03 .98 .08 1466.04** 2 2-Factor 387.32** 43 .06 .97 .09 -- -- 1-Factor 2-Factor 3-Factor 2772.83** 1262.89** 386.56** 65 64 62 .16 .07 .04 .80 .91 .98 .23 .15 .07 -1509.94** 876.33** -1 2 1-Factor 8937.89** 252 .18 .72 .23 -- -- 2-Factor 4-Factor 5-Factor 5558.31** 251 1950.30** 246 1062.02** 242 .12 .06 .05 .83 .94 .97 .17 .09 .06 3379.58** 3608.01** 888.28** 1 5 4 2 Notes. Minimum fit function χ values are reported as recommended by Kline (2011). SRMR = standardized root mean residual. CFI = comparative fit index. RMSEA = root mean square error of approximation. **p < .001. 50 Job Flexibility and Job Permeability. A two-factor model in which job flexibility and job permeability items loaded on separate factors fit the data better than a one-factor model, 2 Δχ (1) = 620.96, p < .001. Further, decomposing job flexibility into the dimensions of timing and 2 location significantly improved fit, Δχ (2) = 1466.04, p < .001. The three-factor model was a good fit to the data (SRMR = .03, CFI = .98, RMSEA = .08). The latent correlations between factors are presented in Table 5. The highest correlation was between flexibility in the timing and flexibility in the location of work at .68. Table 5 Phi Matrix for Job Flexibility/Permeability Confirmatory Factor Analysis 1 2 3 Job Flexibility - Timing Job Flexibility - Location Job Permeability 1 1.00 .68 .53 2 3 1.00 .48 1.00 Work-Nonwork Conflict and Work-Nonwork Interruptions. The conflict and interruptions scales had a sufficient number of items to examine each of the constructs separately as a first step. Based on the work of Fisher et al. (2009), a two-factor solution for the worknonwork conflict scale was tested, which had acceptable fit (SRMR = .06, CFI = .97, RMSEA = .09). A two-factor model of work-nonwork interruptions that distinguished between direction of 2 interruptions fit the data better than a one-factor model, Δχ (1) = 1509.94, p < .001. Further, distinguishing between self-initiated and other-initiated work-to-nonwork interruptions in a 2 three-factor model significantly improved fit, Δχ (2) = 876.33, p < .001. The three-factor model for work-nonwork interruptions fit the data well (SRMR = .04, CFI = .98, RMSEA = .07). 51 In the combined model of conflict and interruptions, distinguishing between direction (i.e., work-to-nonwork and vice versa) significantly improved fit beyond a baseline one-factor 2 model, Δχ (1) = 3379.58, p < .001. Additionally, a four-factor model that parses interruptions 2 from conflict provided even better fit to the data, Δχ (5) = 3608.01, p < .001. Finally, significant improvement was seen after distinguishing between self-initiated and other-initiated work-to2 nonwork interruptions in a five-factor model, Δχ (4) = 888.28, p < .001. Fit of the five-factor model was adequate (SRMR = .05, CFI = .97, RMSEA = .06). The latent correlations between the factors are presented in Table 6. The largest correlation was between the two dimensions of work interruptions of nonwork (self- and other-initiated) at .73. The two directions of interruptions demonstrated sufficient distinctiveness with correlations under .20. Additionally, conflict was distinguishable from interruptions with intercorrelations ranging from .19 to .52. Table 6 Phi Matrix for Interruptions/Conflict Confirmatory Factor Analysis 1 2 3 4 5 W-NW Conflict NW-W Conflict NW-W Interruptions W-NW Interruptions (S) W-NW Interruptions (O) 1 1.00 .15 .19 .46 .52 2 3 4 5 1.00 .42 -.09 -.02 1.00 -.12 -.13 1.00 .73 1.00 Descriptive Statistics and Scale Intercorrelations Descriptive statistics and scale intercorrelations for the major study variables are presented in Table 7. Hypothesis Testing 52 Table 7 Scale Descriptives and Intercorrelations N M SD 1 2 3 4 5 6 7 8 9 10 1 Sex (0=Male, 1=Female) 1118 0.23 0.42 -2 Age -.13 1127 2.53 0.55 -3 Marital/Partner Status .04 -.15 1131 0.86 0.34 -4 Spouse FT Employment .37 -.09 -.08 994 0.44 0.5 -5 Number of Children .33 -.15 1130 2.46 1.15 -.13 -.17 -6 Job Level .02 -.08 .03 .07 1129 2.12 0.71 -.01 -7 Hours Worked .01 -.12 .04 .31 1126 51 8.98 -.07 -.04 -8 Workload .00 -.09 .00 -.02 .05 .16 .33 (.84) 1140 4.11 0.71 9 Family Responsbility .01 .02 .13 -.13 .15 .15 -.04 -.08 1114 28.1 21.9 -10 Job Flexibility - Timing .01 (.92) .07 -.01 -.01 .06 -.02 -.04 -.20 -.20 1140 2.65 0.96 11 Job Flexibility - Location .05 .02 .02 -.10 -.07 .05 .10 -.05 -.02 .63 1137 2.22 0.9 12 Job Permeability .04 .01 .01 .02 -.07 -.13 .00 .45 1129 2.23 0.76 -.01 -.09 13 W-NW Interruptions (S) .00 -.06 .03 -.05 .01 -.09 .09 .23 .34 .42 1141 3.19 0.91 14 W-NW Interruptions (O) .04 -.24 .06 -.05 .10 .26 .37 .43 1141 2.66 0.89 -.12 -.02 15 NW-W Interruptions .06 .01 -.08 -.11 .10 .09 .07 .23 1140 2.21 0.49 -.01 -.05 16 W-NW Conflict .01 -.05 .01 -.04 .05 .01 -.40 .12 .33 .64 1141 3.32 0.79 17 NW-W Conflict .03 .01 -.03 .04 .03 .03 .10 -.02 -.04 .16 1141 1.85 0.56 18 Pref for Seg (W from NW) .03 -.11 -.04 .03 -.16 .07 .13 1133 3.86 0.81 -.02 -.10 -.01 19 Pref for Seg (NW from W) .00 -.06 .00 -.06 -.04 .06 .09 -.11 -.08 1135 3.59 0.79 -.04 20 Boundary Control .03 -.07 .03 -.06 -.02 -.19 -.34 -.03 .12 .36 1127 3.18 0.98 21 Formal Telecommuting .01 -.06 .05 -.04 -.03 -.09 -.10 -.02 .09 .19 1136 0.09 0.28 22 Telecommuting Volume .06 .02 .00 .03 .09 -.01 -.06 -.02 -.07 .11 1126 0.24 0.77 23 Hrs Worked Eve/Weekend 1138 7.05 6.90 .06 -.04 .01 .01 .04 .12 .25 .23 .11 -.06 Notes. Correlations in bold are significant (p < .05). S = Self-Initiated. O = Other=Initiated. W = Work. NW = Nonwork. FT = Fulltime. 53 Table 7 (cont’d) Scale Intercorrelations 11 12 13 14 15 16 17 18 19 20 21 22 23 1 Sex (0=Male, 1=Female) 2 Age 3 Marital/Partner Status 4 Spouse FT Employment 5 Number of Children 6 Job Level 7 Hours Worked 8 Workload 9 Family Responsbility 10 Job Flexibility - Timing 11 Job Flexibility - Location (.89) 12 Job Permeability .40 (.77) 13 W-NW Interruptions (S) .12 -.05 (.87) 14 W-NW Interruptions (O) -.10 -.15 .66 (.92) 15 NW-W Interruptions .17 .43 -.10 -.11 (.80) 16 W-NW Conflict -.26 -.28 .41 .47 -.16 (.90) 17 NW-W Conflict -.01 .06 -.06 -.01 .37 .16 (.85) 18 Pref for Seg (W from NW) .05 (.86) -.20 -.09 -.23 -.07 -.03 .18 19 Pref for Seg (NW from W) .04 .05 -.33 -.07 -.18 .12 -.17 .32 (.88) 20 Boundary Control .25 .28 -.41 -.50 .14 -.49 -.06 -.08 -.03 (.91) 21 Formal Telecommuting .01 -.01 -.05 -.16 -.07 -.04 -.01 .31 .14 .11 -22 Telecommuting Volume .04 .05 .02 -.01 -.08 -.02 -.05 .02 .01 .23 .34 -23 Hrs Worked Eve/Weekend .09 -.19 .06 -.08 -.09 .46 .36 -.09 .26 -.02 -.08 .06 Notes. Correlations in bold are significant (p < .05). S = Self-Initiated. O = Other=Initiated. W = Work. NW = Nonwork. FT = Fulltime. 54 Results overview. An overall model depicting the level of support for hypotheses is provided in Figure 2, followed by a model presenting results for the supported hypotheses in Figure 3. The models are provided to summarize the results, but the individual relationships depicted were tested via separate regression analyses (aside from mediations). Control variables. Control variables were entered into the first step of hierarchical regression analyses. Two variables were chosen for statistical control on the basis of significant correlations with dependent variables. For analyses with conflict or interruptions in the direction of work to nonwork as the dependent variable, job level and number of children were used as control variables. For analyses with conflict or interruptions in the direction of nonwork to work as the dependent variable, only number of children served as a control variable. Job Flexibility and Work-Nonwork Conflict. Hypothesis 1A states that job flexibility is negatively related to work-to-nonwork conflict. As predicted, flexibility in the timing of work was negatively related to work-to-nonwork conflict, β = -.39, p < .001 (overall model F(3,1119) 2 = 76.51, p < .001, R = .17). Similarly, flexibility in the location of work was negatively related 2 to work-to-nonwork conflict, β = -.26, p < .001 (overall model F(3,1116) = 20.32, p < .001, R = .09). Thus, Hypothesis 1A was fully supported. Hypothesis 1B states that job flexibility is negatively related to nonwork-to-work conflict. Flexibility in the timing of work was not a significant predictor of nonwork-to-work 2 conflict, β = .04, p > .05 (overall model F(2,1126) = 6.49, p < .01, R = .01), nor was flexibility 2 in the location of work, β = -.01, p > .05 (overall model F(2,1123) = 5.85, p < .01, R = .01). These results fail to support Hypothesis 1B. 55 Figure 2 Support for Hypothesized Relationships Family + Responsibility - Job Flexibility Composite - Flexibility over Timing > Start/Quit Times > General Autonomy + - Preference for Segmenting Nonwork from Work Nonwork Interruptions of Work + + Nonwork-to-Work Conflict - + + Boundary Control - - Flexibility over Location + - Work Interruptions of Nonwork - Workload Work-to-Nonwork Conflict + + Preference for Segmenting Work from Nonwork + Note. Solid lines represent predicted relationships that were supported. Dotted lines represent predicted relationships that were not supported. Dashed line represents mixed support. The hypothesized mediated relationship of job flexibility to nonwork interruptions of work through job permeability is not represented here though it was supported (see Figure 3). 56 Figure 3 Significant Results of Hypothesis Tests Family Responsibility Job Permeability .35*** .07** Nonwork Interruptions of Work .25*** .36*** Nonwork-to-Work Conflict .36*** Job Flexibility in Timing of Work -.12*** -.39*** Job Flexibility in Location of Work Work-to-Nonwork Conflict -.26*** (-.31***) -.09*** -.11*** Self-Initiated Work Interruptions of Nonwork .10*** Workload .41*** .47*** -.23*** .39*** Other-Initiated Work Interruptions of Nonwork Note. Standardized coefficients are shown. 57 Job Flexibility and Work-Nonwork Interruptions. Hypothesis 2 states that job flexibility is positively related to nonwork interruptions of work. In support of this hypothesis, flexibility in the timing of work was a significantly positive predictor of nonwork interruptions of work, β = 2 .24, p < .001 (overall model F(2,1137) = 40.49, p < .001, R = .07), as was flexibility in the 2 location of work, β = .17, p < .001 (overall model F(2,1122) = 21.99, p < .001, R = .04). Hypothesis 3 states that job flexibility is positively related to work interruptions of nonwork. In support of this hypothesis, flexibility in the location of work was significantly related to self-initiated work interruptions of nonwork in the anticipated direction, β = .10, p < 2 .001 (overall model F(3,1116) = 29.41, p < .001, R = .07). However, flexibility in the location of work was negatively related to other-initiated work interruptions of nonwork. β = -.11, p < 2 .001 (overall model F(3,1116) = 34.79, p < .001, R = .09). Furthermore, flexibility in the timing of work was negatively related to both forms of work interruptions of nonwork: self-initiated, β = 2 -.09, p < .01 (overall model F(3,1119) = 28.17, p < .001, R = .07), and other-initiated, β = -.23, 2 p < .001 (overall model F(3,1119) = 53.43, p < .001, R = .13). These results provide only partial support for Hypothesis 3 and highlight the importance of considering different types of job flexibility and interruptions. Type of Flexibility. Hypothesis 4a states that flexibility in the location of work is more strongly related than flexibility in the timing of work to nonwork interruptions of work. Contrary to this hypothesis, the zero-order correlation between flexibility in the timing of work and nonwork interruptions of work (r = .23) was somewhat larger than the zero-order correlation between flexibility in the location of work and nonwork interruptions of work (r = .17). To test 58 whether these correlations were significantly different from each other, Steiger’s (1980) Z-test for comparing dependent correlations was performed, which suggested that they are in fact different, Z = 2.64, p < .01. Additional evidence in support of this trend is provided by a relative weights analysis including number of children as a control variable (Johnson, 2000), which showed that flexibility in the timing of work explained 4% of the variance in nonwork interruptions of work whereas flexibility in location explained only 2% of the variance. Thus, Hypothesis 4a was not supported. Hypothesis 4b states that flexibility in the location of work is more strongly related than flexibility in the timing of work to work interruptions of nonwork. In the case of self-initiated work interruptions of nonwork, this relationship could not be technically be tested as it was intended (i.e., that one flexibility type was more strongly positively related) since the relationships of these predictors were opposite in sign. Nevertheless, the underlying logic of Hypothesis 4b was supported for self-initiated work interruptions of nonwork in that flexibility in the location of work was a positive predictor whereas flexibility in the timing of work was a negative predictor (i.e., see results above for Hypothesis 3). These results are consistent with the idea that flexibility in where one performs work is more of a facilitator of self-initiated interruptions of personal life than is flexibility in when one performs work. With respect to other-initiated work interruptions of nonwork, however, Hypothesis 4b was not supported. Results were inconsistent with either type of job flexibility facilitating this form of interruption. In contrast, both types of flexibility were significantly negative predictors of other-initiated work interruptions of nonwork (see results above for Hypothesis 3). Hypothesis 5a states that autonomy with respect to starting and ending times for work is less strongly related to nonwork interruptions of work than is more general autonomy over work 59 hours. A test of Hypothesis 5a could not be conducted as planned, because the item analyses did not support a separate factor to measure autonomy over starting/ending times as distinct from more general autonomy over work hours. There were supplemental items in the survey designed to measure very specific forms of schedule flexibility that provide tentative support of this hypothesis. An inspection of correlations between these supplemental items and nonwork interruptions of work shows that two items relating to establishing one’s own starting and ending times were not significantly related to nonwork interruptions of work (rs = -.00 and .05, ps > .05), whereas two items referring to taking longer breaks or choosing whichever hours one wants to work were significantly positively related to nonwork interruptions of work (rs=.12 and .10, ps < .01). Hypothesis 5b makes the same predictions as above but with respect to work interruptions of nonwork. The intention of the hypothesis was to examine whether autonomy over start and end