Accepted Manuscript Title: Too much job control? Two studies on curvilinear relations between job control and eldercare workers’ well-being Author: Bettina Kubicek Christian Korunka Sara Tement PII: DOI: Reference:

S0020-7489(14)00130-8 http://dx.doi.org/doi:10.1016/j.ijnurstu.2014.05.005 NS 2395

To appear in: Received date: Revised date: Accepted date:

21-1-2014 13-5-2014 16-5-2014

Please cite this article as: Kubicek, B., Korunka, C., Tement, S.,Too much job control? Two studies on curvilinear relations between job control and eldercare workers’ well-being, International Journal of Nursing Studies (2014), http://dx.doi.org/10.1016/j.ijnurstu.2014.05.005 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

TOO MUCH JOB CONTROL?

1

Running head: TOO MUCH JOB CONTROL?

Too much job control?

ip t

Two studies on curvilinear relations between job control and eldercare workers’ well-

cr

being

University of Vienna

an

1

us

Bettina Kubicek1, Christian Korunka1, Sara Tement2

2

M

Department of Applied Psychology: Work, Education and Economy

University of Maribor

te

d

Department of Psychology, Faculty of Arts

Ac ce p

Corresponding author: Sara Tement, University of Maribor, Faculty of Arts, Department of Psychology, Koroska 160, SI-2000 Maribor, T: +386 (0)2 22 93 889; E-mail: [email protected]

Funding: The research was partly funded by a grant from the Austrian Science Fund (FWF), project number P23377-G17  

Page 1 of 35

TOO MUCH JOB CONTROL?

2 Abstract

Background: Longer life expectancy and changes in family patterns in Western societies have increased the need for professional caregivers to the elderly. A detailed understanding of

ip t

job resources, especially of job control, may help preserve eldercare workers’ work-related well-being.

cr

Objectives: The aim of the present study was to determine optimal levels of job control for eldercare workers’ work-related well-being.

us

Design: Cross-sectional (Study 1) and longitudinal (Study 2) data were used to assess

an

the linear and non-linear relations between job control and eldercare workers’ work-related well-being. As indicators of well-being, irritation was examined cross-sectionally, whereas

M

burnout and engagement were used in the longitudinal study. In both studies, demographic variables and time pressure were included as controls. In Study 2, results were additionally

d

controlled for baseline levels of the outcome variables.

te

Participants: Study participants were recruited from the nursing staff of 10 publicly financed nursing homes for the elderly (Study 1) and from 18 privately and 20 publicly

Ac ce p

financed nursing homes for the elderly (Study 2). In both studies, eldercare workers (i.e., registered nurses, orderlies, and nursing assistants) were invited to answer paper-pencil questionnaires. In Study 1, 606 persons followed this invitation (response rate: 65%). In study 2, 591 persons provided data on both measurement points (response rate: 35% of Time 1 participants). The samples of both studies were predominantly female and between 20 and 65 years of age.

Methods: Hierarchical multiple regressions with linear and quadratic terms were conducted. Results: Curvilinear effects of job control were found in both studies. In study 1, low as well as high levels of job control were associated with higher levels of irritation compared to medium levels of job control. The longitudinal data from Study 2 supported these results. Page 2 of 35

TOO MUCH JOB CONTROL?

3

Eldercare workers with higher and lower levels of job control reported a higher tendency to depersonalize care recipients and a lower tendency to feel dedicated to, absorbed by and vigorous at work than those with medium levels of job control.

ip t

Conclusions: Higher levels of job control are not necessarily advantageous for work-

cr

related well-being of eldercare workers.

Ac ce p

te

d

M

an

us

Keywords: burnout, engagement, eldercare workers, irritation, job control, curvilinear effects

Page 3 of 35

TOO MUCH JOB CONTROL?

4 Introduction

Due to the increasing life expectancy and a higher ratio of older dependents to the working population (i.e., age dependency ratio) in Western societies, there is an increasing

ip t

need for eldercare workers and other healthcare professionals with geriatric care competencies in international labor markets (e.g., IOM, 2008). However, reports show that an employment

cr

in eldercare is not especially sought after among nurses. Especially younger nurses tend to

prefer to work in positions that do not involve eldercare (Kümmerling, 2011). One possible

us

explanation for such a trend may be that working conditions in eldercare are often

an

characterized by low pay and high emotional demands (e.g., IOM, 2008). In addition, eldercare workers often have to carry high levels of responsibility, as they face important

M

decisions with respect to preparing time and activity schedules or to the welfare of elderly dependents in general.

d

Such taxing working conditions may pose a risk factor for impaired well-being. In fact,

te

studies already showed that eldercare workers are among the most susceptible to impaired work-related well-being such as burnout (Demerouti et al., 2000). It may also be speculated

Ac ce p

that eldercare workers may experience low levels of work engagement, a positive aspect of work-related well-being. One of the crucial steps in attracting new workers, retaining them, and preserving their health is a detailed understanding of what aspects of the work environment could foster work-related well-being. As decision making is inherent to the nature of eldercare work, it may be assumed that the extent of job control or freedom over important decisions may be one of the most salient work characteristics of eldercare and an important determinant of eldercare workers’ well-being (e.g., Schmidt and Diestel, 2011, Willemse et al., 2012). In line with the Job Demands-Control (JD-C) model (Karasek, 1979), past research has frequently operationalized job control in terms of a broad construct including control over work tasks as well as usage of different skills (i.e., skill discretion). As such a broad Page 4 of 35

TOO MUCH JOB CONTROL?

5

conceptualization may lead to unclear results in terms of the relation to work-related wellbeing (Schmidt and Diestel, 2011, Willemse et al., 2012), the present research applies a narrow conceptualization of job control reflecting only decision latitude. In that sense, job

ip t

control is conceptualized as one’s degree of control over work tasks and the methods to execute these tasks as well as control over work schedules. Thus, it encompasses timing as

cr

well as method control (Verhaeghe et al., 2008).

Based on the JD-C model (Karasek, 1979) and the more general Job Demands-

us

Resources (JD-R) model (Bakker and Demerouti, 2007, Demerouti et al., 2001), job control is

an

regarded as a job resource which has exclusively positive effects on work-related well-being. According to these models, job control has a dual-role in relation to work-related well-being.

