A culture of care: for young or for old? : a quantitative study of the influence of informal caretaking on work stress and the roles of

(1)

A culture of care: for young or for old?

A quantitative study of the influence of informal caretaking on work stress and the roles of organizational health culture and age

Evita Dupker (10336761) 22-06-18

University of Amsterdam

MSc Business Administration – Leadership & Management Supervisor: Sofija Pajic

(2)

Statement of originality

This document is written by Evita Dupker. I declare to take full responsibility for the contents of this document. I declare that the work presented in this document is original, and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Acknowledgements

I would like to thank the people who have helped me realizing my thesis. First of all, I want to thank my supervisor, Sofija, for guiding me along the way with much enthusiasm, feedback, and calm when I freaked out about deadlines (or results, or literature, or anything else). Thanks to Karin, Erik-Jan and Jessy – freaking out about things together has never been this much fun. Moreover, thanks to my awesome and creative friend Esmee for decorating my front-page. Thanks to Dorien and Thomas for proofreading this thesis while you could have done thousands of other things with your time. I am lucky to have such sweet and smart people around me. Thanks to all respondents who took the time to fill out our survey, without you there would be little to present. Finally, thanks to my family and other friends for loaning me your network when I was searching for respondents, for occasionally pulling me away from the library to have some drinks in the pub, and for tolerating me when I still managed to talk about my thesis for the rest of those nights. From now on, I promise I will talk about other things again.

(4)

Abstract

Recently, the trend of increasing work stress seems to have become overshadowed by that of the rapidly ageing society. The old-age dependency ratio, or the number of citizens aged 65 or older as a percentage of the total potential working population, will be increasing up to 50 percent at its climax in 2040. Current governmental solutions leave a decreasing working population that has to both deal with increasing work stress and continue working until an older age, while increasing its contribution to informal caretaking. This study examines if high levels of health culture can support workers by reducing work stress while they continue working until an older age and increase their contribution to informal caretaking. Specifically, this study examines the influence of both informal caretaking and organizational health culture on work stress. Moreover, it investigates the role of age in these relationships. The conceptual model was tested using cross-sectional data that has been collected using an online survey. The survey was filled out by 402 workers from 117 different organizations. In order to test the hypotheses of the moderated moderation model, a correlation and a hierarchical multiple linear regression analysis have been performed. The findings show that informal caretaking relates positively but not significantly to work stress. Moreover, organizational health culture relates negatively and significantly to work stress, indicating that higher levels of organizational health culture lead to decreases in work stress. Furthermore, health culture non-significantly moderates the negative effect of informal caretaking on work stress. Also, informal caretaking relates negatively and non-significantly to work stress for older workers. Subsequently, age non-significantly moderates the positive effect of health culture on work stress. Last, there is a positive but non-significant three-way interaction between health culture, informal caretaking and age. From these findings can be derived that a culture of care is not necessarily applicable to young or to old, but to both young and old.

Key words: Work stress; Informal caretaking; Organizational health culture; Age;

(5)

List of tables, figures, appendices, abbreviations

Tables

Table 1. Sample characteristics 25

Table 2. Descriptive statistics 30

Table 3. Means, Standard Deviations, Correlations 32

Table 4. Regression analysis of influences on work stress 34

Figures

Figure 1. Conceptual model 23

Figure 2. Conditional effects of age, health culture and informal

caretaking on work stress 39

Figure 3. Final framework 39

Appendices

Appendix A. Survey questions of the main measures 58

Appendix B. Principal axis factoring analysis 60

Appendix C. Two-way interaction plots 61

Appendix D. SPSS syntax 62

Abbreviations

CBS The Central Bureau for Statistics

COR Conservation of Resources

ISCED International Standard Classification of Education

(7)

1. Introduction

The Dutch National Institute for Public Health and the Environment predicted in 2012 that increasing work stress would be the cause of the highest disease burden in the working population until at least 2020 (Eysink et al., 2012). In 2016, overstrain and burnout due to work stress already formed 74 percent of all reported occupational illnesses (van der Molen et al., 2017). Work stress even relates to a larger functional impairment at work than any physical condition: in 2012, 85 percent of the workers with a mental condition were declared fully disabled as opposed to 62 percent of the workers with a physical condition (Schaufeli & Bakker, 2013). There does not seem to be any direction of this trend turning around soon (Ybema et al., 2011; Weel, 2015; van der Molen et al., 2017; Dennerlein, 2017).

However, recently this trend seems to have become overshadowed by that of the rapidly ageing society. This is due to a decrease of births and an increase of life expectancy (Witkamp et al., 2017). The Central Bureau for Statistics (CBS) expects the percentage of citizens aged 65 or older to increase to 26 percent of the total population in 2040. The so-called old-age dependency ratio, or the number of citizens aged 65 or older as a percentage of the total potential working population, will be increasing to 40 percent in 2025, and even to 50 percent at its climax in 2040 (Witkamp et al., 2017; Stoeldraijer et al., 2017).

The older people get, the more care they need. This is mostly due to both the probability of getting a chronic disease and the number of chronic diseases per person increasing with age (Witkamp et al., 2017; Stoeldraijer et al., 2017). Thus, coherently with the increase of the old-age dependency ratio, the demand for care will be increasing with a yearly average of 4 percent (SER, 2016). Self-evidently, the increase of the old-age dependency ratio and the accompanying increase of the demand for care will lead to increasing social expenses such as retirement funds and health care costs (Witkamp et al., 2017; Stoeldraijer et al., 2017; Polder, 2017).

In order to deal with the increasing social expenses and demand for care, the Dutch government formulated different solutions. First, on January 1st 2018, the granted age for retirement was shifted from 65 to 66.5. From 2022 on, the granted age will be matched with the life expectancy, which means it will be increasing until at least 72 in 2050 (Rijksoverheid, 2015; van Dam et al., 2017; Stoeldraijer et al., 2017). Second, through installing acts such as the Social Support Act, the government has been steering to a shift from the so-called ‘welfare state’ to a ‘participation society’ where there is a focus on all citizens contributing their part in informal caretaking – all long-term care to people with chronic or disabling

(8)

conditions being given outside the context of a paid job (Rijksoverheid, 2013; de Klerk et al., 2017). This increase in contribution to informal caretaking will not just need adjustment in workers’ personal lives but due to factors such as time constraints also in their work lives (Witkamp et al., 2017; de Klerk et al., 2017; Bukman, 2018). Most informal caretakers take care of one or both parents. Moreover, 65 percent of the people who do not yet engage in informal caretaking are planning on doing so for one or both parents in the future (Witkamp et al., 2017; de Klerk et al., 2017). This implies that for the working population, especially for the workers aged 50 or above who often have parents in need of care, an increase in informal caretaking might relate to less free time, time for work, and work-life balance, and therefore to more work stress (Stoeldraijer et al., 2017; Ybema et al., 2017).

