Involuntary vs. Voluntary Flexible Work: Insights for Scholars and Stakeholders

Abstract

Building on insights from the early stages of our research partnership with a U.S. Fortune 500 organization, we came to differentiate between voluntary and involuntary schedule variability and remote work. This differentiation underscores the complexity behind flexible schedules and remote work, especially among white-collar, salaried professionals. We collected survey data among the partner firm’s information technology (IT) workforce to evaluate whether these forms of flexibility had different implications for workers, as part of the larger Work, Family, and Health Network Study. We find that a significant minority of these employees report working variable schedules and working at home involuntarily. Additionally, involuntary variable schedules are associated with greater work-to-family conflict, stress, burnout, turnover intentions, and lower job satisfaction in models that adjust for personal characteristics, type of job, work hours, family demands, and other factors. Voluntary remote work, in contrast, is protective and more common in this professional sample. Employees working at least 20% of their hours at home and reporting moderate or high choice over where they work have lower stress and intentions to leave the firm (as well as higher job satisfaction in some models). These findings point to the importance of both stakeholders and scholars distinguishing between voluntary and involuntary forms of flexibility, even in a relatively advantaged professional and technical workforce.

Introduction

When is workplace flexibility the solution to challenges that contemporary American employees face, and when is it a new source of work-family conflict and strain, contributing to stress and emotional exhaustion? This is an important issue for translational research and outreach, one which could provide much needed insights for employers, managers, and employees. But evidence to date is far from clear. Many scholars claim that flexible work arrangements provide a better fit for today’s workforce, given that all adults in the household are apt to be employed, with no backup on the family care or home front (e.g. King et al., 2012; Moen & Roehling, 2005). However, the very same information and communication technologies enabling working more flexibly can also spawn an “always on, always available” work environment intruding into family life. In fact, some worry that the practices employers promote as flexibility primarily benefit management. For example, management benefits when employees make themselves available for work at times traditionally set aside for family and personal recovery, and/or accept unpredictable schedules with no accompanying wage premium (Clawson & Gerstel, 2014; Gerstel & Clawson, 2001; Henly & Lambert, 2014).

We contend that conversations about, and practices around “workplace flexibility” have been muddled because neither scholars nor stakeholders -- managers, employers, and employees-specify the different forms of flexible work practices at play. Accordingly, the evidence base linking workplace flexibility to employee health and well-being remains conceptually and empirically underdeveloped (Hill et al., 2008). Existing research has not adequately theorized or fully operationalized workplace flexibility, often measured as employees’ perceived flexibility or their sense of control over their schedules. The result is an array of mixed findings. For example, Allen et al’s (2013) meta-analyses on the relationship between flexible work arrangements and work-family conflict (WFC), summarizing findings from 109 independent samples with over 120,000 participants, find inconsistent results. Three meta-analyses find significant negative correlations between flexible work arrangements and WFC, even as two find non-significant relationships. Stakeholders turning to scholarly research confront inconsistent evidence.

In an effort to clarify what is known about flexible work, we engage in a critical examination of existing frameworks and practices around flexibility and conduct an empirical study that specifically analyzes separate stands of flexibility. We follow Allen and colleagues, as well as other scholars, who point to the need to distinguish between flexible work-day, work-week schedules and flexible work locations (specifically, working at home, as well as distinguishing between access to flexibility and actual use of it (Allen, Johnson, Kiburz, & Shockley, 2013; Kelly et al., 2008). To provide an empirical test of a more theoretically- and practice-focused framing, we draw on survey data collected through a collaborative partnership with a large Fortune 500 corporation that we call TOMO. We investigate flexible work practices (FWP) – looking at what people do, specifically, by looking at variable schedules and remote work (or telecommuting) separately. This makes intuitive sense for the stakeholders on the ground, including managers and employees, in that these two practices tended to be offered (or prohibited) separately from one another. Note that different practices may occur in separate teams within a firm like TOMO, or even within teams, as managers reward some workers with flexible work options and demand more traditional work arrangements from others.

We theorize FWP as 1) voluntary, chosen and desired by employees, or 2) involuntary, working over and beyond conventional times and places because managers or employers require it. “Just in time” staffing in hourly service jobs, for example, is a key manifestation of such an involuntary flexibility practice. We contend that such employer-driven flexibility to meet business needs also occurs in professional, white-collar settings, in part because new technologies facilitate working across tiem and space. This theoretical distinction contributes to the field and to translational research and outreach by distinguishing FWP that employees describe as based on their choice from FWP employees report they do not choose or control. We expect that differentiating voluntary and involuntary FWP may help resolve the contradictions in prior research and identify which forms of flexibility are helpful to or challenging for workers.

Our partnership with a single organization, TOMO, permits us to hold constant the broad occupational category, industry characteristics, and organizational policies and benefits of the research site; this allows to us set up a conservative test of our claims about the importance of distinguishing forms of flexibility even in “good jobs.” If both voluntary and involuntary flexible work practices are reported in a single organization and in a relatively advantaged professional and technical workforce, and if those forms of flexibility are associated differently with work attitudes and well-being, then we have clear evidence of the importance of distinguishing different types of flexibility.

We believe most organizations frame flexibility as something desired by their employees. They are less apt to recognize what we call involuntary flexibility: their own management practices, enhanced by new information and communication technologies, that presume work will be accomplished on evenings, early mornings, and weekends in addition to conventional working days, and that assume work will happen at home as well as at the organizational work site. White-collar organizations may be particularly unlikely to recognize or label involuntary flexibility, seeing these “always on” practices as expected, just the reality of the way professional and technical work is done.

We address three related questions: Do professional and technical workers report involuntary, as well as chosen, flexible work practices? Which professional and workers report involuntary schedules and remote work? Are these distinct forms of flexibility associated with work-family conflict, other measures of well-being, and turnover intentions?

Literature Review

Grounded in the job demands-control model, stress process theory, and job demands-resources theory (e.g., Moen et al., 2016; Schieman, 2013), numerous studies find that perceived flexibility or schedule control, i.e. feeling able to shift your work time or work location, is associated with better employee outcomes, including less WFC and better “balance” (e.g. Byron, 2005; Kossek, Lautsch, & Eaton, 2006; Roeters, Van Der Lippe, & Kluwer, 2010; Tausig & Fenwick, 2001). Recent quasi-experimental and experimental studies have also shown that workplace initiatives increasing employees’ control over when and where they work reduce WFC (Kelly et al., 2014; Kelly, Moen, & Tranby, 2011; Pryce, Albertsen, & Nielsen, 2006) and improve employees’ health and well-being (Moen et al., 2016; Moen, Kelly, Tranby, & Huang, 2011; Olson et al., 2015). A recent systematic review argues that there is good evidence on the positive impact of schedule control – which they measure as employees’ say on work time but not work location – on work-life balance and job satisfaction, with less evidence linking schedule control to other health and well-being outcomes (Nijp, Beckers, Geurts, Tucker, & Kompier, 2012).

Yet the relationship between specific flexible work practices – actually working a variable schedule or engaging in remote work, for example – and subjective well-being is less clear. These practices may blur boundaries between work and family and encourage long work hours. One study finds that remote work is linked to better well-being and less conflict only when employees also report having control over when and where they work (Kossek et al., 2006). The benefits of schedule control for lower WFC are partially offset by the fact that more control also encourages interacting with work contacts and bringing work home; those practices, in turn, increase WFC (Schieman & Glavin, 2008; Voydanoff, 2005). Control over when to start and finish work each day is negatively related to WFC, but it is also associated with more “work-family multitasking,” which produces greater WFC (Glavin & Schieman, 2012; Schieman & Young, 2010). Perceived control over work time is beneficial to employees, but some research suggests it facilitates specific practices – including remote work, work-family multi-tasking, and attending to work concerns through calls or email – that may create more stress for employees

Previous research, which generally assumes that flexible work practices among professionals are voluntary and sought by employees, finds that employees engaged in variable schedules and remote work have less WFC (Byron, 2005; Kossek et al., 2006; Moen, Kelly, & Huang, 2008). But research findings are mixed, and flexible schedules seem to have a stronger effect on WFC than is seen with remote work (Allen et al., 2013; Michel, Kotrba, Mitchelson, Clark, & Baltes, 2011).

Voluntary FWP are also expected to be associated with higher job satisfaction (Carlson, Grzywacz, & Kacmar, 2010; Nijp et al., 2012) and turnover expectations (Gajendran & Harrison, 2007; Moen, Kelly, & Hill, 2011). While these flexible work strategies are expected to promote subjective well-being, many of the positive findings regarding stress, psychological distress, and burnout come from studies that measure employees’ perceived control over work time and work location (Duncan & Pettigrew, 2012; Moen, Kelly, & Lam, 2013; Moen, Kelly, Tranby, et al., 2011; Nijp et al., 2012) – meaning what employees feel they can do – rather than analyses of what they actually do. One study, though, found that a flexible schedule itself was associated with lower stress and burnout, with just about half of the benefits tied to perceived flexibility (Grzywacz, Carlson, & Shulkin, 2008).

