Social Indicators Research

, Volume 93, Issue 3, pp 489–508

Work-Life Conflict and Social Inequality in Western Europe

Authors

    • Economic and Social Research Institute
  • Emma Calvert
    • Economic and Social Research Institute
Article

DOI: 10.1007/s11205-008-9433-2

Cite this article as:
McGinnity, F. & Calvert, E. Soc Indic Res (2009) 93: 489. doi:10.1007/s11205-008-9433-2

Abstract

Recent debates on time-use suggest that there is an inverse relationship between time poverty and income poverty (Aguiar and Hurst in Q J Econ C(3):969–1006, 2007), with Hammermesh and Lee (Rev Econ Stat 89(2):374–383, 2007) suggesting much time poverty is ‘yuppie kvetch’ or ‘complaining’. Gershuny (Soc Res Int Q Soc Sci 72(2):287–314, 2005) argues that busyness is the ‘badge of honour’: being busy is now a positive, privileged position and it is high status people who work long hours and feel busy. Is this also true of work-life conflict? This paper explores the relationship between work-life tension and social inequality, as measured by social class, drawing on evidence from the European Social Survey. To what extent is work-life conflict a problem of the (comparatively) rich and privileged professional/managerial classes, and is this true across European countries? The countries selected offer a range of institutional and policy configurations to maximise variation. Using regression modelling of an index of subjective work-life conflict, we find that in all the countries under study, work-life conflict is higher among professionals than non-professionals. Part of this is explained by the fact that professionals work longer hours and experience more work pressure than other social classes, though the effect remains even after accounting for these factors. While levels of work-life conflict vary across the countries studied, country variation in class differences is modest. We consider other explanations of why professionals report higher work-life conflict and the implications of our findings for debates on social inequality.

Keywords

Work-life conflictWork-life balanceSocial classEuropean social surveyComparative research

1 Introduction

Reconciling work-family life has become a critical issue for policy debates in Europe and the US (OECD 2001; Jacobs and Gerson 2004). International evidence points to a growth in the proportion of people feeling rushed and stressed (Bittman 2004), thus supporting Schor’s (1991) ‘Overworked American’ hypothesis. However, there is also research suggesting that perceptions of time stress are strongly related to social class. Gershuny (2005) argues that busyness is a ‘badge of honour’: being ‘busy’ is now a positive, privileged position and it is people with high status work long hours and feel busy. Hammermesh and Lee (2007) echo this point, claiming that complaints about being busy or a lack of time are more commonly expressed by well-off couples, and that less public concern should be given to this relative to other concerns.

Building on this literature, this paper explores the relationship between work-life tension and social inequality, as measured by social class, in eight countries in Western Europe, drawing on evidence from the European Social Survey. More specifically, we ask to what extent is work-life conflict a problem of the (comparatively) rich and privileged professional and managerial classes, and how does this vary across countries? The countries selected for comparison vary significantly in terms of institutional and policy configurations concerning work and family life and include Denmark, Germany, Spain, France, Ireland, The Netherlands, Sweden and the United Kingdom. Note that this paper is concerned with class differences in work-life conflict, and how this varies across countries, rather than with country variation in work-life conflict per se.

2 Relevant Debates: Social Inequality and Work-Life Conflict

Research on social inequality has a long tradition in Western Europe. Social inequality can be measured in terms of poverty (Atkinson 1998) or more broadly social exclusion (Paugam 1998), class (Erikson and Goldthorpe 1993), inequality in education (Shavit and Müller 1998), or in unemployment and its consequences (Gallie and Paugam 2000). There is also a strong overlap between disadvantage in different spheres of life, i.e., between those who experience poverty, low paid jobs, unemployment, health problems and low education.

The concerns posed by high unemployment in the 1980s and 1990s have been superseded by issues around work intensification, increasing labour market participation and the appropriate balance between work and family life. Indeed reconciling work-family life has recently become a critical issue for policy debates in Europe and the US (OECD 2001; Jacobs and Gerson 2004). Under the traditional male breadwinner model competing demands in the work and family sphere were managed by a division of labour between the sexes, whereby men were primarily responsible for paid work, while women assumed responsibility for caring. However, increasingly EU citizens have to combine both caring and employment roles, with consequences for work-family tension within households.