Thus, these solutions leave a decreasing working population that has to not just deal with increasing work stress, but also has to continue working until an older age while increasing its contribution to informal caretaking. As a consequence of these increases and therefore, ironically, the decreasing health of the working population, the necessity of investing in worker health has become apparent (van der Molen et al., 2017; Stoeldraijer et al., 2017; Pronk, 2017; Polder, 2017; Ybema et al., 2017, de Klerk et al., 2017). Organizations are to an increasing extent implementing health practices or programs. However, these are often not implemented as part of a larger system or strategy. This is one of the reasons why both participation numbers and effects of health practices and programs are often small (Loeppke et al., 2015; Rongen, 2015; Ybema et al., 2017).

In order to effectively reduce work stress, we should instead look at organizational health culture: the extent to which health is a central focus in the organization (Ybema et al., 2017). Health culture aims to uncover how health practices and programs and their effects can be not just additive but interactive with other factorsin the organization, such as performance measurement and rewarding. Therefore, it is expected to have larger health-related effects than the sum of its parts. These effects have the potential to translate themselves into actual competitive advantage (Ybema et al., 2011; Loeppke et al., 2015; Pronk & Narayan, 2016; Fabius et al., 2016; Ybema et al., 2017; Schramade, 2017; Boon et al., in press).

However, first, since most health-related research focuses on individual health practices or programs, research on health culture is still limited (Loeppke et al., 2015; Rongen, 2015; Ybema et al., 2017). Since health culture is expected to have larger health-related effects than health practices or programs due to its holistic approach (Loeppke et al., 2015; Pronk & Narayan, 2016; Fabius et al., 2016; Ybema et al., 2017; Schramade, 2017;

(9)

Boon et al., in press), this study takes into account the effect of health culture instead of individual health practices or programs.

Second, most health-related research focuses on the impact of health practices or programs on performance instead of on more personal outcomes such as work stress. When research does look at such outcomes, there is a dominant focus on physical health (Ybema et al., 2017). This study takes into account the effect of health culture on work stress instead of on performance measures in order to examine to what extent work stress declines when the level of health culture increases (Loeppke et al., 2015; Rongen, 2015; Ybema et al., 2017).

Third, since the combination of increasing work stress and increasing need for informal care has only recently come in sight as potentially problematic, research on the effects of informal caretaking on any work-related outcomes is limited. Existing studies mainly focused on effects of informal caretaking on personal outcomes such as mental and physical health (de Klerk et al., 2017). This study aims to contribute to the existing literature by focusing on potential negative effects of an increase of informal caretaking on specifically work stress. This way, the extent to which informal caretaking relates to loss of free time, time for work, health, and work-life balance and therefore adds to work stress can be examined. In addition, the effect of organizational health culture on the relationship between informal caretaking and work stress can be examined: does health culture moderate possible effects of informal caretaking on work stress?

Fourth, this study focuses on age-related differences in both the relationship between informal caretaking and work stress and the relationship between health culture and work stress. Will informal caretaking relate to higher levels of work stress for older workers, since they often have parents in need of care and are therefore expected to spend more time on informal caretaking (Stoeldraijer et al., 2017; Ybema et al., 2017)? Moreover, will health culture relate to a lower increase in work stress for older workers, since, due to their presumed lower physical capacity, gains from health culture would be higher (SER, 2016; Witkamp et al., 2017; Gagnon et al., 2017)?

Thus, this study will examine if high levels of health culture will be able to support workers by reducing work stress, while they continue working until an older age and increase their contribution to informal caretaking. It is important to look at the effects of informal caretaking and health culture on work stress in order to create ground for a discussion about who should contribute to what extent to informal caretaking, and to what extent health culture can serve as a solution that keeps not only the elderly, but also the working population healthy. The research question of this study is formulated as follows: To what extent does

(10)

informal caretaking influence work stress, and to what extent does this influence remain for workers in different levels of health culture and in different ages?

(11)

2. Theoretical framework

In this theoretical background, first the concept of work stress will be introduced shortly. Next, the concepts of informal caretaking and health culture, their relationships with work stress and the hypotheses that stem from these relationships will be examined. Moreover, the moderating effect of health culture on the relationship between informal caretaking and work stress and the hypothesis that stems from it will be examined. Subsequently, the moderating effects of age in the relationship between informal caretaking and work stress and in the relationship between health culture and work stress, and the hypotheses that stem from these relationships will be examined.

2.1 Work stress

Stress is a pattern of physical and emotional responses to so-called stressors, functioning to prepare the human organism for physical activity – to fight or flight. The National Institute for Occupational Safety and Health (NIOSH) defines work stress as the harmful physical and emotional responses that occur when the requirements of the job do not match the capabilities, resources, or needs of the worker (European Commission, 2000).

Not all stress is necessarily destructive. According to Lazarus’ coping theory, stress responses are the consequences of certain interpretations of reality. Stressors interpreted as threatening by one person can be interpreted as irrelevant or as a challenge by another. Stress responses only come into play when a stressor is interpreted as threatening (Schaufeli, 2000).

Moreover, stress increases the chance of survival when in acute danger. Short-term stress acts as an impetus for the immune system: slightly stressed patients for instance seem to recover faster after a surgical operation (de Visser, 2013). However, when people experience levels of stress that are either too high or of chronic nature, stress does become destructive. The hormone cortisol, normally meant to reinstate balance after a stress response by fabricating extra glucose, will fail to leave the body on its own, causing all kinds of physical disturbances (de Visser, 2013; Eysink et al., 2012).

In addition to Lazarus’ coping theory, conservation of resources (COR) theory considers not just internal but also environmental elements in investigating the development and protection of so-called resources in stress-related processes (Schaufeli, 2000; Hobfoll, 2001; Hobfoll, 2010). The theory posits that individuals strive to obtain, retain, protect, and foster resources. These resources are defined as objects, personal characteristics, conditions,

(12)

or energies that are valued by the individual (Hobfoll, 1989). They can for instance consist of manageable workload, personal health, family stability, help with care, or understanding and support from coworkers and management. Stress occurs when after significant resource investment one’s resources are threatened or lost. Moreover, people who lose resources are more vulnerable to losing additional resources and therefore to entering a so-called loss spiral, whereas people who possess resources are more inclined to gain additional resources and therefore to enter a so-called gain spiral. However, since loss is more potent than gain, loss cycles will be more influential and more accelerated than gain cycles (Hobfoll, 1989; Hobfoll, 2001; Hobfoll, 2010).

Different factors can lead to threatened or even lost resources, and therefore to increasing work stress. Examples of these factors are high workload (Weel, 2015), globalization and accelerating organizational changes (Schaufeli, 2000), individualization (Schaufeli, 2000; Eysink et al., 2012), increasing expectations such as development opportunities (van der Molen et al., 2017), and an increase in cases of infringement of the psychological contract. This entails the deteriorating reciprocal relationship between employer and employee. Organizations are asking more of their employees while they are offering less, for example less career path opportunities and less permanent contracts (Schaufeli, 2000; van der Molen et al., 2017).