Some of the strongest evidence of the effects of actually working at home (as opposed to perceived control or flexibility) comes from an experiment conducted with call center employees in a Chinese firm (Bloom, Liang, Roberts, & Ying, 2013). Workers randomized to work at home (exclusively) were compared to workers who had also been interested in working at home but were randomized to the control group (working in the office). In addition to better work performance, those working at home report significantly higher satisfaction and positive affect, significantly lower emotional exhaustion and negative affect, compared with their peers randomized to stay in the office (Bloom et al., 2013, Table 7). Further, employees working at home were about half as likely as the control group to leave their job during the study period (Bloom et al., 2013). Note that this experimental study began with a population of employees interested in working at home; it provides no evidence of the consequences of involuntary FWP.

Locating Involuntary Flexibility

Variable schedules and remote work are not isomorphic with employees’ preferences. Interest in management-driven “flexibility” arose from studies of retail and food service where “just in time staffing” ties work hours tightly to immediate customer demand; for example, workers are sent home – losing expected pay – if the shop floor is quiet and called in or held over without advance notice when there are many customers coming in (Lambert, 2008). These unpredictable schedules create additional stress regarding the logistics of managing family responsibilities and fluctuating income flows that exacerbate financial stress (Henly & Lambert, 2014; Henly, Shaefer, & Waxman, 2006; Lambert, Haley-Lock, & Henly, 2012). Concern about such highly variable schedules is growing. A 2014 New York Times article on Starbucks’ scheduling practices garnered wide attention and led to a change in company policy (Kantor, 2014a, 2014b) and other major U.S. companies have announced a move away from on-call and other unpredictable scheduling (Abrams, 2015). Multiple cities and states have passed “fair scheduling” laws, which require employers to pay employees for a minimum number of hours if they are sent home from work before the end of their shift and the federal Schedules that Work Act has been introduced to Congress repeatedly since 2014 (CLASP, n.d.).

While hourly workers in the service sector have been the primary focus, other workers also face involuntary variable schedules and involuntary remote work. Professionals and managers feel they need to be “always on” and responsive to questions from clients, coworkers, and bosses at any time, with technology facilitating those work patterns (Barley, Meyerson, & Grodal, 2010; Perlow, 2012). If there is a perceived obligation to be available to one’s employer or clients at any time, a variable schedule may reflect the incursion of work and “role blurring,” with negative effects for employees and for family interactions (Chesley, 2005; Glavin, Schieman, & Reid, 2011; MacEachen, Polzer, & Clarke, 2008).

Remote work may be a strategy for coping with high work demands rather than a choice to split one’s regular or reasonable hours between a workplace and elsewhere. This interpretation is supported by the finding that working at home is positively associated with working longer hours; telecommuters are much more likely to work more than 40 hours per week, so their remote work may come on top of full days in the office (Noonan & Glass, 2012). McCrate’s (2012) analysis of the 2004 Work Schedules Supplement to the Current Population Survey reveals that 11.5% of all U.S. workers report limited control over their schedules while simultaneously reporting that their starting and stopping times vary. These “flexible” schedules do not reflect workers’ control over their work time but the way their work is organized by management. Lambert, Fugiel, and Henly (2014) analyze measures from the 2011 wave of the National Longitudinal Survey of Youth 1997 panel and find that fully 38% of younger working Americans (ages 26–32 years old) know their schedule only one week or less in advance, while 74% have their weekly hours fluctuate over the course of a month. These unpredictable and variable work hours are particularly common for hourly and low-wage workers, but occur almost as often (or sometimes more often) for elite professionals, business staff and technical employees (Lambert et al., 2014).

While these nationally representative data provide suggestive information, there is little known about the interplay between perceived control over when and where one works and FWP in white-collar settings, specifically. Professional and technical workforces, like the one we study, face both insecurity and rising demands in the context of globalization and the technological mediation of knowledge work across time and space (Kalleberg, 2011; Lam et al., 2015; MacEachen et al., 2008). Some of these employees may be pushed into (involuntary) variable schedules and substantial remote work because their demanding jobs “require” it. Their variable schedules may reflect expectations that they be available nights, early mornings, and weekends. Similarly, professional and technical workers may do substantial amounts of work at home because they are overloaded, fitting the work in wherever, and whenever, they can (Noonan & Glass, 2012). Involuntary flexible work practices are driven, at least in part, by work overload and, within IT, by the need to coordinate with “off-shore” workers via late night and early morning conference calls. Thus we expect to see some involuntary FWP in addition to chosen FWP, even among these relatively privileged workers.

From the perspective of demands-resources theory, shifting one’s work schedule or doing some work at home may be indicative of either a job demand or a job-related resource. Within this theoretical paradigm, demands are claims, expectations or norms met by the exertion of physical or mental effort, while resources are motivating or rewarding aspects of the job (Demerouti & Bakker, 2007; Voydanoff, 2005). The expectation of being “always on” means (involuntary) FWP is experienced as a demand, while choosing when and where to work gives workers a sense of control, a prototypical job resource (Schieman, 2013).

Although little research specifically examines how involuntary FWP impact employees’ well-being or work attitudes, the job demands-resource paradigm and role-blurring research suggest negative associations. A recent study of retail workers found unstable work schedules, which workers do not choose, are associated with psychological distress, poor sleep quality, and unhappiness (Schneider & Harknett, 2019). In the professional and managerial context, involuntary FWP may also facilitate boundary-spanning work demands (e.g., more work-related calls or emails during personal/family time and multi-tasking on work tasks while at home) that negatively affect employee well-being (Glavin et al., 2011; Voydanoff, 2005). One recent study found greater permeability of work into personal time predicts higher levels of WFC and emotional exhaustion nine months later (Boswell, Olson-Buchanan, & Harris, 2014). A critical perspective suggests that FWP – particularly those occurring with little input from employees – reinforce the intensification of work. If, as MacEachen and colleagues suggest, “flexibility in effect accommodated an increased intensity of work through the merging of home with work, and work with home” (MacEachen et al., 2008, p. 1025), then employees reporting more involuntary FWP should report worse well-being.

Research Questions and Hypotheses

Our first two questions are descriptive: First, do professional, managerial, and technical workers report involuntary, as well as voluntary, FWP? How common are these forms of flexibility? Second, which workers in this highly-educated and salaried professional and managerial workforce report involuntary versus voluntary FWP? Both of these address issues of importance to stakeholders on the ground as well as to work, family, and community scholars.

Our third research question is whether the two forms of flexibility have different associations with employees’ reports of well-being, along with turnover intentions. Here we begin to address prior mixed findings by looking at the relationships between involuntary FWP and WFC, job satisfaction, stress, burnout, and psychological distress. Our general expectation is that voluntary FWP will be linked to better work attitudes and greater subjective well-being for employees, while involuntary FWP may be associated with more negative assessments of one’s job and the organization, along with worse levels of WFC and other measures of subjective wellbeing.

Specifically, we consider six well-being outcomes: WFC, job satisfaction, turnover intentions, emotional exhaustion (burnout), perceived stress, and psychological distress. These are all established measures of well-being, with potentially serious medical consequences. More specifically, WFC is associated with chronic disease and obesity (e.g. Sabbath, Melchior, Goldberg, Zins, & Berkman, 2012), mood disorders (e.g. Frone, 2000), and other indicators of poor health (e.g. van Steenbergen & Ellemers, 2009). Perceived stress predicts many mental and physical health outcomes (Cohen & Williamson, 1988), and psychological distress is predictive of serious mental illness (Kessler et al., 2003). Emotional exhaustion (a dimension of burnout) is associated with turnover intentions and organization commitment (Lee & Ashforth, 1996), as well as mental health outcomes (Wolfram, Bellingrath, Feuerhahn, & Kudielka, 2013). Job satisfaction and turnover intentions are of particular interest to employers, since they are associated with actual turnover (Griffeth, Hom, & Gaertner, 2000), which imposes financial and other costs on companies as well as affecting the attainment and careers of employees.

These questions and hypotheses are summarized as:

R1: To what degree do professional and technical workers report involuntary, as well as voluntary flexible work practices?

R2: Which professional and workers are most likely to report voluntary and involuntary schedules and remote work?

R3: Are these distinct forms of flexibility associated with work-family conflict, other measures of well-being, and turnover intentions?

H1: Compared to those not practicing workplace flexibility, employees who have involuntary variable schedules and engage in involuntary remote work will report greater work-to-family conflict, lower job satisfaction, greater turnover intentions, greater emotional exhaustion, greater perceived stress, and greater psychological stress.

H2: Compared to those not practicing workplace flexibility, employees who have involuntary variable schedules and engage in involuntary remote work lower work-to-family conflict, greater job satisfaction, lower turnover intentions, lower emotional exhaustion, lower perceived stress, and lower psychological stress.

Methods

To test the value of conceptualizing voluntary versus involuntary FWP, we draw on survey data from information technology employees in a large U.S. firm with the pseudonym of TOMO. These data were collected as part of the larger Work, Family, and Health Network study, which partnered with TOMO to investigate the impact of work conditions on work, family life, and health outcomes (see Bray et al., 2013; King et al., 2012). We utilize the baseline data, including a survey of employees and administrative data provided by the Human Resources department of the firm (for job function, tenure, salary, and team size).

