Development and Initial Validation of a Scale to Measure Cognitive Demands of Flexible Work

Prem, Roman; Kubicek, Bettina; Uhlig, Lars; Baumgartner, Vera; Korunka, Christian

doi:10.3389/fpsyg.2021.679471

ORIGINAL RESEARCH article

Front. Psychol., 10 September 2021

Sec. Organizational Psychology

Volume 12 - 2021 | https://doi.org/10.3389/fpsyg.2021.679471

Development and Initial Validation of a Scale to Measure Cognitive Demands of Flexible Work

$\nRoman Prem &#x;$ Roman Prem¹^*^†

Bettina Kubicek¹^†

Lars Uhlig^1,2

Vera Baumgartner²

Christian Korunka²

¹Institute of Psychology, Faculty of Natural Sciences, University of Graz, Graz, Austria
²Department of Occupational, Economic and Social Psychology, Faculty of Psychology, University of Vienna, Vienna, Austria

With globalization, digitalization, and the spread of information and communication technologies, rules regulating work have been softened or completely abolished. Consequently, employees face additional cognitive demands to plan, structure, and coordinate their work. To capture these demands of contemporary work, we constructed and initially validated the Cognitive Demands of Flexible Work (CODE) scale. The scale comprises four subscales (i.e., structuring of work tasks, planning of working times, planning of working places, and coordinating with others). We initially validated the scale in three independent studies (overall N = 1,129) in German and English. Confirmatory factor analyses supported the four-factor structure, as well as scalar invariance, of the different language versions. Moreover, the subscales showed convergent and divergent validity with related constructs such as requirements for problem solving or autonomy. The criterion validity for emotional exhaustion, engagement, positive work rumination, negative work rumination, and problem-solving pondering suggested that cognitive demands of flexible work can be construed as challenge demands. However, relationships with emotional exhaustion were not significant. Overall, the CODE scale was shown to be a reliable and valid instrument to measure cognitive demands of flexible work.

Introduction

The organization of work and the associated working conditions change according to societal and technological developments (Allvin and Movitz, 2017; Flecker et al., 2017). Over the last decades, globalization, individualization, and digitalization have changed working life and resulted in increasingly flexible work environments in terms of when, where, and how work is conducted (Putnam et al., 2014). Moreover, managerial strategies have shifted the responsibility for regulating work from management to employees (Allvin et al., 2011). These developments have led to additional cognitive demands for employees: With more flexibility and responsibility, cognitive demands for structuring tasks arise. In addition, the opportunity to work flexibly in terms of time and place (e.g., Wessels et al., 2019) leads to cognitive demands for planning when and where work is conducted. Furthermore, the need to arrange work with coworkers and clients who may not be present at the same time and place entails cognitive demands for coordinating with others.

Despite the increasing prevalence of cognitive demands of flexible work, there is still no measure that captures the extent of these specific demands to which employees are exposed. Established scales are either too unspecific or have a different focus. For example, the Work Design Questionnaire (WDQ; Morgeson and Humphrey, 2006), one of the most comprehensive measures of work characteristics, covers three facets of autonomy, namely freedom in (1) work scheduling, (2) decision-making, and (3) work methods. However, these refer to decision possibilities and do not measure decision requirements and the specific cognitive demands that stem from such requirements (Frese and Zapf, 1994; Allvin et al., 2011). The WDQ (Morgeson and Humphrey, 2006) also measures initiated and received interdependence but not the demands for coordination with others. Similarly, established measures of flexible work (e.g., Shockley and Allen, 2007) focus either on the availability or use of temporal and spatial flexibility. These measures do not capture the demands for work-time and workplace planning or coordinating work with others that may result from workplace flexibility (Allvin et al., 2011).