times of work was less strongly positively related to work interruptions of nonwork than autonomy over work hours more generally. However, because flexibility in the timing of work was negatively related to both forms of work interruptions of nonwork, it makes little sense to test this hypothesis. Mediating Role of Job Permeability. Hypothesis 6 states that the positive relationship between job flexibility and nonwork interruptions of work, which was established by the results for Hypothesis 2, is mediated by job permeability. In fulfillment of the requirements for mediation, flexibility in the timing of work was a significant predictor of job permeability, β = 2 .45, p < .001 (F(1,1126) = 281.64, p < .001, R = .20), which in turn was a significant predictor of nonwork interruptions of work, β = .42, p < .001 (overall model F(2,1114) = 128.82, p < .001, 2 R = .19). Importantly, when both flexibility in the timing of work and job permeability were 60 entered as predictors in a multiple regression, the size of the coefficient for flexibility in the 2 timing of work was reduced to β =.07, p < .05 (overall model F(3,1112) = 88.20, p < .001, R = .19). The additional criteria required to demonstrate that job permeability mediates the relationship between flexibility in the location of work and nonwork interruptions of work were also satisfied. Flexibility in the location of work was a significant predictor of job permeability, β 2 = .40, p < .001 (F(1,1123) = 218.71, p < .001, R = .16). After adding job permeability as a predictor in a multiple regression, the relationship between flexibility in the location of work and nonwork interruptions of work was no longer significant β =.01, p > .05 (overall model 2 F(3,1109) = 85.50, p < .001, R = .19). The direct and indirect effects were also estimated via a bootstrapping procedure. Nonwork interruptions of work was entered as the dependent variable, job flexibility in the timing of work as the predictor variable, job permeability as the mediator, and number of children as a control variable in the SPSS macro created by Preacher and Hayes (2008). The bootstrap results indicated that the relationship between flexibility in work timing and nonwork interruptions of work (total effect b = .12, p < .001) was reduced when job permeability was included in the model (direct effect b = .03, p < .05). Furthermore, the effect of flexibility in work timing through job permeability was significant (indirect effect b = .09, 95% confidence interval = .07 to .11). Next, the same analysis was conducted with job flexibility in the location of work entered as the predictor variable. Again, the relationship of flexibility in the location of work to nonwork interruptions of work (total effect b = .10, p < .001) was reduced and, in fact, became nonsignificant (direct effect b = .01, p > .05) after entering job permeability in the model. The indirect effect of flexibility in the location of work through job permeability was 61 significant (indirect effect b = .09, 95% confidence interval = .08 to .12). These results indicate that higher job permeability is largely responsible for the positive relationship between flexibility in the timing of work and nonwork interruptions of work. Additionally, the positive relationship between flexibility in the location of work and nonwork interruptions of work can be fully explained by job permeability. Thus, full support is provided for Hypothesis 6. Predictors of Work-Nonwork Interruptions and Moderators of the Job FlexibilityWork-Nonwork Interruptions Relationship. Hypothesis 7a states that workload is positively related to work interruptions of nonwork. This hypothesis was fully supported for both selfinitiated work interruptions of nonwork, β = .39, p < .001 (overall model F(3,1119) = 97.98, p < 2 .001, R = .21), and other-initiated work interruptions of nonwork, β = .39, p < .001 (overall 2 model F(3,1119) = 105.73, p < .001, R = .22). Hypothesis 7b states that the relationship between job flexibility and work interruptions of nonwork is stronger at higher workload levels. Workload significantly interacted with both types of job flexibility in predicting self-initiated work interruptions of nonwork, β = -.06, p < 2 .05 (overall model F(5,1116) = 59.87, p < .001, R = .21) and β = -.05, p < .05 (overall model 2 F(5,1113) = 65.78, p < .001, R = .23) for flexibility in timing and location, respectively. These 2 interactions were trivial in magnitude (ƒ < .01) but explored nonetheless. Because only flexibility in location predicted self-initiated work interruptions of work in the theoretically predicted direction, this interaction was investigated first. The nature of the interaction was counterintuitive and contrary to Hypothesis 7b in that flexibility in the location of work was less strongly positively related to self-initiated work interruptions of nonwork when workload was high. Further, the unexpectedly negative relationship between flexibility in the timing of work 62 and self-initiated work interruptions of nonwork was more strongly negative when workload was high. To facilitate interpretation of these interactions, the independent variables (i.e., job flexibility) and moderator (i.e., workload) were reversed. The results showed that the positive relationship between workload and self-initiated work interruptions of nonwork is attenuated at higher levels of job flexibility (see Figure 4). Workload also significantly interacted with both types of job flexibility to predict otherinitiated work interruptions of nonwork, β = -.06, p < .05 (overall model F(5,1116) = 73.00, p < 2 2 .001, R = .25) and β = -.06, p < .05 (overall model F(5,1113) = 66.55, p < .001, R = .23) for flexibility in timing and location, respectively. Again, these interactions represent trivial effect 2 sizes (ƒ < .01) and the nature of the interactions was not as predicted. Recall, first of all, that both forms of job flexibility were unexpectedly negatively related to other-initiated work interruptions of nonwork. Higher workloads amplified those negative relationships. Similar to above, when the independent and moderating variables were flipped, the results demonstrated that the positive relationship between workload and other-initiated work interruptions of nonwork is attenuated at higher levels of job flexibility (see Figure 5). Hypothesis 8a states that family responsibility is positively related to nonwork 2 interruptions of work. This hypothesis was supported, β = .06, F(2,1100) = 6.37, p = .05, R = .01. The result is qualified, however, by the fact that the effect size would be considered trivial 2 (ƒ < .01). Hypothesis 8b predicted that the relationship between job flexibility and nonwork interruptions of work would be stronger at higher levels of family responsibility. The results failed to provide support for this hypothesis. Family responsibility was not a significant moderator of the relationship between job flexibility and nonwork interruptions of work, β = .03, 63 Figure 4 Job Flexibility Attenuates the Positive Relationship between Workload and Self-Initiated Work Interruptions of Nonwork 5 4.5 4 Low Flexibility in Work Timing 3.5 3 High Flexibility in Work Timing 2.5 2 1.5 1 Low Workload High Workload 5 4.5 4 3.5 Low Flexibility in Work Location 3 High Flexibility in Work Location 2.5 2 1.5 1 Low Workload High Workload 64 Figure 5 Job Flexibility Attenuates the Positive Relationship between Workload and Other-Initiated Work Interruptions of Nonwork 5 4.5 4 Low Flexibility in Work Timing 3.5 3 High Flexibility in Work Timing 2.5 2 1.5 1 Low Workload High Workload 5 4.5 4 3.5 Low Flexibility in Work Location 3 High Flexibility in Work Location 2.5 2 1.5 1 Low Workload High Workload 65 2 p > .05 (overall model F(4,1097) = 20.03, p < .001, R = .07) and β = -.02, p > .05 (overall 2 model F(4,1094) = 11.43, p < .001, R = .04) for flexibility in timing and location, respectively. In support of Hypothesis 9a workload was a significantly negative predictor of nonwork 2 interruptions of work, β = -.12, p < .001 (overall model F(2,1125) = 12.55, p < .001, R = .02. Hypothesis 9b states that job flexibility is less strongly related to nonwork interruptions of work when workload is high. However, workload was not a significant moderator of the relationship between job flexibility and nonwork interruptions of work, β = .00, p > .05 (overall 2 model F(4,1122) = 21.97, p < .001, R = .07) and β = .00, p > .05 (overall model F(4,1119) = 2 14.39, p < .001, R = .05) for flexibility in the timing and location of work, respectively. Hypothesis 10a states that family responsibility is negatively related to work interruptions of nonwork. The results failed to provide support for this hypothesis, β = .00, p > .05, (overall 2 model F(3,1094) = 25.63, p < .001, R = .07), and β = .04, p > .05, (overall model F(3,1094) = 2 30.84, p < .001, R = .08) for self-initiated and other-initiated interruptions, respectively. According to Hypothesis 10b, job flexibility is less strongly related to work interruptions of nonwork when family responsibility is high. However, the interaction between the two variables was not significant in the prediction of self-initiated interruptions, β = -.02, p > .05 (overall 2 model F(5,1091) = 17.30, p < .001, R = .07) and β = -.01, p > .05 (overall model F(5,1088) = 2 17.68, p < .001, R = .08) for flexibility in the timing and location of work, respectively. Neither was the interaction significant in the prediction of other-initiated interruptions, β = -.01, p > .05 2 (overall model F(5,1091) = 32.31, p < .001, R = .13) and β = -.01, p > .05 (overall model 66 2 F(5,1088) = 21.45, p < .001, R = .09) for flexibility in the timing and location of work, respectively. Work-Nonwork Interruptions and Conflict. Hypothesis 11 states that work interruptions of nonwork are positively related to work-to-nonwork conflict. The results offer full support for this hypothesis, with both self-initiated work interruptions of nonwork , β =.41, p < .001 (overall 2 model F(3,1120) = 79.43, p < .001, R = .18), and other-initiated interruptions of nonwork , β 2 =.47, p < .001 (overall model F(3,1120) = 107.66, p < .001, R = .22), emerging as strong positive predictors of work-to-nonwork conflict. Hypothesis 12 states that nonwork interruptions of work are positively related to nonwork-to-work conflict. The results also offer strong support for this hypothesis, β =.36, p < 2 .001 (overall model F(2,1126) = 90.65, p < .001, R = .14). Moderating Role of Preference for Segmentation. Hypothesis 13 states that preference for segmenting work from nonwork moderates the relationship between work interruptions of nonwork and work-to-nonwork conflict, such that the relationship is stronger for those who prefer greater segmentation. Preference for segmenting work from nonwork had a direct positive relationship with work-to-nonwork conflict, β =.20, p < .001, but contrary to Hypothesis 13, it failed to moderate the effects of either type of work interruption of nonwork, β = -.04, p > .05 2 (overall model F(5,1113) = 78.24, p < .001, R = .26) and β = -.02, p > .05 (overall model 2 F(5,1113) = 83.39, p < .001, R = .27) for self-initiated and other-initiated interruptions, respectively. 67 Hypothesis 14 states that preference for segmenting nonwork from work moderates the relationship between nonwork interruptions of work and nonwork-to-work conflict, such that the relationship is stronger for those who prefer greater segmentation. Preference for segmenting nonwork from work had a direct negative relationship with nonwork-to-work conflict (β = -.16, p < .001), but failed to moderate the effect of nonwork interruptions of work, β = .04, p > .05 2 (overall model F(4,1121) = 46.37, p < .001, R = .14). Thus, Hypothesis 14 was not supported. Moderating Role of Boundary Control. Hypothesis 15a states that boundary control moderates the relationship between nonwork interruptions of work and nonwork-to-work conflict, such that the relationship becomes less strongly positive at higher levels of boundary control. Counter to Hypothesis 15a boundary control was not a significant moderator of the effect of nonwork interruptions of work, β = -.03, p > .05 (overall model F(3,1122) = 64.26, p < 2 .001, R = .15). Though there was a significant positive correlation between boundary control and nonwork-to-work conflict (r = .06), the main effect of boundary control after controlling for number of children living at home was nonsignificant as well, β = -.06, p > .05 (overall model 2 F(2,1117) = 7.23, p < .01, R = .01). Hypothesis 15b states that boundary control moderates the relationship between work interruptions of work and work-to-nonwork conflict, such that the relationship becomes less strongly positive at higher levels of boundary control. Boundary control had a strong direct negative relationship with work-to-nonwork conflict (β = -.49, p < .001), but it failed to moderate the effects of either type of work interruptions of nonwork, β = -.04, p > .05 (overall model 2 F(5,1108) =97.70, p < .001, R = .31) and β = .00, p > .05 (overall model F(5,1108) = 101.41, p 2 < .001, R = .31) for self-initiated and other-initiated interruptions, respectively. 68 Inconsistent Mediation. Hypothesis 16a states that work interruptions of nonwork is an inconsistent mediator of the relationship between job flexibility and work-to-nonwork conflict. As hypothesized, the negative relationship between flexibility in the location of work and workto-nonwork conflict, β =-.26, p < .001, became somewhat stronger after controlling for selfinitiated work-to-nonwork interruptions, β =-.31, p < .001 (overall model F(4,1115) = 102.91, p 2 < .001, R = .27). The other conditions required for inconsistent mediation via the Baron and Kenny (1986) procedure were also satisfied; recall that flexibility in the location of work is a significantly positive predictor of self-initiated work interruptions of nonwork (see results for Hypothesis 3), which in turn is a significantly positive predictor of work-to-nonwork conflict (see results for Hypothesis 11). Preacher and Hayes’ (2008) bootstrapping procedure was also used to test this hypothesis. As expected, the negative relationship between flexibility in the location of work and work-to-nonwork conflict (total effect b = -.23, p < .001), became stronger when including self-initiated work-to-nonwork interruptions as a mediator (direct effect b = -.27, p < .001). Furthermore, the indirect effect was statistically significant and in the opposite direction (indirect effect b = .04, 95% confidence interval = .02 to .06) indicating inconsistent mediation. It was not useful to test for inconsistent mediation for the remaining parts of Hypothesis 16a. As previously reported, flexibility in the location of work was a negative predictor of otherinitiated work interruptions of nonwork as well as a negative predictor of work-to-nonwork conflict. An independent variable displaying a relationship with a mediator that is in the same direction as its relationship with the dependent variable is not supportive of inconsistent mediation. Similarly, flexibility in the timing of work was negatively related to self- and otherinitiated work interruptions of nonwork, thus, violating the conditions for inconsistent mediation. 