Across multiple locations, 1182 non-supervisory employees were eligible for the study, and 823 completed the in-person survey (69.6% response rate). We restricted our sample to respondents who did not have missing values for any of the covariates for an analytic sample of 758 employees nested in 207 teams.

Primary Variables

We construct voluntary and involuntary flexible work practices drawing on a combination of variables. For Variable Schedule, respondents were asked “Which of the following best describes your work schedule at this job? Variable Schedule (one that changes from day to day), Regular Daytime Schedule, Regular Evening Shift, Regular Night Shift, Rotating Shift (one that changes regularly from days to evenings or nights), Split Shift (consisting of two distinct periods each day).” We constructed the 0/1 “variable schedule” measure by setting all variable schedule, rotating shift, and split shift responses to 1 and all other responses to 0. Substantial Remote Work is constructed as a 0/1 variable, where respondents who work 20% or more of their typical weekly hours at home are coded as 1. Respondents were asked “About how many hours do you work in a typical week in this job?” as well as “Do you ever work at home or take work-related calls at home on this job?” and “About how many hours/week do you work or take calls from home on this job?” We constructed the variable this way because 98% of employees do some work at home in this tech-savvy population. 20% of weekly hours roughly corresponds to one day per week at homeⁱ. Alternative specifications of remote work were explored in our sensitivity analyses.

To distinguish between voluntary and involuntary FWP, we combine these work practices with specific questions from a schedule control scale (modified from Thomas & Ganster, 1995). We combine variable schedule with “How much choice do you have over when you begin and end each work day?” and substantial remote work with “How much choice do you have over doing some of your work at home or at another location, instead of at TOMO?” Response choices for each of these questions range from 1=Very Little to 5=Very Much [choice]; we dichotomized responses defining “high” control as responses of 4=Much or 5=Very Much choice over those work practices. For simplicity, we refer to these high and low categories as “voluntary” and “involuntary” (e.g. “voluntary variable schedule,” “involuntary remote work”) when describing FWP with high or low control reported, respectively.

We then examine the relationship between voluntary and involuntary FWP and six wellbeing outcomes: WFC conflict, job satisfaction, turnover intentions, emotional exhaustion (burnout), perceived stress, and psychological distress. Each of these is measured via an established scale; details of scale questions and sources are in Appendix A.

Additional Variables: Demographics, Family Status, and Key Job Characteristics

We include demographic and family status measures, including gender, parent of children age 5 or under, parent of children age 6 to 18, marital status, and providing care for adult relatives, as well as a nativity indicator for respondents born in the United States. We also include respondents’ age, associated with both schedule control and WFC in previous research (Schieman, Milkie, & Glavin, 2009)¹. Additional potentially relevant measures of status include race/ethnicity (white, non-Hispanic, Asian or Pacific Islander, and other), education (graduate degree, bachelor’s degree, and high school diploma/some college), salary, and tenure in the organization.

We also incorporate variables in our models capturing individuals’ self-reported job characteristics as they may potential predictors of voluntary and involuntary FWP. These include indicators for working long hours (50 or more hours per week) or fewer hours (40 or fewer). High psychological job demands (see Appendix A for details on this measure) is included as an indicator as well as autonomy, indicated by the decision authority subscale of job control (Karasek & Theorell, 1990).

Additional Variables: Understanding the Occupational and Organizational Context

We also include broad job categories in the analysis. All of the respondents in the sample are technical professionals with computer science skills. Yet there are differences in the interactions required for each job and in the status associated with each job within the firm and the industry more broadly. Software developer is the modal job category; a developer’s core task is to write computer code for the applications and programs the company provides to internal and external clients. Database administrators are developers who focus on writing code that appropriately stores and pulls customer and financial data as needed by different applications and programs. Analysts turn clients’ desires into technical requirements while systems engineers and architects decide how programs will work together, within the broader technical system. Project managers identify appropriate deadlines and monitor progress. Once the code is written (by the software developers), quality assurance staff test the applications and identify “bugs” and other problems. When problems or concerns arise later on, production support staff pinpoint the issue, resolve it, and pull in developers to make more substantial adjustments as needed.

Other team-level characteristics and contexts may also be potential predictors of voluntary and involuntary FWP. Thus we include task interdependence, and an indicator for high job insecurity, as well as a variable indicating whether or not the respondent was surveyed before a merger announcement that may have changed FWP.

Analytic Techniques

For the analysis investigating employees’ engagement in voluntary and involuntary FWP, we fit multinomial logistic regression models with robust clustered standard errors to compare the potential effects of individual, team and manager characteristics on respondents’ voluntary and involuntary FWP. The models we present in tabular form use the “stable schedule” and “low remote work” groups as reference categories, but we comment on other direct comparisons from models with different reference categories.

To examine the relationship between forms of flexibility and employee well-being, we fit multilevel random effects models, including individual-level independent variables that potentially predict individual employees’ well-being. The models also include a random intercept for each manager to represent unobserved heterogeneity between managers. All individual-level variables are centered at the grand mean to make the manager-level random intercept more meaningful.

In simplest notation, the linear models for well-being are of the format

$$Y_{ij}=\alpha +{\mathit{\beta }}_1{\mathit{X}}_{1ij}+{\mathit{\beta }}_2{\mathit{X}}_{2j}+{\zeta }_j+{ϵ}_{ij}$$

where 𝑌_𝑖j is the outcome for an individual i in work team j, 𝛼 is the intercept, ${\mathit{\beta }}_1{\mathit{X}}_{1ij}$ is the vector of individual characteristics for an individual i in team j, and ${\mathit{\beta }}_2{\mathit{X}}_{2j}$ is the vector of work team characteristics for team j. ${\zeta }_j$ is the random intercept and remains constant for all members of a team but potentially varies across teams. ${ϵ}_{ij}$ is the individual specific error component that varies between individuals and work teams.

Results

R1: To what degree do professional and technical workers report involuntary, as well as voluntary flexible work practices?

FWP are fairly common in this professional and technical workforce (Table 1). Over one in ten (13%) employees report a “variable schedule” they have chosen, which we denote as a voluntary variable schedule. Additionally, almost one in ten (9%) report a variable schedule but say they have little or no choice over the timing of their work hours; we denote this an involuntary variable schedule. These flexibility measures are conservative, since respondents had to describe their usual hours as variable, rather than choosing a “regular daytime” or “regular evening” schedule. Survey questions asking whether employees occasionally shift their starting and stopping hours would likely capture more reports of flexible schedules, but were not in this survey.

Table 1:

Sample Means for Outcomes and Characteristics

	Variable Schedule			Remote Work
	Stable	Voluntary	Involuntary	Low	Substantial Voluntary	Substantial Involuntary
Overall Prevalence in Sample (Proportion)	0.784	0.127	0.00	0.545	0.311	0.144
Outcomes
Schedule Variability
Voluntary Variable Schedule				0.00	0.15 *	0.11
Stable Schedule				0.833	0.703 *	0.771
Involuntary Variable Schedule				0.077	0.102	0.110
Remote Work
Substantial Voluntary Remote Work	0.279	0.47 *	0.353
Low Remote Work	0.579	0.385 *	0.471
Substantial Involuntary Remote Work	0.141	0.135	0.176
Wellbeing and Work Attitude Outcomes
Work-to-Family Conflict Scale	3.008	3.231 *	3.610 *	3.103	2.7	3.246
Job Satisfaction Scale	3.967	4.163 *	3.613 *	3.888	4.03 *	3.45
Turnover Intentions Scale	2.21	2.172	2.61 *	2.344	2.076 *	2.307
Emotional Exhaustion (Burnout) Scale	4.211	4.035	5.020 *	4.275	4.174	4.3 8
Perceived Stress	8.530	8.21	9.485 *	8.72	8.216 *	8.780
Psychological Distress	10.802	10.646	12.044 *	10.65	10.576	11.312
Individual Characteristics and Family Demands
Woman	0.32	0.375	0.324	0.337	0.462 *	0.34
Married or Partnered	0.78	0.771	0.77	0.816	0.784	0.725 *
Has Children Age 5 or Under at Home	0.15	0.135	0.162	0.184	0.174	0.211
Has Children Age 6 to 18 at Home	0.37	0.271 *	0.338	0.383	0.335	0.440
Caregiver (at least 3 hours of care per week for adult relative wil	0.221	0.229	0.368 *	0.228	0.250	0.22
Age
Age 24–39	0.266	0.177	0.324	0.262	0.263	0.248
Age 40–49	0.380	0.302	0.24	0.368	0.326	0.422
Age 50–66	0.354	0.521 *	0.382	0.370	0.411	0.330
Race/Ethnicity
White, Non-Hispanic	0.668	0.72	0.706	0.644	0.758 *	0.642
Asian or Pacific Islander	0.224	0.125 *	0.235	0.254	0.131 *	0.22
Other Race/Ethnicity (non-white, non-Asian)	0.108	0.146	0.05	0.102	0.110	0.128
Education
High School / Some College	0.217	0.271	0.147	0.162	0.275 *	0.303 *
College Degree	0.525	0.40	0.676 *	0.51	0.470 *	0.45 *
Graduate Degree	0.258	0.240	0.176	0.247	0.254	0.23
Respondent Born in United States	0.717	0.844 *	0.706	0.683	0.826 *	0.716
Individual Job Demands and Status
Work Hours
Works 40 or Fewer Hours/Week	0.338	0.167 *	0.103 *	0.281	0.301	0.33
Works 41–4 Hours/Week	0.411	0.448	0.353	0.477	0.352 *	0.284 *
Works 50 or More Hours/Week	0.251	0.385 *	0.544 *	0.242	0.347 *	0.376 *
Task Interdependence	4.045	4.083	4.250 *	4.131	3.2 *	4.000
Psychological Job Demands	3.526	3.649	3.873 *	3.592	3.545	3.560
Decision Authority (subscale of Job Control)	3.765	4.167 *	3.833	3.739	4.014 *	3.719
Job Function
Analysts	0.128	0.146	0.118	0.138	0.127	0.101
Database Administrators	0.037	0.04 *	0.103 *	0.024	0.08 *	0.064 *
Software Developers	0.345	0.313	0.324	0.383	0.275 *	0.312
Production Support/Operations	0.044	0.073	0.05	0.017	0.081 *	0.101 *
Project Managers	0.157	0.156	0.11	0.157	0.174	0.138
Quality Assurance	0.152	0.156	0.147	0.167	0.110 *	0.183
System Engineers/Architects	0.138	0.063 *	0.05	0.114	0.144	0.101
Higher Job Level (Leads vs Lower-Level Contributors)	0.56	0.542	0.632	0.632	0.581	0.468 *
Ln(Annual Salary)	4.448	4.458	4.471	4.472	4.43 *	4.36 *
Organizational Tenure in Years	13.424	16.751 *	13.045	12.764	15.883 *	13.22
High Job Insecurity	0.338	0.333	0.441	0.324	0.36	0.385
Interviewed after Merger Announcement and Geographic Move	0.483	0.438	0.338 *	0.416	0.568 *	0.422