Some existing scales deal more specifically with flexible or deregulated work, but these scales have different foci and usually cover other aspects of changing working conditions, such as demands for self-directed career development, learning, and effort regulation. The Intensification of Job Demands Scale (Kubicek et al., 2015), for example, captures intensified job-related planning and decision-making demands. The focus, however, is on the intensification of these demands over a longer period as a result of social acceleration. The Flexibility Requirements Scales (Höge, 2011) focus on individually perceived expectations that an organization conveys to its employees regarding self-directed behavior related to flexibility but not on cognitive demands that could result from these expectations. For example, the requirements for temporal flexibility subscale includes items regarding the employer's expectation that employees work overtime or are flexible as far as their working hours are concerned rather than the cognitive demands of planning one's working times. Finally, the Job Design Demands Scale (Dettmers and Bredehöft, 2020) touches on the idea that more autonomous working conditions lead to additional demands for employees that include structuring and organizing one's work on one's own. This can be regarded as an important first step in providing measures that capture the cognitive demands related to modern working conditions. However, the Job Design Demands Scale does not include measures that focus on planning working times and working places or coordinating with others in flexibly organized work.

Given this lack of available instruments, the goal of the present study is to develop and initially validate a measure of cognitive demands of flexible work. Such a measure is needed because with the changing nature of work, different working conditions prevail, and cognitive demands have become an important feature of flexible work that—to varying degrees from job to job—is relevant to many modern workplaces. An instrument that captures the cognitively demanding aspects of flexible work can shed light on the processes that link flexible work with negative outcomes in employees and help to explain findings on negative or neutral effects of flexible work on employee well-being (Ter Hoeven and van Zoonen, 2015). Moreover, the instrument could help to advance scholarly knowledge about the extent of these demands in today's work settings and improve our understanding of the adverse and favorable effects of modern working conditions on employees' well-being, motivation, and learning (Grant et al., 2011). Without a reliable and valid measure, we are unable to provide a solid empirical basis to integrate these demands into established theoretical models such as the job demand—control model (Karasek and Theorell, 1990), the job demands—resources theory (Bakker and Demerouti, 2017), or the challenge—hindrance stressor framework (LePine et al., 2005).

To advance research and theory on contemporary working conditions, we sought to develop a measure that we call the Cognitive Demands of Flexible Work (CODE) scale. To validate the scale, we will assess construct validity and reliabilities and show that the factor structure of the CODE subscales is the same across German and English versions. We will further test convergent and divergent validity with other job-related factors, as well as criterion validity with work outcomes. Finally, we will investigate whether the means of the CODE subscales differ between groups of employees that should be exposed differently to flexible work environments.

Cognitive Demands of Flexible Work

The last decades have brought significant growth in globalization and global operations (Castells, 2010), as well as enhanced use of information technology (Cascio and Montealegre, 2016). As a consequence of these transformations, work has become more flexible in various respects reflected in changes in the regulatory conditions of work. According to Allvin et al. (2011), “rules and norms that have traditionally defined and directed work, in various manners and to varying degrees, are now being softened up or simply abolished” (p. 218).

Although the term “flexible work” is sometimes used to refer to less-regulated ways of how labor is contractually organized (e.g., crowd work, gig work, and contingent work), we will focus in the following on the aspects of flexible work that reflect deregulated working conditions in terms of how work is conducted. Many employees in today's companies and organizations are exposed to different forms of flexible work, such as flexible working hours, telework (e.g., DiMartino and Wirth, 1990), and an increased projectification of organizations (e.g., Midler, 1995).

Overall, a growing number of jobs can be described as low-regulated or unregulated, in the sense that employees are expected to organize and plan their work more or less on their own. For example, in flextime and flexplace arrangements, the time and place of work are no longer solely determined by the organization. Further, with the increased use of projects and project-like work forms (e.g., agile forms of work organization), there has been a shift from standard operating procedures to goal-oriented performance, where work activities and coordination with others are also no longer determined only by the organization.

Such deregulation of work does not necessarily mean that control in organizations has become irrelevant. Rather, employees, instead of management, are required to take control. Thus, control has become increasingly indirect by being transferred to the employees (Flecker et al., 2017). This shift from direct control toward indirect control has created flexibility regarding how, when, where, and with whom people work. In addition, it has increased employees' responsibility for regulating their work (Allvin et al., 2011).