69 Hypothesis 16b states that nonwork interruptions of work is an inconsistent mediator of the relationship between job flexibility and nonwork-to-work conflict. The conditions for testing this hypothesis were met – both types of flexibility were positively related to nonwork interruptions of work (see results for Hypothesis 2), which in turn was positively related to nonwork-to-work conflict (see results for Hypothesis 12). The nonsignificant relationship between flexibility in the timing of work and nonwork-to-work conflict became significantly negative after controlling for nonwork interruptions of nonwork, β = -.06, p = .05 (overall model 2 F(3,1124) = 61.12, p < .001, R = .14). Additionally, the nonsignificant relationship between flexibility in the location of work and nonwork-to-work conflict became significantly negative after controlling for nonwork interruptions of nonwork, β = -.07, p < .05 (overall model 2 F(3,1121) = 63.01, p < .001, R = .14). Again, bootstrapping was also conducted to test for inconsistent mediation. The results were similarly supportive. The nonsignificant relationship between flexibility in the timing of work and nonwork-to-work conflict (total effect b = .02, p > .05) became significantly negative when controlling for nonwork interruptions of work (direct effect b = -.03, p = .05). The indirect effect of flexibility in the location of work on nonwork-towork conflict through nonwork interruptions of work was significant (indirect effect b = .05, 95% confidence interval = .04 to .07). The same pattern of results was observed for flexibility in the location of work. That is, the nonsignificant relationship between flexibility in the location of work and nonwork-to-work conflict (total effect b = -.00, p > .05) became significantly negative when controlling for nonwork interruptions of work (direct effect b = -.05, p < .05). The indirect effect was significant (indirect effect b = .04, 95% confidence interval = .03 to .06). Therefore, the results provide full support for Hypothesis 16b. 70 Analysis with Significant Other Reports. Before testing hypotheses using significant other data, confirmatory factor analyses were performed to ensure that the scales remained factorally distinct when completed with respect to another person. Replicating the findings from the self report data, the results (see Table 8) provided support for two dimensions of job flexibility that are distinct from permeability, as well as three dimensions of work-nonwork interruptions that are distinct from work-nonwork conflict. Table 8 Confirmatory Factor Analysis Results for Significant Other Reports 2 Fit indices Constructs 2 χ 2 Δχ -- Work-Nonwork Interruptions Work-Nonwork Interruptions and WorkNonwork Conflict SRMR .10 CFI .88 RMSEA .25 2-Factor 380.17** 43 .08 .90 .23 83.42** 1 3-Factor 76.75** 41 .03 .99 .06 303.42** 2 2-Factor 150.68** 43 .06 .96 .11 -- -- 640.57** 65 .16 .85 .23 -- -- 2-Factor 246.58** 64 .06 .95 .13 393.99** 1 3-Factor Work-Nonwork Conflict 463.59** df 44 1-Factor Job Flexibility and Job Permeability Models 1-Factor χ -difference test 112.41** 62 .04 .99 .06 134.17** 2 1-Factor 2175.07** 252 .18 .78 .24 -- -- 2-Factor 1528.07** 251 .16 .85 .19 647.00** 1 4-Factor 550.50** 246 .06 .97 .08 977.57** 5 5-Factor 416.00** 242 .05 .98 .06 134.50** 4 2 df -- Notes. Minimum fit function χ values are reported as recommended by Kline (2011). SRMR = standardized root mean residual. CFI = comparative fit index. RMSEA = root mean square error of approximation. **p < .001. 71 An examination of the correlations between the self and other reports (see Table 9) suggested that employees and their significant others provided convergent ratings of the employees’ work interruptions of nonwork (rs = .49 and .57 for self-initiated and other-initiated, respectively) as well as work-to-nonwork conflict (r = .57). In contrast, they had more discrepant viewpoints regarding the nonwork-to-work direction of interruptions and conflict (r = .24 and r = .15, respectively). The relative size of these correlations is consistent with past research suggesting spouses can more reliably report on the work-to-nonwork direction of conflict than the nonwork-to-work direction (Green, Bull Schaefer, MacDermid, & Weiss, 2011). Based on these findings, the decision was made to only use significant other reports in the work-tononwork direction. Presumably, significant others have more opportunity to observe the work-tononwork phenomena since they unfold at home and, therefore, such reports should have greater validity. Significant other reports of work-to-nonwork conflict and work interruptions of nonwork were used to corroborate the findings from the self-report data involving these variables. Selfreports of job flexibility were used to minimize same source bias. To test for inconsistent mediation, which involves both variables, only the significant other reports of work interruptions of nonwork were used. Regression analyses were conducted hierarchically with the control variables in the first step. A test of Hypothesis 1 revealed that only flexibility in the timing of work was a significantly negative predictor of work-to-nonwork conflict, β = -.27, p < .0001 (overall model 2 F(3,189) = 7.41, p < .001, R = .11). The relationship between job flexibility in the location of work was in the negative direction, as hypothesized, but was not statistically significant, β = -.11, 2 p > .05 (overall model F(3,187) = 2.95, p < .05, R = .05). 72 Table 9 Intercorrelation between Self and Significant Other Reports 1 2 3 4 5 6 7 8 Job Flexibility - Timing Job Flexibility - Location Job Permeability W-NW Interruptions (S) W-NW Interruptions (O) NW-W Interruptions W-NW Conflict NW-W Conflict Self Report N M SD 1140 2.65 .96 1137 2.22 .90 1129 2.23 .76 1141 3.19 .91 1141 2.66 .89 1140 2.21 .49 1141 3.32 .79 1141 1.85 .56 α .92 .89 .77 .87 .92 .80 .90 .85 Significant Other Report N M SD α 196 2.45 .88 .92 195 1.96 .76 .91 194 2.17 .70 .77 194 3.1 .96 .88 194 2.74 .99 .94 195 2.3 .54 .85 194 2.8 .97 .93 194 1.46 .46 .85 1 .36 .63 .45 -.09 2 .63 .39 .40 .12 3 .49 .44 .32 -.05 4 -.24 .02 -.17 .49 5 -.28 -.09 -.27 .76 6 7 .28 -.36 .18 -.20 .54 -.35 -.29 .58 8 -.07 -.01 .05 -.07 -.24 .23 -.40 .03 -.10 .17 -.26 -.01 -.15 .43 -.28 .06 .66 -.10 .41 -.06 .57 -.11 .47 -.01 -.33 .67 .24 -.36 -.16 .57 .37 .16 -.06 .29 .22 .15 Notes. Correlations in bold are significant (p < .05). Correlations along the diagonal (shaded) are between self reports and significant other reports. Correlations above the diagonal are between significant other reports only. Correlations below the diagonal are between self reports only. S = Self-Initiated. O = Other=Initiated. W = Work. NW = Nonwork. 73 In support of Hypothesis 3, flexibility in the location of work was a significantly positive predictor of self-initiated work interruptions of nonwork, β = .16, p < .05 (overall model 2 F(3,187) = 4.11, p < .01, R = .06). However, consistent with the findings using all self report data, flexibility in the timing of work was a significantly negative predictor of other-initiated 2 work interruptions of nonwork, β = -.18, p < .05 (overall model F(3,189) = 4.26, p < .01, R = .06). No other relationships between job flexibility and work interruptions of nonwork were significant. In support of Hypothesis 7a, self-reported workload was a significantly positive predictor of both self-initiated work interruptions of nonwork, β = .26, p < .001 (overall model F(3,189) = 2 7.16, p < .001, R = .10) and other-initiated work interruptions of nonwork, β = .33, p < .001 2 (overall model F(3,189) = 10.17, p < .001, R = .14). Hypothesis 11 was also supported using the significant other reports of interruptions. Both types of work interruptions of nonwork were significantly positive predictors of work-to2 nonwork conflict, β = .30, p < .001 (overall model F(3,189) = 7.64, p < .001, R = .11) and β = 2 .41, p < .001 (overall model F(3,189) = 14.48, p < .001, R = .19) for self-initiated and otherinitiated interruptions, respectively. Because the relationship between job flexibility and work interruptions of nonwork was only significantly positive with respect to flexibility in location and self-initiated interruptions, inconsistent mediation (Hypothesis 16a) was only plausible using these variables as the independent variable and mediator, respectively. Consistent with an inconsistent mediation effect, the negative relationship between self-reported flexibility in the location of work and 74 work-to-nonwork conflict became more strongly negative after entering self-initiated work interruptions of nonwork into the regression equation, β = -.32, p < .001 (overall model F(4,186) 2 = 12.45, p < .001, R = .21). The same conclusions were reached via Preacher and Hayes’ (2008) bootstrapping procedure. The direct effect of flexibility in work location was significantly negative (direct effect b = -.15, p = .05), whereas there was a significantly positive relationship through self-initiated work interruptions of nonwork (indirect effect b = .05, 95% confidence interval = .01 to .11). Exploratory Research Questions. The exploratory research questions concern how job flexibility relative to the job flexibility of one’s significant other relates to nonwork interruptions of work and division of household labor. Essentially, an exploration of these questions requires consideration of a three dimensional space that maps the joint relationship of the focal employee’s and significant other’s job flexibility with outcomes. Polynomial regression, an analytical technique that has become a popular method for examining the relationship of personenvironment fit with other variables, was identified as the most appropriate method of investigation to answer the exploratory research questions for several reasons. First and foremost, it permits one to model three dimensional response surfaces. Second, it avoids some of the well-known problems associated with difference scores. Third, Kristof-Brown, Zimmerman, and Johnson (2005) suggested that more complicated relationships emerge in polynomial regression compared to more traditional approaches (e.g., interactions, difference scores). For example, researchers have observed differences in the effects of fit between two variables (e.g., supplies and preferences) depending on whether the two variables are high or low and asymmetries in the effects of misfit (e.g., more supplies are better than fewer, Ostroff, 2007). In the present study, misfit was of particular interest (i.e., when an employee has more or less job 75 flexibility than his/her significant other). Misfit is represented by the line of incongruence, where X = -Y on the response surface graph (Shanock, Baran, Gentry, Pattison, & Heggestad, 2010). Different effects were expected depending on the direction of misfit. Employees were expected to report the most frequent nonwork interruptions of work if their job flexibility exceeded that of their significant other, and the fewest nonwork interruptions of work if their significant other reported relatively greater job flexibility. A positive slope along the line of incongruence (calculated via the formula provided by Shanock et al., 2010) would support this pattern of results. Finally, self- and significant other reports of job flexibility fit the types of data amenable to polynomial regression since the two measures were commensurate (i.e., assess the same content dimension) and used the same numeric scale (Edwards, 2002). Matched self-reports of job flexibility from employees and their significant others were used to test the relationships with employee-reported nonwork interruptions of work. The intercorrelations and descriptive statistics for these variables are presented in Table 10. Because not all significant others were employed, the sample for analyses was limited to 134 and 132 employees and their significant others for flexibility in the timing and location of work, respectively. When conducting exploratory polynomial regression analysis, Edwards (2000) recommends that equations of a progressively higher order (linear, quadratic, etc.) be estimated until the increment in variance explained is not significant. Significance of equations with higher order terms indicates that response surface methodology is useful in understanding the variable interrelationships. In the first step of a hierarchical regression analysis, the employee and significant other job flexibility variables were entered as predictors (i.e., the linear equation). In the second step, the squared terms and the cross-product were entered as predictors (i.e., the 76 Table 10 Correlations with Significant Others’ Job Flexibility N 1 2 3 4 5 6 Nonwork interruptions of work Division of Household a Labor Flexibility in timing of work Flexibility in place of work Significant other flexibility in timing of work Significant other flexibility in location of work M SD 1 1140 2.21 .49 (.80) 191 .37 .21 .08 1140 2.65 .96 .23 1137 2.22 .90 .17 134 2.63 132 2.05 2 3 4 5 6 -** .08 (.92) .13 .63 1.29 .03 -.13 .02 -.06 (.95) 1.28 .10 -.11 -.04 -.10 .76 ** ** (.89) ** (.97) a Notes. Coefficient alphas are presented along the diagonal. Calculated as employee's percentage of the total time spent on household tasks between the employee and his/her significant other quadratic equation). The predictor variables (i.e., employee job flexibility and significant other job flexibility) were first centered relative to the scale midpoints as recommended by Edwards (1994). Two such regressions were performed for the analysis of nonwork interruptions of work, the first with respect to flexibility in the timing of work and the second with respect to flexibility in the location of work. In both analyses, the first step of the regression equations explained significant variance and employees’ job flexibility was a significantly positive predictor (see Tables 11 and 12). In neither analysis was significant others’ job flexibility a significant predictor. Additionally, the second step of each regression analysis failed to explain significant increment in variance. These results failed to support the idea that the discrepancy between employees’ and significant others’ job flexibility affects the frequency with which their nonwork 77 Table 11 Polynomial Regression Results for Employees’ and Significant Others’ Flexibility in the Timing of Work Predicting Employees’ Nonwork Interruptions of Work 2 β SE B 2 R ΔR .12 B .12* .15 .03 1 Flexibility in timing (employee) Flexibility in timing (s.o.) F(2,131) = 8.98, p < .001 .15** .04 .01 .03 Flexibility in timing (employee) squared Flexibility in timing (s.o.) squared Flexibility in timing (employee) x Flexibility in timing (s.o.) F(5,128) = 4.37, p < .01 -.06 -.01 .00 .33 .03 2 .03 .02 .03 -.16 -.03 .00 Note: Coefficients shown are from the final step of the model. N = 134. ** p < .01. Table 12 Polynomial Regression Results for Employees’ and Significant Others’ Flexibility in the Location of Work Predicting Employees’ Nonwork Interruptions of Work 2 β SE B 2 Flexibility in location (employee) Flexibility in location (s.o.) F(2,127) = 9.04, p < .001 .17* .06 .07 .05 Flexibility in location (employee) squared Flexibility in location (s.o.) squared Flexibility in location (employee) x Flexibility in location (s.o.) F(5,124) = 3.75, p < .01 -.02 -.00 .02 R ΔR .13 B .13* .13 .00 1 .36 .19 2 .04 .03 .04 -.05 -.02 .09 Note: Coefficients shown are from the final step of the model. N = 130. * p < .05. 