Profiling a spectrum of mental job demands and their linkages to employee outcomes

Journal for Person-Oriented Research 2020; 6(1): 55-71 Published by the Scandinavian Society for Person-Oriented Research Freely available at https://journals.lub.lu.se/jpor and https://www.person-research.org https://doi.org/10.17505/jpor.2020.22046 55 Profiling a Spectrum of Mental Job Demands and their Linkages to Employee Outcomes Saija Maunoa,b & Jaana Minkkinena a Tampere University, Faculty of Social Sciences (Psychology), Finland b University of Jyväskylä, Faculty of Education and Psychology (Psychology), Finland Corresponding author: Dr. Jaana Minkkinen, Tampere University, Faculty of Social Sciences (Psychology), Kalevantie 5, FI-33014 Tampere, Finland, Tel: + 358 50 318 7671, Fax: + 358 3 213 4473, Email: [email protected] ORCID 0000-0002-9457-9599 To cite this article: Mauno, S., & Minkkinen, J. (2020). Profiling a spectrum of mental job demands and their linkages to employee outcomes. Journal for Person-Oriented Research, 6(1), 55-71. https://doi.org/10.17505/jpor.2020.22046 Abstract: Working life is becoming more mentally demanding and intense due to technological acceleration. The present study explored employees’ experiences of different mental job demands (MJDs) and their outcomes (job burnout, job performance, and meaning of work). We focused on intraand inter-individual variations and possible harmful combinations of MJDs, which we explored via latent profile analysis (LPA). To identify harmful combinations of MJDs, we also investigated how the profiles of MJDs related to the outcomes of interest. The study was based on a diverse sample of Finnish employees (n = 4,583). LPA showed that both intra-individual and inter-individual variation characterized MJDs as we identified five latent profiles of MJDs. The most harmful profile, which predicted the most negative outcomes (particularly job burnout), was characterized by employees’ scoring high on all MJDs. A profile characterized by low learning demands and moderate level of other MJDs was also a harmful combination in terms of outcomes. In contrast, a profile characterized by moderate level of learning demands and low level of other MJDs did not relate to negative outcomes. Altogether, the findings suggest that different MJDs may co-occur implying risks to employee well-being and performance. However, MJDs simultaneously form a complex spectrum that may differ within and between individuals. Keywords: mental job demands, work intensification, job burnout, job performance, meaning of work, latent profile analysis Contemporary working life is characterized by rapid technological acceleration in the form of increasing digitalization, robotization, and artificial intelligence, which are changing working conditions in many ways (see Chesley, 2014; Mustosmäki, 2017; Rosa, 2003; Paškvan et al., 2016). One hallmark of these changes is an intensification of work referring to work processes and work cultures, where the work effort required of employees has become more intense and efficacy-focused in terms of time and quality (e.g., Green, 2004; Kubicek et al., 2015; Mauno et al., 2019b; Mauno et al., 2020). In this study, we approach intensified working life from the perspective of mental job demands (henceforth MJDs), referring to a spectrum of recently identified mental job demands which have intensified and increased due to technological and structural changes in working life and also to empowerment-focused management practices (see more, Chesley, 2014; Galy et al., 2012; Kubicek et al., 2015; Rosa, 2003; Mauno et al., 2019b; Mauno et al., 2020). Specifically, we investigate whether Finnish blueand white-collar workers (N = 4.583) experience MJDs in qualitatively different ways. To achieve this, we first examine how different MJDs combine by analyzing latent profiles (LPA) of MJDs, the method which enables us to model MJDs as multi-faceted and complex phenomena at the intra-individual and inter-individual levels (Laursen & Mauno & Minkkinen (2020) Profiling a spectrum of mental job demands 56 Hoff, 2006; Muthén, 2001; Spurk et al., 2020). LPA enables to identify homogeneous and heterogeneous groups (profiles) of individuals as regards the phenomena of interest (here MJDs), revealing also typical and atypical configurations/patterns of the constructs (see Bergman & Lundh, 2015; Spurk et al., 2020). Second, as MJDs typically entail stressors for employees with detrimental employee outcomes (Chesley, 2014; Fletcher et al., 2018; Franke, 2015; Kubicek et al., 2015), we also examine whether and how the profiles of MJDs relate to certain employee outcomes, that is, job burnout, job performance, and meaning of work. These outcomes were selected as they represent qualitatively different consequences and profile differences in them would also validate the profiles of MJDs (supporting criterion validity) (see Spurk et al., 2020). The main contribution of our study is two-fold. First, our research model includes various self-rated MJDs (described below), which have so far been studied only rarely due to the novelty of these demands. Second, if studied at all, MJDs have typically been analyzed as separate constructs without paying attention to their potential integrated properties or inter-relationships (at intraand inter-individual level), which is focused here. Theoretical underpinnings of MJDs Overall, MJDs refer to the mental effort and thinking required at work to accomplish the (mental) tasks and to perform adequately at work (Galy et al., 2012; Zapf et al., 2014; Warm et al., 2008). However, no job demands occur in a vacuum but are typically inter-linked and additive (see e.g., Galy et al., 2012; Sweller, 1988). This may be particularly true regarding MJDs as such demands typically tax the same psychophysiological systems, e.g., shortand long-term memory, hence also causing cognitive load (e.g., Dillard et al., 2019; Sweller, 1988). Despite this systemic similarity, different cognitive load factors can be distinguished. Indeed, Galy et al. (2012) have distinguished three cognitive load factors, that is, task difficulty, time pressures, and arousal/alertness, each of these being embedded in cognitive load theory (CLT) (Sweller, 1988), which is also applicable in the context of work. Specifically, CLT suggests that heavy mental workload requires the individual to allocate extra (mental) resources, which, in turn, impairs information processing efficiency and performance, and can also be distressing