Although flexibility is associated with more discretion for employees (Gajendran and Harrison, 2007), it also puts additional cognitive demands on them because it increases the decisions to be made (Allvin et al., 2013; Höge and Hornung, 2015; Dettmers and Bredehöft, 2020). In flexibly organized work, employees not only receive more decision latitude but are oftentimes also required to use it efficiently, therefore facing additional job demands. In a way, flexible work entails aspects of job demands and job control (cf. Karasek and Theorell, 1990). Although choice is important for people and allows us to control our lives, having more choices is not always beneficial. As Schwartz (2016) points out in his seminal book The Paradox of Choice, a variety of options can overburden people and take up time and energy that should be devoted to other matters. In this sense, flexible work creates new demands for employees to self-organize and self-regulate their work activities (Pongratz and Voß, 2003; Allvin et al., 2013; Höge and Hornung, 2015). With fewer directives from management and higher individual responsibility to regulate work, employees must devote time and energy to structuring, planning, and coordinating their work. In line with this argument, chronic flextime use has been shown to require high levels of self-organization and coordination because employees cannot rely on an established daily temporal routine (Spieler et al., 2017). Moreover, initial evidence suggests that at least some of the demands that more flexible and autonomous work entails could also help to explain the adverse effects on burnout and job-related cognitions during leisure time (Dettmers and Bredehöft, 2020).

To conceptualize cognitive demands of flexible work, we rely on the areas of flexible or deregulated work that Allvin et al. (2011) put forward: performance, time, place, and social relations. Hence, we suggest that cognitive demands of flexible work entail structuring work tasks, planning (working times and working places), and coordinating work with others. Such cognitive demands of flexible work require mental effort to process information, solve problems, and make decisions regarding ordering work tasks and monitoring work progress, scheduling when and where work tasks should be dealt with, and coordinating with supervisors, coworkers, and clients. In the following we will describe each sub-dimension of cognitive demands of flexible work in more detail.

Structuring of Work Tasks

With the deregulation of work, the logic of corporate labor control has changed. Direct control, characterized by work tasks and the working order that stipulate the process of work being defined and prescribed by management, is reduced, and requirements for autonomous planning and structuring work increase (Pongratz and Voß, 2003; Kubicek et al., 2015). Employees are increasingly expected to take a self-directed and outcome-oriented approach in fulfilling their tasks and achieving their work goals (Höge and Hornung, 2015). Consequently, employees must individually define, structure, and take responsibility for their performance and the completion of work tasks (Allvin et al., 2011; Dettmers and Bredehöft, 2020). Cognitive demands for structuring work refer to requirements on the part of the employee to not only define individual work steps but also determine the ordering of the work steps and monitor the progress of work.

Planning of Working Times

The planning of working times concerns the temporal aspects of work arrangements. With the increase in temporal flexibility, fixed working hours are less common. This trend is evident, for example, in the decline of classic “nine-to-five jobs” in favor of flexible and trust-based working hours (Höge and Hornung, 2015). In such working arrangements, employees must individually decide and plan when, for how long, and how much they work (Hill et al., 2008). This not only includes deciding the start and end points of a working day but also includes being able to adjust the length of the working day. Often, employees have to keep track of their working hours and are expected to offset overtime and missing hours over longer periods while adjusting to workload demands. Thus, temporal flexibility increases the number of decisions that employees must make (Allen et al., 2013). Hence, cognitive demands for work time planning include having to decide on one's own when to start, pause, and end one's workday, how long one works on workdays, and when one plans to take breaks.

Planning of Working Places

In analogy with work time planning, the planning of working places concerns the spatial aspect of work arrangements. With advances in information and communication technologies, organizations have moved from using traditional offices with permanent workplaces to more flexible forms of working that enable many employees to work from different working places inside (e.g., project rooms, open office space, or silent areas, e.g., Wohlers and Hertel, 2017) and outside the employers' premises (e.g., working from home, co-working spaces, or on the go). In the absence of a predefined workplace, employees have to decide individually where to work and which work venue best fits their tasks and needs (Wessels et al., 2019). Cognitive demands for workplace planning thus entail planning and deciding where to work on certain tasks, ensuring that all relevant work materials are available, and adjusting the workplace to the specific requirements of the task and individual needs.