A related vein of research literature is concerned with increased time pressure. International evidence points to a growth in the proportion of people feeling rushed and stressed (Bittman 2004). This growth in time pressure is associated with health problems and a deteriorating quality of life. From this evidence, there seems to be general support for Schor’s ‘Overworked American’ thesis (Schor 1991). While there is controversy over whether hours of paid work have actually increased (Robinson and Godbey 1997; Gershuny 2000; Bittman 2004), there are indications that time poverty is particularly pronounced among the more privileged in society. Using US time-use data, Aguiar and Hurst (2007) find that in the last 40 years the largest increase in leisure has been for those with lower educational qualifications: the income poor are ‘time rich’ while the income rich are ‘time poor’.

Gershuny (2005), building on Becker’s (1965) argument that time and goods are substitutable, argues that the reason people are feeling busier is because there is now a positive view of busyness and lack of leisure. People with higher earning power will work more and concentrate on ‘goods intensive’ leisure to maximise utility; lower earners with lower purchasing power will favour ‘time intensive’ leisure and purchase fewer commodities. Gershuny stresses the importance of paid work relative to leisure for privileged social positions. With the emergence of mass unemployment and other social changes, the historical association of ‘idleness’ with high status was displaced: busyness is now the position of privilege. In addition, high status individuals often have intrinsically more rewarding jobs than the lower classes. It is now those with high status who work long hours—in rewarding, well-paid jobs—and feel busy.

In support of this, Hammermesh and Lee (2007) argue that complaints about lack of time are more commonly expressed by well-off couples. They find that households with higher earnings perceive more time stress for the same amount of time spent in market work and household work. They claim that complaints about insufficient time come disproportionately from higher full-income families partly because their members choose to work more hours, partly too because they have higher incomes to spend during the same amount of non-work time. They conclude that at least some of this complaining is ‘yuppie kvetch’ and should not be of major policy concern, particularly relating to reducing inequalities in society. While it is mentioned, one line of explanation not pursued by Hammermesh and Lee is that high earners in professional positions suffer more time stress because their work is more demanding and comes with greater responsibility; we explore evidence supporting this in the following section (Section 3).

The key question in this paper is therefore: is the inverse relationship between social inequality and time stress also true of work-life conflict? Is it the comparatively rich and privileged who report high work-life conflict? How much of this can be explained by known factors such as long working hours and stressful jobs? It seems less plausible that work-life conflict has the same positive connotations as busyness, but we can reflect on the extent to which higher professionals choose to work more hours, either for money or personal fulfilment. They may also have higher expectations of their leisure time, more money to spend in it, thus want more leisure and report this as ‘work-life conflict’.

A second key question in this paper then asks: do professionals across the different countries under study exhibit the same patterns with respect to work-life conflict? We compare a range of countries with different institutional settings—labour markets and welfare states—and ask whether an inverse relationship between work-life conflict and social inequality is found across the population. We then explore whether these class differences remain across countries, after accounting for other important factors, such as job demands, longer working hours, as well as household characteristics. Previous research highlights the gendered nature of the experience of work-life conflict, so we also consider gender differences (Van der Lippe et al. 2006).

Essentially work-life conflict refers to when meeting demands in one domain makes it difficult to meet the demands in the other domain (Greenhaus and Singh 2003). In the work-life conflict research literature ‘work’ usually implies paid work and there is tension as to whether ‘family/life’ refers to caring, or leisure or both (MacInnes 2006). However, in time-use research, work is typically defined as both paid and unpaid work (Hammermesh and Lee 2007; Gershuny 2000). To be consistent with the bulk of previous research on work-life conflict, and the measures used in the data, this paper defines ‘work’ as paid work and ‘life’ as caring and/or leisure.