and mentally draining (e.g., Dillard et al., 2019; Sweller, 1988). Moreover, cognitive load factors can be divided into intrinsic and extrinsic (Sweller, 1988). Task difficulty belongs to the former, whereas time pressures and arousal belong to the latter category, although this distinction is not so strict in reality (Galy et al., 2012). Actually, these cognitive load factors stand in reciprocal relation to each other, and their effects are mostly additive, that is, the more cognitive load factors co-occur, the more distressed an individual is (Galy et al., 2012). In line with this assumption, empirical studies have already shown that it is the interaction of these cognitive load factors which matters most. For example, Galy et al. (2012) showed that individuals’ performance and mental efficiency were poorer when both task difficulty and time pressures were high and when their alertness was low. Furthermore, other studies have found that cognitive load not only impairs our performance but is also distressing (Dillard et al., 2019). Inspired by these findings, it has been suggested that research should continue screening different cognitive load factors and their multiple outcomes (Galy et al., 2012). In the present study, cognitive load factors include five particular indicators of mental workload, which we introduce next. Defining the MJDs of the present study In the present study, the MJDs comprise five specific job demands, namely work intensification, intensified planningand decision-making demands in relation to one’s job or career, intensified skilland knowledge-related learning demands at work, illegitimate tasks and interruptions at work. We will evaluate these MJDs through employees’ cognitive appraisals/self-reports as employees’ cognitive appraisal of their work environment is decisive when assessing how the work environment affects employees’ well-being and performance (Lazarus & Folkman, 1984). All these MJDs are relatively new in work psychology and their self-report assessment has only recently been developed (see Fletcher et al., 2018; Kubicek et al., 2015; Semmer et al., 2015). Consequently, our study is one of the first attempts to investigate how a spectrum of perceived MJDs combines at two levels (intraand inter-individual levels). Intensified job demands (IJDs) refer to recently launched job stressors developed to characterize and assess the consequences of accelerated and intensified working life on employees’ appraised mental workload (Korunka et al., 2015; Kubicek et al., 2015; Paškvan et al., 2016; Mauno et al., 2019b; Mauno et al., 2020). IJDs are an offshoot of social acceleration theory (Rosa, 2003), which claims that our activities in all life spheres, including working life, have accelerated, and the primarily fueling phenomenon underlying this is technological acceleration. Consequently, IJDs are currently highly relevant mental job stressors as technological acceleration in the form of digitalization, robotization, and artificial intelligence renders working life more intense and mentally demanding (Chesley, 2014; Mustosmäki, 2017). Specifically, IJDs manifest as three following inter-related job demands: (1) work intensification, (2) intensified planningand decision-making demands in relation to one’s work and career, and (3) intensified knowledgeand skill-related learning demands. Work intensification describes the intensification of workload over time, including increased time-related demands throughout the working day, such as intensified pace of work, lack of breaks, and multitasking requirements at Journal for Person-Oriented Research, 6(1), 55-71 57 work (see Green, 2004; Franke, 2015; Kubicek et al., 2015; Paškvan et al., 2016). We define work intensification as one form of MJDs as working hard, performing multitasking, and skipping breaks require a lot of mental effort from an employee. In the framework of CLT (Galy et al., 2012; Sweller, 1988), work intensification corresponds to time pressures as an indicator of cognitive load (at work). Intensified joband career-related planning and decisionmaking demands refer to the increased requirements for employees to autonomously plan and pursue their work goals and daily work tasks (i.e., job-related demands) and to take greater individual responsibility for their career management and employability (i.e., career-related demands). Indeed, employees may experience increased autonomy as a requirement to make individual decisions on setting and achieving work-related goals too frequently or as a need to perform their work too independently overall. Moreover, freedom to make self-directed choices concerning one’s career development may impose excessive personal responsibility for employees to be able to maintain their attractiveness in the labor market (Korunka et al., 2015; Kubicek et al., 2015; Mauno et al., 2019b; Mauno et al., 2020). Such self-directness regarding working or career management might be stressful as it implies higher mental workload for employees. The acceleration characterizing contemporary working life may also increase employees’ experiences of mental workload in the form of intensified learning demands referring to a need to continuously update old information and acquire new work-relevant knowledge (Korunka et al., 2015; Kubicek et al., 2015; Paškvan et al., 2016; Mauno et al., 2019b; Mauno et al., 2020). Such pressures to adopt the latest professional knowledge exemplify the intensified knowledge-related learning demands. However, not only is there a need to constantly update one’s work-relevant knowledge, but also one’s skills, for example by learning new competencies that enable effective job performance in the face of intensified skill-related learning demands. Thus, learning demands are, by definition, mental demands to be included in the spectrum of MJDs. Viewed in the framework of CLT (Galy et al., 2012; Sweller, 1988), intensified joband career-related planning and decision-making demands and learning demands illustrate intrinsic task difficulty (at work) but also share some features of time pressures due to intensification. There is some empirical evidence to show that these IJDs are also sources of stress at work associated with impaired well-being and health (e.g., Franke, 2015; Kubicek et al., 2015; Paškvan et al., 2016; Mauno et al., 2019b; Mauno et al., 2020). In addition to these IJDs, workers may also experience other kinds of MJDs, to which we now turn. There may be tasks at work, which employees experience as