M

It counteracts job demands and their physiological or psychological costs (i.e., prevents the health-impairment process). Additionally, it helps to obtain work goals and fosters motivation

d

as well as personal growth (i.e., fosters the motivational process). In fact, numerous studies

te

found that job control is negatively related to burnout (e.g., Taris et al., 2005) as well as emotional distress and physical health complaints (e.g., Spector, 1986). On the other hand, job

Ac ce p

control was found to be positively related to motivation, job satisfaction, and organizational commitment (e.g., Schmidt and Diestel, 2011, Spector, 1986). In sum, it seems from the above that job control has only positive effects. However, research insights also suggest that job control might not be an exclusively valuable resource. Conversely, job control is also associated with increased planning requirements and

problems with predictability of one’s work and, in turn, may reduce employees’ psychological capabilities and decrease work-related well-being (Korunka and Kubicek, 2013). Authors speculate that “the more autonomy employees are given over organizing their time in work seems to mean that they are spending longer and longer at work or working” (Brannen, 2005, p. 115). In the case of eldercare workers where everyday decisions are made with, for, and on behalf of the elderly dependents in order to preserve their well-being or manage pain (e.g., Page 5 of 35

TOO MUCH JOB CONTROL?

6

Thompson et al., 2013), higher levels of job control are coupled with higher responsibility levels. For instance, when making end-of-life decisions where nurses often enact the role of information brokers transmitting information to physicians, family members, or mediating in-

ip t

between parties (Adams et al., 2011), the decision on how to communicate, what to say, and when to say it may put a huge responsibility on the nurse. Therefore, it can be argued that

cr

high levels of job control may not always be considered as a resource. Rather, they may also

become demanding. As the relations between job characteristics and work-related well-being

us

are more complex than initially presumed by several prominent theories and models

an

(Karanika-Murray, 2010), the focus should be on the optimal levels of job control in specific occupations.

M

The curvilinear effects of job control In contrast to the JD-C model (Karasek, 1979) and the JD-R model (Bakker and

d

Demerouti, 2007), the Vitamin model (Warr, 1990, 1994) challenges the assumption that job

te

control is generally beneficial. The model assumes that the role of job characteristics can be analogous to vitamins. More precisely, Warr (1994, p. 88) states that “the intake of vitamins is

Ac ce p

important for physical health up to, but not beyond, a certain level” and that excessive vitamin consumption can even be harmful for a person’s health. Certain features of the work environment are proposed to have similar properties as vitamins, such that after a certain point, an increase in beneficial job characteristics may not have a beneficial effect on workrelated well-being (i.e., constant effect) or may even be detrimental (i.e., additional decrement). Among the job characteristics that are assumed to suffer from additional decremental effects are job control, skill use, and interpersonal contact. Warr (1994) provides several explanatory mechanisms for such a curvilinear effect, all of which may be applied to job control. First, high levels of job control may not be considered as a “nicety” at the workplace but more as a workplace “necessity”; i.e., rather inhibiting action than fostering it. Second, high levels of job control may be associated with high levels Page 6 of 35

TOO MUCH JOB CONTROL?

7

of several undesirable job characteristics, such as high workload. Third, high levels of job control may be associated with lower levels of other desirable job characteristics, such as workplace support. For instance, if one has the freedom on how to approach a patient, his or

turn could lead to lower levels of mutual understanding and support.

ip t

her coworkers may not be familiar with such an approach or may not engage in it, which in

cr

Several studies tested the assumptions of the Vitamin model, especially focusing on job

control and different aspects of work-related well-being (e.g., burnout, job satisfaction). Warr

us

(1990), for instance, found support for the propositions made by the Vitamin model in a

an

heterogeneous sample. By focusing on job control only, the study found that the quadratic term of job control significantly predicted job satisfaction. However, instead of resembling an

M

inverted U-shape, the relation was rather a concave downward with a smaller increase in job satisfaction with higher values of job control. In a sample of civil servants, cross-sectional as

d

well as longitudinal evidence indicated a J-shaped relation between job control and job

te

satisfaction where higher levels of job control were associated with higher job satisfaction (Rydstedt et al., 2006). In a study among health care workers, the effect for job control was

Ac ce p

again not in line with the assumptions of the Vitamin model (De Jonge and Schaufeli, 1998). Instead of a U-shaped curve, an inverted U-shaped curve was found where low and high levels of job autonomy were related to a lower level of exhaustion. Moreover, several studies failed to find non-linear associations between job control and work-related outcomes (Jeurissen and Nyklíček, 2001, Mäkikangas et al., 2007). In sum, previous studies provide some support for non-linear relations between job control and aspects of work-related wellbeing. In general, however, the results are inconsistent. The inconsistencies can be explained by four different characteristics of studies to date. The first possible explanation is the utilization of a broader job control measure, confounding job control and job variety (e.g., Mäkikangas et al., 2007, Rydstedt et al., 2006). Furthermore, studies also focused more or less on Warr’s (1990, 1994) affective well-being Page 7 of 35

TOO MUCH JOB CONTROL?

8

conceptualization (i.e., pleasure-displeasure, anxiety-comfort, depression-enthusiasm) (De Jonge and Schaufeli, 1998, Mäkikangas et al., 2007). Different outcomes, for instance work engagement, should be considered when striving to detect non-linear relations (e.g., Karanika-

ip t

Murray, 2010). Thirdly, non-linear effects of job characteristics on outcomes are likely to remain undetected in longitudinal studies with very long time lags in-between measurements

cr

(e.g., 3 years Mäkikangas et al., 2007 or 5.8 years Rydstedt et al., 2006). Finally, job control may have a different meaning depending on different types of job. Therefore, results from

us

non-nursing samples can hardly be generalized.

an

Aim and overview of the studies

Based on all mentioned theoretical deliberations as well as empirical limitations, the

M

aim of the present research is to examine the non-linear effect of job control. In two separate studies we tested whether the non-linear effects on eldercare workers’ well-being can be

d

found cross-sectionally as well as longitudinally. By taking the same predictors into account,

te

the second (longitudinal) study builds on the first, but differs in terms of well-being indicators. The indictors were chosen based on the theoretical deliberations as well as calls for

Ac ce p

examining positive organizational behaviors (Bakker and Schaufeli, 2008). This paper extends the knowledge on the effect of job control and benefits the literature in three points. First, we applied a narrow conceptualization of job control which has rarely been done previously. Second, we gave careful considerations to the studied outcomes with examining indicators of rather short-term as well as long-term well-being. Third, along with aspects of work-related well-being such as burnout which were examined more frequently, we assessed work engagement as an aspect of work-related well-being which, to our knowledge, has not been explored to date. In Study 1, we tested cross-sectionally whether job control had a non-linear effect on irritation (i.e., an indicator of work-related well-being). Irritation is considered to be a workrelated state of mental impairment involving rumination (i.e., cognitive aspect of irritation) as Page 8 of 35

TOO MUCH JOB CONTROL?