78 interrupts their work life. Likewise, for the analysis of division of household labor two regressions were performed to analyze each type of flexibility separately. The first step of the regression was only significant for the analysis involving flexibility in the timing of work (see Tables 13 and 14). Flexibility in the timing of work was a significantly positive predictor of the focal employee’s share of household labor. The addition of the quadratic terms in the second step did not explain incremental variance for either analysis. These results indicate that employees with greater flexibility in the timing of their work have a larger share of household responsibilities. These results are inconsistent, however, with the notion that job flexibility relative to the job flexibility of one’s significant other affects division of household labor. Table 13 Polynomial Regression Results for Employees’ and Significant Others’ Flexibility in the Timing of Work Predicting Division of Household Labor 2 β SE B 2 Flexibility in timing (employee) Flexibility in timing (s.o.) F(2,128) = 3.92, p < .05 .04* -.02 .02 .02 Flexibility in timing (employee) squared Flexibility in timing (s.o.) squared Flexibility in timing (employee) x Flexibility in timing (s.o.) F(5,125) = 1.94, p > .05 -.02 -.00 .01 R ΔR .06 B .06* .07 .01 1 .20 -.12 2 .02 .01 .02 -.11 -.01 .04 Note: Coefficients shown are from the final step of the model. N = 131. * p < .05. 79 Table 14 Polynomial regression results for employees’ and significant others’ flexibility in the location of work predicting division of household labor B SE B Flexibility in location (employee) Flexibility in location (s.o.) F(2,124) = 2.41, p > .05 .02 -.02 .04 .02 Flexibility in location (employee) squared Flexibility in location (s.o.) squared Flexibility in location (employee) x Flexibility in location (s.o.) F(5,121) = 1.55, p > .05 -.02 -.02 .00 β 2 R 2 ΔR 1 .09 -.13 .04 .04 .06 .02 2 Note: Coefficients shown are from the final step of the model. 80 .02 .01 .02 -.12 -.11 .01 DISCUSSION The results of the present study can be most easily understood by considering work-tononwork and nonwork-to-work phenomena separately. The first section of the discussion focused on summarizing the main findings is organized accordingly. Work-to-Nonwork There were hypothesized to be challenging aspects of job flexibility, specifically in the form of increased work interruptions of nonwork. This idea did receive some support. As hypothesized, flexibility in the location of work was associated with more frequent self-initiated interruptions of nonwork. That is, individuals who were free to work at their location of choice were more likely to work during their personal time (e.g., on vacation, during family activities). Furthermore, the positive relationship of flexibility in the location of work with these types of interruptions partially obscured its negative relationship with work-to-nonwork conflict. These findings suggest that giving employees freedom in where to work would be more beneficial for their personal lives if that freedom did not lead to work spilling over into their personal lives. Despite the above findings, the majority of results highlight the benefits of job flexibility for employees’ personal lives. Perhaps most importantly, employees with greater job flexibility had lower work-to-nonwork conflict. Additionally, however, the results are suggestive of some unexpected benefits. Both flexibility in the timing and location of work were related to less frequent interruptions by colleagues during personal time. Employees with more flexibility in when they work were also less likely to engage in self-initiated interruptions of their personal time. Potential Explanations for the Unexpected Benefits of Job Flexibility There are a few potential explanations for the negative relationship between job flexibility and other-initiated work interruptions of nonwork. First, it may be that the coworkers 81 of employees who work flexible hours or at home are accustomed to not knowing when those employees are working. Thus, they may be less likely to contact the person for fear of disturbing their personal time. Second, coworkers may make assumptions about the extent to which employees can sacrifice their personal time. Based on the fact that an employee had the need or desire for job flexibility in the first place, coworkers may assume he or she is protective over personal time. This explanation is consistent with some of the observed relationships in the present study – employees with higher job flexibility worked slightly fewer hours, reported lower workloads, and stronger boundary control. However, the results do not support the idea that this is the explicit preference of the employee -- the more flexible an employee’s job, the weaker was his or her preference for segmenting work from nonwork. Another explanation of a different ilk is a perceptual bias. Employees with high job flexibility may be no less likely than their low-flexibility counterparts to experience interruptions from their coworkers during their personal time, but they may perceive fewer interruptions. Though boundary theory stipulates that stronger role integration increases the propensity for interruptions, it also proposes that it decreases their affective impact (Ashforth et al., 2000). Employees with flexible jobs may have an easier time accommodating other-initiated interruptions when they occur because of relaxed spatial or temporal boundaries (e.g., they already have the work materials with them), resulting in less strong reactions. They may also be less likely to view their lives as neatly divided into work time/space and personal time/space and thus such incidents may not be as salient to them. The finding that employees with greater flexibility in the timing of work report fewer self-initiated work interruptions of nonwork may also be due to such a perceptual bias. However, having control over the temporal boundaries of one’s work could also have a meaningful effect 82 wherein one uses that schedule control to plan work around personal time so that it does not interfere. Having a sense of ownership over one’s work time may also encourage employees to use their work time more efficiently, which could minimize spillover and interruptions of personal time. This is not what was hypothesized, but the data are consistent with this explanation. Job flexibility was positively related to perceptions of boundary control. Employees with greater job flexibility felt a stronger sense of control over whether their work and personal lives were kept separate. Nonwork-to-Work The findings for the nonwork-to-work direction were mostly in line with predictions. As expected, employees with more freedom in when and where they work reported more frequent nonwork interruptions of work. Mediation analyses support the idea that higher levels of job permeability are largely responsible for this finding. Recall permeability means that an employee is capable of being physically located in one role but psychologically or behaviorally involved in a different role. Employees with flexible schedules had more permeable jobs, presumably because they have greater decision latitude about the timing of their work and therefore can stop working to take care of personal matters more easily. Likewise, employees with flexibility in work location had more permeable jobs, which is most likely due to an increased practical ability to take care of home-related matters as and less restrictive social norms about doing so. The results suggest that any negative relationship between job flexibility and nonwork-towork conflict may be obscured by the tendency for job flexibility to facilitate nonwork interruptions of work. However, accounting for these interruptions only brought the nonsignificant relationship between job flexibility and nonwork-to-work conflict to modestly negative. The implication of these findings is that job flexibility does not appear to be nearly as 83 helpful in reducing nonwork-to-work conflict as it appears to be for work-to-nonwork conflict. Previous findings by other researchers reveal a similar asymmetry. For example, Shockley and Allen (2007) found that flexibility in the timing and location of work were more strongly negatively related to work-to-family conflict than they were to family-to-work conflict for a sample of women. Similarly, in a meta-analysis by Byron (2005) the negative relationship of schedule flexibility with work-to-family conflict was nearly twice as large as its relationship with family-to-work conflict. Another example of the differential relationship of job flexibility with the two different directions of conflict is provided by Golden, Veiga, and Simsek (2006) who found that greater telecommuting volume was associated with lower work-to-family conflict but higher family-towork conflict. Golden et al. suggested that future research should investigate how altered work patterns mediate the impact of flexible work arrangements on work-family conflict and the present research speaks to this issue. The results of the present study are consistent with the idea that job flexibility indirectly contributes to higher nonwork-to-work conflict through increased nonwork interruptions of work. However, the indirect positive relationship between job flexibility and nonwork-to-work conflict was counteracted by a negative albeit small direct relationship between the two (leading to an overall null relationship). The overall positive relationship between job flexibility and nonwork-to-work conflict found by Golden et al. may be due to the unique nature of their sample—all research participants were telecommuters with an average of more than two days spent telecommuting per week. In contrast, the participants in the present sample worked from home less than half a day per week on average. It may be that job flexibility only leads to more nonwork-to-work conflict (i.e., the costs outweigh the advantages) at more extensive levels of flexibility. The data from the present study do not necessarily support 84 this idea (i.e., curvilinear trends did not fit the data), but there was some range restriction – for example, employees spent a maximum of three days per week telecommuting, which could limit the ability to detect such an effect. Alternatively, the reasons that job flexibility is not as strongly related to nonwork-towork conflict as it is to work-to-nonwork conflict may be less about cost and more about lack of benefit. Job flexibility helps employees resolve nonwork-to-work conflicts (e.g., take care of a sick child, meet the cable company technician), but it cannot necessarily prevent them from arising. Employees who have job flexibility may be better able to meet their personal needs, but this does not guarantee that their personal needs will not arise at inopportune times and that their work will not be affected. Even if they flex their work role to accommodate personal needs, they may miss out on a meeting, have to delay a project, or return to work tired or preoccupied with personal concerns. The small body of literature on decision making and attributions made in the context of work-family conflict also provides insight into the seemingly weak relationship between job flexibility and nonwork-to-work conflict. Researchers have found that irrespective of work and family centrality employees are more likely to blame the work role than the family role for their work-family conflicts (Poposki, 2011). Thus, when employees are faced with a conflict between work and nonwork they will experience one of two likely outcomes. They may lack the job flexibility required to tend to their personal need, in which case they are likely to blame work and perceive work as interfering with nonwork. Or, they may have sufficient job flexibility to accomodate the personal need, in which case they will not blame work and not perceive work interfering with nonwork. According to this construal of events, job flexibility can only benefit the work-to-nonwork direction of interference. The person may still recognize the event as an 85 incident of nonwork-to-work conflict, but this perception is unlikely to be affected by job flexibility. Questioning Assumptions: Are Interruptions Undesirable? The purpose of this dissertation was to move away from black-and-white thinking about job flexibility (i.e., is it helpful?) toward a more nuanced consideration of its benefits and costs. Work-nonwork interruptions were recognized as a potential drawback of job flexibility, because they represent distractions that impede an individual’s ability to be fully immersed in a role (Ashforth et al., 2000). However, it is important to not lose sight of the overall picture of an individual’s well-being, which depends on both directions of interruptions and how the person feels about the interruptions. Work interruptions of nonwork may not be so detrimental to personal life if it is also possible to interrupt work with nonwork concerns. A study by Kossek and colleagues (2012) identified different clusters of boundary management styles based on levels of work interruptions of nonwork, nonwork interruptions of work, work and family centrality, and perceived boundary control. Interestingly, the only group with high levels of work interrupting nonwork that had relatively positive outcomes (e.g., lower psychological distress, turnover intentions) also had high levels of nonwork interrupting work. The cluster of individuals who had high work interruptions of nonwork but low nonwork interruptions of work had relatively negative outcomes. Allard, Haass, and Hwang (2007, p. 480) suggested that “managers working for organizations with boundless flexible time cultures offering flextime will be better able to combine work and family if borders at home also are weak (flexible and permeable).” This sort of “give and take” allows an individual to meet both work and personal needs. 