After reviewing previous comparative research findings on work-life conflict (Section 3), we then consider the different institutional contexts of the countries under study (Section 4). Section 5 considers hypotheses on work-life conflict and social inequality, Section 6 reviews methodology and data. Section 7 discusses our findings, while Section 8 concludes by reflecting on possible explanations and their implications.

3 Previous Research on Work-Life Conflict

Work-life conflict has attracted a growing amount of academic, as well as policy attention, resulting in an increasing body of research, from single country studies (Fagnani and Letablier 2004) to occupational groups (Greenhaus et al. 2003). While cross-national studies have also been undertaken, based on qualitative research (Abrahamson 2007) and smaller scale quantitative surveys (Fine-Davis et al. 2002), cross-national research, particularly studies which consider the individual and household level, as well as the interface between paid work and family life at a societal level, are relatively limited.

In terms of paid work, the level of work-life conflict is related to the demands in terms of time, intensity and scheduling. Long working hours and unsocial hours have been found to be positively associated with work-life conflict (Dex and Bond 2005; Crompton and Lyonette 2006; Van der Lippe et al. 2006; Scherer and Steiber 2007). There is also evidence to suggest that work-life conflict increases with higher levels of work stress (Scherer and Steiber 2007). Forms of flexibility which allow employees to vary their schedule to accommodate their family lives tends to reduce work-life conflict (Fagan 2003); flexibility which benefits employers, like working overtime at short notice, may increase work-life conflict. While we might expect that the longer hours and greater work stress experienced by professionals may increase their work-life conflict, employee flexibility is more prevalent among higher-status occupations (Fagan 2003), which may ameliorate some of these effects.

The level of work-life conflict will also relate to time commitments and demands within the home. Previous research has found that the presence of children generally increases work-life conflict (Cousins and Tang 2004): high total paid work hours (i.e. of both partners) has also been found to contribute to work-life conflict (Scherer and Steiber 2007; Jacobs and Gerson 2004). While the ability of high-earning professionals to alleviate work-life conflict through paying for childcare or domestic labour, non-professionals may be more likely to opt out of the labour market if faced with financial difficulties. Previous research has also emphasised the importance of the gendered division of labour within the household in explaining work-life conflict, as well as gender role attitudes and consistency between attitudes and behaviour (Crompton and Lyonette 2006; Scherer and Steiber 2007). Here, we might expect professionals to exhibit more egalitarian attitudes with regards the household division of labour within the household; if these expectations are not met then this could increase work-life conflict.

At the ‘societal level’, research has emphasised the importance of institutional-level factors, e.g. welfare regime and the extent of reconciliation or ‘family-friendly’ policies such as parental leave, which may explain variations in work-life conflict across countries (Strandh and Nordenmark 2006; Van der Lippe et al. 2006; Crompton et al. 2007). Research found that, controlling for work and family factors, Finland and Norway, relatively ‘family-friendly’ countries, had on average lower levels of work-life conflict, compared to Great Britain, France and Portugal (Crompton and Lyonette 2006). However, Scherer and Steiber (2007) argue that the lack of well-developed reconciliation policies implies the reduction of women’s working hours and the possibility of more traditional combinations of paid and unpaid work, thus potentially resulting in lower perceptions of work life conflict. In support of this, studies have found that some countries with very well-developed reconciliation policies have higher levels of work-life conflict, compared to those without (Cousins and Tang 2004; Van der Lippe et al. 2006). Labour market regulations and employment regimes are also important, e.g. collective control over working time and flexibility (Scherer and Steiber 2007; Tomlinson 2007): we might expect that strict limits on maximum working hours reduce work-life conflict. It is therefore important to consider both the policy context of the countries under study, as well as to characterise them more generally in terms of the labour market structure and welfare regime.

4 Welfare States, Labour Markets and Policies to Facilitate Work and Family Life

Following Esping-Andersen (1990), and the gender critique of his work (e.g. Lewis 1992), it is widely argued that welfare regimes affect how individuals engage in paid work, caring and unpaid work, and how these are combined over the life course. It is also valuable to distinguish between ‘coordinated’ and ‘liberal’ market economies (Hall and Soskice 2001) when considering how working time and employment regulations affect work-life conflict. While (female) labour market participation is strongly associated with welfare regimes, state supports for caring varies within regime type: therefore it is also useful to consider how specific policies, such as parental leave facilitate the combination of working and caring.