9

well as irritability (i.e., emotional aspect of irritation) (Mohr et al., 2006). Based on the incentive-disengagement cycle (Klinger, 1975), the rumination aspect is associated with increased efforts toward goal-achievement, whereas the irritability aspect is related to a lost

ip t

incentive for achieving important goals. Both aspects can be seen as indicators of irritation but also as antecedents of more serious impairments such as depression or anxiety (e.g., Dormann

cr

and Zapf, 2002). It can also be argued that the incentive-disengagement cycle along with

irritation is related to a burnout process where after increased efforts for goal-attainment, a

us

state of vital exhaustion and apathy may occur (Burisch, 2006). Therefore, irritation can be

an

viewed as immediate reaction to the working environment, whereas burnout is a long-term phenomenon. As job control may be useful when attaining work-related goals (e.g., Bakker

M

and Demerouti, 2007), it may decrease irritation. Yet, very high levels of control may again increase irritation due to higher responsibilities at work. In line with the propositions of the

d

Vitamin model (Warr, 1990, 1994), we therefore hypothesized a non-linear relation between

te

job control and irritation. More precisely, we predicted a U-shaped relation between job control and irritation (Hypothesis 1).

Ac ce p

In Study 2, we tested whether long-term non-linear effects of job control can be found.

We used burnout, a prolonged work-related process characterized by vital exhaustion (i.e., emotional exhaustion), a depersonalized attitude towards one’s care recipients (i.e., depersonalization), and an inability to perform (i.e., reduced professional efficacy), as an indicator of work-related well-being (Maslach et al., 1996, Maslach et al., 2001). The second indicator of work-related well-being in this study, which may also reflect a long-term process, is work engagement, a more persistent and pervasive positive state of mind characterized by high levels of energy (i.e., vigor), a strong sense of pride, significance and enthusiasm (i.e., dedication) and by high levels of concentration (i.e., absorption) (e.g., Bakker et al., 2008). In line with the Vitamin model (Warr, 1990, 1994), we presumed that job control is related to both outcomes in a non-linear manner following an additional decrement pattern. In other Page 9 of 35

TOO MUCH JOB CONTROL?

10

terms, after a certain point, job control may no longer be beneficial in decreasing burnout and increasing work engagement, rather it may start to increase burnout and decrease work engagement. Thus, we hypothesized that the relations between job control and the three

ip t

burnout dimensions, emotional exhaustion (Hypothesis 2a), depersonalization (Hypothesis 2b), and reduced professional efficacy (Hypothesis 2c) are U-shaped. We further hypothesized

cr

to find inverted U-shaped relations between job control and the three engagement dimensions, vigor (Hypothesis 3a), dedication (Hypothesis 3b), and absorption (Hypothesis 3c).

an

Method

us

Study 1: Cross-sectional curvilinear relations between job control and irritation

Participants and procedure

M

A cross-sectional study was conducted in publicly financed nursing homes for the elderly. After the nursing directors and the managers of the nursing homes had approved the

d

study, the eldercare workers (i.e. registered nurses, orderlies, and nursing assistants) received

te

paper-pencil questionnaires. Participants were introduced to reflect on their work situation while answering the questionnaire. In total, 934 questionnaires were distributed and 606 were

Ac ce p

returned (response rate: 65%). Of the participating eldercare workers, 91% were female. The respondents’ mean age was 42.41 years (SD = 9.59) and on average, they worked 35.17 hours per week (SD = 8.73). Measures

Job control was assessed with a 5-item scale from the instrument for stress-related job

analysis (ISTA, Semmer et al., 1995). For example, respondents had to answer the following question: “Considering your work activity in general, how much opportunity is there for you to make your own decisions?” The response format ranged from 1 = very little to 5 = very much. Irritation was assessed with an 8-item scale developed by Mohr and colleagues (2006). A sample item is: “I get irritated easily, although I don’t want this to happen.” For each item, Page 10 of 35

TOO MUCH JOB CONTROL?

11

respondents indicated their level of irritation on a 7-point rating scale ranging from 1 = not at all true to 7 = completely true. Control variables. Age, gender, and working hours were included as socio-demographic

ip t

control variables and time pressure was included as a job-related control variable. Time pressure was measured using 4 items from the instrument for stress-related job analysis

response format ranged from 1 = very seldom to 5 = very often.

us

Analytical strategy

cr

(ISTA, Semmer et al., 1995). A sample item is: “How often are you pressed for time?” The

an

Hierarchical multiple regression analysis using SPSS was conducted to test whether the relation between job control and irritation was non-linear. We controlled for gender, age,

M

working hours and time pressure in the first step of the model. In the second and third step of the regression model, we added the linear and the quadratic term of job control, respectively.

d

The quadratic term of job control was constructed by squaring the centered linear term.

Descriptive statistics

te

Results

Ac ce p

Means, standard deviations, internal consistencies, and correlations between the study

variables are presented in Table A1. Correlations between study constructs were medium in size and in the expected direction. All scales yielded satisfactory internal consistencies. Regression analysis

As shown in Table 1, control variables accounted for significant portions of variance in

the outcome. A closer look at the regression coefficients reveals that this effect stems solely from time pressure which was positively related to irritation (β = .36, p < .001). The linear term of job control was negatively related to irritation (β = - .24, p < .001) and was associated with a significant increment in variance as well. The quadratic term was positively related to irritation (β = .09, p = .02), indicating a curvilinear relation between job control and irritation. This finding, which is further illustrated in Figure 1, suggests that an increase in job control is Page 11 of 35

TOO MUCH JOB CONTROL?

12

initially related to lower levels of irritation, but the relation becomes weaker and even changes direction at high scores of job control. The inflection point (i.e., the score at which the curve starts changing direction; zinflection = - b1/2b2) was estimated at 1.33 SD above the mean level

ip t

of job control. However, at high scores of job control levels of irritation do not reach the starting point (at low levels of job control) indicating a more J-shaped relation. Nevertheless,

cr

such a finding is in line with the (inverted) additional decrement argument. Thus, these findings provide support for Hypothesis 1.

an

engagement

us

Study 2: Longitudinal curvilinear relations between job control and burnout and

Method

M

Participants and procedure

A two-wave longitudinal study with a time lag of 16 months was realized. Apart from

d

assessing the effects of job control, the study focused on emotional demands (Kubicek and

te

Korunka, in prep., Kubicek et al., 2013) and acceleration-related demands (Korunka, et al., in press, Kubicek et al., 2013). The study was conducted in privately (N = 18) and publicly

Ac ce p

(N = 20) financed nursing homes for the elderly and was approved by the nursing directors and the managers of the nursing homes. At Time 1, 1697 people filled out usable questionnaires, at Time 2, 1555 questionnaires were returned, resulting in response rates of 54% (Time 1) and 46% (Time 2). Using a personal code, 591 people, i.e., 35% of the respondents at Time 1, could be matched over the two measurement points. Of these respondents, 89% were female, 45% were middle-aged (41 to 50 years), and 54% worked full-time (more than 30 hours per week). Measures Job control was assessed using a German self-report instrument for work analysis in hospitals (TAA; Büssing and Glaser 2002). The scale consisted of seven items (e.g., “One can

Page 12 of 35

TOO MUCH JOB CONTROL?