The personal and societal damage of work stress is enormous. In addition to the physical disturbances, work stress relates to a decrease in personal wellbeing, in overstrain and in burnout through for example decreasing job satisfaction and engagement (Schaufeli, 2000; de Visser, 2013; Weel, 2015). As stated before, workers with a mental condition (e.g., burnout) even have higher chances on being declared fully disabled than workers with a physical condition. Thus, work stress does not just relate to a relatively big outflow of the working population and accompanying negative personal effects, but also to decreasing organizational health. This is the state in which an organization is able to operate and adapt efficiently, improve intrinsic worker competence and satisfaction, and be profitable (Schaufeli & Bakker, 2013; Meng et al., 2014).

Subsequently to outflow of the working population there are associated costs. Collective expenses due to chronic work stress in the Netherlands amount around fifty billion euros per year. These costs are passed on to the government, employers, health insurance companies, and thus to all people paying taxes and premiums. Looking at both the perspective of increasing work stress and its substantial personal and societal damage, stress prevention should be a top priority (van Winkelhof & Bakker, 2017).

(13)

2.2 Informal caretaking

The rapidly ageing society and the accompanying awaited increase in contribution to informal caretaking add to the list of arguments for prioritizing stress prevention (Stoeldraijer et al., 2017). Informal caretaking can be defined as all long-term care to people – both known and unknown – with chronic or disabling conditions being given outside the context of a paid job. This is a broad definition containing both volunteering through care organizations and providing any form of help to dependent people in the social environment, such as family or neighbors (de Klerk et al., 2017).

In 2017, 36 percent of Dutch citizens above 16 engaged in informal caretaking (CBS, 2018a). However, the need for care is expected to increase strongly until at least 2040. Additionally, the willingness to engage in informal caretaking is increasing to a much lesser extent than expected (de Klerk et al., 2017; Bukman, 2018). Since 2014, the number of people that started engaging in informal caretaking has been increasing slightly among citizens between the age of 55 and 64, but as opposed to the goal of the supportive government policies, little else has changed (de Klerk et al., 2015; de Klerk et al., 2017; Bukman, 2018). This can be explained using social network theory, from which can be derived that the willingness to help becomes larger the closer people in need of care are linked to others in their social network (de Klerk et al., 2015; Liu et al., 2017). Most informal caretakers for instance help a parent. Moreover, 65 percent of the people not engaging in informal caretaking state to be planning on helping a parent in the future (de Klerk et al., 2017; CBS, 2018a). People also take care of or are planning on taking care of people who are less closely linked to themselves, such as other family members, friends, or neighbors, but to a much lesser extent (de Klerk et al., 2015; CBS, 2018a; de Klerk et al., 2017; Bukman, 2018). Therefore, the number of people engaging in informal caretaking will relatively decrease, leaving more ageing citizens without the care they need (de Klerk et al., 2017; Bukman, 2018).

People have different reasons for not engaging in, or planning on engaging in informal caretaking. 50 percent of the people not engaging in informal caretaking attributes this to time constraints, mostly due to a fulltime job and/or raising children. Moreover, 20 percent deals with health issues of their own. Other reasons are a long travel time to or a bad relationship with the people in need of care. Subsequently, 63 percent of the people not engaging in informal caretaking take the view that the primary responsibility for informal caretaking is that of the government, which is why they do not prioritize it. Coherently, the number of people that says people in need of care should receive that care as much as possible from the

(14)

social environment has been rapidly decreasing from 41 percent in 2010, to 23 percent in 2016 (de Klerk et al., 2017; Bukman, 2018). These statements imply that an increasing number of people take the view that there should be more limits to both the care expected to be provided by the social network, and to whom that care should be provided. This is opposed to the government’s efforts to get more people involved in providing informal care of not just parents, but also of other people in their social network (Rijksoverheid, 2013; Witkamp et al., 2017; de Klerk et al., 2017; Bukman, 2018).

An underlying reason for unwillingness to engage in informal caretaking might be the fact that it can have serious health-related consequences. According to Lazarus’ coping theory, informal caretaking can be seen as threatening and therefore as contributing to stress when for example one experiences it as emotionally heavy or when one engages in it because there is no one else able to do it (Schaufeli, 2000; de Klerk et al., 2015). Moreover, in line with COR theory, engaging in informal caretaking requires significant investment of resources such as time and energy. This accounts especially for workers who are dealing with other factors that are threatening resource loss, and are therefore increasing work stress already (Hobfoll, 2001; Hobfoll, 2010).

Such factors are for instance the number of hours per week one engages in informal caretaking, the number of hours per week one works, the amount of recognition one gets for helping out, and the ability to take days off on short notice, to determine working hours, or to work from home (de Klerk et al., 2015; de Klerk et al., 2017; Stoeldraijer et al., 2017). Since 83 percent of informal caretakers are employed, and 72 percent of the male and 28 percent of the female informal caretakers work 32 hours per week or more, especially work-related factors are shown to contribute largely to the loss of free time, time for work, work-life balance, and therefore to increasing work stress. 11 percent of the informal caretakers for instance say they need to interrupt their work multiple times a week in order to help someone (CBS, 2018a; de Klerk et al., 2017; Stoeldraijer et al., 2017).

Thus, investing in informal caretaking can threaten resources such as free time, time for work, family stability, and personal health. Therefore, workers who invest in informal caretaking are at even higher risk of work stress and of triggering a loss spiral than workers in general (Hobfoll, 2001; Hobfoll; 2010). These risks are reflected in health reports of informal caretakers. Negative health-related consequences have been steadily increasing since 2016. Sick leave of informal caretakers is both higher and lasts longer than sick leave of workers who do not engage in informal caretaking (Josten & de Boer, 2015; Bukman, 2018). 10 percent of the informal caretakers say to feel too heavily ballasted, and 6 percent say to be

(15)

overstrained. Percentages regarding feelings of being too heavily ballasted even increase to 16 percent for people taking care of a partner and to 20 percent for people taking care of a child. Similar percentages account for informal caretakers of terminally or psychologically ill people (de Klerk et al., 2017).

Thus, in order to instigate people to engage in informal caretaking, either the government or organizations themselves should facilitate processes or policies that aim to integrate work and care. Only this way, negative health- and performance-related consequences of informal caretaking can be reduced (Rijksoverheid, 2013; de Klerk et al., 2017; Loeppke et al., 2015; Rongen, 2015; Ybema et al., 2017).

As engaging in informal caretaking can threaten resources such as free time, time for work, and family stability, and therefore relate to increasing work stress, the following hypothesis is presented:

H1: Informal caretaking positively relates to work stress. Engaging in informal caretaking

relates to higher levels of work stress for workers.