86 There is some evidence from the present study to suggest that nonwork interruptions of work are not as problematic as work interruptions of nonwork. First, the former seem to occur less frequently than the latter. The mean for nonwork interruptions of work corresponded to slightly above “rarely” on the response scale, whereas the means for both types of work interruptions of nonwork were closer to “sometimes”. Second, it appears that nonwork interruptions of work are not as undesirable to employees as are work interruptions of nonwork. Employees whose work lives interrupted their personal lives were more likely to feel a lack of control over their boundaries. Work interruptions of nonwork was negatively correlated with boundary control (rs = -.41 and -.50 for self- and other-initiated, respectively) whereas nonwork interruptions of work was positively albeit modestly correlated with boundary control (r = .14). Kossek et al. (2012) observed a similar pattern of relationships between these variables and furthermore, found boundary control is a significantly negative predictor of psychological distress and turnover intentions. Thus, work interruptions of nonwork would seem to put employees more at risk of burning out in their jobs. Kossek et al. identified a group of individuals with high nonwork interruptions of nonwork and low work interruptions of nonwork that had positive outcomes, suggesting that an asymmetry in favor of nonwork may be more conducive to overall well-being than the reverse. Flexibility in the Timing versus Location of Work Shockley and Allen (2007) found that flextime, but not flexplace, was negatively associated with work interference with family, presumably because flexplace makes maintaining boundaries difficult. Some scholars have regarded flextime as a relatively segmenting policy as discussed previously (Rau & Hyland, 2002; Rothbard et al., 2005). The findings from this study corroborate that view at least in the work-to-nonwork direction. It was hypothesized in the 87 present study that flexibility in work location would be more strongly positively related than flexibility in work timing to work interruptions of nonwork. Support for this prediction was demonstrated in that flexibility in the timing of work was negatively related to both forms of work to nonwork interruptions, whereas flexibility in the location of work was modestly positively related to self-initiated work interruptions of nonwork. Indeed, the results demonstrate that if it were not for the more frequent self-initiated interruptions of nonwork, flexibility in the location of work and work-to-nonwork conflict would be more strongly negatively related. Overall, the results support Shockley and Allen’s (2007) contention that flexibility in the timing of work is more efficacious than flexibility in the location of work. Not only was flexibility in the timing of work more strongly negatively related to work-to-nonwork conflict than was flexibility in the location of work (zero order correlations of -.40 versus -.26, respectively, Steiger’s Z = 5.87, p < .001), but only flexibility in the timing of work was significantly correlated with significant other reports of work-to-nonwork conflict suggesting that it is a more robust relationship. The two types of flexibility appear equally inefficacious with respect to the nonwork to work direction of interference. Limitations A common shortcoming in work-life research is overreliance on self-report data (Casper, Eby, Bordeaux, Lockwood, & Lambert, 2007). One strength of the present study is that other reports were collected and used to corroborate findings. The results for the work-to-nonwork direction of interruptions and conflict were consistent whether self or other reports were used. However, the significant other reports of the nonwork-to-work direction of conflict and interruptions failed to display strong correlations with self-reported data. This finding could be due to employees failing to provide honest self reports because they want to respond in a socially 88 desirable manner, employees and their significant others focusing on different behaviors, or family members’ limited opportunity to observe the nonwork-to-work phenomena unfold. Future studies could utilize coworker reports of nonwork interruptions of work, which may be more reliable since coworkers have the opportunity to observe these types of interruptions happening. Nevertheless, the fact that some of the findings in this study were replicated using other reports suggests that common method variance is an unlikely explanation for the results. The cross-sectional design of the present study precludes causal inferences—one can only say that the data are consistent with job flexibility reducing work-to-nonwork conflict, for example. Though a true experiment is not feasible, a quasi-experimental study in which measurements are obtained before and after implementation of a flexible work arrangement may be informative in this regard. The present study did highlight one issue that could be researched in a laboratory—given an incident of work-nonwork conflict, how does the ability to interrupt one role to accommodate the other make people feel (i.e., perceptions of conflict, satisfaction with ability to meet role demands, etc.)? The present study suggests that it increases the perception of conflict from the interrupting role but in the case of nonwork interrupting work may decrease perceptions of conflict from the role being interrupted (i.e., because work is not interfering with one’s ability to take care of personal demands when they arise). These questions could be addressed in an experimental design using vignettes. An additional limitation worth mentioning concerns the sample characteristics. Respondents worked in white collar occupations in a corporate setting. The results are unlikely to generalize to jobs that require shift work or to work environments in which work-nonwork interruptions are impractical. Also the sample was predominantly male, a reflection of the 89 workforce population sampled. Though gender was not strongly correlated with any of the study variables it is possible that the lesser representation of women could have influenced the results. Future Research Directions The results of the present study suggest that giving employees autonomy over their work hours and location can alleviate the impact of a demanding job on their personal lives but may not necessarily prevent (and may exacerbate) personal matters interfering with their work. Job flexibility is a resource organizations can offer to help deal with nonwork-to-work conflict when it occurs (Behson, 2002), but it is disconnected from the actual cause of the conflict. Byron’s (2005) meta-analysis indicates that the antecedents for work-to-family conflict and family-towork conflict differ, such that the strongest predictors of work-to-family conflict reside in the work domain whereas the strongest predictors of family-to-work conflict reside in the family domain. Job stress and schedule flexibility were the predictors most strongly correlated with work-to-family conflict, whereas family stress was most strongly correlated with family-to-work conflict. More direct means of prevention may be through higher quality childcare and more spousal support (Aycan & Eskin, 2005) yet relatively little research in the organizational literature has focused on these factors. The reality may be that the prevention of nonwork-towork conflict is outside the control of employers. Future research should identify the key levers for change in the family environment. Organizations are likely to take different stances on their role in employees’ lives with some willing to be more involved in affecting change than others. Furthermore, there is a delicate balance with some employees viewing any intervention from the organization as an invasion of their privacy (Hall & Richter, 1989). On the one hand, organizations have to decide how far they are willing to go in their support for employees’ 90 personal lives and on the other hand, they must gauge employees’ reactions to work-life policies before implementing them to ensure they are received positively. More research is needed to determine the extent to which using job flexibility to accommodate personal needs negatively impacts job performance. Such an assessment is important to make if one is to build a stronger business case for job flexibility and alleviate management concerns. The results of the present study suggest that the more frequently employees experience nonwork interruptions of work, the more likely they are to feel as though their nonwork life conflicts with their work life (e.g., difficulty getting work done, work suffering). Future research could investigate whether this finding extends to supervisor reports of job performance and the magnitude of these effects. As mentioned previously, nonwork to work interruptions were relatively rare, on average, in which case the overall effect may be negligible. Some of the results in the present study were unexpected, such as the negative relationship between job flexibility and other-initiated work interruptions of nonwork. This finding is inconsistent with boundary theory, which stipulates that flexible boundaries uniformly encourage greater integration between roles. Ashforth et al. proposed that loose boundaries create confusion among individuals and members of their role sets about which role should be more salient at any given time. Though this finding needs to be replicated, it emphasizes the importance of distinguishing between self- and other-initiated interruptions. This distinction was made by Hall and Richter (1989) but has since been lost in more current formulations of boundary theory (e.g., Ashforth et al., 2000). Additional research is clearly needed to better understand how employees’ coworkers form perceptions of their availability. Practical Implications 91 Job flexibility is not a panacea. There are tradeoffs that employees and organizations must consider when establishing the strength of boundaries around the work role. As we improve our understanding of how boundaries between work and nonwork affect individuals, organizations can play the important role of educating their employees about the various tradeoffs of job flexibility. Workshops may aid employees in making more thoughtful decisions about how to combine their work and personal life, as well as provide an opportunity for brainstorming ways to minimize the downsides and maximize the benefits of job flexibility. Importantly, both forms of job flexibility were positively associated with boundary control suggesting that giving employees more autonomy over the timing and location of their work has positive psychological effects overall. Most of the “boundary work” tactics that have been proposed in the literature as a way for employees to manage their work-life balance involve some degree of autonomy over when and where individuals work. For example, Kreiner, Hollensbe, and Sheep (2009) found that individuals use time banking (e.g., working extra work hours to be able to take time off from work later) and selectivity about where certain types of work (e.g., creative versus busywork) are performed. Employees and their managers have a joint responsibility to set expectations about the acceptable types and frequency of interruptions between work and nonwork, and how these should be dealt with when they arise. An understanding is often developed indirectly by managers interpreting an employee’s reaction to being contacted at home or asked to work overtime, and by employees interpreting manager’s reactions to their personal conflicts that arise with work. Kreiner et al. (2009) identified incidents of boundary violations as opportunities for active dialogue. In addition, organizations can provide diagnostic tools for employees to decide if and when interruptions between work and nonwork become problematic, and managers should 92 check in with employees periodically to discuss what is and is not working about any flexible working arrangements they may use. 93 APPENDICES 94 APPENDIX A Focal Employee Informed Consent and Survey 95 Welcome! You are invited to participate in a study of work-life balance being conducted by researchers at Michigan State University. All data collected from this survey will be used for research purposes only. This study is being conducted by: Jessica Keeney, Doctoral Candidate, Department of Psychology, Michigan State University Ann Marie Ryan, Professor, Department of Psychology, Michigan State University Procedure: The purpose of this study is to learn more about work-life balance. If you agree to be in this study, you will be asked to do the following: 1. Complete a confidential web-based survey about your work and personal responsibilities, the ways in which you manage the work-nonwork boundary, and your work-life balance. 2. Provide contact information for a significant other with whom you live (e.g., spouse or close family member) so that we may invite him/her to complete a survey about how your work and personal life interact from his/her perspective. Participation by significant others is entirely voluntary. The reasons we are seeking this data are to enhance the reliability of our results (asking the same questions from multiple points of view helps us do this), and to research interesting issues such as whether having a significant other with a more flexible job improves one’s work-life balance. Conditions for Eligibility: In order to participate, you must be: (1) a manager lLL6 and above, (2) U.S.