Coordinating With Others

In traditional and bureaucratically ruled organizations, cooperation between individuals is more or less specified by their organizational positions and facilitated through shared working hours and places. In flexible work, however, conditions for cooperation are to a large extent the responsibility of the individual (Allvin et al., 2011). Instead of relying on a given group of colleagues and a formal hierarchy, employees must establish and maintain the social contacts necessary for their work themselves and must coordinate their work with others. They must independently initiate and organize cooperation with the people needed to carry out the task at hand. This requirement may even become more demanding as colleagues and supervisors also work flexibly, which reduces shared working hours and places and makes spontaneous communication less likely (Hinds and Mortenson, 2005). Hence, cognitive demands for coordinating work with others entail having to agree upon work schedules, as well as the procedures and approaches needed to accomplish work tasks.

Study 1: Scale Development, Reliabilities, and Construct Validity

The goal of the first study was to develop a scale to operationalize cognitive demands of flexible work that require employees to structure their work tasks, plan their working times and working places, and coordinate with others. Further, this study aimed to gather information on the construct validity of the newly developed scale.

To get a better understanding of how cognitive demands to structure, plan, and coordinate manifest in employees' working lives, we initially conducted four focus groups with employees working in flexible work environments. We asked participants to identify aspects of flexible work that they perceived as demanding. The focus group participants were working at different office locations of a large multi-national corporation that offered flexible working times and the option to work from home and in which employees were often involved in various projects with changing teams. The participants were recruited via a contact person from the human resource department. Participation was voluntary, and a small donation was made to a local charity project for each participant. Each group consisted of five to eleven employees with and without leadership responsibilities. At the end of each focus group, the results were summarized and clustered by a trained moderator and discussed again within the group.

Based on the information from the focus groups and the literature, we developed a pool of potential items. Because we intended to measure demands from the job rather than degrees of latitude or autonomy, we made sure that items were worded in terms of requirements (e.g., “My job requires me to …” or “At work, I have to …”). All items were positively worded to avoid artificial factor solutions, which could occur when including negatively worded items alongside positively worded items within the same scale (Dalal and Carter, 2015). Overall, we initially created an item pool of 24 items (i.e., 6 items for each of the 4 subscales) that was formulated in German.

We expected that all four subscales of the CODE scale would share positive relationships with other cognitive demands, such as information processing and problem solving (Hypothesis 1). Information processing at work refers to requirements to monitor and process data or other information in general, whereas problem solving is characterized by more specific requirements to actively process information, come up with ideas and solutions to solve non-routine problems, and deal with errors (Morgeson and Humphrey, 2006). Structuring, planning, and coordinating work will, in most cases, require action regulation at the intellectual level (cf. Zacher and Frese, 2018). Therefore, employees confronted with cognitive demands of flexible work should more often be required to monitor and process information about the progress of their work, how their schedule and work venue fit their work tasks, and how their work is aligned with that of their coworkers and other people at work. Moreover, when identifying discrepancies or inconsistencies, they will have to actively solve problems and come up with solutions.

Moreover, we expected that the subscale for the structuring of work tasks would be positively related to decision-making autonomy and work methods autonomy (Hypothesis 2). Both decision-making autonomy and work methods autonomy can be described as facets of general job autonomy that reflect how much freedom, independence, and discretion employees have at work (Morgeson and Humphrey, 2006). We assumed that possibilities for employees to make decisions as part of their work and/or in choosing their work methods would be necessary for cognitive demands related to the structuring of work to arise.

Further, we expected that the subscale for the planning of working times would be positively related to work scheduling autonomy and the availability of flextime (Hypothesis 3). Work scheduling autonomy is also considered an aspect of general job autonomy (Morgeson and Humphrey, 2006), whereas flextime refers to flexibility in the timing of work typically characterized by the possibilities of employees to vary the start and end times, as well as the length, of their workdays (Shockley and Allen, 2007; Kattenbach et al., 2010). We assumed that it would be necessary for employees to have autonomy regarding their work scheduling and/or to be provided with flextime possibilities for cognitive demands regarding the planning of working times to arise.