2.3 Health culture

Thus, different factors that might increase work stress are expected to play an increasing role in society (van der Molen, 2017; Stoeldraijer et al., 2017; de Klerk et al., 2017). However, work stress can be diminished effectively. Any practice aimed at the so-called relaxation-response leads to a decrease in metabolism, a slower heartbeat, a lower blood pressure, higher cognitive performance, and more relaxed muscles and respiration (van Winkelhoven & Bakker, 2017).

In 2014, the Ministry of Social Affairs and Employment introduced the so-called yearly ‘Week of Work Stress’ with the goal of increasing national awareness of work stress by organizing events about reducing work stress in collaboration with different organizations. However, due to – ironically – workload and time constraints, the Ministry handed off the responsibility for organizing this week to other businesses (Corré, 2017). In 2017, the Ministry of Health, Wellbeing and Sport announced a National Prevention Agreement in which it will collaborate with care organizations, health insurance companies, municipalities, sports and other societal organizations in order to reduce the number of preventable illnesses (van Elst, 2018). However, the focus of the agreement appears to be limited to smoking, obesity and alcohol use. When one takes into account first the personal and societal damage

(16)

of work stress, and second the fact that smoking, obesity and alcohol use issues are all three already declining in the Netherlands, this focus can be criticized as incomplete. Thus, regarding stress prevention, little is to be expected of the government in the near future (van Winkelhoven & Bakker, 2017; van Elst, 2018).

However, organizational focus on worker health and more specifically on stress prevention is growing (Loeppke et al., 2015; Rongen, 2015; Ybema et al., 2017). Often pushed by increasing health care costs, organizations first started installing specific health practices such as healthy lunch menus or on-site sports facilities (Pronk, 2012; Loeppke et al., 2015; Rongen, 2015). Nowadays, an increasing number of organizations extend their approach to worker health by introducing overarching health programs such as stress prevention or fitness programs. Many of such programs are shown to improve physical activity (Pronk, 2012; Loeppke et al., 2015; Rongen, 2015), healthy eating (Pronk et al., 2010; Pronk, 2012; Rongen, 2015), weight status (Rongen, 2015), job satisfaction (Loeppke et al., 2015; Fabius et al., 2016; Ybema et al., 2017), productivity (Pronk et al., 2010; Pronk, 2012; Loeppke et al., 2015; Rongen, 2015; Fabius et al., 2016), and performance (Cascio, 2006; Loeppke et al., 2015). In addition, they reduce work stress (Loeppke et al., 2015; Fabius et al., 2016; Ybema et al., 2017), presenteeism (Loeppke et al., 2015; Fabius et al., 2016), absenteeism (Loeppke et al., 2015; Fabius et al., 2016; Ybema et al., 2017), burnout (Fabius et al., 2016; Ybema et al., 2017), depression (Pronk, 2012; Fabius et al., 2016), and chronic illness (Pronk et al., 2010; Loeppke et al., 2015; Rongen, 2015). On the organizational level, these factors relate to decreasing turnover and health care costs (Fabius et al., 2016; Women INC., 2016; Ybema et al., 2017), improving performance (Loeppke et al., 2015; Fabius et al., 2016), a positive return on investment (Fabius et al., 2016; Pronk, 2017), profit (Fabius et al., 2016), and a good reputation (Pronk, 2012; Women INC., 2016). Moreover, according to COR theory, people who invest in their resources by engaging in such health programs are more inclined to gain additional resources – the gain spiral. This implies that the factors stated above reinforce each other (Hobfoll, 1989; Hobfoll, 2010).

However, both participation numbers and effects of health practices and programs are often small and do not last over time (Loeppke et al., 2015; Rongen, 2015; Fabius et al., 2016; Ybema et al., 2017). This has different reasons. First, health practices and programs are often implemented as isolated measures instead of as part of a larger system or strategy. Therefore, organizations often focus too much on only one or a few aspects of worker health and fail striking the root of the problem (Loeppke et al., 2015; Rongen, 2015; Ybema et al., 2017).

(17)

Second, many health practices and programs promote and focus on individual responsibility, while the social environment is of crucial importance in practices and programs that require both changing and maintaining changed behavior (Pronk, 2012; Rongen, 2015; Pronk & Narayan, 2016; Fabius et al., 2016; Sorensen et al., 2016; Critelli, 2017). Health programs that mainly rely on self-control for instance relate to smaller participation numbers and effects than programs that include active coaching and check-ups. Moreover, programs that include not just the worker but also family members are shown to be even more effective due to their incorporation of the social environment (Critelli, 2017). This social environment can also include coworkers: healthy behavior and participation of coworkers in health practices or programs are shown to be large predictors of behavior and participation of the workers around them (Critelli, 2017; Bot et al., 2018).

Third, research shows discrepancy between employer and worker opinions regarding the extent and quality of health practices and programs in their organizations. This discrepancy might be due to insufficient implementation of those practices or programs – organizations often seem to underestimate the time, the steps or the number of people necessary for behavior to change in a sustainable manner. For instance: 95 percent of human behavior is controlled by unconscious, automatic processes (van Vuuren & Marcelissen, 2017). Therefore, organizations should bear in mind that creating awareness of the need for, and the availability of health programs is an important first step in increasing health. This step is often passed over (Pronk, 2012; Rongen, 2015; van Vuuren & Marcelissen, 2017;Bot et al., 2018).

Fourth, the workers participating in health practices and programs are not necessarily the ones most in need of improving their health. While participation by anyone is already a gain, participation by the workers most in need of health improvement conducts the largest gains (Rongen, 2015; Fabius, 2016; Ybema et al., 2017; Gagnon et al., 2017). However, not everyone possesses similar capacity to change behavior (van Vuuren et al., 2016; WRR, 2017; Ybema et al., 2017). Differences in capacity of behavioral change are shown to be dependent on differences in personality, sex, education level, length and type of employment contract, and living conditions (van Vuuren et al., 2016; WRR, 2017).

In order to overcome these issues and effectively reduce work stress, new research based on systems science proposes that we should look at health culture instead of at individual health practices or programs (Schramade, 2017; Ybema et al., 2017). Health culture can be defined as the extent to which health is a central focus in the organization (Ybema et al., 2017). Health should be recognized as complex in its biological, social,

(18)

psychological and societal determinants (Pronk & Narayan, 2016). It is not merely the absence of disease but a state in which these determinants are converged and satisfied (WHO, 2017). Employers need to recognize that the workplace is a fruitful setting for health investment due to the presence of natural social networks, the possibility of reaching a large population, and the amount of time that workers spend in their organization. Simultaneously, due to these same factors the workplace should be seen as a complex social environment (Rongen, 2015; Pronk & Narayan, 2016). Systems science provides an opportunity to assess the holistic nature of the workplace as a complex social environment and all its components working together to fulfill some purpose that cannot be accomplished by looking at individual parts in isolation. A sharper focus on systems science will uncover unexplored potential in improving workplace solutions to work stress (Rongen, 2015; Pronk & Narayan, 2016; Fabius, 2016; Ybema et al., 2017; Schramade, 2017; Boon et al., in press).