inappropriate, irrelevant or unfair. Semmer et al. (2015) have called these tasks illegitimate tasks. Examples are when a nurse is required to write a report on a computer instead of caring for the patient, or when a teacher is struggling how to use new software instead of teaching students. Illegitimate tasks threaten employees’ work identity or core work roles and are thus self-threatening and often also include feelings of unfairness as expressed in the feeling that “I should not be doing this or nobody should be doing this”, thereby, constituting a source of stress for employees (Eatough et al., 2016; Ma & Peng, 2019; Semmer et al., 2015). Actually, there are two types of illegitimate task, namely those which are unreasonable and those which are unnecessary. The former refers to tasks that an employee perceives to be incompatible with his/her work role and which should be done by someone else, whereas the latter refers to tasks which are simply a waste of time and resources and nobody should be doing those (Semmer et al., 2015). We take the view that illegitimate tasks include cognitive load (Galy et al., 2012; Sweller, 1998) as they require complex cognitive appraisal and evaluation processes from an employee, thereby capturing the essence of intrinsic task difficulties (at work). Furthermore, they may also contain unwanted external stimulation, which is taxing an employee’s alertness/vigilance and also constitutes one hallmark of cognitive load (distracting attention from core tasks). Finally, illegitimate tasks may also contain a time pressure component of cognitive load as often core tasks need to be performed alongside with extra-role tasks. Illegitimate tasks have been found to relate to poorer well-being and job performance (see e.g., Eatough et al., 2016; Ma & Peng, 2019; Semmer et al., 2015), signifying that they are seriously taken job stressors. Moreover, it is possible that an ongoing “technological tsunami” at work may even increase illegitimate tasks as employees’ attention will be increasingly needed in technological aspects of the work, which they may consider illegitimate, especially if core work tasks require other kinds of attention or behavior, e.g. human interaction, care, or creative thinking. Interruptions at work have been defined in several ways, but this MJD typically refers to external or internal stimuli distracting a worker’s mental resources from the primary work task towards disruptive stimuli, thereby also inhibiting progress in the primary task (Jett & George, 2003; Fletcher et al., 2018). Examples at the workplace are various, but include at least distracting noises, smells, images, conversations, information flow, or computer problems that may distract employees’ attention from the core task at hand. The principles of effective work rely on employees’ ability to engage freely in the mental actions needed at work and to allow employees to focus on primary tasks without interruptions or distractions (Liebl et al., 2012; Sander et al., 2019). Viewed in the light of CLT (Galy et al., 2012; Sweller, 1988), interruptions clearly contain cognitive load as they typically include attention-split/vigilance difficulties, which again deplete an employee’s mental resources and efficiency (Hancock, 2017). Furthermore, interruptions may also involve time pressure, a core element of cognitive load, as work tasks need to be done in spite of interruptions. On this ground, interruptions are naturally Mauno & Minkkinen (2020) Profiling a spectrum of mental job demands 58 stressful, considering that they typically also inhibit progress in primary work tasks, which also may cause extra stress if the work goals are not achieved as expected (Sander et al., 2019; Seddigh et al., 2014). There is empirical evidence indicating that perceived interruptions at work are stressors resulting in poorer well-being and job performance (Fletcher et al., 2018; Liebl et al., 2012; Lin et al., 2013; Sander et al., 2019; Seddigh et al., 2014). We assume that the acceleration occurring in working life right now may increase interruptions at work as it encourages open offices, multitasking ideology, and global connectivity (Green, 2004; Sander et al. 2019), all of which may increase interruptions at (core) tasks, culminating ultimately in higher mental load at work. Aims and hypotheses The first aim of this study is to examine how the five above-described MJDs combine intraand inter-individual levels (via LPA) and reveal qualitatively different configurations at both levels (see Spurk et al., 2020). As MJDs form a spectrum of mental demands arising from cognitive load at work, we may expect at least some degree of interdependence. This assumption is also consistent with the CLT (Galy et al., 2012; Sweller, 1998), which argues that cognitive load factors (e.g., intrinsic and extrinsic cognitive properties of the tasks) may also co-occur or accumulate. Consequently, we hypothesize that we shall find a profile (group) of employees who score either high or low on all five MJDs defined above (H1). However, it is also possible to identify more diverse employee profiles in LPA; for example, those who score high on some dimensions of MJDs but low on others. LPA, like person-centered analysis methods more generally, are data-driven methods, implying that it is difficult to predict what kinds of profiles/clusters will emerge from the data, particularly if firm theoretical assumptions on profile characteristics are lacking. However, these more explorative data analysis methods allow us to better understand how different phenomena may combine within and between individuals (e.g., Muthén, 2001; Spurk et al., 2020). Actually, there are also theoretical reasons to expect individual variation in the profiles of MJD. Because stress appraisal is a crucial element in the stress process (see Brem et al., 2017; Lazarus & Folkman, 1984), there may be individual differences in the extent to which work characteristics (here MJD) are appraised as stressful or accumulating by an individual. Viewed in this light, LPA is one appropriate tool to explore typical and atypical configurations of job demands at intraand interindividual level (see also Spurk et al., 2020; Woo et al., 2018) The second aim of this study is to investigate how the profiles of MJDs relate to three specific employee outcomes, i.e., job burnout, job performance, and meaning of work. These selected outcomes also form important criteria for the profiles of MJDs; the profiles should show meaningful variation