13

decide autonomously, which tasks to perform”) with a response format ranging from 1 = no, not at all to 5 = yes, definitely. Burnout was assessed using the German version of the Maslach Burnout Inventory

ip t

(MBI; Büssing and Perrar, 1992). Emotional exhaustion (e.g., “I feel totally exhausted in my job”) comprised 9 items; depersonalization (e.g., “I don’t really care what happens to some

cr

recipients”) consisted of 5 items; and reduced professional efficacy comprised 6 items (e.g., I feel confident that I am effective at getting things done“; recoded). The response format

us

ranged from 1 = never to 6 = very often.

an

Engagement was assessed using the German short version of the Utrecht Work Engagement Scale (UWES-9; Schaufeli and Bakker, 2003). The questionnaire measures the

M

three sub-dimensions of work engagement, i.e. vigor, dedication, and absorption, with three items each. A sample item of the dedication subscale reads as follows: “I am enthusiastic

d

about my job”. For each item the response format ranged from 1 = never to 7 = always.

te

Control variables. As in Study 1, age, gender, and working hours were included as socio-demographic control variables and time pressure was included as job-related control

Ac ce p

variable. Time pressure was measured using the respective subscale from the German selfreport instrument for work analysis in hospitals (TAA; Büssing and Glaser 2002). The scale consists of three items (e.g., “Even in a hurry I often cannot finish all my work”) with a response format ranging from 1 = no, not at all to 5 = yes, definitely. Analytical strategy

To test our predictions regarding the non-linear relations between job control and

burnout as well as work engagement, we conducted several multiple regression analyses and calculated separate regression models for all burnout and engagement dimensions at Time 2. Due to the longitudinal nature of the study, we controlled not only for gender, age, working hours, and time pressure, but also for the respective Time 1 measure of each outcome in the regression models. Page 13 of 35

TOO MUCH JOB CONTROL?

14 Results

Descriptive statistics Means, standard deviations, internal consistencies, and correlations between the study

ip t

variables are presented in Table A2. Correlations between job control and the outcome variables were small in size and pointed by and large in the expected direction. Again, all

cr

scales yielded satisfactory internal consistencies.

Assuming potential selection problems due to panel loss, we compared participants who

us

provided data only at Time 1 (N = 1,106) with respondents who participated in both waves of

an

data collection (N = 591). No selective dropout was found for gender, working hours, job control, reduced professional efficacy, dedication, or absorption. Those who participated at

M

Time 1 and Time 2 differed slightly in age (χ2 (5) = 14.27; p = .01), emotional exhaustion (M = 2.84 vs. 3.04; t(1605) = -3.74; p < .001), depersonalization (M = 2.12 vs. 2.28; t(1596) = -

d

3.16; p = .002), and vigor (M = 5.18 vs. 5.00; t(1624) = 2.94; p = .003) from those who

te

dropped out. However, the effect sizes of the differences in the outcomes were very small (r = -.09 for emotional exhaustion; r = -.08 for depersonalization; r = .07 for vigor) (see Cohen,

Ac ce p

1988). Thus, it can be concluded that the longitudinal sample differed only slightly from the Time 1-only sample and selective dropout was not a serious problem within this study. Regression analysis

Table 2 presents the final step of the regression analyses for each of the outcomes.

When looking at the burnout dimensions first, apart from the Time 1 measures of the respective outcome, only few control variables were related to the three burnout outcomes. Being up to 30 years of age was positively related to reduced professional efficacy (β = .12, p = .018), whereas time pressure was positively related to depersonalization (β = .10, p = .014). The linear term of job control was negatively related only to reduced professional efficacy (β = -.10, p = .013). The quadratic term of job control was significantly related to depersonalization (β = .07, p = .049), but not to the other two burnout outcomes. The positive Page 14 of 35

TOO MUCH JOB CONTROL?

15

sign of the beta coefficient indicates a U-shaped relation between job control and depersonalization at Time 2. Figure 2 illustrates that higher levels of job control decreased depersonalization only up to a certain point after which depersonalization started to increase.

ip t

As the inflection point for the curve was estimated at 0.29 SD above the mean, job control was related to slightly higher depersonalization levels already around middle levels of job

cr

control. Thus, Hypothesis 2b was supported, whereas Hypotheses 2a and 2c were not supported.

us

Similarly to the burnout dimensions, all the engagement dimensions at Time 2 were

an

predicted by their respective values at Time 1. Of the control variables, gender was significantly related to vigor (β = .09, p = .021), dedication (β = .09, p = .011), and absorption

M

(β = .08, p = .047), with women exhibiting higher levels of each dimension. Interestingly, the linear term of job control was not significantly related to any of the engagement outcomes at

d

Time 2. The quadratic term of job control, on the other hand, was significantly related to

te

dedication (β = -.07, p = .037) and absorption (β = -.12, p < .001). The regression term for vigor approaches statistical significance (β = -.06, p = .089). The relations between job control

Ac ce p

and all the engagement outcomes resembled an inverted U-shape. In other terms, an initial increase in job control was related to higher levels of all three engagement outcomes but only up to a certain point (i.e., inflection point) after which higher levels of job control led to lower engagement levels. An inspection of the estimated inflection points revealed similar effects for all three outcomes (zinflection (absorption) = 0.27; zinflection (vigor) = 0.42; zinflection (dedication) = 0.36). Apparently, the expected change in direction of the relation occurred at middle levels of job control for absorption, vigor, and dedication. Figure 3 further illustrates these findings for all three engagement dimensions. Thus, the results lend support to Hypotheses 3b and 3c. As the relation between job control and vigor has the expected shape, Hypothesis 3a could be partially confirmed as well. Discussion Page 15 of 35

TOO MUCH JOB CONTROL?

16

The present paper adds to existing literature on the positive and potentially negative effects of job control by analyzing cross-sectional and longitudinal data. Following Warr’s (1994) Vitamin model, we argued that the beneficial effects of job control disappear and are

ip t

even reversed when job control exceeds a certain threshold. The results generally support the hypothesized curvilinear effect of job control on eldercare workers’ well-being.

cr

Results of Study 1, using a cross-sectional design, supported the hypothesis that job control exerts non-linear effects on eldercare workers’ irritation. Hierarchical multiple

us

regression analyses indeed showed that the quadratic term of job control was positively

an

related to irritation. Medium levels of job control were associated with low values of irritation, whilst low and high levels of job control were associated with high values of

M

irritation.

Because the negative effects of high levels of job control may disappear in the long run

d

or may even be attenuated, we realized a longitudinal design in Study 2. In addition to

te

assessing the long-term effect of job control on burnout as an indicator of long-term workrelated well-being, we took the effect of job control on work engagement as a positive

Ac ce p

persistent work-related state of mind into account. In concordance with Study 1, we found non-linear effects of job control on burnout and work engagement. Even in the long run, eldercare workers with high levels of job control showed decrements in well-being. In fact, they reported a higher tendency to depersonalize care recipients and a lower tendency to feel dedicated to, absorbed by, and vigorous at work than eldercare workers with medium levels of job control.