Thus, it is not enough to install practices or programs aiming to prevent stress. The focus should be on creating an overarching health culture in order to uncover how effects of health practices or programs can be not just additive but interactive with other factors in the organization, such as performance measurement and rewarding. This health culture contains strategic value placed on not just stress prevention but on all aspects of worker health (Loeppke et al., 2015; Rongen, 2015; Ybema et al., 2017). Moreover, it shifts individual responsibility to organizational responsibility due to a strong focus on the environment. It ensures that everybody takes part: not just workers, but also their coworkers and their families. This is done indirectly by practices meant to increase autonomy and decrease work-life conflict such as flexible work hours, a flexible work schedule, and a flexible workspace (Ybema et al., 2011; Loeppke et al., 2015; Pronk & Narayan, 2016; Ybema et al., 2017), and directly through family-inclusive programs and through programs and processes that facilitate peer support (Critelli, 2017; Bot et al., 2018). Subsequently, an overarching health culture enables management to both signal importance of and nudge workers to better health (Loeppke et al., 2015; Pronk & Narayan, 2016; Polder, 2017).

According to COR theory, health culture does not only have the potential to neutralize work stress, but also to facilitate a gain spiral in which workers replenish and gain new resources (Hobfoll, 2001; Hobfoll, 2010). It is expected to increase positive effects of health practices and programs due to relative increases in positive relationships with coworkers and management, job satisfaction, and engagement (Rongen, 2015; Loeppke et al., 2015; Pronk & Narayan, 2016; Fabius, 2016; Ybema et al., 2017). These relative increases can be seen as actual competitive advantage. Therefore, health culture creates more value than the sum of its

(19)

parts. It relates to a sustainable change of behavior – system change due to holistic thinking instead of box ticking (Rongen, 2015; Pronk & Narayan, 2016; Fabius, 2016; Ybema et al., 2017; Schramade, 2017; Boon et al., in press). Moreover, these relative increases in positive relationships with coworkers and management, job satisfaction, and engagement show to be specifically important in contributing to stress prevention. Positive relationships with and support of coworkers and management are for instance important because coworkers or management can help reduce stress in various ways. They can show support or provide distraction, but also help by providing feedback or by taking over some work (Pronk & Narayan, 2016; Sorensen, 2017; Mol & de Vries, 2009). These facts lead to the following hypothesis:

H2: Health culture negatively relates to work stress. Higher levels of organizational health

culture relate to lower levels of work stress for workers.

2.4 The moderating role of health culture

On the one hand, investing in informal caretaking can threaten resources such as free time, time for work, family stability, and personal health (de Klerk et al., 2015; de Klerk et al., 2017; Stoeldraijer et al., 2017). Health culture, on the other hand, is expected to negatively relate to work stress due to its potential to both neutralize resource threat or loss and facilitate a gain spiral in which workers replenish and gain new resources (Hobfoll, 2001; Hobfoll, 2010).

Foremost, this is due to work-related resources such as the ability to take days off on short notice, to determine working hours, or to work from home (Ybema et al., 2017; de Klerk et al., 2017; Stoeldraijer, 2017). Subsequently, this is due to health-related resources such as the integrated practices and programs that relate to various positive health-related outcomes, among which decreasing work stress (Loeppke et al., 2015; Fabius et al., 2016; Ybema et al., 2017). Moreover, this is shown in the relative increases in job satisfaction, engagement, and positive relationships with coworkers and management. These are expected to reinforce the positive effects of all health-related outcomes, but especially to decrease stress due to their ability to provide support in various ways (Rongen, 2015; Loeppke et al., 2015; Pronk & Narayan, 2016; Fabius, 2016; Ybema et al., 2017; Sorensen, 2017; Mol & de Vries, 2009).

Moreover, according to COR theory, stress occurs when after significant resource investment one’s resources are threatened or lost (Hobfoll, 2001; Hobfoll, 2010). Informal

(20)

caretaking requires investment of resources such as free time, time for work, and energy (de Klerk et al., 2017; Stoeldraijer et al., 2017). Therefore, informal caretakers are at higher risk of work stress and of triggering a loss spiral than workers in general. This implies that in a health culture that does not just neutralize resource threat or loss, but also facilitates resource gain, higher health risks might relate to higher health-related gains (Hobfoll, 2001; Hobfoll, 2010; Gagnon et al., 2017). Therefore, health culture is expected to be particularly beneficial for informal caretakers in decreasing work stress. This leads to the following hypothesis:

H3: Health culture moderates the negative effect of informal caretaking on work stress.

Informal caretaking relates to a lower increase in work stress for people working in higher levels of organizational health culture.

2.5 The moderating roles of age

Due to the ageing society and accompanying government solutions, the working population is ageing as well (Rijksoverheid, 2015; van Dam et al., 2017; Stoeldraijer et al., 2017). Moreover, 63 percent of informal caretakers are between the age of 50 and 75 (CBS, 2018a). Since people are expected to both work until an older age and increase engagement in informal caretaking, not just the age, but also the number of informal caretakers in the working population will be increasing (Rijksoverheid, 2015; van Dam et al., 2017; Stoeldraijer et al., 2017; Rijksoverheid, 2013; de Klerk et al., 2017).

This combination might have implications for work stress. More specifically, informal caretaking might relate to higher levels of work stress for older workers. First, this is because older workers appear to invest more of their resources (e.g., time) in informal caretaking than younger workers (Stoeldraijer et al., 2017; de Klerk et al., 2017).

Additionally, older workers deal with increasing resource threat or loss due to age (Hobfoll, 2001; Hobfoll, 2010). They for instance increasingly deal with health issues such as chronic diseases (SER, 2016; CBS, 2018a; Witkamp et al., 2017). Also, they deal with a slight reduction in mental capacity due to the slowing of cognitive processes (de Lange et al., 2005; Salas et al., 2012; de Klerk et al., 2017). These issues are reflected in absence from work: workers aged between 55 and 64 are shown to be absent almost twice as much as workers between 25 and 34 (Kompier & Geurts, 2015). Moreover, female workers in the age category between 41 and 50, and male workers in the age category between 51 and 60 are

(21)

shown to be the highest risk groups for occupational illnesses due to work stress (van der Molen et al., 2017).

Moreover, older workers are often stereotyped as less flexible and less tolerant in respect of changes at work. Due to this, older workers are involved in activities that connote change less often. This leads to a negative relationship between worker age and impediments due to changes at work, which relates to increasing work stress (de Lange et al., 2005; Schreurs et al., 2012).