-based, (3) employed full-time (i.e., 35 hours or more per week). Time Requirements: We estimate that the survey will take about 15 minutes to complete. Confidentiality: At no point will identifiers (e.g., name, e-mail, or IP address) be linked to any of your survey responses. No identifying information you provide will be disclosed to anyone other than the trained researchers (Dr. Ryan and Jessica Keeney). Furthermore, in any sort of report we might publish, it will not be possible to identify you as a subject in this research. The ratings you and your significant other provide will be combined with those of other participants. Your privacy will be protected to the maximum extent allowable by law. All survey responses that we receive will be treated confidentially and stored on a secure server. However, given that the survey can be completed from any computer (e.g., personal, work, school), we are unable to guarantee the security of the computer on which you choose to enter your responses. Risks, Benefits, & Incentives: There are no foreseeable physical, psychological, or economical risks associated with this research. By participating in this study, you will be helping to further knowledge about work-life balance. In addition, you will gain access to an online resource with research- based tips for work-life balance. There is no financial compensation for participating. Voluntary Nature of the Study: Your participation in this study is strictly voluntary and may be 96 discontinued at anytime without penalty. You are free to decline to answer any question. If you have any questions or concerns about the study, contact Jessica Keeney, 302 Psychology Building, Michigan State University, East Lansing, MI 48824 (jkeeney@msu.edu). Consent to Participate: By continuing, you agree that you have read the above description and voluntarily agree to participate in this study. 97 The first set of questions will help us understand your work and personal life responsibilities. [Workload] My job requires me to work very fast. My job requires me to work very hard. My job leaves me with little time to get things done. There is a great deal of work to be done in my job. 2. On average, how many hours per week do you spend on the following nonwork responsibilities? [Family responsibility] Housework, including cooking, cleaning, laundry, picking up dry cleaning, shopping, etc., or arranging for any of these types of tasks to be done by others. Child care and/or parent care, including transportation of children and/or parents or arranging child and/or parent care. Household maintenance, including lawn care, gardening, household repairs and improvements, painting, remodeling, fixing appliances, fixing the car, etc., or arranging for these types of tasks to be done by others. Support of spouse or partner, including providing emotional support, attending his/her work or social functions, entertaining his/her friends or coworkers. 98 Very Often Often Sometimes Occasionally Rarely 1. Please answer how frequently the following occurs in your job. [Work-to-nonwork conflict] I have to miss out on important personal activities due to the amount of time I spend doing work. I neglect my personal needs because of the demands of my work. My personal life suffers because of my work. I come home from work too tired to do things I would like to do. My job makes it difficult to maintain the kind of personal life I would like. [Nonwork-to-work conflict] My personal life drains me of the energy I need to do my job. My work suffers because of everything going on in my personal life. I am too tired to be effective at work because of things I have going on in my personal life. When I am at work, I worry about things I need to do outside of work. I have difficulty getting my work done because I am preoccupied with personal matters. I devote less time to work because of everything I have going on in my personal life. 99 Almost All the Time Often Sometimes Occasionally Rarely 3. The next set of questions concerns how your work and personal life influence one another. Please indicate how frequently the following occurs. There are no right or wrong answers. You should answer the questions as honestly as possible. [Flexibility in the timing of work] I am free to work the hours that are best for my schedule. I have control over when I work. I am able to arrive and depart from work when I want. I have choice in determining when I begin and end each workday. [Flexibility in the location of work] My job permits me to decide on my own where the work is done. I have the freedom to work wherever is best for me — either at home or at work. I can work from almost anywhere. I have flexibility in selecting the location of where I work. [Job permeability] My employer allows me to carry out non-work projects during spare time at work. While at work, I can stop what I am doing to meet responsibilities related to my family and personal life. I could conduct personal errands during business hours if necessary. 100 Very Great Extent Considerable Extent Moderate Extent Limited Extent Not at All 4. These questions deal with different aspects of work. Please indicate to what extent these aspects appear in your job. 5. Approximately how many hours per week do you typically spend working at home during the evening or on weekends? 6. Do you have a formal arrangement set up with your employer (i.e., following company policy) to work from home one or more days per week? Yes No 7. In a typical work week (Monday through Friday during standard business hours), how many days do you work from home? Please report in increments of .5 days if you work half days at home. 8. Which of the following describe your work schedule? Check all that apply.  You establish your starting and ending times but maintain the same schedule each day.  You must work during specified core hours but may adjust starting and ending times each day.  You may take a longer than scheduled break if you make up the time by starting work earlier or working later.  You must work a specified number of hours, but can choose whichever hours you want. 101 [Nonwork interruptions of work] I stop what I am working on because a family member or friend requires my attention. I allow family/friends to interrupt me when I am working. My work is brought to a halt due to unexpected contact (e.g., emails, texts, and phone calls) from a family member or friend. I take care of personal or family needs during work. My family members or friends make unexpected requests of me while I am working. [Self-initiated work interruptions of nonwork] I take care of work issues during my personal time away from work. I bring work home. I work during my vacations. I bring work materials with me when I attend personal or family activities. [Other-initiated work interruptions of nonwork] My personal life is interrupted by unexpected contact (e.g., emails, texts, and phone calls) from a colleague. I receive calls from co-workers or my supervisor during the evening or weekend. I stop whatever personal activity I am engaged in because a colleague requires my attention. My colleagues make unexpected requests of me during my leisure time. 102 Very Often Often Sometimes Rarely Never 9. People have different styles of handling their work and nonwork responsibilities. Please remember that there are no right or wrong answers, and you should answer the following questions as honestly as possible. How often does the following typically occur? [Preference for segmenting nonwork from work] I don’t like to think about non­work issues while I am at work. I prefer to keep non-work life at home. I don’t like non­work issues creeping into my work life. I like being able to leave non-work issues behind when I go to work. [Preference for segmenting work from nonwork] I don't like to think about work issues while I am at home. I prefer to keep work life at work. I don't like work issues creeping into my home life. I like being able to leave work issues behind when I go home. [Boundary control] I control whether I am able to keep my work and personal life separate. I control whether I have clear boundaries between my work and personal life. I control whether I combine my work and personal life activities throughout the day. 103 Strongly Agree Agree Neither Agree Nor Disagree Disagree Strongly Disagree 10. The following questions ask you about your preferences and beliefs regarding the interaction between your work and personal life. Please indicate how strongly you agree or disagree with each statement. Answers to the following questions will be used to help describe the people who participated in this survey. Any research publications that result from this study will only report this information in the aggregate. 11. What is your job level?     Individual Contributor (IC) Manager (supervise ICs only) Manager of other managers Director/Executive and above 12. How many hours per week do you spend in paid employment? Please provide a single number, e.g., 35. If your hours vary, give your best estimate of the average. 13. On average, how frequently must you schedule a meeting outside of regular business hours (e.g., early morning or evening) because of time zone differences?       Never Less than once per month 1 to 3 days per month 1 to 2 days per week 4 days per week 5 or more days per week 14. What is your marital status?     Married or in a domestic partnership Single, never married Divorced Widowed 15. Does your spouse/partner work?     Yes, full time Yes, part time No N/A (no spouse/partner) 104 16. How many children are living at home with you?      0 1 2 3 4+ 17. What is the age of the youngest child living at home with you (in years)? For children under one year, please write “infant”. 18. For how many people do you provide eldercare (e.g., your or your partner’s parents)?      0 1 2 3 4+ 19. What is your sex?  Male  Female 20. What is your age?     65 & older 47-64 yrs old 34-46 yrs old 33 or younger You are almost finished! 105 Significant Other Participation We would appreciate you allowing us to contact your significant other (e.g., spouse or other close family member with whom you live) to invite him/her to complete a brief questionnaire. WE PROTECT YOUR ANONYMITY IN TWO WAYS: (1) Your significant other’s contact information will be kept in a different computer file than the baseline questionnaire that you filled out. You will be re-directed to a completely new survey to provide his/her e­mail. (2) We will link your responses and your significant other’s responses together using a unique identifier that you will create below. The reasons we are seeking his/her responses are to enhance the reliability of our results (asking the same questions from multiple points of view helps us do this), and to research interesting issues such as whether having a significant other with a more flexible job improves one’s work­life balance. Because the survey is anonymous, we need a unique way to link your survey responses with the responses of your significant other. We have devised a code system for this. Please create a 6-digit code below consisting of the first two letters of the street you and your significant other live on, your 2­digit birthday month and your significant other’s 2-digit birthday month. Your significant other will be asked to create the same code. For example, if you live on Sixth Street, were born in July, and your significant other was born in March, your code would be si0703. 24. Enter code here: In order to contact your significant other, we will need his/her e-mail address. In order to protect the anonymity of the responses you have already provided, you will now be re-directed to a new survey to enter this information. Any information you enter hereafter will not be linked to your previous responses. Please click ‘continue’ below. 106 APPENDIX B Significant Other Informed Consent and Survey 107 The person that is asking you to rate her/him is participating in a research study of work-life balance. This form is intended to give you a short description of the study, to indicate how your ratings will be used, and to ask for your consent to use your ratings in this research project. All data collected from this survey will be used for research purposes only. This study is being conducted by: Jessica Keeney, Doctoral Candidate, Department of Psychology, Michigan State University Ann Marie Ryan, Professor, Department of Psychology, Michigan State University Background Information: The purpose of the study is to learn more about work-life balance. The aim of this research is to help people better balance their lives and organizations develop better work-life policies. If you agree to be in this study, you will be asked to do the following: 1. Answer questions about your significant other’s work and personal life responsibilities and the way he/she manages the work-nonwork boundary. 2. Answer questions about your own work and personal life responsibilities. Time Requirements: We estimate that the survey will take 15 minutes or less to complete. Risks, Benefits, & Incentives: There are no foreseeable physical, psychological, or economical risks associated with this research. To thank you for your participation, you will receive access to an online resource with research-based tips for managing work-life balance. In addition, you will be entered into a drawing for a $100 gift certificate to Amazon.com. Odds of winning depend on the total number of survey responses received. Confidentiality: Your responses are anonymous. At no point in time will personal identifiers (e.g., email address) be linked to your survey responses. In any sort of report we might publish, it will not be possible to identify you as a subject in this research. The ratings you and your significant other provide will be combined with those of other participants. Your privacy will be protected to the maximum extent allowable by law. All survey responses that we receive will be treated confidentially and stored on a secure server. However, given that the survey can be completed from any computer (e.g., personal, work, school), we are unable to guarantee the security of the computer on which you choose to enter your responses. Voluntary Nature of the Study: Your participation in this study is strictly voluntary and may be discontinued at any time without penalty. You are free to decline to answer any question. If you have any questions or concerns about the study, contact Jessica Keeney, 302 Psychology Building, Michigan State University, East Lansing, MI 48824 (jkeeney@msu.edu). Consent to Participate: By continuing, you agree that you have read the above description and voluntarily agree to participate in this study. Your significant other participated in a survey about work-life balance and provided your email address so that we could ask you some questions about how you think they balance work and personal life, and how you do it. 108 Because your responses to this survey are anonymous (your email address is not linked), we need to use a unique coded created by you and your significant other to link together your responses. How the code works is that you both should be able to come up with the same answers to the following questions. 1. What are the first two letters of the street you live on? NOTE: If the street is actually a number, please spell out the first two letters. For example, "si" for 6th St., "te" for 10th Ct., "tw" for 29th Blvd., "on" for 127th Ave., etc. 2. What is your significant other’s birthday month? Please enter in the form of two numbers (ex. “03” for March). 