Similarly, we expected that the subscale for the planning of working places would be positively related to the availability of telework possibilities to work from home and the availability of telework possibilities to from other locations outside the employer's premises (Hypothesis 4). Both working from home and telework from other locations can be considered aspects of flexplace that describe alterations in the physical boundaries around work (Munsch et al., 2014; Thompson et al., 2015). We refer to working from home as the possibility to work from one's place of residence, whereas telework from other locations is considered working from locations other than one's place of residence that are outside the employers' premises (e.g., mobile telework). We assumed that for cognitive demands relating to the planning of workplaces to arise, it would be necessary for at least some spatial flexibility to be available to employees.

Finally, we expected that the subscale for coordinating with others would be positively related to the interdependence of tasks at work (Hypothesis 5). Interdependence can be described as the extent to which a job is connected with that of others. It is oftentimes differentiated into initiated interdependence, i.e., the extent to which others depend on one to finish one's work, and received interdependence, i.e., the extent to which one's work depends on others to finish their work (Morgeson and Humphrey, 2006). We assumed that interdependence would be necessary for cognitive demands with regard to coordinating with others to arise.

Method

Procedure

Data on the German CODE scale were collected using an online questionnaire. The procedure and materials of this study did not undergo examination by an ethics committee, as the measures and procedures of our study followed standard procedures in applied psychological research, and we did not touch on sensitive topics (e.g., sexual orientation). Our protocol fully complied with the ethical principles of psychologists and the code of conduct of the American Psychological Association (2017). Participation was voluntary, and participants were free to withdraw whenever they wanted.

Sample

The sample consisted of employees from Germany working at least 30 contractual hours per week that were accessed via an ISO 26362-certified online panel (www.respondi.com). This panel recruits its participants via offline and online methods and ensures high quality through minimizing participation frequency and focusing on intrinsic motivation instead of financial dependency.

In total, 303 employees participated in study 1. About half (51.5%) of the sample was female. The average age of the participants was 43.13 years (SD = 11.11); they worked on average 41.70 h per week (SD = 6.09) and had worked for an average of 11.37 years (SD = 10.11) for their current company. About half (49.2%) of the participants held an academic degree.

Measures

Participants responded to the initial set of 24 items generated for the CODE scale. They were asked to indicate on a 5-point scale how the statements of the generated items applied to their work in general (1 = not at all to 5 = fully).

Most job-related factors for evaluating construct validity were measured using the German translation of the WDQ of Stegmann et al. (2010). Information processing and problem solving were each measured with 4 items, and decision-making autonomy, work methods autonomy, work scheduling autonomy, initiated interdependence, and received interdependence were each measured with 3 items. All WDQ items were measured on 5-point rating scales (1 = strongly disagree to 5 = strongly agree). The English version of the items can be found in the Appendix of Morgeson and Humphrey (2006).

Availability of flextime was measured with two self-created items (“I can decide on my own when to start and end a working day” and “I can decide on my own how many hours I work on a given day”) on a 7-point rating scale (1 = does not apply at all to 7 = completely applies). As both items were highly correlated (r = 0.85), we combined them to create a score for the availability of flextime.

Availability of working from home and availability of telework from other locations were each measured using a single self-created item. Participants were asked to indicate whether their employer allowed them to work from home or other locations outside their employer's premises. If they indicated that they were allowed to do so, they were also asked to indicate for how many hours per week they were allowed to work from home or other locations outside their employer's premises. We divided these numbers by the contractual working hours to get percentage scores that indicated how much of the participants' working time was available to work from home or telework from other locations.

Results

Selection of Items Based on Exploratory Factor Analyses

We conducted exploratory factor analyses in Mplus 8.3 (Muthén and Muthén, 1998–2017) on the initial set of items to shorten the subscales and select items for the final version. Given that the changes in chi-square values between models with different numbers of non-orthogonal factors supported our assumption of four correlated subscales, we relied on the four-factor solution for selecting the final items. Overall, we selected 12 items (i.e., 3 items for each of the 4 subscales) for the final version of the scale that adequately loaded on their respective factor (i.e., factor loading > 0.60), did not show substantial cross-loading (i.e., cross-loadings on other factors <0.10), and logically fitted the relevant construct best (cf. Supplementary Table 1).