Furthermore, older workers increasingly value qualitative over quantitative social relationships (de Lange et al., 2005; Schreurs et al., 2012). Therefore, they prefer to invest in fewer strong ties, often more in the family sphere, over developing an extended network full of weaker ties that is often more in the professional sphere (Seibert et al., 2001). A decrease of investment in the professional network leads to a decrease of coworker and management support, which is shown to be one of the most important sources for neutralizing work stress (de Lange et al., 2005; Schreurs et al., 2012; Baeten, 2014).

Thus, older workers first appear to invest more time in informal caretaking than younger workers. Second, due to health-, change- and network-related reasons, older workers deal with increasing resource threat or loss. This combination is expected to put older workers under greater strain, which might trigger further resource loss and result in higher levels of work stress (Hobfoll, 2010; Stoeldraijer et al., 2017; de Klerk et al., 2017). This leads to the following hypothesis:

H4: Age moderates the negative effect of informal caretaking on work stress. Informal

caretaking relates to higher levels of work stress for older workers.

However, two side notes need to be made. First, in these theories ‘older’ and ‘younger’ workers are not defined in clear age categories. Moreover, it is presumed that all older workers are in both a worse physical and a worse mental state than younger workers (de Lange et al., 2005; Schreurs et al., 2012). These side notes should be taken into consideration in the analysis of all potential moderating roles of age.

As opposed to engaging in informal caretaking, working in organizations with high levels of health culture might relate to a lower increase in work stress for older workers. First, next to the reasoning that older workers deal with increasing resource threat or loss due to the health-, change- and network-related reasons stated above, they are also presumed to dispose of compensating strategies. They for example have more life experience, which can be

(22)

applied on resource investment and on dealing with resource threat (de Lange et al., 2005; Schreurs et al., 2012). This can be done in the form of efficiently combining high workload with informal caretaking. Additionally, older workers may dispose of a higher emotional maturity causing them to be more selective in resource investment. They are for instance more selective in what they want to achieve at work and how to achieve it (de Lange et al., 2005; Schreurs et al., 2012). Health culture contains aspects that facilitate flexibility and adaptability, such as the ability to take days off on short notice, to determine working hours, or to work from home. This offers older workers the space to apply their compensating strategies, which will reinforce its positive effects (Ybema et al., 2017; Gagnon et al., 2017).

Second, since older workers in general increasingly deal with health issues, gains due to high levels of health culture might also be higher (Rongen, 2015; Fabius et al., 2016; Witkamp et al., 2017; de Klerk et al., 2017; Ybema et al., 2017; Gagnon et al., 2017). Moreover, health culture facilitates a gain spiral in which older workers can replenish the resources such as free time or time for work they invested in for instance balancing workload and informal caretaking. This implies that older workers stay healthier and would therefore have even more resources left to invest in informal caretaking (Hobfoll, 2010).

Third, intrinsic work motives appear to gain value with age. Older workers for example are less interested in training and advancement or in status, and more in security, health, and the earlier mentioned qualitative social relationships (de Lange et al., 2005; Ybema et al., 2011; Schreurs et al., 2012; Baeten, 2014). Health culture is shown to increase not just health-related factors but also to lead to relative increases in positive relationships with coworkers and management, job satisfaction, and engagement (Rongen, 2015; Loeppke et al., 2015; Pronk & Narayan, 2016; Fabius, 2016; Ybema et al., 2017; Sorensen, 2017). These can be seen as factors that are important in neutralizing work stress – especially for older workers who tend to invest less in their professional network (de Lange et al., 2005; Schreurs et al., 2012; Baeten, 2014; Ybema et al., 2017; Mol & de Vries, 2009). This reasoning leads to the following hypothesis:

H5: Age moderates the positive effect of health culture on work stress. Health culture relates

to a lower increase in work stress for older workers.

Thus, age is expected to both moderate the negative effect of informal caretaking on work stress and the positive effect of health culture on work stress. Therefore, a three-way

(23)

interaction between health culture, informal caretaking, and age is expected (Agresti & Finlay, 2009). This leads to the following hypothesis:

H6: There is a three-way interaction between health culture, informal caretaking and age.

The interaction between health culture and informal caretaking is stronger among older workers.

(24)

3. Methods

3.1 Procedure

This study is of explanatory nature. In order to answer the research question, a quantitative research with a correlational design is performed. In a group of four Master of Business Administration students, we collected cross-sectional multi-level data using a survey that we administered online. We created the survey by combining sets of adopted measurement items from other questionnaires. Due to our target population and sample, we chose to provide both an English and a Dutch survey option. Therefore, we contacted all authors we used measurement items from of which we needed translations. We made use of parallel translation for the questions we did not receive translations of (Toepoel, 2016).

After creating the survey, we performed a pilot test by selecting 10 people to fill out and reflect on the survey. This feedback proved to be useful and resulted in some changes regarding structure, length and translation of questions, and in the addition of some clarifying examples of statements.

In the period between April 5 and May 4we conducted research using non-probability sampling, mainly in the form of volunteer sampling (snowball sampling). We aimed to reach a sample size of 80 organizations and 400 workers. We sent out e-mails to organizations both in- and outside of our personal networks and asked between four and ten employees of every organization to take part in our survey. We sent out reminders around one week after our first approach. Some of us additionally used convenience sampling by asking organizations to take part in the survey by posting messages on social media such as Facebook and LinkedIn. In order to incentivize workers, we used gift vouchers that we randomly divided over the respondents after closing the survey. Due to the distribution of the survey on social media and to not all group members tracking their sample, the overall response rate is hard to predict. When accounted for my individual contribution to the snowball sampling, the organizational response rate would be 78%.

3.2 Sample

The focus in the literature used in this study is on Dutch workers. However, an ageing population, an increasing workload and an increasing demand for informal caretaking all show to be trends that are relevant not just in the Netherlands, but also elsewhere in Europe.

(25)

Therefore, the population of interest for this study consists of European workers (European Commission, 2014; WHO, 2017).

The outcomes of the survey relate to a sample of 402 workers who were employed in 117 organizations. Since the target population consists of European workers (European Commission, 2014; WHO, 2017), only respondents working in European countries were selected. In addition, based on the international CBS-definition of the working population, respondents under the age of 18 and respondents working less than 12 hours per week were filtered out (Dirven & Janssen, 2012). This approach is aimed to separate possible information of student jobs above 12 hours per week from information of part-time jobs. Moreover, only respondents who answered all questions used in this study were selected. This relates to a valid sample of 366 workers from 114 organizations.

Table 1: Sample characteristics

Variable Min Max M SD N

Informal caretaking (yes = 1) 0 1 0.19 0.39 396 Age

Sex (male = 1) Low education level Medium education level High education level Working hours 0 0 0 19 0 0 0 0 12 65 1 1 1 1 76 33.86 0.46 0.08 0.11 0.81 33.37 12.04 0.50 0.28 0.31 0.39 12.78 388 395 395 395 395 366 Note: Valid N = 366.