N=758 employees.

^*indicates the mean for variable schedules (or substantial remote work) compared to stable schedules (or low remote work) is statistically different at a significance level of p<.05

Our measure of remote work is also conservative. In this professional and technical workforce, over 95% of employees report doing some work at home, reflecting the realities of digital communication and information technologies. We categorize those who work at least 20% of their usual weekly hours at home or another offsite location as doing substantial remote work. Three in ten (31%) of this sample does at least this much remote work and reports having choice over when and how much work they do off site; we call this substantial voluntary remote work. Another 14% report doing substantial remote work but say they have little or no choice over taking work home, defined as involuntary remote work, driven by management expectations, perceived job requirements (such as responding to counterparts working in other countries), or high workloads.

Voluntary variable schedules and voluntary remote work are associated with one another other; 20% of those who choose to work at home extensively also report a voluntary variable schedule, while only 9% who do less remote work and 12% of those who do substantial involuntary remote work have a voluntary variable schedule.ⁱⁱ

R2: Which professional and workers are most likely to report voluntary and involuntary schedules and remote work?

We next consider who engaged in voluntary and involuntary flexible work. One possibility is that these work patterns are pursued primarily in some jobs and are rarely found in others. However, job functions are not strong predictors of voluntary and involuntary flexible work practices, as reported in Appendix B. Correlations between job categories and the forms of flexibility are sometimes significant, but all less than 0.20. In each job, the work can conceivably be done with a flexible schedule and off-site work. Indeed, a substantial proportion of employees in all job functions were working in geographically-distributed teams (coordinating across time zones) and all could access the company’s systems securely from home or elsewhere.

The pressure to be “always on” and available for work, which encourages involuntary flexible work practices, does vary some by job function. Production support teams must be responsive to problems that arise at any time; not surprisingly, they are less likely to do limited work at home and more likely to report involuntary remote work. In some teams and under some managers, production support staff may be free to choose to work at home during the day and for routine work (so voluntary remote work is seen among some). Employees in jobs with functions involving the most proactive planning and fewer urgent tasks – systems engineers and architects -- are somewhat more likely have stable schedules. Software developers and quality assurance staff – two roles that work closely together – are less likely to do substantial voluntary remote work; employees in these jobs may do more work in the office in part to coordinate.

Jobs are far from determinative, however. In addition to managerial approval for voluntary practices, involuntary flexibility also reflects managers’ and team’s norms about how quickly employees are expected to respond to work-related questions and how many tasks are considered “urgent.”

Looking beyond the job categories, Table 1 provides information on who is more likely to report particular work practices. Women are overrepresented among employees who do substantial voluntary remote work. Women are 46% of those who choose to work at home at least 20% of their total work hours, as compared to 33% of the workers in the low remote work category (and 38% of this sample, overall). Gender and family variables do not differentiate those reporting involuntary variable schedules from those with stable schedules or those doing substantial involuntary remote work from those doing less remote work.

Younger workers are overrepresented among those with involuntary variable schedules, while older workers (ages 50–66) are overrepresented among those working a voluntary variable schedule. This pattern suggests younger workers may be asked to adjust their schedules in response to specific work duties (e.g., being on call to deal with technical problems at night) while older workers are more likely to be able to choose their schedules. This is supported by the fact that those with voluntary variable schedules have significantly longer tenure at the firm, while those working involuntary variable schedules have significantly shorter tenure than employees reporting stable schedules.

Employees who identify as Asian or Pacific Islander are significantly less likely to have voluntary variable schedules or to do voluntary remote work. White employees are overrepresented among those doing substantial voluntary remote work (and underrepresented among those working remotely involuntarily). Respondents born in the U.S. are significantly more likely to have voluntary variable schedules and less likely to have involuntary variable schedules. Bivariate analysis (Table 1) shows technical professionals who have some college are overrepresented among those doing substantial voluntary remote work, while those with a college degree are underrepresented in substantial voluntary remote work and overrepresented in involuntary variable schedules. This likely reflects education’s association with job functions, which are weakly associated with voluntary and involuntary work practices (see correlations in Appendix B). Technical workers without a college degree are more likely to be analysts, in production support, and in quality assurance. Two of these jobs (production support and quality assurance) routinely involve late night and early morning work in this firm that is done at home. Additionally, it may be that employees with a college degree and better prospects for advancement are more cautious about doing voluntary remote work and more willing to make themselves available to work at any hour (thus reporting involuntary variable schedules) in hopes of getting ahead.

Critical perspectives on FWP suggest that workers pursue these arrangements as a strategy for coping with very long hours. We see in Table 1 that 54% of those working involuntary variable schedules work 50+ hours, as compared to only 25% of those with stable schedules. Work hours also predict substantial voluntary remote work. Consistent with research on perceived flexibility (e.g., Schieman et al., 2009), those working a voluntary variable schedule or voluntary remote work also tend to have more decision authority (i.e. control over how to do the work). Decision authority is also significantly higher among those with involuntary variable schedules than those with stable schedules. Additionally, those working involuntary variable schedules report significantly higher job demands (as measured by working hard, working fast, and having too much to do) than those working stable schedules. These findings, in combination with the college degree findings reported in Table 1, lend credence to the “stress of higher status” claim that some workers in higher-status jobs are also under greater strain and may be overwhelmed by high work demands (e.g., Schieman et al. 2009). However, neither job level nor salary predicts involuntary schedules or remote work within the restricted range of non-managerial, technical professional jobs captured in this sample.

We also considered two contextual factors that may have affected these workers’ FWP. During the period of data collection, the firm announced it would be part of a major merger and employees surveyed after that announcement reported higher job insecurity (Lam et al., 2015). Employees may be wary of pursuing FWA – choosing flexible schedules and/or remote work – when they face job precarity. Choosing to work differently than one’s peers may signal lower commitment to one’s work (Minnotte, Cook, & Minnotte, 2010; Munsch, Ridgeway, & Williams, 2014). Moreover, employees who anticipate organizational restructuring and downsizing with the merger may want to maximize their “face time” and visibility to managers. Involuntary variable schedules and involuntary remote work may thus be accepted by employees who feel insecure in their jobs. Those working involuntary variable schedules and those doing substantial involuntary remote work report higher job insecurity than others, though the difference is marginally significant (p<.10). At about the same time as the merger announcement, the firm also decided to move employees within the same city. Some employees were assigned to a work site quite far from their homes and some managers relaxed constraints about working at home. Employees surveyed after those changes were significantly more likely to be doing substantial remote work, particularly voluntary remote work.

R3: Are these distinct forms of flexibility associated with work-family conflict, other measures of well-being, and turnover intentions?

In support of our hypotheses, we find substantial evidence that voluntary and involuntary flexibility are differentially associated with well-being outcomes. Bivariate relationships (Table 1) reveal that the highest WFC is reported by those with involuntary variable schedules while those with voluntary variable schedules have significantly higher WFC than those with stable schedules. .Moreover, employees engaged in substantial involuntary remote work have significantly higher WFC than those in the low remote work category. The lowest WFC is reported by those doing substantial voluntary work at home.