in the outcomes or otherwise their criterion validity might be insufficient (see Spurk et al., 2020). In this respect, we are particularly interested in identifying risk profiles of MJDs (co-occurrence of MJDs), which, in turn, should relate to negative employee outcomes (i.e., more burnout, poorer performance and meaning of work). Indeed, if MJDs are negative stressors at work, they should relate to negative employee outcomes, a proposition consistent with many job stress models (e.g., Karasek & Theorell, 1990; Siegrist, 1996; Warm et al., 2008; Zapf et al., 2014). As we expected to find a profile (group) of employees scoring high on all dimensions of MJDs (H1), we further hypothesize that belonging to this “high-risk group” would likely predict more job burnout, perceiving one’s job performance poorer and one’s work to be less meaningful (H2). Nevertheless, as already stated, it is equally possible to find more diverse configurations of MJDs as stress appraisal is also individualistic (Brem et al., 2017; Lazarus & Folkman, 1984). However, it is difficult to predict beforehand how these profiles would look like. Consequently, it makes no sense to pose hypotheses on their relations to employee outcomes. Materials and methods Participants and procedure The present study is part of a larger research project (IJDFIN) examining MJDs and employee outcomes in Finland. Participants were sampled via trade unions as, of all Finnish employees, 73% belonged to some trade union in 2017 (Ministry of Employment and the Economy, 2018). Data were collected during spring-summer 2018 from the Trade Union of Education (OAJ), the Industrial Union (TL), Service Union United (PAM), and Trade Union Pro (Pro). The participants were chosen from among currently working members on the register of each labor union using random sampling with a total of 5,000 individuals per union. Participation in the survey study was voluntary; participants were adults and no physiological or health data was gathered. The survey was filled out online and tested before data collection. A total of 4,583 respondents participated in the study (nOAJ = 2,434, nTL = 647, nPAM = 857, nPro = 645). The mean response rate was 24% (OAJ members 48%, TL members 14%, PAM members 19%, Pro members 13%). More women (69%) participated in the study (womenOAJ = 79%, womenTL = 26%, womenPAM = 75%, womenPro = 64%) than men, but compared to the trade unions’ respective memberships the distribution was significantly different only in TL and PRO. Over 50-year-olds were overrepresented for the OAJ and Pro in relation to membership (57% and 49% vs. 43% and 15% respectively), whereas under 20-year-olds and over 61-year-olds (2% and 4% vs. 9% and 15% respectively) were underrepresented for PAM and respondents under the age of 40 were underrepresented for TL (74% vs. 55%). The sample in our analyses consisted of those 3,294 em- Journal for Person-Oriented Research, 6(1), 55-71 59 ployees who had responded to five indicators of MJDs (i.e., work intensification, intensified joband career-related planningand decision-making demands, learning demands, illegitimate tasks, and interruptions at work). Of these respondents, 69% were women, their ages varied from 20 to 66 years (M = 46.8, SD = 11.4). A total of 51% were white-collar workers and 12% worked in managerial positions. Information about employees’ level of education, working hours in week, and type of employment contract are described under control variables (see more in next section). These control variables – in addition to gender, age, occupational group, and managerial position – were included in the analyses if they had significant bivariate correlations with a dependent variable in regression analyses (see more in Results). Measures IJDs were measured using the Intensification of Job Demands Scale developed and validated by Kubicek and colleagues (2015). Respondents were asked to assess changes in mental job demands in their work organization during the last five years (or less, if a participant had been working less than five years). It is noteworthy that as IJDs try to capture a societal process of acceleration occurring in particular job demands in recent years (Rosa, 2003), a time-frame of this scale focuses on perceived changes in IJDs that have occurred in the past (Kubicek et al., 2015; Mauno et al., 2019b; Mauno et al., 2020). In this study, we used the following three subscales of IJDs: 1) work intensification (WI) including five items (e.g., “…ever more work has to be completed by fewer and fewer employees”), 2) intensified job-related and career-related planning and decision-making demands (IJCPDs) including five items concerning intensified job-related demands (e.g., “one increasingly has to check independently whether the work goals have been reached”) and three items concerning intensified career-related demands (e.g., “one is increasingly required to maintain one’s attractiveness for the job market, e.g., through advanced education, networking”), 3) intensified learning demands (ILDs) including six items (e.g., “one has to update one’s knowledge level more frequently” and “one increasingly has to familiarize oneself with new work processes”). The response scale was a five-point Likert-scale (1 = not at all, 5 = completely), higher scores reflecting more frequent/higher intensified job demands (WI: M = 3.66, SD = 1.07; IJCPDs: M = 3.39, SD = .87; ILDs: M = 3.74, SD = 1.00). Cronbach’s alpha coefficients for WI, IJCPDs, and ILDs were .89, .88, and .95 respectively. Illegitimate tasks were assessed using eight items from the Bern Illegitimate Tasks Scale (Semmer et al., 2010). The scale includes four items describing unnecessary tasks (e.g., “Do you have work tasks to take care of which keep you wondering if they have to be done at all?”) and four items characterizing unreasonable tasks (e.g., “Do you have work tasks to take care of which you believe should be done by someone else?”). Answers were given on a five-point Likert scale (1 = never, 5 = always), higher scores reflecting more illegitimate tasks (M = 2.95, SD = .80). Cronbach’s alpha coefficient was .90. Interruptions at work was evaluated via distractions, which were assessed using five items measuring distractions from the Interruption Scale developed by Fletcher and colleagues (2018; e.g., “It was hard to keep my attention on my work because of distractions in my workplace”, “A noise or other distraction interrupted my workflow”.) The sub-scale of distractions was selected to