As shown in Table 1, the distribution of sex in this sample is fair: 53.9% of the respondents identify as female. Moreover, respondent age ranges from 19 to 65, with an average age of 33.86. In addition, 8.4% of the respondents obtained a low education level, indicating completion of high school or lower. 10.6% of the respondents obtained a medium education level, indicating completion of lower vocational educational training. 81% of the respondents obtained a high education level, indicating completion of higher vocational educational training or a university degree (bachelor, master or PhD). These variables are recoded using international ISCED-guidelines (CBS, 2018b).

Furthermore, 94.2% of the respondents are working in The Netherlands. Other countries respondents work in are Croatia (0.3%), Czech Republic (1%), France (0.3%), Germany (1.5%), Greece (0.3%), Montenegro (0.3%), Turkey (1.3%), and the United Kingdom (1%). Most respondents work in the health science industry (15.5%), followed by marketing, sales and service (12.6%) and finance (9.7%). Respondents work between 12 and 76 hours per

(26)

week (including overtime), with an average of 33.37 hours.They have been working for their organizations for 6.84 years on average.

Moreover, 19.2% of the respondents engage in informal caretaking. 63.2% of the informal caretakers identify as female. Age ranges from 20 to 65, with an average age of 44.63. 5.3% of the informal caretakers obtained a low education level, 9.2% obtained a medium education level, and 85.5% obtained a high education level.Informal caretakers work between 12 and 76 hours per week (including overtime), with an average of 34.12 hours. They have been working for their organizations for 10.6 years on average. They spend on average 5.60 hours per week on informal caretaking.

3.3 Measures

The main variables used in this study are work stress, informal caretaking, health culture and age. Work stress (α = 0.80) is measured with the four items of Motowidlo et al. (1986). Two of them are indicative and two of them are counter indicative. These items are chosen on account of the NIOSH definition of work stress as the harmful physical and emotional responses that occur when the requirements of the job do not match the capabilities, resources, or needs of the worker (European Commission, 2000). An example item is ‘My job is extremely stressful’. All items are measured on a 5-point Likert scale.

Informal caretaking is measured with the binary item ‘Do you engage in informal caretaking?’ of de Klerk et al. (2015). This item is chosen on account of their definition of informal caretaking as all long-term care to people – both known and unknown – with chronic or disabling conditions being given outside the context of a paid job (de Klerk et al., 2017).

Health culture (α = 0.87) is measured using seven items of Ybema et al.’s (2017) Health Policy Scale. These items are chosen on account of their definition of health culture as the extent to which health is a central focus in the organization. An example item is ‘In my organization, sufficient efforts are made to protect the health of employees’. All items are measured on a 5-point Likert scale.

Age is measured using the standard ratio item ‘What is your age?’ The variable is filtered for respondents aged between 18 and 75, according to the international CBS-definition of the working population(Dirven & Janssen, 2012).

(27)

3.3.1 Control variables

In order to take into account possible spurious relationships while examining the hypotheses, a few other variables are controlled for. These control variables are sex, education level, working hours, and workability. Both the main variables and the control variables are shown in Table 1.

The variable ‘sex’ is recoded into a binary variable (male = 1). Due to the structural disadvantages women experience at the work floor, they are expected to experience higher levels of work stress than men. Examples of these structural disadvantages are masculine stereotypes at work causing career blocks, a less positive experience of the work environment, lower salaries, and less autonomy. Subsequently, there still is the stereotype of women being socialized with the idea that they should approach stressful situations in a more insecure manner than men, resulting in a larger number of stress symptoms. Besides these factors women are expected to experience more role conflict and overload, since they often still bear most of the responsibility for managing the household and raising children (Jick & Mitz, 1985; Matud, 2004). The fact that women engage in informal caretaking more often than men is in line with this (de Klerk et al., 2015; de Klerk et al., 2017; CBS, 2018a).

The variable ‘education level’ is recoded into the three binary variables ‘low education level’ (8.4%), ‘medium education level’ (10.6%) and ‘high education level’ (81%) on account of international ISCED-guidelines. A low education level indicates completion of high school or less, a medium education level indicates completion of lower vocational educational training, and a high education level indicates completion of higher vocational educational training or a university degree (bachelor, master or PhD) (CBS, 2018b). According to Maslach, Schaufeli and Reiter (2001) there is a positive relationship between education level and work stress. First, this is due to the increasing number of responsibilities that go hand in hand with an increase in education level. These responsibilities imply a need for resource investment that can relate to increases in work stress levels. Second, higher educated workers appear to have higher expectations of their job, which relates to work stress when these expectations are not met. Therefore, increases in education level are expected to relate to increases in work stress.

The variable ‘working hours’ entails the number of working hours per week including overtime, ranging from 12 to 76. On account of the international CBS-definition of the working population that is used, the minimum number of working hours is set at 12 per week (Dirven & Janssen, 2012). The number of working hours per week is expected to positively relate to work stress. The more hours one has to invest in working, the more tired and

(28)

possibly stressed one becomes. In addition, with more hours invested in work, fewer hours will be left for recovering from the effort put in at work, or for reinstating work-life balance. The increasing necessary effort, and the lack of time for recovery and for reinstating work-life balance can cause work stress (Schaufeli et al., 2008; de Klerk et al., 2017; Stoeldraijer et al., 2017).

The variable ‘work ability’ is measured using the item ‘Is your ability to work reduced because of illness, accident, or being worn down?’ and recoded into a binary variable (no = 1). Work stress is expected to relate to decreases in work ability. Moreover, decreases in work ability are expected to relate to increases in work stress, initiating a loss spiral (Bethge et al., 2009; Li et al., 2016; Hobfoll, 2010).

(29)

4. Results

The following section reports the results of the analyses performed in this study. First, the results of a principal factor analysis will be presented. Second, the descriptive statistics, sample distribution and normality checks will be presented. Third, the results of a correlation analysis will be presented. Last, the results of the hierarchical multiple linear regression analysis will be presented. The direct and conditional effects of the used variables and the six hypotheses will be discussed.

4.1 Factor analysis

In order to inspect the measurement model of the self-rated variables used in this study, a principal component analysis (PCA) was conducted on the items measuring ‘work stress’ and ‘health culture’. The Kaiser-Meyer-Olkin measure verified the sampling adequacy for the analysis (KMO = .843). Bartlett’s Test of Sphericity (χ2 (55) = 1730.576, p < .001) indicated that correlations between items were sufficiently large for PAF (Agresti & Finlay, 2009). An initial analysis was run to obtain eigenvalues for each component. Two components had eigenvalues over Kaiser’s criterion of 1 and in combination explained 59.23% of the variance. Four items representing work stress loaded highly on the factor ‘work stress’ (factor loadings between .729 and .835) and seven items representing health culture loaded highly on the factor ‘health culture (factor loadings between .686 and .818 – see Appendix A for all factor loadings). These results support work stress and health culture as two distinct constructs and all items used in these scales as measuring either the one or the other.