Compared to those with stable schedules, job satisfaction is significantly higher for those with voluntary variable schedules and significantly lower for those with involuntary variable schedules. Substantial voluntary remote work is also associated with higher job satisfaction than seen among those in the low remote work category. Involuntary variable schedules are associated with significantly higher turnover intentions, burnout, stress, and psychological distress compared to employees working stable schedules. These relationships provide preliminary support for the value of distinguishing between voluntary and involuntary flexible work practices; further insight is provided in the multivariate models described below.

Table 2 reports multivariate, multilevel models in which we examine whether FWP predict wellbeing and work attitudes. Model 1 for each outcome reports the focal flexible work variables; Model 2 adds extensive controls. The coefficients for control variables largely show relationships predicted by previous studies. Women report more WFC, stress, and psychological distress but do not differ from male peers in terms of job satisfaction, turnover intentions, or burnout. Parents of school-aged children are more stressed and burnout but having young children or caregiving responsibilities is not associated with these outcomes. Specific work conditions matter too. Long work hours are associated with worse WFC, burnout, stress and psychological distress. Greater decision authority is always significantly associated with better wellbeing outcomes and work attitudes.

Table 2:

Multilevel Models of the Relationship Between Flexible Work Practices and Employee Well-Being

	Work to Family Conflict				Job Satisfaction				Turnover Intentions				Emotional Exhaustion				Perceived Stress				Psychological Distress
	Mode l		Mode 2		Mode l		Mode 2		Mode l		Mode 2		Mode l		Mode 2		Mode l		Mode 2		Mode l		Mode 2
	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error
Flexible Work Practices
Variable Schedule (stable schedule omitted)
Voluntary Variable Schedule	0.200 *	(0.099)	0.211 **	(0.082)	0.128	(0.084)	−0.052	(0.075)	0.015	(0.114)	0.209 *	(0.105)	−0.196	(0.162)	−0.151	(0.138)	−0.218	(0.293)	0.240	(0.278)	−0.057	(0.356)	0.366	(0.333)
Involuntary Variable Schedule	0.566 ***	(0.115)	0.239 *	(0.095)	−0.365 ***	(0.098)	−0.355 ***	(0.087)	0.502 ***	(0.132)	0.396 **	(0.122)	0.799 ***	(0.188)	0.408 *	(0.160)	0.973 **	(0.339)	0.747 *	(0.324)	1.270 **	(0.411)	0.722 +	(0.387)
Remote Work (less than 20% of work is remote omitted)
Substantial Voluntary Remote Work	−0.100	(0.078)	−0.055	(0.064)	0.144 *	(0.066)	0.061	(0.058)	−0.275 **	(0.087)	−0.147 +	(0.081)	−0.047	(0.126)	0.012	(0.108)	−0.443 *	(0.223)	−0.362 +	(0.218)	−0.396	(0.266)	−0.277	(0.258)
Substantial Involuntary Remote Work	0.119	(0.099)	0.103	(0.080)	0.043	(0.084)	0.079	(0.073)	−0.055	(0.112)	−0.064	(0.102)	0.047	(0.160)	0.075	(0.134)	0.049	(0.288)	−0.140	(0.272)	0.311	(0.346)	0.017	(0.323)
Covariates
Woman			0.190 **	(0.060)			−0.049	(0.054)			−0.125	(0.076)			0.137	(0.101)			0.567 **	(0.203)			0.710 **	(0.243)
Married or Partnered			0.039	(0.070)			0.083	(0.064)			−0.114	(0.090)			−0.067	(0.118)			−0.270	(0.237)			−0.870 **	(0.285)
Has Children Age 5 or Under at Home			0.113	(0.078)			−0.028	(0.071)			0.164	(0.100)			0.128	(0.131)			0.182	(0.265)			0.086	(0.316)
Has Children Age 6 to 18 at Home			0.037	(0.058)			−0.017	(0.053)			0.099	(0.075)			0.195 *	(0.098)			0.606 **	(0.198)			0.307	(0.238)
Caregiver (at least 3 hours of care per week for			−0.032	(0.062)			0.068	(0.057)			−0.005	(0.080)			−0.089	(0.105)			0.251	(0.212)			0.529 *	(0.255)
Race/Ethnicity (White, Non-Hispanic omitted)
Asian or Pacific Islander			0.110	(0.114)			0.064	(0.104)			−0.062	(0.146)			−0.223	(0.192)			−0.153	(0.390)			0.694	(0.464)
Other Race/Ethnicity (non-white, non-Asian)			−0.165 +	(0.089)			0.177 *	(0.081)			0.022	(0.114)			−0.286 +	(0.150)			−1.196 ***	(0.303)			−0.819 *	(0.362)
Education (College Degree omitted)
High School / Some College			−0.127	(0.077)			−0.010	(0.071)			−0.001	(0.099)			−0.076	(0.130)			0.310	(0.263)			0.178	(0.315)
Graduate Degree			−0.053	(0.067)			0.090	(0.061)			−0.075	(0.086)			−0.214 +	(0.112)			−0.010	(0.227)			−0.325	(0.272)
Respondent Born in United States			0.136	(0.104)			−0.058	(0.095)			0.132	(0.134)			0.382 *	(0.176)			−0.536	(0.356)			−0.150	(0.425)
Age (Age 40–49 omitted)
Age 24–39			0.032	(0.076)			−0.081	(0.070)			0.154	(0.098)			0.051	(0.128)			0.401	(0.259)			0.334	(0.310)
Age 50–66			0.031	(0.068)			0.067	(0.063)			−0.119	(0.088)			0.027	(0.116)			0.042	(0.233)			−0.333	(0.280)
Work Hours (41–49 Hours/Week omitted)
Works 40 or Fewer Hours/Week			−0.247 ***	(0.067)			−0.007	(0.062)			−0.095	(0.086)			−0.078	(0.114)			0.215	(0.230)			0.412	(0.274)
Works 50 or More Hours/Week			0.394 ***	(0.066)			−0.063	(0.060)			0.138	(0.085)			0.296 **	(0.111)			0.457 *	(0.225)			0.878 **	(0.268)
Task Interdependence			−0.020	(0.036)			0.064 *	(0.033)			−0.009	(0.046)			0.035	(0.060)			−0.258 *	(0.122)			−0.106	(0.145)
Psychological Job Demands			0.474 ***	(0.042)			−0.081 *	(0.038)			0.043	(0.053)			0.824 ***	(0.070)			0.698 ***	(0.141)			0.859 ***	(0.168)
Decision Authority (subscale of Job Control)			−0.342 ***	(0.039)			0.516 ***	(0.036)			−0.473 ***	(0.050)			−0.545 ***	(0.066)			−0.840 ***	(0.134)			−0.904 ***	(0.160)
Job Function (Software Developers omitted)
Analysts			−0.173 +	(0.094)			0.186 *	(0.083)			−0.001	(0.115)			−0.097	(0.154)			0.258	(0.317)			0.410	(0.363)
Database Administrators			0.294 *	(0.140)			0.432 ***	(0.121)			−0.081	(0.165)			−0.134	(0.226)			−0.405	(0.471)			−0.355	(0.522)
Production Support/Operations			−0.044	(0.139)			−0.068	(0.124)			0.045	(0.171)			0.138	(0.230)			−0.062	(0.471)			0.454	(0.543)
Project Managers			−0.080	(0.089)			−0.072	(0.078)			−0.013	(0.108)			0.054	(0.146)			0.042	(0.300)			0.463	(0.342)
Quality Assurance			−0.089	(0.094)			0.163 * (0.082)	(0.082)			−0.241 *	(0.111)			−0.362 *	(0.153)			−0.333	(0.317)			0.070	(0.353)
System Engineers/Architects			−0.041	(0.097)			0.108	(0.087)			−0.240 *	(0.120)			−0.061	(0.161)			0.686 *	(0.330)			0.866 *	(0.381)
Ln(Annual Salary)			0.220	(0.165)			−0.309 * (0.150)	(0.150)			0.578 **	(0.210)			0.310	(0.277)			−0.143	(0.562)			0.237	(0.666)
Organizational Tenure in Years			−0.001	(0.004)			0.004	(0.003)			−0.019 ***	(0.005)			−0.000	(0.006)			0.005	(0.013)			0.003	(0.015)
High Job Insecurity			0.073	(0.058)			−0.052	(0.053)			0.257 ***	(0.074)			0.076	(0.098)			0.376 +	(0.198)			0.742 **	(0.236)
Interviewed after Merger Announcement and			−0.097	(0.063)			0.050	(0.055)			0.163 *	(0.075)			−0.171 +	(0.102)			−0.330	(0.211)			−0.424 +	(0.239)
BIC	2036		1848		1786		1711		2239		2225		2774		2639		3668		3706		3962		3976

^***p < 0.001,

^**p < 0.01,

^*p < 0.05,

⁺p < 0.1. N=758 employees

Even after controlling for other factors such as work hours, job demands, and family demands, an involuntary variable schedule is associated with significantly greater WFC, lower job satisfaction, and greater turnover intentions, emotional exhaustion, perceived stress, and psychological distress (with distress only marginally significant in Model 2). Unexpectedly, involuntary remote work is not clearly linked to these outcomes. These findings provide partial support for H1, since employees with involuntary variable schedules fare worse on all outcomes but those doing involuntary remote work do not.