describe interruptions at work as it indicated the most consistent relationships with the employee outcomes in a validation study (Fletcher et al., 2018). Answers were given on a six-point Likert scale (1 = never, 6 = very frequently), higher scores reflecting more distractions (M = 3.34, SD = 1.11). Cronbach’s alpha coefficient was .88. Job burnout refers to a health impairment in response to chronic stressors at work including the dimensions of exhaustion, cynicism, and (lower) professional efficacy (Maslach Schaufeli, & Leiter, 2001). In the present study, burnout was evaluated via job exhaustion and cynicism, both of which were assessed with three items from the Bergen Burnout Indicator-9, the reliability and validity of which have been shown to be high in Finland (Feldt et al., 2014; Salmela-Aro et al., 2011). The items were rated on a six-point Likert scale (1 = completely disagree, 6 = completely agree), higher scores reflecting greater job exhaustion (M = 3.29, SD = 1.19) and more cynicism (M = 2.78, SD = 1.28). Cronbach’s alpha coefficient for the exhaustion scale was .75 and for the cynicism scale .87. Job performance refers to employees’ behaviors and actions related to the goals of their work organization (Campbell, 1990). Job performance was operationalized via task performance, which was assessed with four items from the Individual Work Performance Questionnaire (e.g., “I was able to plan my work so that I finished it on time”; Koopmans et al., 2016). The items were rated on a fivepoint Likert scale (1 = rarely, 5 = always), higher scores reflecting better performance (M = 3.58, SD = .73). Cronbach’s alpha coefficient for the task performance scale was .79. Meaning of work refers to an individual interpretation of what work or the role of work signifies in the life context influenced by the social context (Rosso et al., 2010). In this study, meaning of work was assessed with four items from a positive meaning of work-scale based on the Work and Meaning Inventory Questionnaire (e.g., “I have found a meaningful career”; Steger et al., 2012). The items were rated on a seven-point Likert scale (1 = completely disagree, 7 = completely agree), higher scores reflecting more positive meaning of work (M = 5.29, SD = 1.27) Cronbach’s alpha coefficient was .90. 60 Table 1 Intercorrelations between the Study Variables Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1. WI – 2. IJCPDs .54*** – 3. ILDs .41*** .46*** – 4. Illeg. tasks .56*** .39*** .28*** – 5. Distractions .43*** .29*** .30*** .45*** – 6. Exhaustion .58*** .32*** .29*** .46*** .42*** – 7. Cynicism .31*** .16*** .02 .40*** .33*** .50*** – 8. Performance -.28*** -.08*** -.07*** -.30*** -.31*** -.39*** -.37*** – 9. Work meaning -.05** .03 .23*** -.17*** -.10*** -.16*** -.63*** .29*** – 10. Gender -.15*** -.06*** -.20*** -.02 -.18*** -.18*** .01 -.01 -.16*** – 11. White-collars .16*** .09*** .36*** .11*** .15*** .17*** -.17*** -.01 .48*** -.25*** – 12. Contract type .06** .01 .08*** .07*** .05** .03 .10*** -.03 -.09*** .05** -.17*** – 13. Manager .02 .07*** .05** .03 -.02 .04* -.04* .00 .09*** .05** .03* .09*** – 14. Education .15*** .16*** .35*** .13*** .18*** .16*** -.07*** -.01 .30*** -.19*** .62*** -.09*** .06*** – 15. Working hours .19*** .16*** .09*** .14*** .07*** .21*** .03 -.08*** .02 .14*** .05** .11*** .18*** .02 – 16. Age .03* .02 .21*** .03 .09*** -.01 -.03 .02 .16*** -.01 .21*** .26*** .06** .08** .04* Note. WI = work intensification; IJCPDs = intensified job-related and career-related planning and decision-making demands; ILDs = intensified learning demands; illeg. tasks = illegitimate tasks; gender: 0 = women, 1 = men; white-collars: 0 = no, 1 = yes; contract type: 0 = temporary employment contract, 1 = permanent employment contract; manager = managerial position: 0 = no, 1 = yes; education: 1 = further vocational qualification or matriculation examination certificate, 2 = specialist vocational qualification, 3 = higher vocational level qualification, 4 = polytechnic qualification or bachelor degree, 5 = university degree, 6 = university postgraduate degree; working hours = working hours in week. * p < .05, ** p < .01, *** p < .001, two-tailed. 61 Control variables of gender (0 = female, 1 = male), occupational group, type of employment contract, managerial position, education, hours worked per week, and age were included in the regression analyses when there was a significant correlation between the control variable and the dependent variable (see Table 1). Occupational group was coded as 0 = not white-collar worker (49%), 1 = white-collar worker (51%). The type of employment contract was coded as 0 = temporary (87%), 1 = permanent (13%). Managerial position was coded as 0 = no (88%), 1 = yes (12%). Education was coded as follows: 1 = vocational qualification or matriculation examination certificate (5%), 2 = specialist vocational qualification (25%), 3 = higher vocational level qualification (6%), 4 = polytechnic qualification or bachelor’s degree (19%), 5 = university degree (42%), 6 = university postgraduate degree; licentiate or doctoral degree (3%). The average hours worked per week were 37.7 (SD = 7.7). The mean age was 46.8 years (SD = 11.4). Data analysis The main analytical tools in this study were latent profile analysis (LPA) and structural equation modeling (SEM). Specifically, LPA is a person-centered method of analysis enabling the identification of homogeneous and heterogeneous groups (profiles, patterns) of individuals as regards the phenomenon of interest (here MJDs) by utilizing mean value information at both intra-individual and interindividual levels (see Muthén, 2001; Spurk et al., 2020; Woo et al., 2008). We perceive that one benefit of LPA is actually practice-oriented; LPA allows to find also smaller and unpredicted (atheoretical/atypical) groups of individuals in relation to the analyzed phenomena, which again might have important implications for those individuals, e.g., higher risks for health problems. Here, LPA was implemented using Mplus statistical package (version 8; Muthén & Muthén, 1998–2017) to identify the number of latent profiles of respondents based on their individual responses to the five indicators of MJDs: WI, IJCPDs, ILDs, illegitimate tasks, and distractions. In LPA, participants sharing the same profile have similar mean estimates in the selected MJDs (Muthén, 2001; Tein et al., 2013). We applied models with the local independence and homogeneity of variance and used maximum likelihood robust estimation in order