Theoretical and practical implications Although the incremental validity gained by adding the quadratic effect was somewhat low, the study makes an important theoretical contribution. It sheds light on the validity of opposing theoretical positions regarding the effect of job control on work-related well-being. The JD-C model (Karasek, 1979) as well as the JD-R model (Bakker and Demerouti, 2007, Page 16 of 35

TOO MUCH JOB CONTROL?

17

Demerouti et al., 2001) postulate that job control is related to work-related well-being in a linear manner. That is, the more control employees have, the less exhausted and the more engaged they feel. The Vitamin model (Warr, 1994), by contrast, posits that too little and too

ip t

much job control is unfavorable, whereas medium levels of job control are beneficial for work-related well-being. The additional decrement in well-being may be explained by the fact

cr

that at high levels, job control is no longer an “opportunity” but becomes an “unavoidable requirement” (Warr, 1994, p. 89).

us

Furthermore, basic psychological studies from the field of motivation suggest that

an

“either-or-decisions” which can be mainly found in the work environment do not imply true autonomy over one’s work tasks (e.g., Moller et al., 2006). For instance, a nurse may decide

M

on how to approach a patient but cannot decline approaching him or her in general. Having to deal with such decisions may not necessarily lead to improved engagement and may even be

d

straining beyond a certain point (e.g., Moller et al., 2006). Our findings from two

te

heterogeneous samples of eldercare workers support this line of reasoning and lend support to the assumptions of the Vitamin model rather than the JD-C or the JD-R model. In the cross-

Ac ce p

sectional as well as in the longitudinal study we found evidence that too little as well as too much job control might be detrimental for eldercare workers’ well-being at work. Given the numerous studies in support of the JD-C and the JD-R model (e.g., Spector,

1986, Taris et al., 2005), we would not dare to challenge their validity based on our results. Yet our findings suggest that the prepositions of the JD-C and the JD-R model may not hold for all occupational groups. For eldercare workers, the effects of job control on well-being seem to follow a curvilinear shape. Reasons for this finding may reside in the specific type of work caregivers to the elderly have to perform and in the responsibilities associated with their job. Eldercare workers have to make everyday decisions with, for, and on behalf of elderly dependents in order to preserve their well-being or manage pain (e.g., Thompson et al., 2013). Thus, higher levels of job control are coupled with higher responsibility levels. Taking such Page 17 of 35

TOO MUCH JOB CONTROL?

18

decisions autonomously without consulting colleagues or supervisors may put pressures on eldercare workers that make them ruminate about daily events in the job in the short run and reduce their work engagement and their personal relationship with care recipients in the long

ip t

run. Depersonalizing care recipients may help to reduce the emotional heaviness of the decisions and may be a way of coping with the seriousness of everyday decisions – though a

cr

dysfunctional one (e.g., Lee and Ashforth, 1996). Reducing one’s work engagement may also

be a way of detaching oneself from work in order to deal with too many responsibilities due to

us

too high levels of job control.

an

In addition to these theoretical considerations, the underlying shape of the relation between job control and eldercare workers’ well-being leads to important practical

M

implications for job redesign. Rather than maximizing job control for all eldercare workers, the curvilinear relation found in this study implies optimizing the exposure to job control.

d

Other than the population approach to prevention which is appropriate when job

te

characteristics and well-being are associated in a linear manner, J-shaped or U-shaped relations call for identifying risk groups (Adams and White, 2005). In this specific case,

Ac ce p

increasing job control would harm those eldercare workers who are already exposed to high levels of job control. Thus, workplace interventions are required to identify eldercare workers with low and high levels of job control and improve their working conditions by implementing optimal levels of job control. By following this line of reasoning, it may be crucial to establish occupation-specific cutoffs or norms for low, middle and high levels of job control which could be further used for identifying risk groups or for detailed job analysis and subsequent workplace interventions. As to date there are no such cutoff points available, we refrain from giving recommendation about optimal levels of job control. This, however, is an issue which should be examined in validation studies of job analysis instruments addressing job control. Another practical recommendation can be given. An innovative approach may be to train eldercare workers in job crafting behaviors where they can proactively modify their Page 18 of 35

TOO MUCH JOB CONTROL?

19

working environment in order to align their job and levels of job control with the personal preferences and competencies (Demerouti and Bakker, 2014). As studies found that feelings of competence or self-efficacy may alleviate possible

ip t

negative side effects of higher job control (e.g., Salanova et al., 2002), eldercare workers could benefit from competence trainings and on-job training programs. Especially beneficial

professionals are often undereducated in that regard (IOM, 2008).

us

Limitations and future research

cr

may be geriatric competence trainings, as the vast majority of nurses and health care

an

In our research, we focused on eldercare workers. For this professional group, we were able to show that job control has positive effects only up to a certain threshold, after which

M

increases in job control are associated with decrements in well-being. As such, this finding is important for designing eldercare work. However, it should be noted that there may be the

d

possibility of undetected effects regarding differences between different groups of eldercare

te

workers (e.g., nurses and nurses' assistants). Besides testing whether the curvilinear effect is contingent on the different eldercare workers professions, it would also be interesting to see

Ac ce p

whether this effect holds true across a variety of professional groups. It may be the case that the curvilinear effect of job control is observable only in occupations where job control is associated with high levels of responsibilities (for others), whereas a linear effect of job control dominates in occupations with fewer responsibilities. Thus, future research should replicate the finding among other professional groups to determine the generalizability of the curvilinear effect of job control. On a more general note, future studies among eldercare workers should also examine linear and curvilinear relations between other work characteristics (e.g., support) and eldercare workers’ well-being including information on patients or elderly dependents and their impairments in order to get a broader picture on wellbeing determinants.

Page 19 of 35

TOO MUCH JOB CONTROL?

20

In both studies, self-report measures were used to assess the study constructs. Using only one source of data collection may contribute to common method biases. By realizing not only a cross-sectional but also a longitudinal study design, we can, however, rule out that our

ip t

results are just due to methodological artifacts. The temporal separation of the assessment of the predictor and the outcome variables in Study 2 minimizes common method biases. It

cr

should, however, once again be stressed, that the results of the cross-sectional and

longitudinal study are not directly comparable due to different, yet theoretically related,

us

outcomes. Future studies could benefit from examining irritation longitudinally as well as

an

from testing Klinger’s (1975) incentive-disengagement cycle along with the process from irritation to burnout. Notwithstanding several important contributions of our study, future

M

research should also use objective (e.g., observational) methods to assess eldercare workers’

Ac ce p

te

d

job control.

Page 20 of 35

TOO MUCH JOB CONTROL?