Similarly, we find partial support for H2, again with distinct differences for variable schedules as compared to remote work. In line with our hypothesis, voluntary remote work is associated with greater job satisfaction, lower turnover intentions, and less stress, although these relationships are attenuated in the full models. Surprisingly, voluntary variable schedules predict greater WFC and higher turnover intentions. Voluntary variable schedules may be a response to particularly high WFC, where responsibilities for children or other caregiving duties lead employees to interrupt their work for substantial portions of the day or split their work shifts; employees may value the voluntary variable schedules but also experience high WFC and wonder whether staying in this job is feasible. Another possibility is that voluntary variable schedules may be pursued to allow an employee to take classes or go on interviews, as part of planning to leave one’s job.

Discussion and Implications for Translational Research

This study sheds new light for employers, managers and employees on the mixed evidence to date regarding FWPs. It does so by distinguishing between voluntary and involuntary FWP, as well as between flexibility in the timing and location of work. Importantly, we foreground a form of flexibility often neglected by these stakeholders as well as by scholars – the fact that management practices often require white-collar employees to be “flexible” by working at home, at night, in the early mornings, and on weekends as part of “normal” work arrangements and as ways of meeting deadlines. The analytic design we use in this study reflects how our view of FWP changed through our being immersed in this research partnership with TOMO. We went from viewing flexibility as an unambiguous good to seeing it through a more nuanced lens.

Our findings indicate the value of choosing one’s work locations and the real risks of being pressured to work variable hours – even for a professional, salaried workforce. Substantial voluntary remote work seems to benefit employees in terms of job satisfaction, turnover intentions, and lower stress. The benefits of voluntary variable schedules are less clear. Employees reporting voluntary variable schedules are significantly more satisfied with their jobs, but they also report higher WFC than those with stable schedules, although the direction may be reversed, with high WFC leading to variable schedules. When employees have a variable schedule that they do not choose, they fare significantly worse (as compared to those with stable schedules) on all six outcomes we study.

What are the implications for translational research and outreach? First, we have identified the prevalence and apparent consequences of involuntary FWP for professional and technical white-collar employees, a group previously presumed to have more control over where and when they work. Involuntary FWP are evident, even among these employees working in good jobs and for a good employer. Involuntary variable schedules are reported by 10% of this sample and involuntary remote work is reported by 13% of respondents; our conservative measures suggest that this is not an especially common situation but we see that it is a consequential one.

Second, this study shows the importance of distinguishing between forms of flexibility with regard to who has control (choice or say) and with regard to specific practices. It is important to distinguish both the specific work practice and whether the practice is chosen or involuntary. The four forms of flexibility studied here are differentially associated with the employee outcomes. Especially important: Involuntary variable schedules are associated with significantly higher WFC, lower job satisfaction, higher turnover intentions, higher emotional exhaustion (burnout), higher perceived stress, and greater psychological distress (marginally significant at p<.10). These findings are net of personal characteristics, socioeconomic status, family demands, work hours, and numerous other measures of the work environment.

In contrast, substantial voluntary remote work is protective and it is also quite common (reported by 32% of employees in this professional, technical workforce). Voluntary remote work is associated with significantly lower turnover intentions, perceived stress, and psychological distress (marginally significant at p<.10), although it is not associated with WFC or job satisfaction. In other words, employees who choose to work at home for a significant portion of the work week are no more or less satisfied with their jobs nor more or less conflicted by the way work and non-work come together, but they feel better and are more interested in staying with the firm. This contradicts a strong version of role-blurring theory, where WFC should increase with more work at home, regardless of whether or not it is voluntary. One intriguing finding is that voluntary variable schedules are associated with unintended, negative consequences for WFC and turnover intentions. This suggests that, even when perceived as chosen, interrupting and shifting schedules substantially is experienced as problematic.

Conclusion

Our analysis has real implications for translational research and practice. While previous discussions of the consequences of involuntary flexibility have focused on low-wage workers, we show involuntary flexibility in work schedules and work location is fairly common and has some negative consequences even for workers in largely “good” jobs at a large firm. We find involuntary variable schedules, in particular, are associated with professionals’ and managers’ greater work-to-family conflict, stress, burnout, turnover intentions, and lower job satisfaction. Voluntary remote work is protective with regard to stress, psychological distress, and intentions to leave the firm, with no negative evident repercussions.

Our findings demonstrate the need for stakeholders to strategize carefully about how to provide flexibility as a resource versus flexibility in the form of pressure for always-on availability. “Flexibility” is a vague term that allows advocates in human resources as well as supportive managers to push for change, even as top management may expect changes that benefit the organization, in terms of squeezing even more work out of their employees. Stakeholders may be using the same term while simultaneously pursuing vastly different goals. We believe there are ways to foster a win-win arrangement, but only when employees have some control and are able to bound their work time.

These analyses also reveal how a highly competitive global economy in tandem with rapid advances in communication and information technologies encourage management practices that establish involuntary variable schedules and involuntary work at home as new reality far beyond retail and other service sector jobs. Moreover, involuntary remote work may be built into work arrangements as a way of reducing space needs, and hence real estate costs. This brings up a related issue: the need to assess the costs to businesses as well as to employees of involuntary flexible work practices. Executives assume benefits to organization of unbounded work and employee responsiveness whenever and wherever. But there are real costs too, in the form of intentions to leave the firm and burnout, in addition to WFC. Globalized and virtual knowledge work means professionals are increasingly likely to experience variable schedules and blurred boundaries imposed as part of the way business is done.

In sum, distinction between voluntary and involuntary flexible work practices is a key and consequential one. Policy discussions and re-evaluations of management practices should also return to feasible ways to limit work demands and avoid overload, such as by ferreting out low-value activities, setting more realistic deadlines, and having sufficient staff to get necessary work accomplished. What we learned from our partnership with TOMO is that work is being redesigned in teams and in boardrooms by technology and by efforts to stay competitive, without much deliberation as to the nature of these changes or their unintended negative consequences. And “flexibility” has become a catch-all phrase that means different things to different stakeholders. Scholars can do much to promote better understanding of these challenges and support the well-being of workers, but that requires that employees and advocated are focused on the “right” kinds of flexibility and, importantly, avoiding compelled, involuntary flexibility.

Author Bio:

Anne Kaduk received her Ph.D. in sociology from the University of Minnesota and works as a research analyst at Facebook. Katie Genadek is a research associate at the Institute for Behavioral Science at the University of Colorado - Boulder and an economist with the U.S. Census Bureau. Erin Kelly is the Sloan Distinguished Professor of Work and Organization Studies at the MIT Sloan School of Management and an affiliate of the Institute for Work and Employment Research at MIT. Phyllis Moen is the McKnight Presidential Chair in Sociology at the University of Minnesota and directs the Life Course Center there.

Appendix A:

Description of Scales/Questions

Scale	Source	Variable Description	Cronb ach's Alpha	Range
Work-to-Family Conflict	Netemeyer 1996	The demands of your work interfere with your family or personal time.	0.92	1–5
		The amount of time your job takes up makes it difficult to fulfill your family or personal responsibilities.
		Things you want to do at home do not get done because of the demands your job puts on you.
		Your job produces strain that makes it difficult to fulfill your family or personal duties.
		Due to your work-related duties, you have to make changes to your plans for family or personal activities.
		Response Choices (reversed): 1=Strongly Disagree, 2=Disagree, 3=Neither, 4=Agree, 5=Strongly Agree
Job Satisfaction	Camman et al 1983	In general, you like working at your job.	0.86	1–5
		In general, you are satisfied with your job
		You are generally satisfied with the kind of work you do in this job.
		Response Choices (reversed): 1=Strongly Disagree, 2=Disagree, 3=Neither, 4=Agree, 5=Strongly Agree
Turnover Intentions	Boroff & Lewin 1997	You are seriously considering quitting [compan y name] for another employer.	0.87	1–5
		During the next 12 months, you will probably look for a new job outside [company name].
		Response Choices (reversed): 1=Strongly Disagree, 2=Disagree, 3=Neither, 4=Agree, 5=Strongly Agree
Burnout (Emotional Exhaustion)	Maslach & Jackson 1986	You feel emotionally drained from your work. How often do you feel this way?	0.89	1– 7
		You feel burned out by your work. How often do you feel this way?
		You feel used up at the end of the workday. How often do you feel this way?
		Response Choices (reversed): 1=Never, 2=A few times a year or less, 3=Once a month or less, 4=A few times a month, 5=Once a week, 6=A few times a week, 7=Every day
Perceived Stress	Cohen, Kamarck & Mermelstein 1983	During the past 30 days, how often have you fe lt that you were unable to control the important things in your life?	0.76	4–20
		During the past 30 days, how often have you felt confident about your ability to handle your personal problems?
		During the past 30 days, how often have you felt that things were going your way?
		During the past 30 days, how often have you felt difficulties were piling up so high that you could not overcome them?
		Response Choices (not reversed): 1=Very often, 2=Fairly often, 3=Sometimes, 4=Almost never, 5=Never
Psychological Distress	Kessler et al 2003	During the past 30 days, how much of the time did you feel so sad nothing could cheer you up?	0.76	6–30
		During the past 30 days, how much of the time did you feel nervous?
		During the past 30 days, how much of the time did you feel restless or fidgety?
		During the past 30 days, how much of the time did you feel hopeless?
		During the past 30 days, how much of the time did you feel that everything was an effort?
		During the past 30 days, how much of the time did you feel worthless?
		Response Choices (reversed): 1=None of the time, 2=A little of the time, 3=Some of the time, 4=Most of the time, 5=All of the time
Adult Care		During the past 6 months have you provided at least 3 hours of care per week to an adult relative inside or outside your home? This could include help with shopping, medical care, or assistance in financial/budget planning.		0–1
		No/Yes
Task Interdependence		How often does your job require you to work closely with others when doing your work?		1–5
		Response Choices: 1=Never, 2=Rarely, 3=Some of the time, 4=Most of the time, 5=All of the time
Psychological Job Demands Scale	Karasek et al 1998	You do not have enough time to get your job done.	0.576	1– 5
		Your job requires very fast work.
		Your job requires very hard work.
		Response Choices (reversed): 1=Strongly Disagree, 2=Disagree, 3=Neither, 4=Agree, 5=Strongly Disagree
Job Control Scale (Decision Authority)	Karasek et al 1998	Your job allows you to make a lot of decisions on your own.	0.718	1– 5
		On your job, you have very little freedom to decide how you do your work.
		You have a lot of say about what happens on your job.
		Response Choices (not reversed): 1=Strongly Agree, 2=Agree, 3=Neither, 4=Disagree, 5=Strongly Disagree
Job Insecurity		Thinking about the next 12 months, how likely do you think it is that you will lose your job or be laid off?		1–4
		Response Choices: 1=Very Likely, 2=Fairly Likely, 3=Not too likely, 4=Not at all likely (1 or 2 = "High Job Insecurity”)
Manager Views of Flexiblity on Productivity	Kossek, Barber & Winters 1999	You worry that allowing more flexibility around hours and working from home would make it more difficult for your employees to reach their objectives	0.87	1–5
		If your allow your employees to use flexible work schedules it would make it harder to get work done.
		Response Choices: 1=Strongly Agree, 2=Agree, 3=Neither Agree Nor Disagree, 4=Disagree, 5=Strongly Disagree