Denmark and Sweden typify the ‘social democratic’ model, with a high level of state intervention in both family life and the labour market, high taxes and well-developed provision of support for caring. Female full-time labour market participation is high, resulting in a high proportion of dual-earner couples and social partners play a strong role in regulating working time (O’Reilly 2003). These countries are characterised by low income inequality as well as little gender difference in terms of labour market engagement (Esping-Andersen 1990, 1999).

While France, Germany and The Netherlands fall into the ‘conservative’ welfare regime and all link benefits strongly to engagement in paid work, they differ in terms of institutional support for the combination of working and caring, and labour market co-ordination (Hall and Soskice 2001). In Germany, the tax and welfare system provides strong incentives for the traditional male breadwinner division of labour (or at most a modified breadwinner model, with the female working part-time; Daly 2000). Childcare is supported, though much of it is part-time and there is poor provision for very young children (Plantenga and Remery 2005). In The Netherlands, tax and welfare provision tends to support the male breadwinner model, although also strongly encourages part-time work (for men and women) and is less socially conservative than the German model: in fact some commentators argue that The Netherlands could be classed as social democratic (Goodin et al. 1999). France is also often classified as a conservative welfare state, although women’s full-time employment is facilitated and there is more state support for combining paid work and caring; however, childcare costs are high (Lewis 1992; Plantenga and Remery 2005). All of these countries are also characterised by medium to low class inequality, although gender inequality varies by country. In Germany and The Netherlands, social partners have negotiated reductions in working time; in France, state legislation has reduced hours but some employers implement their own ‘flexible strategies’ (O’Reilly 2003).

The UK and Ireland are often classified as ‘liberal’ welfare regimes, with very little market interference, low taxes and low state support for caring. Childcare costs are an issue, particularly in the UK and Ireland, with very little subsidies. There are relatively low rates of parental leave take-up in Ireland and the UK, compared to practically universal take-up in other EU countries (Plantenga and Remery 2005). Both the UK and Ireland are characterised by high wage inequality, and class inequality generally. However, there are more women working in the UK (although many work part-time) than in Ireland, despite changes in recent years. Low statutory working-time regulation and weak industrial relations allows employers more power in shaping working time and this has led to high polarisation of working hours, especially in Britain (Fagan 2003).1

Spain is often classified as belonging to the Mediterranean model with the ‘family as breadwinner’ (Gallie and Paugam 2000). The shortfall in welfare provision is assumed to be met by the family, with women in particular taking on the responsibility for caring. Low availability of part-time work makes it particularly difficult for women to combine work and family life in the absence of state supports.

However, as already outlined, it is important to bear in mind that policy and institutional supports for combining work and caring can influence work-life conflict in divergent ways, by both easing the successful combination of work and family life as well as influencing the nature of female labour participation (Gornick and Meyers 2003). Traditional solutions, like in Spain, actually mean less work for women—and possibly less work-life conflict for them—though this may have implications for gender inequality. What matters for us of course is how work-life conflict varies across classes: we return to this point in the next section.

5 Research Hypotheses

5.1 Work-Life Conflict and Social Inequality

Our first hypothesis is that, as with time stress, work-life conflict will be higher among the higher and lower professional classes (hypothesis 1). Secondly, we examine time demands and work-life conflict. Do professionals experience higher work-life conflict because of greater demands on time, either from work or home? Are they ‘time poor’ and thus more prone to work-life conflict (hypothesis 2)? We also consider how family commitments, housework and time spent in paid work influence work-life conflict and affect differences between class groups. We also look at how time is allocated, i.e., the flexibility with which individuals can organise their paid work time and whether jobs require overtime or unsocial hours, which may affect work-life conflict more than the hours of work per se.