21 References

Adams, J., White, M., 2005. When the population approach to prevention puts the health of individuals at risk. International Journal of Epidemiology 34 (1), 40-43.

ip t

Adams, J.A., Bailey, D.E., Anderson, R.A., Docherty, S.L., 2011. Nursing roles and strategies in end-of-life decision making in acute care: A systematic review of the literature.

cr

Nursing Research and Practice, 2011, 15 pages. doi:10.1155/2011/527834.

Bakker, A.B., Demerouti, E., 2007. The job demands-resources model: State of the art.

us

Journal of Managerial Psychology 22 (3), 309-328.

an

Bakker, A.B., Schaufeli, W.B., 2008. Positive organizational behavior: Engaged employees in flourishing organizations. Journal of Organizational Behavior 29 (2), 147-154.

M

Bakker, A.B., Schaufeli, W.B., Leiter, M.P., Taris, T.W., 2008. Work engagement: An emerging concept in occupational health psychology. Work & Stress 22 (3), 187-200.

d

Brannen, J., 2005. Time and the negotiation of work-family boundaries: autonomy or

te

illusion? Time & Society 14 (1), 113-131. Burisch, M., 2006. Das Burnout-Syndrom: Theorie der inneren Erschöpfung [The burnout-

Ac ce p

syndrome: a theory of inner exhaustion]. Springer Medizin Verlag, Heidelberg.

Büssing A., Glaser, J., 2002. Das Tätigkeits- und Arbeitsanalyseverfahren für das Krankenhaus - Selbstbeobachtungsversion (TAA-KH-S) [Activity and work analysis in hospitals - self-report version (TAA-KH-S)]. Hogrefe, Göttingen.

Büssing, A., Perrar, K.-M., 1992. Die Messung von Burnout. Untersuchung einer deutschen Fassung des Maslach Burnout Inventory (MBI-D). [The measurement of burnout]. Diagnostica 38 (4), 328-353. Cohen, J., 1988. Statistical power analysis for the behavioral sciences. Lawrence Erlbaum Associates, Hillsdale, NJ.

Page 21 of 35

TOO MUCH JOB CONTROL?

22

De Jonge, J., Schaufeli, W.B., 1998. Job characteristics and employee well-being: A test of Warr’s Vitamin model in health care working using structural equation modelling. Journal of Organizational Behavior 19 (4), 387-407.

ip t

Demerouti, E., Bakker, A.B., 2014. Job crafting. In Peeters, M.C.W., de Jonge, J., Taris, T.W. (Eds.), An introduction to contemporary work psychology. Wiley-Blackwell,

cr

Chichester, pp. 414-433

Demerouti, E., Bakker, A.B., Nachreiner, F., Schaufeli, W.B., 2000. A model of burnout and

us

life satisfaction amongst nurses. Journal of Advanced Nursing 32 (3), 545-464.

an

Demerouti, E., Bakker, A.B., Nachreiner, F., Schaufeli, W.B., 2001. The job demandsresources model of burnout. Journal of Applied Psychology 86 (3), 499-512.

M

Dormann, C., Zapf, D., 2002. Social stressors at work, irritation, and depressive symptoms: Accounting for unmeasured variables in a multi-wave study. Journal of Occupational

d

and Organizational Psychology 75 (1), 33-58.

te

IOM (Institute of Medicine), 2008. Retooling for an aging America: Building the health care workforce. The National Academies Press, Washington, DC

Ac ce p

Jeurissen, T., Nyklíček, I., 2001. Testing the Vitamin model of job stress in Dutch health care workers. Work & Stress 15 (3), 254-264.

Karanika-Murray, M., 2010. Work and health: Curvilinearity matters. In: Houdmont, J., Leka, S. (Eds.), Contemporary occupational health psychology: Global perspectives on research, education, and practice. Wiley-Blackwell, Chichester, pp. 151-168.

Karasek, R.A., 1979. Job demands, job decision latitude, and mental strain: Implications for job redesign. Administrative Science Quarterly 24 (2), 285-314. Klinger, E., 1975. Consequences of commitment to and disengagement from incentives. Psychological Bulletin, 82, 1-25. Korunka, C., Kubicek, B., 2013. Beschleunigung im Arbeitsleben: Neue Anforderungen und deren Folgen [Acceleration at work: New demands and their consequences]. In: Page 22 of 35

TOO MUCH JOB CONTROL?

23

Junghanns, G., Morschhäuser, M. (Eds.), Immer schneller, immer mehr: Psychische Belastung bei Wissens- und Dienstleistungsarbeit. Springer Fachmedien, Wiesbaden, pp. 17-39.

ip t

Korunka, C., Kubicek, B., Paškvan, M., Ulferts, H., in press. Changes in work intensification and intensified learning: Challenge or hindrance? Journal of Managerial Psychology.

cr

Kubicek, B., Hoelzl, E., Korunka, C., 2013. Emotion regulation requirements and affective

forecasts regarding expected organizational changes. Horizons of Psychology, 22, 1-

us

13.

rule dissonance on employee burnout.

an

Kubicek, B., Korunka, C., in prep. Job complexity mitigates the negative effect of emotion-

M

Kubicek, B., Ulferts, H., Korunka, C., 2013. Acceleration in the care of older adults: New demands as predictors of employee burnout and engagement. Journal of Advanced

d

Nursing 69 (7), 1525-1538.

te

Kümmerling, A., 2011. “And who cares for the carer?” Elderly care work in Germany (Walqing social partnership series 2011.16). http://www.walqing.eu/?id=64 accessed

Ac ce p

26/8/2013

Lee, R.T., Ashforth, B.E., 1996. A meta-analytic examination of the correlates of the three dimensions of job burnout. Journal of Applied Psychology 81 (2), 123-133.

Mäkikangas, A., Feldt, T., Kinnunen, U., 2007. Warr’s scale of job-related affective wellbeing: A longitudinal examination of its structure and relationship with work characteristics. Work & Stress 21 (3), 197-219.

Maslach, C. Jackson, S.E., Leiter, M.P., 1996. MBI: The Maslach Burnout Inventory manual (3rd ed.). Palo Alto, CA: Consulting Psychologists Press. Maslach, C., Schaufeli, W.B., Leiter, M.P., 2001. Job burnout. In: Fiske, S.T., Schachter, D.L., Zahn-Waxer, C. (Eds.), Annual Review of Psychology, 53, 397-422.

Page 23 of 35

TOO MUCH JOB CONTROL?

24

Mohr, G., Müller, A., Rigotti, T., Aycan, Z., Tschan, F., 2006. The assessment of psychological strain in work contexts: Considering the structural equivalency of nine language adaptations of the irritation scale. European Journal of Psychological

ip t

Assessment 22 (3), 198-206. Moller, A.C., Deci, E.L., Ryan, R.M., 2006. Choice and ego-depletion: The moderating role

cr

of autonomy. Personality and Social Psychology Bulletin 32 (8), 1024-1036.