Appendix B:

Correlations

		1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45
1	Voluntary Variable Schedule	1
2	Stable Schedule	−0.7247*	1
3	Involuntary Variable Schedule	−0.1195*	−0.5974*	1
4	Substantial Voluntary Remote Work	0.1380*	−0.1310*	0.0282	1
5	Low Remote Work	−0.1219*	0.1310*	−0.0468	−0.7357*	1
6	Substantial Involuntary Remote Work	−0.0091	−0.0129	0.0292	−0.2756*	−0.4484*	1
7	Work-to-Family Conflict Scale	0.057	−0.1665*	0.1735*	−0.0667	0.0143	0.0677	1
8	Job Satisfaction Scale	0.0985*	0.0167	−0.1387*	0.1140*	−0.1005*	−0.0078	−0.3479*	1
9	Turnover Intentions Scale	−0.0303	−0.0662	0.1307*	−0.1149*	0.0925*	0.0204	0.2905*	−0.5548*	1
10	Emotional Exhaustion (Burnout) Scale	−0.0572	−0.0634	0.1579*	−0.039	0.0102	0.037	0.6027*	−0.4123*	0.2845*	1
11	Perceived Stress	−0.051	−0.0329	0.1068*	−0.0907*	0.0624	0.0312	0.3790*	−0.3945*	0.2587*	0.4187*	1
12	Psychological Distress	−0.0292	−0.054	0.1118*	−0.066	0.024	0.053	0.3665*	−0.3506*	0.2596*	0.4507*	0.6718*	1
13	Woman	−0.007	0.0327	−0.039	0.1078*	−0.1065*	0.0089	0.0995*	−0.0185	−0.0927*	0.1085*	0.1018*	0.1438*	1
14	Married or Partnered	−0.0207	0.024	−0.0104	−0.0149	0.0624	−0.0689	0.018	0.0527	−0.02	−0.0165	−0.0323	−0.1125*	−0.1990*	1
15	Has Children Age 5 or Under at Home	−0.0484	0.0519	−0.0186	−0.019	−0.0019	0.0278	0.0172	−0.0334	0.1128*	−0.0351	0.0398	0.0087	−0.1380*	0.2265*	1
16	Has Children Age 6 to 18 at Home	−0.0827*	0.0838*	−0.0245	−0.0573	0.0149	0.0545	0.0447	−0.0309	0.0722*	0.0624	0.1159*	0.0528	0.0201	0.2018*	0.1219*	1
17	Caregiver (at least 3 hours of care per week for adult rel	−0.0051	−0.0642	0.0984*	0.0241	−0.0187	−0.0053	0.0354	0.0203	0.0121	0.0052	0.0692	0.1078*	0.0810*	−0.0087	−0.0231	−0.0381	1
18	Age 24–39	−0.0719*	0.0265	0.0455	0.0043	0.004	−0.0114	0	−0.0783*	0.1184*	−0.0368	0.0647	0.0739*	−0.0843*	−0.0386	0.3768*	−0.0315	−0.0799*	1
19	Age 40–49	−0.0481	0.07	−0.0448	−0.0511	0.0119	0.0505	0.0109	−0.0314	0.0571	0.027	0.0107	0.0342	0.0363	0.0674	−0.0197	0.3036*	0.0416	−0.4471*	1
20	Age 50–66	0.1128*	−0.0933*	0.0033	0.0468	−0.0155	−0.0398	−0.0108	0.1020*	−0.1638*	0.0065	−0.0692	−0.1007*	0.0403	−0.032	−0.3214*	−0.2727*	0.0311	−0.4613*	−0.5874*	1
21	White, Non-Hispanic	0.0406	−0.0451	0.0178	0.1139*	−0.0829*	−0.0327	0.0642	−0.0278	−0.0296	0.1889*	0.0181	−0.0387	0.0191	−0.0232	−0.2121*	−0.1087*	−0.0329	−0.2439*	−0.0049	0.2256*	1
22	Asian or Pacific Islander	−0.0814*	0.0535	0.0176	−0.1332*	0.1119*	0.017	−0.0046	−0.042	0.0736*	−0.1619*	0.0811*	0.1033*	−0.1513*	0.0664	0.2599*	0.1230*	0.0471	0.3248*	−0.0095	−0.2845*	−0.7560*	1
23	Other Race/Ethnicity (non-white, non-Asian)	0.0462	−0.0027	−0.0499	0.0043	−0.0228	0.0267	−0.0904*	0.0970*	−0.0525	−0.0706	−0.1340*	−0.0779*	0.1705*	−0.0526	−0.0235	0.0015	−0.0126	−0.0611	0.0199	0.0356	−0.5070*	−0.1809*	1
24	High School / Some College	0.0491	−0.0023	−0.0537	0.0941*	−0.1470*	0.0845*	−0.0908*	0.0662	−0.1209*	0.0005	−0.0049	0.0144	0.2015*	−0.0854*	−0.1440*	−0.0927*	0.0321	−0.1668*	−0.019	0.1698*	0.0815*	−0.2505*	0.2074*	1
25	College Degree	−0.0341	−0.0345	0.0895*	−0.0862*	0.1239*	−0.0621	0.0813*	−0.1093*	0.1000*	0.0764*	0.0104	0.023	−0.0896*	0.0449	0.0422	0.0203	−0.0693	0.1311*	−0.0161	−0.1026*	0.0954*	−0.0131	−0.1261*	−0.5650*	1
26	Graduate Degree	−0.0074	0.0421	−0.052	0.0097	−0.0027	−0.009	−0.0071	0.0629	0	−0.0887*	−0.0073	−0.0403	−0.0890*	0.0297	0.0888*	0.065	0.0494	0.0079	0.0368	−0.0437	−0.1881*	0.2545*	−0.0525	−0.3029*	−0.6152*	1
27	Respondent Born in United States	0.0959*	−0.0645	−0.0187	0.1429*	−0.1220*	−0.0154	0.0176	0.0261	−0.0602	0.1719*	−0.0825*	−0.0811*	0.1405*	−0.0739*	−0.2726*	−0.1149*	−0.0656	−0.2937*	−0.0022	0.2679*	0.6443*	−0.8150*	0.1051*	0.2974*	−0.0211	−0.2597*	1
28	Works 40 or Fewer Hours/Week	−0.1075*	0.1788*	−0.1325*	0.0079	−0.0351	0.0395	−0.3407*	0.0624	−0.0944*	−0.2278*	−0.0413	−0.0574	−0.0059	−0.0114	0.1239*	0.0345	−0.0041	0.0645	−0.0016	−0.0568	−0.1251*	0.1302*	0.0165	0.0717*	−0.0793*	0.0231	−0.1111*	1
29	Works 41–49 Hours/Week	0.0291	0.0019	−0.0366	−0.0801*	0.1484*	−0.1049*	−0.028	0.0174	−0.0125	0.0019	−0.0746*	−0.1092*	−0.0022	−0.0502	−0.0929*	−0.0661	−0.0382	−0.0907*	0.0009	0.0811*	0.1419*	−0.1578*	−0.0056	−0.024	−0.017	0.0426	0.1592*	−0.5402*	1
30	Works 50 or More Hours/Week	0.0762*	−0.1811*	0.1722*	0.0786*	−0.1250*	0.0737*	0.3713*	−0.0813*	0.1080*	0.2260*	0.1219*	0.1753*	0.0083	0.0656	−0.0238	0.0368	0.0453	0.0333	0.0006	−0.0307	−0.028	0.0399	−0.0105	−0.0459	0.0978*	−0.0691	−0.0607	−0.4181*	−0.5385*	1
31	Task Interdependence	0.0071	−0.0559	0.0723*	−0.0658	0.0863*	−0.0357	0.1225*	−0.0027	0.0571	0.1344*	0.0245	0.0839*	0.0724*	0.0238	−0.0328	−0.0088	0.0940*	0.0133	0.0144	−0.0263	0.0024	0.0367	−0.0519	0.0149	−0.0597	0.0547	−0.057	−0.1702*	0.0091	0.1606*	1