to take into account the skewness of analyzed variables. The LPA included the participants who had full data for all five MJDs (n = 3,294). The choice of the number of profiles followed the established procedure (Celeux, & Soromenho, 1996; Nylund et al., 2007; Tein et al., 2013). In the absence of general consensus of the best criteria for determining the number of profiles (Nylund et al., 2007), we based our decision on several statistical tests and reasonable content in profiles including adequate disparity of profiles, as the number of profiles may be overestimated in LPA (Bauer & Curran, 2003). We used likelihood ratio statistical tests (LoMendell-Rubin tests; p < 0.05; Celeux & Soromenho, 1996), information criterion tests (Bayesian Information Criterion, BIC, and Akaike’s Information Criterion, AIC), the estimates of which are smaller when the model fits better data comparison with the alternative model, and entropy-based criterion (scale 0–1, good entropy > 0.80 Celeux and Soromenho 1996). One benefit of LPA over more traditional person-centered analysis methods (e.g., cluster analysis) is that LPA provides these statistical rigorous tests to compare the number of profiles in the data. We used two variables, both of which reflect latent profiles of MJDs. In LPA executed by Mplus, each participant gets a posterior probability (henceforth PP) to belong to each one of the latent profiles, thus the number of PP variables is equal to that of the latent profiles (Muthén & Muthén, 1998–2017). For example, from the analysis including five latent profiles, every participant gets five PPs which represent participant’s probability (0–100) to belong to each profile and each of these probabilities can be used as a separate PP variable. Thus, PPs offer more information about each participant than one simple categorical clustering variable which was the main reason why we used PPs as separate continuous variables in SEM modeling as explanatory variables. The second variable which reflected latent profiles of MJDs in our analyses was a categorical clustering variable which represented a respondent’s most likely latent profile membership (henceforth MLP). A participant’s MLP was determined by comparing his/her PPs to belong in each profile and choosing the profile which had the highest probability. The scale of MLP was 1–5 as the LPA solution included five latent profiles (see Results). MLP was used for naming profiles and descriptive analyses. Each latent profile was interpreted and named after its most prominent content comparing the standardized sample means of five MJDs for the profile. The associations between MJDs and control variables were studied using Chi-square tests for dichotomous variables (gender, occupational group, type of employment contract, managerial position) and equality tests of means across profiles among variables modeled as continuous variables (education, hours worked per week, and age) using the modified BCH method in Mplus (Asparouhov & Muthén, 2018). Before SEM analyses we also examined the correlations of the variables studied including control variables (see Table 1). Descriptive and correlation analyses were conducted using IBM SPSS Statistics (Version 25). Next, SEM was performed to analyze the relationships between the (MJDs) profiles and dependent variables (three employee outcomes as latent constructs). We used the SEM latent variable framework as it takes into account measurement errors which are associated with observed variables (Kline, 2011). These SEM analyses would also validate our profile solution: the (MJDs) profiles should show meaningful and significant associations with the employee outcomes studied, otherwise their criterion validity would be insufficient (see also Spurk et al., 2020). In SEM, we used separate PP variables (see the description above), each representing one MJDs’ profile, as explanatory variables. We used PPs as they yielded more information about each Mauno & Minkkinen (2020) Profiling a spectrum of mental job demands 62 participant compared to one categorical MLP variable. When PP variables were included in SEM, one of them was dropped out due to statistical limitations, as including all PPs in the same SEM caused unidentifiable model. The reason for this is that high correlation among the explanatory variables violates the assumption for linear regression (the absence of multicollinearity) and leads to numerical problems (Tabachnick & Fidell, 2013). This was the case here, as PPs were dependent on each other and the sum of the probabilities from all PPs was 100 for each participant. However, this dependency between PPs also signified that the information of the dropped profile was still affecting statistics in the SEM, even though PP in question was removed from the analysis. As the LPA solution included five latent profiles (see Results), four PPs representing MJDs’ profiles were entered simultaneously into the SEM model as explanatory variables. We dropped a latent profile, which was mostly characterized by low MJDs, as we hypothesized that higher MJDs would be particularly stressful for employees (e.g., Galy et al., 2012; Karasek & Theorell, 1990; Zapf et al., 2014). Thus, latent profiles characterized by higher MJDs were more relevant for our purposes. Specifically, four SEM models were executed, e.g., one model for each dependent variable (job exhaustion, cynicism, task performance, meaning of work) using maximum likelihood robust estimation in Mplus. We further compared the magnitude of the significant effects of the MJDs’ profiles (four PP variables) on dependent variables with each other in the same SEM using the absolute values of the confidence intervals of the standardized regression coefficients (b*). The effects of MJDs’ profiles were interpreted as statistically significantly different if the confidence intervals of 95% did not overlap. Control variables were included in the SEM models if they had a significant bivariate correlation with a dependent variable (p < .05; see Table 1). For SEM models, we applied the missing data approach using Mplus statistical package (Version 8) which handles missing values through full information maximum likelihood procedure (FIML; see Muthén & Muthén, 1998– 2017). The missing data percentages in the dependent variables varied from 2.3% (task performance) to 5.1% (work meaning). The corresponding proportions for control variables were 0.2% in gender, 0% in occupational group, 10.5% in type of employment contract, 9.7% in managerial