Rydstedt, L.W., Ferrie, J., Head, J., 2006. Is there support for curvilinear relationships

us

between psychosocial work characteristics and mental well-being? Cross-sectional and

an

long-term data from the Whitehall II study. Work & Stress 20 (1), 6-20. Salanova, M., Peiró, J.M., Schaufeli, W.B., 2002. Self-efficacy specificity and burnout among

M

information technology workers: An extension of the job demands-control model. European Journal of Work and Organizational Psychology 11 (1), 1-25.

d

Schaufeli, W.B., Bakker, A.B., 2003. Utrecht Work Engagement Scale: Preliminary Manual.

te

Utrecht: Department of Psychology, Utrecht University, The Netherlands. http://www.schaufeli.com accessed 12/3/2012

Ac ce p

Schmidt, K.-H., Diestel, S., 2011. Differential effects of decision latitude and control on the job demands-strain relationship: A cross-sectional survey study among elderly care nursing staff. International Journal of Nursing Studies 48 (3), 307-317.

Semmer, N.K., Zapf, D., Dunckel, H., 1995. Assessing stress at work: A framework and an instrument. In: Svane, O., Johansen, C. (Eds.), Work and health. Office for Official Publications of the European Communities, Luxembourg, pp.105-113.

Spector, P.E. (1986). Perceived control by employees: A meta-analysis of studies concerning autonomy and participation at work. Human Relations 39 (11), 1005-1016. Taris, T.W., Stoffelsen, J., Bakker, A.B., Schaufeli, W.B., Van Dierendonck, D., 2005. Job control and burnout across occupations. Psychological Reports 97 (3), 955-961.

Page 24 of 35

TOO MUCH JOB CONTROL?

25

Thompson, C., Aitken, L., Doran, D., Dowding, D., 2013. An agenda for clinical decision making and judgement in nursing research and education. International Journal of Nursing Studies 50 (12), 1720-1726.

ip t

Verhaeghe, R., Vlerick, P., De Backer, G., Van Maele, G., Gemmel, P., 2008. Recurrent changes in the work environment, job resources and distress among nurses: A

cr

comparative cross-section survey. International Journal of Nursing Studies 45 (3), 382-392.

us

Warr, P., 1990. Decision latitude, job demands and employee well-being. Work & Stress 4

an

(4), 285-294.

Warr, P., 1994. A conceptual framework for the study of work and mental health. Work &

M

Stress 8 (2), 84-97.

Willemse, B.M., De Jonge, J., Smit, D., Depla, M.F.I.A., Pot, A.M., 2012. The moderating

d

role of decision authority and coworker- and supervisor support on the impact of job

te

demands in nursing homes: A cross-sectional study. International Journal of Nursing

Ac ce p

Studies 49 (7), 822-833.

Page 25 of 35

i cr

26

us

TOO MUCH JOB CONTROL?

an

Table 1

Linear and Quadratic Effects of Job Control on Irritation (Study 1)

Variable

β

B

B

0.879*

β

Step 3 β

B

1.062**

1.074**

-0.03

-.01

-0.03

-.01

-0.03

-.01

Age

-0.01

-.04

-0.01

-.05

-0.01

-.05

0.01

.05

0.01†

.08

0.01

.07

ce

Gender: female

pt

ed

Intercept

Step 2

M

Step 1

.36

0.41***

.29

0.40***

.28

-0.36***

-.24

-0.35***

-.24

0.15*

.09

Working hours

Job control Job control² Δ R2  Total R2

Ac

Time pressure

0.50***

.13***

.05***

.01*

.13***

.18***

.19***

Note. N = 509; *p < .05, **p < .01, *** p < .001.

Page 26 of 35

i cr

27

us

TOO MUCH JOB CONTROL?

an

Table 2 Linear and Quadratic Effects of Time 1 Job Control on Time 2 Burnout and Time 2 Engagement (Study 2) Burnout (Time 2)

Engagement (Time 2)

Ac

ce

pt

ed

M

Reduced Emotional professional exhaustion Depersonalization efficacy Vigor Dedication Absorption (N=525) (N=518) (N=514) (N=536) (N=535) (N=536) Variable β β β β β β B B B B B B Intercept 0.97*** 0.64** 0.92*** 2.14*** 2.24*** 2.60*** Time 1 measure 0.59*** .56 0.51*** .51 0.46*** .47 0.55*** .52 0.59*** .58 0.50*** .50 Gender: female -0.10 -.03 -0.11 -.04 -0.02 -.01 0.32* .09 0.36* .09 0.31* .08 a † Up to age 30 0.10 .03 0.05 .02 0.22* .12 -0.22 -.06 -0.20 -.05 -0.31 -.08 Age 31-40 -0.08 -.03 0.09 .04 0.07 .05 -0.10 -.04 -0.17 -.06 -0.14 -.05 Age 41-50 0.14 .07 0.08 .05 0.05 .04 -0.07 -.03 -0.12 -.05 -0.09 -.04 b † Working hours 0.03 .02 0.12 .06 0.10 .08 -0.01 -.00 -0.03 -.02 -0.07 -.03 † Time pressure (Time 1) 0.10 .08 0.12* .10 0.05 .06 -0.05 -.03 -0.09 -.05 -0.06 -.04 Job control (Time 1) -0.07 -.04 -0.05 -.04 -0.09* -.10 0.08 .04 0.10 .05 0.14 .07 † Job control² (Time 1) 0.03 .02 0.12* .07 0.02 .02 -0.12 -.06 -0.16* -.07 -0.28*** -.12 2 Total R .37*** .32*** .29*** .32*** .40*** .33*** a b Note. Results of the final step of the regression analyses are presented. Age 50+ served as reference category; 1 = full-time employment, 0 = parttime employment; *p < .05, **p < .01, *** p < .001.

Page 27 of 35

i cr

28

Ac

ce

pt

ed

M

an

us

TOO MUCH JOB CONTROL?

 

Figure 1. Curvilinear effect of job control on irritation (Study 1).

Page 28 of 35

i cr

29

Ac

ce

pt

ed

M

an

us

TOO MUCH JOB CONTROL?

Figure 2. Curvilinear effect of Time 1 job control on Time 2 depersonalization (Study 2).

 

Page 29 of 35

i cr

30

Ac

ce

pt

ed

M

an

us

TOO MUCH JOB CONTROL?

Figure 3. Curvilinear effects of Time 1 job control on Time 2 work engagement (vigor, dedication, absorption are shown separately) (Study 2).

Page 30 of 35

i cr

31

us

TOO MUCH JOB CONTROL?