32	Psychological Job Demands	0.0408	−0.1247*	0.1322*	−0.0263	0.0299	−0.0077	0.4992*	−0.1425*	0.0947*	0.4964*	0.2251*	0.2436*	0.1560*	−0.0078	−0.0648	0.0167	0.0481	−0.0545	0.0235	0.026	0.1612*	−0.1538*	−0.0397	0.029	−0.0448	0.0241	0.1437*	−0.3225*	0.0596	0.2586*	0.2421*	1
33	Decision Authority (subscale of Job Control)	0.1870*	−0.1546*	0.0051	0.1841*	−0.1287*	−0.0603	−0.2847*	0.4933*	−0.3529*	−0.2866*	−0.2677*	−0.2597*	−0.0316	0.045	−0.0358	−0.0126	−0.0516	−0.0311	−0.0431	0.0708	0.0727*	−0.1286*	0.0601	0.0488	−0.0422	0.0021	0.1083*	−0.021	0.0295	−0.0108	−0.0248	−0.0724*	1
34	Analysts	0.0188	−0.0076	−0.0109	−0.0043	0.0285	−0.0347	−0.037	0.0613	−0.0038	0.0289	0.0302	0.0188	0.1243*	0.0223	−0.0821*	−0.0393	0.0277	−0.049	0.0036	0.0408	0.1299*	−0.1424*	−0.0076	0.0731*	−0.0186	−0.0483	0.1443*	−0.0255	0.0383	−0.0158	0.0812*	0.0323	0.0044	1
35	Database Administrators	0.0761*	−0.1142*	0.0760*	0.1197*	−0.1300*	0.0265	0.0782*	0.1169*	−0.0242	−0.0238	−0.0451	−0.0523	−0.0695	−0.0616	−0.0159	−0.0535	0.0724*	−0.0534	−0.0476	0.0956*	0.0153	−0.0011	−0.0216	−0.0333	−0.0522	0.0920*	−0.0112	0.0235	−0.0196	−0.0024	−0.1581*	0.0048	0.0468	−0.0885*	1
36	Software Developers	−0.0214	0.0244	−0.0103	−0.0904*	0.1006*	−0.0235	0.018	−0.0806*	0.0795*	−0.015	0.0009	−0.0558	−0.1757*	0.0910*	0.1762*	0.1114*	−0.0352	0.0522	0.0508	−0.0976*	−0.1290*	0.2072*	−0.0790*	−0.1752*	0.1156*	0.0339	−0.1899*	0.0492	−0.0365	−0.0098	−0.0624	−0.0872*	−0.0128	−0.2760*	−0.1645*	1
37	Production Support/Operations	0.0426	−0.0445	0.0146	0.0989*	−0.1618*	0.0991*	−0.0543	0.0115	−0.026	−0.0041	−0.0535	−0.0323	−0.0278	0.01	−0.0447	−0.0494	−0.0244	−0.0086	0.0201	−0.0121	−0.0018	−0.0877*	0.1182*	0.0734*	−0.0954*	0.04	0.0955*	0.0277	0.0226	−0.0522	−0.0897*	−0.0275	0.0778*	−0.0873*	−0.052	−0.1622*	1
38	Project Managers	−0.0035	−0.0159	0.027	0.0259	−0.0067	−0.0246	0.0768*	−0.1520*	0.0915*	0.1164*	0.0516	0.1080*	0.1077*	−0.1061*	−0.1053*	−0.0186	0.039	0.0538	−0.0442	−0.0049	−0.0402	0.0291	0.0222	−0.003	0.0386	−0.0418	−0.0211	−0.0928*	0.0026	0.0901*	0.1221*	0.1097*	−0.1425*	−0.1679*	−0.1001*	−0.3122*	−0.0987*	1
39	Quality Assurance	0.0048	−0.0011	−0.0041	−0.0779*	0.0468	0.0363	−0.0422	0.0746*	−0.1312*	−0.0952*	−0.0445	−0.0118	0.1501*	−0.0198	−0.0023	−0.0474	−0.0174	−0.0242	0.0174	0.0046	−0.0404	−0.0308	0.1013*	0.1245*	−0.0549	−0.0555	0.0066	0.0404	−0.0163	−0.0229	−0.0275	−0.0063	0.0147	−0.1630*	−0.0972*	−0.3029*	−0.0958*	−0.1843*	1
40	System Engineers/Architects	−0.0687	0.0972*	−0.0601	0.0467	−0.0254	−0.0257	−0.0443	0.0566	−0.0366	−0.0153	0.0241	0.0097	−0.0940*	0.0205	−0.0103	0.0201	−0.0343	−0.0176	−0.0284	0.0441	0.1341*	−0.0942*	−0.0774*	0.0193	−0.0337	0.0204	0.0879*	−0.0194	0.0267	−0.0095	0.0651	−0.0079	0.0751*	−0.1432*	−0.0854*	−0.2662*	−0.0842*	−0.1620*	−0.1572*	1
41	Higher Job Level (Leads vs Lower-Level Contributors)	−0.0393	0.014	0.0256	−0.0162	0.0882*	−0.1038*	0.1291*	−0.0639	0.0971*	0.1401*	−0.028	−0.0198	−0.0076	0.0994*	0.0212	0.062	0.0099	−0.1022*	0.0229	0.0697	0.0859*	−0.0287	−0.0914*	−0.0894*	−0.0264	0.1159*	−0.0046	−0.1629*	0.0698	0.0878*	0.1336*	0.1612*	0.0674	−0.0004	0.0552	−0.0013	−0.0613	0.0757*	−0.2030*	0.1439*	1
42	Ln(Annual Salary)	0.0129	−0.0323	0.0315	−0.0402	0.1183*	−0.1147*	0.1683*	−0.0441	0.0985*	0.1160*	0.0062	−0.0187	−0.1553*	0.1467*	−0.0105	0.053	0.0029	−0.2265*	0.0471	0.1582*	0.1036*	−0.0253	−0.1222*	−0.2048*	0.026	0.1656*	−0.011	−0.2421*	0.1067*	0.1273*	0.1290*	0.1578*	0.0339	−0.0742*	0.1120*	0.2048*	−0.1744*	−0.0632	−0.2735*	0.1910*	0.6299*	1
43	Organizational Tenure in Years	0.1212*	−0.0798*	−0.026	0.1508*	−0.1240*	−0.023	−0.0126	0.1428*	−0.2393*	0.0274	−0.0529	−0.0554	0.1992*	0.0423	−0.2291*	−0.1482*	0.0736*	−0.3713*	−0.1465*	0.4813*	0.1867*	−0.3295*	0.1533*	0.4501*	−0.2115*	−0.1858*	0.3541*	−0.063	0.0225	0.0388	0.0005	0.0941*	0.1540*	0.0711	0.0532	−0.0773*	0.0012	−0.0575	0.0581	0.0033	0.0712	0.0829*	1
44	Interviewed after Merger Announcement and Geograpl	ic-0M.0o2v0e5	0.0717*	−0.0794*	0.1394*	−0.1051*	−0.0348	−0.1243*	0.0687	0.0357	−0.0958*	−0.0604	−0.0419	0.0156	−0.0332	0.0135	−0.0347	0.0645	−0.021	−0.0204	0.0392	0.0388	−0.0049	−0.0518	0.0729*	−0.0958*	0.041	0.0374	0.1158*	−0.0453	−0.0671	0.0902*	−0.0708	0.0478	−0.0513	−0.02	−0.1305*	−0.0636	0.0925*	−0.1283*	0.3344*	0.1211*	0.0731*	0.0636	1
45	High Job Insecurity	−0.0109	−0.0343	0.0621	0.0306	−0.0517	0.033	0.059	−0.0628	0.1359*	0.0716*	0.0730*	0.1260*	0.0572	−0.0036	−0.0541	−0.0222	0.0866*	−0.1792*	0.0783*	0.0844*	−0.0041	−0.0736*	0.1030*	0.1595*	−0.0973*	−0.04	0.0966*	−0.0165	0.0062	0.0099	0.0913*	0.0492	−0.1098*	0.0165	−0.0404	−0.0830*	0.0278	0.0455	−0.007	0.0686	−0.0439	−0.0212	0.0990*	0.2049*	1