4.2 Descriptive statistics

As can be seen in Table 2, the average respondent work stress level is 2.95 out of 5. Moreover, respondents work in organizations with an average health culture level of 3.13 out of 5. For informal caretakers, the average work stress level is slightly higher: 3.05 out of 5. Furthermore, informal caretakers work in organizations with an average health culture level that is slightly lower: 3.11 out of 5.

In order to analyze the normality of the distribution of the sample, skewness and kurtosis of the non-binary variables were explored. As can be seen in Table 2, skewness and kurtosis are relatively close to 0 for all variables. The positive skewness value of ‘age’

(30)

indicates that the distribution of the sample regarding this measure is slightly clustered towards the lower values. The negative skewness value of ‘health culture’ indicates that the distribution of the sample for this measure is slightly clustered towards the higher values. The positive kurtosis value of ‘working hours’ indicates that the distribution of the sample for this measure is slightly clustered around the center, whereas the negative kurtosis value of ‘age’ indicates that the distribution of the sample for this measure is slightly clustered around the extremes (Agresti & Finlay, 2009; Pallant, 2010).

Table 2: Descriptive statistics

Variable Min Max M SD Skewness (SE) Kurtosis (SE)

Work stress 1 5 2.95 0.80 -0.15 (.12) -0.37 (.24) Health culture 1 5 3.13 0.53 -0.32 (.12) -0.10 (.24) Age Working hours 19 12 65 76 33.86 33.37 12.04 12.78 0.78 (.12) -0.11 (.13) -0.77 (.25) 0.68 (.25)

Note: Valid N = 366. SD = Standard Deviation. SE = Standard Error.

4.3 Correlation analysis

In order to get a first overview of the relationships between the variables used in this study, a correlation analysis was conducted. The descriptive statistics and correlation coefficients (r) are shown in Table 3.

All correlations point in the expected direction, but not all of them are strong and/or significant. Informal caretaking for instance relates positively to work stress, indicating that informal caretakers on average have higher stress levels than workers who do not engage in informal caretaking. However, its coefficient is very small and not significant (r = .06, ns).

Furthermore, work stress relates negatively to work ability (r = -.17, p < .01). This indicates the association between higher levels of work stress with reduced work ability.

Health culture relates negatively to work stress (r = -.13, p < .05), indicating the association between higher levels of health culture in organizations with lower levels of work stress among workers. Moreover, health culture relates positively to work ability (r = .11, p < .05) and to working hours (r = .17, p < .01), indicating that higher levels of health culture go hand in hand with high work ability and a higher number of working hours per week.

Age positively relates to informal caretaking (r = .43, p < .01), indicating that informal caretakers are on average older than workers who do not engage in informal caretaking.

(31)

Furthermore, age negatively relates to low education level (r = -.16, p < .01), indicating that older workers on average have obtained higher education.

Sex relates negatively to work stress (r = -.18, p < .01), indicating lower average work stress levels for male than for female workers. Sex relates positively to working hours (r = .24, p < .01) and to health culture (r = .12, p < .05), indicating that male workers on average work a higher number of hours per week, and perceive higher levels of health culture in their organizations.

A low education level relates negatively to work stress (r = -.27, p < .01), indicating that workers with a low education level experience lower levels of work stress than workers with a medium or a high education level. A medium education level relates negatively to working hours (r = -.17, p < .01), indicating that workers with a medium education level work a lower number of hours per week than workers with a low or a high education level. A high education level relates positively to work stress (r = .22, p < .01) and to working hours (r = .22, p < .01), indicating that workers with a high education level experience higher levels of work stress and work a higher number of hours per week than workers with a medium or a low education level.

Last, working hours relate positively to work stress (r = .20, p < .01), indicating the association between working a higher number of hours with higher levels of work stress.

(32)

Table 3: Means, Standard Deviations, Correlations

Variable M SD 1 2 3 4 5 6 7 8 9 10

1. Work stress 2.95 0.80 (.80)

2. Informal caretaking (ref. no) 0.19 0.39 .06 -

3. Health culture 3.13 0.53 -.13* -.01 (.87)

4. Age 33.86 12.04 .10 .43** .03 -

5. Sex (ref. female) 0.46 0.50 -.18** -.09 .12* -.03 -

6. Low education level 0.08 0.28 -.27** -.06 -.07 -.16** .02 -

7. Medium education level 0.11 0.31 -.04 -.02 -.04 .07 -.06 -.10* -

8. High education level 0.81 0.39 .22** .06 .08 .06 .04 -.62** -.71** -

9. Working hours 33.37 12.78 .20** .04 .17** .08 .24** -.07 -.17** .19** -

10. Work ability (ref. reduced) 0.12 0.32 -.17** -.06 .11* -.08 .07 -.00 -.05 .05 .05 -

Note: N = 366.

* Correlation is significant at the 0.05 level (two-tailed). ** Correlation is significant at the 0.01 level (two-tailed).

(33)

4.4 Regression analysis

The primary focus of this study is on individual workers’ experience of work stress, and its relationship with informal caretaking, perceptions of organizational health culture, and age. However, because of the nested structure of the data (i.e. employees were nested within companies), prior to testing the hypotheses, inter-class correlation (ICC) was calculated to estimate the amount of variance in work stress that is attributable to company-level factors. The ICC score for work stress is .19, indicating that 19 percent of the variance in work stress can be attributed to company level factors. As this is below the recommended threshold of ICC > .25 (Nezlek, 2008), we opted for testing the hypotheses using conventional single level regression rather than multilevel modeling. Therefore, a hierarchical multiple linear regression analysis has been performed.

First, prior to the analysis, the numerical independent variables ‘age’ and ‘health culture’ were mean centered. In addition, the four interaction variables needed for testing H3,

H4, H5, and H6 were created (‘healthculture*informalcaretaking’, ‘age*informalcaretaking’,

‘age*healthculture’, healthculture*informalcaretaking*age’).

Second, the regression assumptions were tested. Evaluation of the regression assumptions of normality, homoscedasticity, linearity, and the absence of multicollinearity were satisfactory (Bryman, 2012).

Third, in order to test each of the six hypotheses, hierarchical multiple regression analysis has been performed in three different steps. In every step there has been controlled for sex, education level, number of working hours per week, and work ability. The effects of informal caretaking, health culture, age, and the control variables on work stress levels are shown in Table 4. This table contains three different regression models. Since all cases with missing data were excluded prior to doing any analyses, the full sample of 366 respondents has been used in every model.

In model 1, the relationships between the control variables and work stress were looked into. The effects of sex, education level, number of working hours per week, and work ability on work stress are shown. The model is statistically significant F(5, 358) = 14.84; p <.01 and explains 17% of the variance in work stress.

In model 2, the focal independent variables used in this study were added in order to test H1 and H2. After controlling for age, sex, education level, working hours and work ability,

A culture of care: for young or for old? : a quantitative study of the influence of informal caretaking on work stress and the roles of