Appendix

0.90

0.30 9.48

Gender: female

2

Age

42.28

3

Working hours

34.98

4

Time pressure

3.28

5

Job control

6

Job control²

7

Irritation

3

4

5

6

7

.01

8.76 -.29*** -.03 0.81

.07

.14**

.05

(.80)

3.33

0.75 -.05

-.06

.11*

-.25*** (.80)

0.58

0.75

.01

.09*

.03

.18*** -.30***

2.48

1.13

.02

.01

.06

.35*** -.30*** .21*** (.89)

ce

Ac

2

pt

1

1

M

SD

ed

M

an

Table A1 Means, Standard Deviations and Internal Consistencies of as well as Bivariate Correlations between Variables of Study 1

Note. N = 509; Internal Consistencies as indicated by Cronbach’s Alpha are presented in parentheses in the diagonal; *p < .05, **p < .01, ***p < .001.

Page 31 of 35

i cr

32

us

TOO MUCH JOB CONTROL?

3

5

6

10

.06

.02

.05

.07

.43*** -.12** -.01

(.89)

.04

-.00

.08

.31*** -.14** .02

.60*** (.91)

.11** -.01

-.04

-.03

.11** .25*** -.04

.11*

0.92 -.06

.06

.02

-.02

-.04

.12** .21*** -.08

.14** .37*** .58*** .55*** (.80)

SD

2

4

9

M

1

an

Table A2 Means, Standard Deviations and Internal Consistencies of as well as Bivariate Correlations between Variables of Study 2 7

8

Gender: female

0.89

0.31

2

Up to age 30

0.11

0.32 -.07

3

Age 31-40

0.26

0.44

.01

-.21***

4

Age 41-50

0.45

0.50

.01

-.32***-.54***

5

Age 50+

0.17

0.38

.04

-.16***-.27***-.41***

6

Working hoursa T1

0.42

0.49 -.27*** .14** -.15*** .02

.03

7

Time pressure T1

3.01

0.77

.15*** -.04

.01

-.02

.04

-.07

(.82)

8

Job control T1

3.26

0.63

.02

.01

-.01

-.02

.03

.01

-.11*

(.78)

9

Job control² T1

0.40

0.57 -.05

.05

-.04

-.01

.02

-.05

10 Emotional Exhaustion T1

2.84

0.96

.03

.01

.00

-.05

11 Emotional Exhaustion T2

2.93

1.03

.00

.02

-.06

2.12

0.90 -.06

2.12

Ac

13 Depersonalization T2

ed

pt

ce

12 Depersonalization T1

11

12

13

14

15

16

17

18

19

20

21

M

1

.52*** .34*** (.76)

14 Reduced prof. efficacy T1

2.17

0.64 -.08

.13** .01

-.06

-.04

.06

.03

-.13** .07

.28*** .20***. .41*** .25*** (.77)

15 Reduces prof. efficacy T2

2.16

0.64 -.09*

.13** .03

-.06

-.06

.12** .06

-.14** .05

.32*** .35***. .37*** .46*** .51*** (.79)

16 Vigor T1

5.18

1.11

.09*

-.04

-.01

.04

.00

-.05

-.15*** .23*** -.06

-.43***-.34***-.29***-.30***-.43***-.34*** (.78)

4.99

1.16

.14** -.07

-.01

.04

.02

-.07

-.09*

-.36***-.52***-.24***-.38***-.33***-.51*** .55*** (.78)

18 Dedication T1

5.55

1.23

.10*

-.02

-.06

.07

-.00

-.04

-.16*** .27*** -.05

-.46***-.38***-.36***-.35***-.45***-.36*** .76*** .51*** (.87)

19 Dedication T2

5.39

1.26

.14** -.04

-.04

.04

.03

-.07

-.13** .23*** -.11*

-.37***-.52***-.30***-.44***-.34***-.49*** .49*** .79*** .61*** (.88)

20 Absorption T1

5.26

1.29

.16*** -.02

-.04

.01

.04

-.06

-.11*

.25*** -.05

-.34***-.29***-.31***-.29***-.41***-.37*** .75*** .49*** .81*** .51*** (.88)

21 Absorption T2

5.08

1.29 .17*** -.07

-.01

.01

.05

-.09

-.09*

.21*** -.15***-.32***-.44***-.26***-.35***-.41***-.55*** .50*** .77*** .53*** .81*** .55*** (.87)

17 Vigor T2

.18*** -.10*

Page 32 of 35

i cr

33

us

TOO MUCH JOB CONTROL?

Ac

ce

pt

ed

M

an

Note. N = 556-587; pairwise deletion; a 1 = full-time employment, 0 = part-time employment; T1 = Time 1, T2 = Time 2; internal Consistencies as indicated by Cronbach’s Alpha are presented in parentheses in the diagonal; *p < .05, **p < .01, ***p < .001.

Page 33 of 35

i cr

34

us

TOO MUCH JOB CONTROL?

an

Toomuchjobcontrol? Twostudies on curvilinearrelationsbetweenjobcontrolandeldercareworkers’ well-being

•

M

Whatthispaperadds

Usingcross-sectional as well as longitudinaldata,thisstudyexamineswhetherjobcontrolexertslinear (as suggestedbythe Job Demands-

ed

Resources model) or curvilineareffects (as suggestedbythe Vitamin model) on eldercareworkers’ well-being. Cross-sectionally as well as longitudinally, curvilinearrelationswerefoundbetweenjobcontrolandeldercareworkers’ well-being.

•

Lowandhighlevelsofjobcontrolwereassociatedwithhigherlevelsofirritationcross-sectionally. In thelongitudinalstudy,

pt

•

Ac

 

ce

lowandhighlevelsofjobcontrolwereassociatedwithhigherburnout-levelsandlowerengagement-levels.

Page 34 of 35

i cr

35

us

TOO MUCH JOB CONTROL?

an

Too much job control?

What is already known about this topic?

Job control, defined as an employee’s freedom to decide over work tasks and work schedules as well as over the methods to execute these

ed

•

M

Two studies on curvilinear relations between job control and eldercare workers’ well-being

tasks is one of the most important job resources in almost every job.

pt

Yet, the effect of job control on work-related well-being is still a matter of debate. The Job Demands-Resources model proposes that job control has a generally motivational potential and is functional in maintaining employee well-being. However, the Vitamin model

ce

postulates that the beneficial effect of job control holds only up to a certain threshold. Once this threshold is exceeded, job control is believed to become detrimental for employee well-being.

•

Ac

What this paper adds

Using cross-sectional as well as longitudinal data, this study examines whether job control exerts linear (as suggested by the Job Demands-Resources model) or curvilinear effects (as suggested by the Vitamin model) on eldercare workers’ well-being.

•

Cross-sectionally as well as longitudinally, curvilinear relations were found between job control and eldercare workers’ well-being.

•

In the short term, low and high levels of job control were associated with higher levels of irritation. In the long term, low and high levels of job control were associated with higher burnout-levels and lower engagement-levels.

Page 35 of 35