position, 0% in education, 11.4% in hours worked per week, and 0.1% in age. The model fit for SEM models was evaluated using Chi-square values (χ2), comparative fit index (CFI), Tucker-Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). The cutoff values were .95 for CFI and TLI, .06 for RMSEA, and .08 for SRMR (Hu & Bentler, 1999). Results Identifying the profiles of MJDs: LPA analysis Several LPA models were executed each with a different number of latent profiles following the established procedure (Celeux & Soromenho, 1996; Nylund et al., 2007; Tein et al., 2013). According to the Lo-Mendell-Rubin tests, the model including five latent profiles fitted better than the model with four profiles (VLMR and LMR, p < .001; Table 2) and the model with six profiles was not a better solution than the five profiles model (VLMR and LMR, p = .776; Celeux & Soromenho, 1996; Tein et al., 2013), which supported to the choice of five profiles. The entropy-based criterion for five latent profiles was also slightly better (.94) than for six profiles (.93). Importantly, from the viewpoint of our research aim, the solution with five latent profiles also had a meaningful content sorting out profiles including higher MJDs from the profile of lower MJDs. These findings supported the choice of five profiles, which was selected for further analyses. It is also noteworthy that even though the test values of log-likelihood, BIC, and AIC were slightly better for the model of six profiles than that of five profiles, the six profile solution did not indicate any new meaningful profiles as regards the content. Actually, the sixth profile was identical in content to the solution of five profiles, the only difference being slightly higher levels of means for each MJD. We named the latent profiles after their most prominent content based on the standardized sample means of five MJDs for each profile as follows (see Figure 1): Low mental demands (LMD), Moderate learning demands and low other mental demands (MLD), Low learning demands and moderate other mental demands (LLD), Moderately high IJDs and moderate illegitimate tasks and distractions (MHIJD), and High mental demands (HMD). The most likely latent profile membership for respondents was MHIJD (29.7%) and the most unlikely membership was LMD (14.0%) according to the estimated posterior probabilities. The corresponding shares were 21.2% for MLD, 18.8% for LLD, and 16.4% for HMD. Comparing the latent profiles with each other, LMD and HMD were easily distinguished due to their distinct quantitative differences for every MJD (see Figure 1). Fewest MJDs accumulated in the profile of LMD and the greatest number of MJDs for HMD. MHIJD was characterized the second highest MJD except for illegitimate tasks, which were at the second highest level in the profile of LLD. Among MLD, LLD, and MHIJD we identified different combinations in experiencing MJDs. Specifically, LLD was characterized by low learning demands (about 1 SD below mean) when MLD and MHIJD were close to their means. Thus, employees having in their MJD profile LLD or LMD did not report high learning demands in their jobs. MHIJD was characterized by moderately high IJDs (about 0.5 SD above the mean), slightly more distractions than average Journal for Person-Oriented Research, 6(1), 55-71 63 and average level of illegitimate tasks. MLD was characterized by low WI, illegitimate tasks, and distractions (all at least 0.5 SD below mean), slightly fewer IJCPDs than average, and slightly higher learning demands than average. In sum, the LPA results revealed different combinations of experiencing MJDs. Table 2 The Latent Profiles Based on Their Most Likely Latent Class Membership Number of profiles VLMR LMR LogL BIC AIC Entropy n (%) 1 - - -190311 381221 380769 - 3294 (100) 2 .000 .000 -175104 351115 350432 .95 1414 (42.9), 1880 (57.1) 3 .003 .003 -170259 341732 340817 .94 675 (20.5), 1183 (35.9), 1436 (43.6) 4 .001 .001 -166361 334245 333099 .94 526 (16.0), 645 (19.6), 961 (29.2), 1162 (35.3) 51) .000 .000 -163964 329758 328379 .94 464 (14.1), 539 (16.4), 617 (18.7), 693 (21.0), 981 (29.8) 6 .776 .776 -162527 327193 325582 .93 359 (10.9), 434 (13.2), 488 (14.8), 527 (16.0), 690 (20.9), 796 (24.2) Note. VLMR = Vuong-Lo-Mendell-Rubin likelihood ratio test, LMR = Lo-Mendell-Rubin adjusted lrt test, LogL= Log-likelihood, BIC = Bayesian information criterion, AIC = Akaike’s information criterion. 1) The selected profile solution for further analyses. Figure 1. Standardized Sample Means of the Job Mental Demands for the Latent Profile Variables Based on Respondents’ Most Likely Latent Class Member. Note. LMD = Low mental demands, MLD = Moderate learning demands and low other mental demands, LLD = Low learning demands and moderate other mental demands, MHIJD = Moderately high IJDs and moderate illegitimate tasks and distractions, HMD = High mental demands; WI = work intensification; IJCPDs = intensified job-related and career-related planning and decision-making demands; ILDs = intensified learning demands. Mauno & Minkkinen (2020) Profiling a spectrum of mental job demands 70 Karasek, R. A., & Theorell, T. (1990). Healthy work, stress, productivity and the reconstruction of working life. New York, NY: Basic Books. Kline, R. B. (2011). Principles and Practice of Structural Equation Modeling (3th ed.). New York: The Guilford Press. Koopmans, L., Bernaards, C. M., Hildebrandt, V. H., Lerner, D., de Vet, H. C., & van der Beek, A. J. (2016). Cross-cultural adaptation of the Individual Work Performance Questionnaire. Work, 53(3), 609–619. https://doi.org/10.3233/WOR-152237 Korunka, C., Kubicek, B., Paškvan, M., & Ulferts, M. (2015). Changes in work intensification and intensified learning: Challenge or hindrance demands? Journal of Managerial Psychology, 30(7), 786–800. https://doi.org/10.1108/JMP-02-2013-0065 Kubicek, B., Paškvan, M., & Korunka, C. (2015). Development and validation of an instrument for assessing job demands arising from accelerated change: The intensification of job demands scale (IDS). European Journal of Work and Organizational Psychology, 24(6), 898–913. https://doi.org/10.1080/1359432X.2014.979160 Laursen, B. P., & Hoff, E. (2006). Person-centered and variable-centered approaches to longitudinal data. Merrill-Palmer Quarterly, 52(3), 377–389. Lazarus, R. & Folkman, S. (1984). Stress, appraisal, and coping. New York, NY: Springer. Liebl, A., Haller, B., Jödicke, H., Baumgartner, S., Schlittemeier, J., & Hellbrück, J. (2012). Combined effects of acoustic