3. What is your birthday month? Please enter in the form of two numbers (ex. “03” for March). Thank you! We will create a 6-digit code using the answers provided, which will allow us to link up your responses with those of your significant other, while protecting your anonymity. Please proceed to the next page to begin the questionnaire. 109 The first part of the questionnaire deals with different aspects of your significant other’s work. [Flexibility in the timing of work] S/he is free to work the hours that are best for his/her schedule. S/he has control over when s/he works. S/he is able to arrive and depart from work when s/he wants. S/he has choice in determining when s/he begins and ends each workday. [Flexibility in the location of work] His/her job permits him/her to decide on his/her own where the work is done. S/he has the freedom to work wherever is best for him/her— either at home or at work. S/he can work from almost anywhere. S/he has flexibility in selecting the location of where s/he works. [Job permeability] His/her employer allows him/her to carry out nonwork projects during spare time at work. While at work, s/he can stop what s/he is doing to meet responsibilities related to family and personal life. S/he could conduct personal errands during business hours if necessary. 110 Very Great Extent Considerable Extent Moderate Extent Limited Extent Not at All 4. Please indicate to what extent these aspects appear in your significant other's job. [Nonwork interruptions of work] S/he stops what s/he is working on because a family member or friend requires his/her attention. S/he allows family/friends to interrupt him/her when s/he is working. His/her work is brought to a halt due to unexpected contact (e.g., emails, texts, and phone calls) from a family member or friend. S/he takes care of personal or family needs during work. His/her family members or friends make unexpected requests of him/her when s/he is working. [Self-initiated work interruptions of nonwork] S/he takes care of work issues during personal time away from work. S/he brings work home. S/he works during vacations. S/he brings work materials with him/her when s/he attends personal or family activities. [Other-initiated work interruptions of nonwork] His/her personal life is interrupted by unexpected contact (e.g., emails, texts, and phone calls) from a colleague. S/he receives calls from coworkers or his/her supervisor during the evening or weekend. S/he stops whatever personal activity s/he is engaged in because a colleague requires his/her attention. His/her colleagues make unexpected requests of him/her during his/her leisure time. 111 Very Often Often Sometimes Rarely Never 5. People have different styles of handling their work and nonwork responsibilities. Please remember that there are no right or wrong answers, and you should answer the following questions as honestly as possible. How often does the following typically occur for your significant other? [Work-to-nonwork conflict] S/he has to miss out on important personal activities due to the amount of time s/he spends doing work. S/he often neglects his/her personal needs because of the demands of his/her work. His/her personal life suffers because of his/her work. S/he comes home from work too tired to do things s/he would like to do. His/her job makes it difficult to maintain the kind of personal life s/he would like. [Nonwork-to-work conflict] His/her personal life drains him/her of the energy needed to do his/her job. His/her work suffers because of everything going on in his/her personal life. S/he is too tired to be effective at work because of things s/he has going on in his/her personal life. When s/he is at work, s/he worries about things s/he needs to do outside of work. S/he has difficulty getting his/her work done because s/he is preoccupied with personal matters. S/he devotes less time to work because of everything s/he has going on in his/her personal life. 112 Almost All the Time Often Sometimes Occasionally Rarely 6. The next set of questions concerns how your significant other’s work and personal life influence one another. Please indicate how frequently the following occurs. There are no right or wrong answers. You should answer the questions as honestly as possible. Now we would like to ask you a few questions about yourself. 7. On average, how many hours per week do you spend on the following nonwork responsibilities? [Family responsibility] Housework, including cooking, cleaning, laundry, picking up dry cleaning, shopping, etc., or arranging for any of these types of tasks to be done by others. Child care and/or parent care, including transportation of children and/or parents or arranging child and/or parent care. Household maintenance, including lawn care, gardening, household repairs and improvements, painting, remodeling, fixing appliances, fixing the car, etc., or arranging for these types of tasks to be done by others. Support of spouse or partner, including providing emotional support, attending his/her work or social functions, entertaining his/her friends or coworkers. 8. Are you currently employed?     Yes, full time Yes, part time No These questions deal with different aspects of work. Please indicate to what extent these aspects appear in your job. 113 [Flexibility in the timing of work] I am free to work the hours that are best for my schedule. I have control over when I work. I am able to arrive and depart from work when I want. I have choice in determining when I begin and end each workday. [Flexibility in the location of work] My job permits me to decide on my own where the work is done. I have the freedom to work wherever is best for me — either at home or at work. I can work from almost anywhere. I have flexibility in selecting the location of where I work. [Job permeability] My employer allows me to carry out non-work projects during spare time at work. While at work, I can stop what I am doing to meet responsibilities related to my family and personal life. I could conduct personal errands during business hours if necessary. 114 Very Great Extent Considerable Extent Moderate Extent Limited Extent Not at All 9. These questions deal with different aspects of work. Please indicate to what extent these aspects appear in your job. [Workload] My job requires me to work very fast. My job requires me to work very hard. My job leaves me with little time to get things done. There is a great deal of work to be done in my job. 11. Approximately how many hours per week do you typically spend working at home during the evening or on weekends? 12. Do you have a formal arrangement set up with your employer (i.e., following company policy) to work from home one or more days per week? Yes No 13. In a typical work week (Monday through Friday during standard business hours), how many days do you work from home? Please report in increments of .5 days if you work half days at home. 14. Which of the following describe your work schedule? Check all that apply.  You establish your starting and ending times but maintain the same schedule each day.  You must work during specified core hours but may adjust starting and ending times each day.  You may take a longer than scheduled break if you make up the time by starting work earlier or working later.  You must work a specified number of hours, but can choose whichever hours you want. 115 Very Often Often Sometimes Occasionally Rarely 10. Please answer how frequently the following occurs in your job. [Work-to-nonwork conflict] I have to miss out on important personal activities due to the amount of time I spend doing work. I neglect my personal needs because of the demands of my work. My personal life suffers because of my work. I come home from work too tired to do things I would like to do. My job makes it difficult to maintain the kind of personal life I would like. [Nonwork-to-work conflict] My personal life drains me of the energy I need to do my job. My work suffers because of everything going on in my personal life. I am too tired to be effective at work because of things I have going on in my personal life. When I am at work, I worry about things I need to do outside of work. I have difficulty getting my work done because I am preoccupied with personal matters. I devote less time to work because of everything I have going on in my personal life. 116 Almost All the Time Often Sometimes Occasionally Rarely 15. The next set of questions concerns how your work and personal life influence one another. Please indicate how frequently the following occurs. There are no right or wrong answers. You should answer the questions as honestly as possible. [Preference for segmenting nonwork from work] I don’t like to think about non­work issues while I am at work. I prefer to keep non-work life at home. I don’t like non­work issues creeping into my work life. I like being able to leave non-work issues behind when I go to work. [Preference for segmenting work from nonwork] I don't like to think about work issues while I am at home. I prefer to keep work life at work. I don't like work issues creeping into my home life. I like being able to leave work issues behind when I go home. [Boundary control] I control whether I am able to keep my work and personal life separate. I control whether I have clear boundaries between my work and personal life. I control whether I combine my work and personal life activities throughout the day. 117 Strongly Agree Agree Neither Agree Nor Disagree Disagree Strongly Disagree 16. The following questions ask you about your preferences and beliefs regarding the interaction between your work and personal life. Please indicate how strongly you agree or disagree with each statement. 17. What is your sex?  Male  Female 18. How many children are living at home with you?      0 1 2 3 4+ 19. For how many people do you provide eldercare (e.g., your or your partner’s parents)?      0 1 2 3 4+ 20. What is your age?     65 & older 47-64 yrs old 34-46 yrs old 33 or younger 118 REFERENCES 119 REFERENCES Ahrentzen, S. B. (1990). Managing conflict by managing boundaries: How professional homeworkers cope with multiple role at home. Environment and Behavior, 22, 723-752. Allard, K., Haas, L., & Hwang, P. (2007). Exploring the paradox: Experiences of flexible working arrangements and work-family conflict among managerial fathers in Sweden. Community, Work and Family, 10, 475-493. Allen, T. D. (2001). Family-supportive work environments: The role of organizational perceptions. Journal of Vocational Behavior, 58, 414-435. Allen, T. D., Herst, D. E. L., Bruck, C. S., & Sutton, M. (2000). Consequences associated with work-to-family conflict: A review and agenda for future research. Journal of Occupational Health Psychology, 5, 278-308. Ammons, S. K. (2009). Boundaries at work: A study of work-family boundary stability within a large organization. (Unpublished doctoral dissertation). University of Minnesota, Minneapolis. Ammons, S. K., & Markham, W. T. (2004). Working at home: Experiences of skilled white collar workers. Sociological Spectrum, 24, 191-238. Ashforth, B. E., Kreiner, G. E., & Fugate, M. (2000). All in a day’s work: Boundaries and micro role transitions. Academy of Management Review, 25, 472-491. Aycan, Z., & Eskin, M. (2005). Relative contributions of childcare, spousal support, and organizational support in reducing work-family conflict for men and women: The case of Turkey. Sex Roles, 53, 453-471. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182. Batt, R., & Valcour, P. M. (2003). Human resource practices as predictors of work-family outcomes and employee turnover. Industrial Relations, 42, 189-220. Behson, S. J. (2002). Coping with family-to-work conflict: The role of informal work accommodations to family. Journal of Occupational Health Psychology, 7, 324-341. Berman, L. M. (1997). When flexibility "works" and when it "fails": An in-depth analysis of alternatives to the nine-to-five work week. Unpublished dissertation. Bulger, C. A., Hoffman, M. E., & Matthews, R. A. (2011). Predicting health symptoms: Understanding domain transitions, segmentation preferences, and work/personal life interference. Manuscript submitted for publication. 120 Bulger, C. A., Matthews, R. A., & Hoffman, M. E. (2007). Work and personal life boundary management: Boundary strength, work/personal life balance, and the segmentationintegration continuum. Journal of Occupational Health Psychology, 12, 365-375. Byron, K. (2005). A meta-analytic review of work-family conflict and its antecedents. Journal of Vocational Behavior, 67, 169-198. Caplan, R. D., Cobb, S., French, J. R. P., Jr., Harrison, R. V., & Pinneau, S. R., Jr. (1975). Job demands and worker health: Main effects and occupational differences. Washington, DC: Department of Health, Education, and Welfare. Cardenas, R. A., Major, D. A., & Bernas, K. H. (2004). Exploring work and family distractions: Antecedents and Outcomes. International Journal of Stress Management, 11, 346-365. Carlson, D. S., Kacmar, K. M., & Williams, L. J. (2000). Construction and initial validation of a multidimensional measure of work-family conflict. Journal of Vocational Behavior, 56, 249276. Casper, W. J., Eby, L. T., Bordeaux, C., & Lambert, D. (2007). A review of research methods in IO/OB work-family research. Journal of Applied Psychology, 92, 28-43. Chen, Z., Powell, G. N., & Greenhaus, J. H. (2009). Work-to-family conflict, positive spillover, and boundary management: A person-environment fit approach. Journal of Vocational Behavior, 74, 82-93. Christensen, K. E., & Staines, G. L. (1990). Flextime: A viable solution to work/family conflict? Journal of Family Issues, 11, 455-476. Clark, S. C. (2000). Work/family border theory: A new theory of work/family balance. Human Relations, 53, 747-770. Clark, S. C. (2002). Communicating across the work/home border. Community, Work & Family, 5, 23-48. D’Abate, C. P. (2005). Working hard or hardly working: A study of individuals engaging in personal business on the job. Human Relations, 58, 1009-1032. Desrochers, S., Hilton, J. M., & Larwood, L. (2005). Preliminary validation of the work-family integration-blurring scale. Journal of Family Issues, 26, 442-466. Edwards, J. R. (1994). The study of congruence in organizational behavior research: Critique and proposed alternative. Organizational Behavior and Human Decision Processes, 58, 51100. Edwards, J. R. (2002). Alternatives to difference scores: Polynomial regression analysis and response surface methodology. In F. Drasgow & N. W. Schmitt (Eds.), Advances in measurement and data analysis (pp. 350-400). San Francisco: Jossey-Bass. 121 Edwards, J. R., & Rothbard, N. P. (1999). Work and family stress and well-being: An examination of person-environment fit in the work and family domains. Organizational Behavior and Human Decision Processes, 77, 85-129. Edwards, J. R., & Rothbard, N. P. (2000). Mechanisms linking work and family: Clarifying the relationships between work and family constructs. The Academy of Management Review, 25, 178-199. Fenner, G. H., & Renn, R. W. (2010). Technology-assisted supplemental work and work-tofamily conflict: The role of instrumentality beliefs, organizational expectations, and time management. Human Relations, 63, 63-82. Fisher, G. G., Bulger, C. A., & Smith, C. S. (2009). Beyond work and family: A measure of work/nonwork interference and enhancement. Journal of Occupational Health Psychology, 14, 441-456. Friedman, S. D., & Greenhaus, J. H. (2000). Work and family—allies or enemies? What happens when business professionals confront life choices. New York: Oxford University Press. Ganster, D. C., & Fusilier, M. R. (1989). In C.L. Cooper & I.T. Robertson (Eds.), International Review of Industrial and Organizational Psychology (p. 235-280). Oxford: John Wiley & Sons. Golden, T. D., Veiga, J. F., & Simsek, Z. (2006). Telecommuting’s differential impact on workfamily conflict: Is there no place like home? Journal of Applied Psychology, 91, 13401350. Grandey, A. A., Cordeiro, B., & Crouter, A. (2005). A longitudinal and multi-source test of thework-family conflict and job satisfaction relationship. Journal of Occupational and Organizational Psychology, 78, 305-323. Grandey, A. A., & Cropanzano, R. (1999). The conservation of resources model applied to workfamily conflict and strain. Journal of Vocational Behavior, 54, 350-370. Green, S. G., Bull Schaefer, R. A., MacDermid, S. M., Weiss, H. M. (2011). Partner reactions to work-to-family conflict: Cognitive appraisal and indirect crossover in couples. Journal of Management, 37, 744-769. Greenhaus, J. H., & Beutell, N. J. (1985). Sources of conflict between work and family roles. The Academy of Management Review, 10, 76-88. Greenhaus, J. H., Parasuraman, S., Granrose, C. S., Rabinowitz, S., & Beutell, N. J. (1989). Sources of work-family conflict among two-career couples. Journal of Vocational Behavior, 34, 133-153. Hall, D. T., & Richter, J. (1988). Balancing work life and home life: What can organizations do to help? Academy of Management Executive, 11, 213-223. 