Reliability and Factor Structure

To check for the internal consistency of the final scale, we calculated Cronbach's alphas for all four subscales (shown in the diagonal of Table 1). Cronbach's alphas ranged between 0.87 and 0.92 and thus were all satisfactory (Nunnally and Bernstein, 1994).

TABLE 1

Table 1. Descriptive statistics, Cronbach's alphas, and correlations for study 1.

Next, we analyzed the factorial structure of the CODE scale by using confirmatory factor analysis (CFA) techniques based on maximum likelihood estimation in Mplus 8.3 (Muthén and Muthén, 1998–2017). Table 2 shows the factor loadings of the items, as well as the means, standard deviations, and corrected item-total correlations. The four-factor model with the items of each subscale loading on separate factors showed an excellent fit (cf. Hu and Bentler, 1999), as the comparative fit index (CFI) and Tucker-Lewis index (TLI) were above 0.95, the root mean square error of approximation (RMSEA) was below 0.06, and the standardized root mean squared residual (SRMR) was below 0.08 (compare Table 3).

TABLE 2

Table 2. Item wording, means and standard deviations, corrected item-total correlations, and standardized factor loadings from confirmatory factor analysis.

TABLE 3

Table 3. Fit indices for measurement models in and measurement invariance tests across the samples of the three studies.

We also compared the four-factor measurement model against different three-factor models, in which we set the items of two subscales to load on the same factor. We calculated six different three-factor models to test all possible combinations of any two subscales loading on the same factor. The best-fitting three-factor model is also shown in Table 3. To compare this model to the four-factor model, we calculated chi-square differences, and because chi-square-difference tests are known to be biased in larger sample sizes, we also calculated CFI differences. A CFI difference between two models of 0.01 or smaller is considered to indicate that both models fit equally well (Cheung and Rensvold, 2002). The best-fitting three-factor model (i.e., the model with the items of structuring of work tasks and planning of working times loading on the same factor) fitted worse than the four-factor model based on a significant chi-square-difference test and showed a CFI difference considerably larger than 0.01 (compare Table 3). Thus, the four-factor structure of the CODE scale was initially supported.

Convergent and Divergent Validity

To test for the convergent validity of the CODE scale, we investigated the correlations of the four subscales with other working conditions that were measured with established scales. To make sure that we did not capitalize on chance when testing our hypotheses, we divided the threshold for significance by the number of correlations being tested for the respective hypothesis. When testing Hypothesis 1, the Bonferroni-corrected α was 0.05/8 = 0.00625 because Hypothesis 1 included 8 correlations. When testing Hypotheses 2 through 5, the Bonferroni-corrected α was 0.05/2 = 0.025 because Hypotheses 2 through 5 each included 2 correlations.

All subscales, except planning of working places, showed significant positive associations with information processing (compare Table 1) that remained significant after Bonferroni correction. Moreover, in line with our assumption, all four subscales showed significant positive associations with problem solving (compare Table 1) that again remained significant after Bonferroni correction. Thus, Hypothesis 1 was largely supported.

Further, in line with Hypotheses 2 through 5, structuring of work tasks showed significant positive associations with decision-making autonomy and work methods autonomy; planning of working times showed significant positive associations with work scheduling autonomy and the availability of flextime; planning of working places showed significant positive associations with the availability of working from home and the availability of telework from other locations; and coordinating with others showed significant positive associations with initiated interdependence and received interdependence (compare Table 1). All of these correlations remained significant after Bonferroni correction, with the exception of the correlation between planning of working places and the availability of working from home. Thus, Hypotheses 2, 3, and 5 were fully supported, and Hypothesis 4 was partly supported.