Our third hypothesis is that it may not be time commitments per se that influence work-life conflict; it is rather because professionals/managers have stressful jobs and responsibilities that are more prone to ‘spillover’ into family life (hypothesis 3a). In addition, we might expect that professional workers are more likely to ‘devalue housework’ and thus find it stressful, therefore contributing to higher levels of work-life conflict (hypothesis 3b). We also look at the role of attitudes to the gender division of paid and unpaid labour.

5.2 Class Differences in Work-Life Conflict: Cross Country Variation

We also examine how the relationship between social class and work-life conflict varies across countries. Are there differences in the overall pattern, and how is this affected when we consider working hours and work pressure? Our point of departure is that of no difference between countries, i.e., professionals will experience higher work-life conflict, whatever the institutional and policy setting (hypothesis 4a). We also expect this difference to be reduced in each country when we account for factors such as work demands and work pressure (hypothesis 4b).

Alternative hypotheses suggest that class differences will vary by country. Firstly, from the overall consideration of differences between welfare regimes and labour markets we expect that class differences will not be as pronounced in the Nordic countries (Sweden and Denmark). Class differences will be moderate in The Netherlands, and slightly higher in France and Germany. The most marked class differences will be in the UK and Ireland (hypothesis 5).

Hypothesis 6 concerns the relationship between state reconciliation policies and work-life conflict and the implications of such policies for class differences in work-life conflict and gender differences in class variation. One possibility is that in countries with low state support for caring, it is a select group of high-earning professional women, on a career track, who participate in the labour market in spite of low support, and who are therefore prone to experiencing higher work-life conflict. Non-professional women either do not participate in the labour market, or work part-time, and experience low work-life conflict (hypotheses 6). In countries with high support for caring, there is higher participation overall. We therefore expect less difference between professionals and non-professionals with respect to work-life conflict in countries with high support for combining paid work and caring, than in countries with low support (hypothesis 6).

6 Methodology, Measurement and Data

6.1 Measuring Inequality

Social inequality is measured in a number of ways. Gershuny (2005) uses education as a measure of human capital which, while closely correlated with job quality, is not the case for some groups, e.g., women working part-time are often overqualified for their positions. Economists such as Hammermesh and Lee (2007) and Aguiar and Hurst (2007) use personal income. Personal income is a good indicator of ‘being rich and privileged’, but is not well measured in the present survey (one-third of the employed have missing information on income). Social class performs equally well, we would argue, with higher professional classes enjoying a privileged position in the occupational hierarchy, and is measured more effectively.

The Erikson and Goldthorpe (1993) class schema differentiates between higher professionals, lower professionals, routine service class, skilled manual workers and unskilled manual workers. Given that the primary interest in the present paper is to explore the relationship between professionals and work-life conflict, we collapsed higher and lower professionals into a new “professionals” category, while the rest formed the “non-professionals” category. While much theoretically-guided class analysis focuses on the manual/non-manual or blue/white collar division, the focus of this paper is on professionals versus non-professionals.

6.2 Measuring Work-Life Conflict

Some authors take ‘objective’ indicators of work-life conflict by assuming that long working hours (Gornick and Meyers 2003) or long hours of both paid and unpaid work (Bittman 2004) is inimical to work-life balance. However, a more common approach in the literature is using a ‘subjective’ indicator of work-life conflict, based on the individual’s own assessment. In the present survey the questions are: ‘How often do you keep worrying about work problems when you are not working?’, ‘How often do you feel too tired after work to enjoy the things you would like to do at home?’ and ‘How often do you find that your job prevents you from giving the time you want to your partner or family?’ Responses are coded 1 (never) through to 5 (always). A fourth question was considered (‘How often do you find that your partner or family gets fed up with the pressure of your job?’) but was very differently distributed to the others, with a lower mean, so was excluded.2 In this paper individual scores are combined for each of these three items and the mean is used. The index thus varies from 1 (never) to 5 (always). Where there is item non-response on any of the questions, the mean of the others is calculated. The items are highly correlated: this index has a Cronbach’s Alpha of 0.68 for the pooled sample.3 This reduces the number of missing cases overall in the index: creating an additive scale means any case with a missing item is excluded.4