122 Hammer, L. B., Neal, M. B., Newsom, J. T., Brockwood, K. J., & Colton, C. L. (2005). A longitudinal study of the effects of dual-earner couples’ utilization of family-friendly workplace supports on work and family outcomes. Journal of Applied Psychology, 90, 799-810. Hecht, T. D., & Allen, N. J. (2009). A longitudinal examination of the work-nonwork boundary strength construct. Journal of Organizational Behavior, 30, 839-862. Hilbrecht, M., Shaw, S. M., Johnson, L. C., & Andrey, J. (2008). ‘I’m home for the kids’: Contradictory implications for work-life balance of teleworking mothers. Gender, Work and Organization, 15, 454-476. Hill, E. J., Erickson, J. C., Holmes, E.K., & Ferris, M. (2010). Workplace flexibility, work hours, and work-life conflict: Finding an extra day or two. Journal of Family Psychology, 24, 349-358. Hill, E. J., Grzywacz, J. G., Allen, S., Blanchard, V. L., Matz-Costa, C., Shulkin, S., & PittCatsouphes, M. (2008). Defining and conceptualizing workplace flexibility. Community, Work & Family, 11, 149-163. Hobfoll, S. E. (1989). Conservation of resources: A new attempt at conceptualizing stress. American Psychologist, 44, 513-524. Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55. Hudson, J. M., Christensen, J., Kellogg, W. A., & Erickson, T. (2002). ‘I’d be overwhelmed, but it’s just one more thing to do:’ Availability and interruption in research management. Computer Human Interaction, 4, 97-104. Ilies, R., Wilson, K. M., & Wagner, D. T. (2009). The spillover of job satisfaction onto employees’ family lives: The facilitating role of work-family role integration. Academy of Management Journal, 52, 87-102. Jett, Q. R., & George, J. M. (2003). Work interrupted: A closer look at the role of interruptions in organizational life. Academy of Management Review, 28, 494-507. Johnson, J. W. (2000). A heuristic method for estimating the relative weight of predictor variables in multiple regression. Multivariate Behavioral Research, 35, 1-19. Judd, C. M., & Kenny, D. A. (1981). Estimating the effects of social interventions. New York: Cambridge University Press. Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek, J. D., & Rosenthal, R. A. (1964). Organizational stress: Studies in role conflict and ambiguity. New York: Wiley. Karasek, R. A. (1979). Job demands, job decision latitude, and mental strain: Implications for job redesign. Administrative Science Quarterly, 24, 285-311. 123 Kelly, E. L., Moen, P., & Tranby, E. (2011). Changing workplaces to reduce work-family conflict: Schedule control in a white-collar organization. American Sociological Review, 76, 265-290. Kingston, P. W. (1988). Studying the work-family connection: Atheoretical progress, ideological bias, and shaky foundations for policy. Journal of Social Behavior and Personality, 3, 55-60. Kline, R. B. (2011). Principles and Practice of Structural Equation Modeling (3rd ed.). New York: The Guilford Press. Konrad, A. M. (2003). Family demands and job attribute preferences: A 4-year longitudinal study of women and men. Sex Roles, 49, 35-46. Kossek, E. E., Baltes, B. B., & Matthews, R. A. (2011). How work-family research can finally have an impact in the workplace. Industrial and Organizational Psychology: Perspectives on Science and Practice, 4, 352-369. Kossek, E. E., & Lautsch, B. A. (2008). CEO of me: Creating a life that works in the flexible job age. Upper Saddle River, NJ: Wharton School Publishing. Kossek, E. E., Lautsch, B. A., & Eaton, S. C. (2005). Flexibility enactment theory: Implications of flexibility type, control, and boundary management for work-family effectiveness. In E. E. Kossek & S. J. Lambert (Eds.), Work and life integration: Organizational, cultural, and individual perspectives (pp. 243-261). Mahwah, NJ: Lawrence Erlbaum Associates. Kossek, E. E., Lautsch, B. A., & Eaton, S. C. (2006). Telecommuting, control, and boundary management: Correlates of policy use and practice, job control, and work-family effectiveness. Journal of Vocational Behavior, 68, 347-367. Kossek, E. E., Pichler, S., Bodner, T., & Hammer, L. B. (2011). Workplace social support and work-family conflict: A meta-analysis clarifying the influence of general and workfamily-specific supervisor and organizational support. Personnel Psychology, 64, 289313. Kossek, E. E., Ruderman, M. N., Braddy, P. W., & Hannum, K. M. (2012). Work-nonwork boundary management profiles: A person-centered approach. Journal of Vocational Behavior, 81, 112-128. Kreiner, G. E. (2006). Consequences of work-home segmentation or integration: A personenvironment fit perspective. Journal of Organizational Behavior, 27, 485-507. Kreiner, G. E., Hollensbe, E. C., & Sheep, M. L. (2009). Balancing borders and bridges: Negotiating the work-home interface via boundary work tactics. Academy of Management Journal, 4, 704-730. Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal and coping. New York: Springer. 124 Kristof-Brown, A. L., Zimmerman, R. D., & Johnson, E. C. (2005). Consequences of individuals’ fit at work: A meta-analysis of person-job, person-organization, persongroup, and person-supervisor fit. Personnel Psychology, 58, 281-342. MacKinnon, D. P., Krull, J. L., & Lockwood, C. M. (2000). Equivalence of the mediation, confounding and suppression effect. Prevention Science, 1, 173-181. Matthews, R. A., & Barnes-Farrell, J. L. (2010). Development and initial validation of an enhanced measure of boundary flexibility for the work and family domains. Journal of Occupational Health Psychology, 15, 330-346. Matthews, R. A., Barnes-Farrell, J. L. , & Bulger, C. A. (2010). Advancing measurement of work and family domain boundary characteristics. Journal of Vocational Behavior, 77, 447-460. Matthews, R. A., & Winkel, D. E. (2011). Boundary theory: Modeling longitudinal effects of role overload, inter-domain transitions, and work-family conflict. Manuscript submitted for publication. Mesmer-Magnus, J. R., & Viswesvaran, C. (2006). How family-friendly work environments affect work/family conflict: A meta-analytic examination. Journal of Labor Research, 4, 555-574. Methot, J. R., & LePine, J. A. (2009). Too close for comfort: Effect of role segmentation preferences in proactive decisions regarding the boundary between work and nonwork domains. Paper presented at the 24th Annual Conference of the Society of Industrial and Organizational Psychology, New Orleans, LA. Middleton, C. A. (2007). Illusions of balance and control in an always-on environment: A case study of BlackBerry Users. Ted Rogers School of Information Technology Management Publications and Research. Paper 4. http://digitalcommons.ryerson.ca/trsitm/4 Mirchandani, K. (1998). Protecting the boundary: Teleworkers insights on the expansive concept of work. Gender and Society, 12, 168-187. Musson, G., & Marsh, K. (2008). Managing the emotional boundaries of telework. In S. Fineman (Ed.) The Emotional Organization. Oxford, UK: Blackwell Publishing. Nippert-Eng, C. (1996). Calendars and keys: The classification of “home” and “work.” Sociological Forum, 11, 563-582. O’Connell, N. (2008). Interruption overload. Strategic Direction, 24, 3-5. Olson-Buchanan, J. B., & Boswell, W. R. (2006). Blurring boundaries: correlates of integration and segmentation between work and nonwork. Journal of Vocational Behavior, 68, 432445. 125 Ostroff, C. (2007). General methodological and design issues. In C. Ostroff & T. A. Judge (Eds.) Perspectives on organizational fit (pp. 352-387). New York: Lawrence Erlbaum Associates. Parasuraman, S., Greenhaus, J. H., & Granrose, C. S. (1992). Role stressors, social support, and well-being. Journal of Organizational Behavior, 13, 339-356. Piotrkowski, C. S. (1979). Work and the family system. New York: Macmillan. Pleck, J. H. (1977). The work-family role system. Social Problems, 24, 417-427. Pleck, J. H., Staines, G. L., & Lang, L. (1980). Conflicts between work and family life. Monthly Labor Review, 103, 29-32. Poelmans, S. A. Y. (2001). A multi-level, multi-method study of work-family conflict: A managerial perspective (Unpublished dissertation). University of Navarra, Spain. Poposki, E. M. (2011). The blame game: Exploring the nature and correlates of attributions following work-family conflict. Group & Organization Management, 36, 499-525. Poppleton, S., Briner, R. B., & Kiefer, T. (2008). The roles of context and everyday experience in understanding work-non-work relationships: A qualitative diary study of white- and blue-collar workers. Journal of Occupational and Organizational Psychology, 81, 481502. Powell, G. N., & Greenhaus, J. H. (2010). Sex, gender, and the work-family interface: Exploring negative and positive interdependencies. Academy of Management Journal, 53, 513-534. Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879-891. Rafnsdottir, G. L., & Heijstra, T. M. (in press). Balancing work-family life in academia: The power of time. Gender, Work and Organization. Rau, B. L., & Hyland, M. M. (2002). Role conflict and flexible work arrangements: The effects on applicant attraction. Personnel Psychology, 55, 111-136. Rothbard, N. P., Phillips, K. W., & Dumas, T. L. (2005). Managing multiple roles: Work-family policies and individuals’ desires for segmentation. Organization Science, 16, 243-258. Sadler, K., Robertson, , T., Kan, M., & Hagen, P. (2006). Balancing work, life and other concerns: A study of mobile technology use by Australian freelancers. Proceedings of the th 4 Nordic Conference on Human-Computer Interaction. Sawyer, K. (2009). Breaking the “iron” barrier: Implications of deviance neutralization theory on the division of household labor and health (Unpublished master’s thesis). Pennsylvania State University, Pennsylvania. 126 Schieman, S., & Young, M. (2010). Is there a downside to schedule control for the work-family interface? Journal of Family Issues, 31, 1391-1414. Shanock, L. R., Baran, B. E., Gentry, W. A., Pattison, S. C., & Heggestad, E. D. (2010). Polynomial regression with response surface analysis: A powerful approach for examining moderation and overcoming limitations of difference scores. Journal of Business and Psychology, 25, 543-554. Shockley, K. M. (2010). You can’t always get what you want, but does it matter? The relationship between pre-child preferences and post-child actual labor division fit and well-being (Unpublished doctoral dissertation). University of South Florida, Florida. Shockley, K. M., & Allen, T. D. (2007). When flexibility helps: Another look at the availability of flexible work arrangements and work-family conflict. Journal of Vocational Behavior, 71, 479-493. Society for Human Resource Management. (2011). 2011 Employee Benefits. Standen, P., Daniels, L., & Lamond, D. (1999). The home as a workplace: Work-family interaction and psychological well-being in telework. Journal of Occupational Health Psychology, 4, 368-381. Steiger, J.H. (1980). Testing pattern hypotheses on correlation matrices: Alternative statistics and some empirical results. Multivariate Behavioral Research, 15, 335-352. Thatcher, S. M. B., & Zhu, X. (2006). Changing identities in a changing workplace: Identification, identity enactment, self-verification, and telecommuting. Academy of Management Review, 31, 1076-1088. Thomas, L. T., & Ganster, D. C. (1995). Impact of family-supportive work variables on workfamily conflict and strain: A control perspective. Journal of Applied Psychology, 80, 615. Towers, I., Duxbury, L., Higgins, C., & Thomas, J. (2006). Time thieves and space invaders: technology, work and the organization. Journal of Organizational Change, 19, 593-618. van Echtelt, P. E., Glebbeek, A. C., & Lindenberg, S. M. (2006). The new lumpiness of work: Explaining the mismatch between actual and preferred working hours. Work, Employment and Society, 20, 493-512. Voydanoff, P. (2005). Consequences of boundary-spanning demands and resources for work-tofamily conflict and perceived stress. Journal of Occupational Health Psychology, 10, 491-503. Westman, M., & Piotrkowski, C. S. (1999). Introduction to the special issue: Work-family research in occupational health psychology. Journal of Occupational Health Psychology, 4, 301-306. 127 Williams, K. J., & Alliger, G. M. (1994). Role stressors, mood spillover, and perceptions of work-family conflict in employed parents. Academy of Management Journal, 37, 837868. Zedeck, S. (1992). Work, families, and organizations. San Francisco: Jossey-Bass. Zeruvabel, E. (1991). The fine line: Making distinctions in everyday life. New York: Free Press. Zijlstra, F. R. H., & Sonnentag, S. (2006). After work is done: Psychological perspectives on recovery from work. European Journal of Work and Organizational Psychology, 15, 129-138. 128