To show that the four subscales were not only significantly related to similar constructs but also measured constructs that were still distinct from these similar constructs (i.e., to test for divergent validity), we conducted a series of confirmatory factor analyses (compare Table 4). For each of the four subscales, we calculated a model that included the focal subscale, the scales for information processing and problem solving, and the other two respective measures that had been used to show convergent validity above. For each subscale, we further estimated four four-factor models that each combined the focal subscale with one of the four other measures in the analysis. The results from the chi-square-difference tests and CFI differences indicated that for all four subscales the five-factor models fitted the data best. This shows that all subscales measured constructs that were empirically separable from similar constructs (compare Table 4).

TABLE 4

Table 4. Fit indices of models to test whether each subscale measures a construct that is different from other similar constructs.

Study 2: English Translation and Criterion Validity

The goal of the second study was to translate the newly created scale for measuring cognitive demands of flexible work from German to English and check for invariance between the two versions. Further, this study aimed to gather information on the criterion validity of the newly developed scale.

Drawing on challenge—hindrance stressor literature (e.g., LePine et al., 2005; Crawford et al., 2010), we assumed that the cognitive demands of flexible work are to be considered challenge demands. In the challenge—hindrance stressor literature, it is asserted that job demands can be differentiated into challenge demands and hindrance demands (e.g., Cavanaugh et al., 2000). Although both types of job demands require effort and are, therefore, associated with strain, their associations with motivational outcomes differ (e.g., Crawford et al., 2010). Challenge demands show positive associations with motivational outcomes, whereas hindrance demands share negative associations with motivational outcomes (e.g., Crawford et al., 2010). These differential relationships with motivation are usually explained by the assertion that individuals are likely to believe that the effort expended in coping with challenge demands will help them to attain work goals and other valued outcomes (e.g., LePine et al., 2005).

We expected the cognitive demands of flexible work to be associated with strain-related and motivational work outcomes. As structuring, planning, and coordinating work will, in most cases, require deliberate cognitive operations, such as information processing, decision-making, or problem-solving, based on action-regulation theory (cf. Zacher and Frese, 2018), we assumed that dealing with them will require cognitive effort. Because the conditions of work vary in flexible work environments, it is difficult for employees to use predefined schemas, learned procedures, or routines for dealing with the cognitive demands of flexible work. Rather, the cognitive demands of flexible work require conscious attention and analysis of the respective situation. Therefore, when dealing with cognitive demands of flexible work, employees expend cognitive effort and experience psychological costs (Hockey, 1997; Zacher and Frese, 2018). In the research on job demands, such psychological costs are usually operationalized with indicators of burnout, such as emotional exhaustion (e.g., feeling burned out or used up from one's work). With regard to motivational work outcomes, we focused on work engagement (i.e., feeling vigorous, dedicated, and absorbed during work). We assumed that the cognitive demands of flexible work would be positively related to work engagement, as the effort invested in these demands would likely be helpful in attaining work goals and provide opportunities to experience competence, autonomy, and connectedness (cf. Gagné and Deci, 2005). We expected that all four subscales of the CODE scale would share positive relationships with emotional exhaustion and engagement (Hypothesis 6).

Further, as previous literature has shown that flexibly organized work tends to blur the boundaries between work and non-work domains (e.g., Spieler et al., 2017), we expected that structuring, planning, and coordinating would also be associated with employees' cognitions during off-job times. Therefore, cognitive demands of flexible work should be associated not only with traditional indicators of strain and motivation but also with indicators of rumination during leisure time. Having to make decisions about how to structure, plan, and coordinate one's work should provide employees with both positive and negative experiences that cognitively occupy them during off-job time. For example, if a person manages to successfully structure a complex work task and create a plan for how to tackle it, they might reflect on this result, even when they are away from their work. Similarly, if they fail to come to an agreement with other people regarding a common approach, employees might be more likely to ruminate on their work experiences. Additionally, it seems rather likely that employees will ponder on unsolved problems that arose from structuring, planning, and coordinating their work and try to solve them during their leisure time. Consequently, we expected that all four subscales of the CODE would share positive relationships with positive work rumination, negative work rumination, and problem-solving pondering (Hypothesis 7).