and visual distraction on cognitive performance and well-being. Applied Ergonomics, 43(2), 424–434. https://doi.org/10.1016/j.apergo.2011.06.017 Lin, B. C., Kain, J. M., & Fritz, C. (2013). Don’t interrupt me! An examination of the relationship between intrusions at work and employee strain. International Journal of Stress Management, 20(2), 77–94. https://doi.org/10.1037/a0031637 Ma, J., & Peng, Y. (2019). The performance costs of illegitimate tasks: The role of job identity and flexile role orientation. Journal of Vocational Behavior, 110, 144–154. https://doi.org/10.1016/j.jvb.2018.11.012 Mauno, S., Ruokolainen, M., De Bloom, J., & Kinnunen, U. (2017). Does recovery buffer against emotional labor in terms of motivational outcomes at work? Analyzing age differences among Finnish health care professionals. Applied Nursing Research, 36, 88–94. https://doi.org/10.1016/j.apnr.2017.06.007 Mauno, S., Kubicek, B., Minkkinen, J., & Korunka, C. (2019a). Antecedents of intensified job demands: evidence from Austria. Employee Relations, 41(4), 694–707. https://doi.org/10.1108/ER-04-2018-0094 Mauno, S., Minkkinen, J., Tsupari, H., Huhtala, M., & Feldt, T. (2019b). Do older employees suffer more from work intensification and other intensified job demands? Evidence from upper white-collar workers. Scandinavian Journal of Work and Organizational Psychology, 4(1): 3, 1–13. https://doi.org/10.16993/sjwop.60 Mauno, S., Kubicek, B., Feldt, T., & Minkkinen, J. (2020). Intensified job demands and job performance: Does SOC strategy use make a difference? Industrial Health, 58(3), 224–237. https://doi.org/10.2486/ind Maslach, C., Schaufeli, W. B., & Leiter, M. P. (2001). Job burnout. Annual Review of Psychology, 52(1), 397–422. Mustosmäki, A. (2017). The intensification of work. In Tammelin, M (Ed.), Family, work and well-being: emergence of new issues (pp. 77-91). Switzerland: Springer. Muthén, B. O. (2001). Latent variable mixture modeling. In G. A. Marcoulides, & R. E. Schumacker (Eds.), New developments and techniques in structural equation modeling (pp. 1–33). Mahwah, NJ: Lawrence Erlbaum Associates. Muthén, B, & Muthén, L. K. (2000). Integrating person-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes. Alcoholism: Clinical and Experimental Research, 24, 882–891. https://doi.org/10.1097/00000374-200006000-00020 Muthén, L. K., & Muthén, B. O. (1998–2017). Mplus User’s Guide. Eighth Edition. Los Angeles, CA: Muthén & Muthén. Nylund, K., Asparouhov, T., & Muthén, B. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14(4), 535–569. https://doi.org/10.1080/10705510701575396 Paškvan, M., Kubicek, B., Prem, R., & Korunka, C. (2016). Cognitive appraisal of work intensification. International Journal of Stress Management, 23(2), 124–146. Rosa, H. (2003). Social acceleration: Ethical and political consequences of a desynhcronized high-speed society. Constellations, 10, 3–33. https://doi.org/10.1111/1467-8675.00309 Rosso, B. D., Dekas, K. H., & Wrzesniewski, A. (2010). On the meaning of work: A theoretical integration and review. Research in Organizational Behavior, 30, 91–127. https://doi.org/10.1016/J.RIOB.2010.09.001 Salmela-Aro, K., Rantanen, J., Hyvönen, K., Tilleman, K., & Feldt, T. (2011). Bergen Burnout Inventory: Reliability and validity among Finnish and Estonian managers. International archives of occupational and environmental health, 84(6), 635–645. https://doi.org/10.1007/s00420-010-0594-3 Sander, E., Caza, A., & Jordan, P. J. (2019). Psychological perceptions matter: Developing the reactions to the physical work environment scale. Building and Environment, 148, 338–347. https://doi.org/10.1016/j.buildenv.2018.11.020 Seddigh, A., Berntson, E., Danielson, C. B., & Westerlund, H. (2014). Concentration requirements modify the effect of office type on indicators of health and performance. Journal of Environmental Psychology, 38, 167–174. https://doi.org/10.1016/j.jenvp.2014.01.009 Semmer, N. K., Tschan, F., Meier, L. L., Facchin, S., & Jacobshagen, N. (2010). Illegitimate tasks and counterproductive work behavior. Applied Psychology, 59(1), 70–96. https://doi.org/10.1111/j.1464-0597.2009.00416.x Semmer, N. K., Jacobshagen, N., Meier, L. L., Elfering, A., Beehr, T. A., Kälin, W., et al. (2015). Illegitimate tasks as a source of stress. Work & Stress, 29(1), 32–56. https://doi.org/10.1080/02678373.2014.1003996 Siegrist, J. (1996). Adverse health effects of high-effort/low-reward conditions. Journal of Occupational Health Psychology, 1(1), 27–41. Journal for Person-Oriented Research, 6(1), 55-71 71 https://doi.org/10.1037/1076-8998.1.1.27 Sonnentag, S., Venz, L., & Casper, A. (2017). Advances in recovery research: What have we learned? What should be done next? Journal of Occupational Health Psychology, 22(3), 365–380. https://doi.org/10.1037/ocp0000079 Spurk, D., Hirschi, A., Wang, M., Valero, D., & Kauffeld, S. (2020). Latent profile analysis: A review and “how to” guide of its application within vocational behavior research. Journal of Vocational Behavior (online first). https://doi.org/10.1016/j.jvb.2020.103445 Steger, M. F., Dik, B. J., & Duffy, R. D. (2012). Measuring meaningful work: The Work and Meaning Inventory (WAMI). Journal of Career Assessment, 20(3), 322–337. https://doi.org/10.1177/1069072711436160 Sweller, J. (1988). Cognitive load theory during problem solving: effects on learning. Cognitive Science, 12, 257–285. https://doi.org/10.1207/s15516709cog1202_4 Tabachnick, B. G. & Fidell, L. S. (2013). Using Multivariate Statistics (6th Ed.). Boston, MA: Pearson Education. Tein, J-Y., Coxe, S., & Cham, H. (2013). Statistical power to detect the correct number of classes in latent profile analysis. Structural Equation Modeling: A Multidisciplinary Journal, 20(4), 640–657. https://doi.org/10.1080/10705511.2013.824781 Warm, J. S., Parasuraman, R., & Matthews, G. (2008). Vigilance requires hard mental work and is stressful. Human Factors, 50(3), 433–441. https://doi.org/10.1518/001872008X312152 Woo, S. E., Jebb, A. T., Tay, L., & Parrigon, S. (2018). Putting the “person” in the center: Review and synthesis of person-centered approaches and methods in organizational science. Organizational Research Methods, 4, 814–845. https://doi.org/10.1177/1094428117752467 Zapf, D., Semmer, N., & Johnson, S. (2014). Qualitative demands at work. In Peeters, M., de Jonge, J., & Taris, T. (Eds.), An introduction to contemporary work psychology (pp. 144-169). Chichester, UK: Wiley Blackwell. Mauno & Minkkinen (2020) Profiling a spectrum of mental job demands 72