6.3 Data

The analysis in this paper draws on the European Social Survey held in 2004. There was a high priority placed on equivalence of samples and equivalence of questions (Jowell et al. 2007), and the survey is excellently suited to measuring work-life conflict from a comparative perspective. Combining data from the special module on work-life conflict with the main data set provides a series of questions on work-life conflict, respondents’ and their partners weekly working hours, the extent of unsocial hours, work pressure, housework time and division of housework in the household, family composition, gender role attitudes, educational attainment and occupational status, from which we derive our measure of social class.

Housework is measured in hours per household per week, while paid hours are individual weekly hours worked. Flexibility in schedule is measured by a question ‘I can decide the time I start and finish work’, answers ‘not at all true’ through to ‘very true’ (1–4). The extent an individual works unsocial hours is measured as a composite index, the mean of three questions: ‘working evenings or nights’, ‘having to work overtime at short notice’ and ‘working at weekends’ and coded ‘never’ through to ‘every week’ (1–5). Job pressure is a combination of ‘working hard’, ‘not having enough time to finish work’ and ‘finding work stressful’, and coded ‘disagree strongly’ through to ‘agree strongly’ (1–5). Two questions are used to measure housework stress: ‘There are so many things to do at home, I often run out of time before I get them all done’ and ‘I find my housework stressful’. The housework stress measure is coded the same as job pressure. It is excluded from the country models as there are many missing cases, and those who answer the question differ from those who answer the work stress questions. Gender role attitudes are also measured using an index, which takes the mean of answers to the following questions: ‘A woman should be prepared to cut down on her paid work for the sake of her family’, ‘Men should take as much responsibility as women for the home and children’ and ‘When jobs are scarce, men should have more right to a job than women’, coded the same as job pressure, with higher values indicating more egalitarian attitudes.

As the work-life conflict index, described above, is continuous and has a normal distribution of residuals, the effect of the covariates is modelled using linear regression. In the first step the eight countries are pooled (Table 2), and factors expected to influence work-life conflict among professionals entered in a stepwise fashion (employees only). In the second part of the analysis, country interactions are specified to test whether class differences vary across countries (Table 3).

7 Results: Work-Life Conflict and Social Inequality

7.1 Work-Life Conflict for all Employees

First, we explore overall patterns of work-life conflict for different social classes (Table 1). Higher professionals report the highest levels of work-life conflict, followed by lower professionals. The routine non-manual and skilled manual groups report lower levels of work-life conflict, with the non-skilled group reporting the lowest levels overall. This broadly supports insights from the time-use debate: time pressure and work-life conflict is most keenly experienced by the privileged. It is also clear from Table 1 that both higher and lower professionals differ from both routine non-manual workers and from skilled to non-skilled manual workers: this also ties in with our substantive hypotheses.
Table 1

Mean work-life conflict scores by class

 

Mean

Standard deviation

Higher professional

2.91

0.79

Lower professional

2.78

0.80

Routine non-manual

2.51

0.84

Skilled manual

2.50

0.86

Non-skilled

2.43

0.87

Table 2 presents a series of linear regression models of work-life conflict to examine our first three hypotheses on work-life conflict and social inequality. These models are based on the same number of base cases, so that they can be directly compared. Model 1 simply looks at class differences in work-life conflict among the employed, with the non-professionals group forming the reference category. As previously shown in Table 1, we can see that the professionals group, comprising higher and lower professionals, report higher levels of work-life conflict, consistent with hypothesis 1. Subsequent models add time demands (from work and home); then flexibility of scheduling/timing of paid work and then home and work stress.
Table 2

Linear regression models—work-life conflict (employees in all countries)

 

Class differences

Class differences & time demands

Class differences with time demands, flexibility & stress

Model 1

Model 2

Model 3

Coefficient

Coefficient

Coefficient

Constant

2.52***

1.58***

0.54***

Professional

0.33***

0.25***

0.18***

Female

 

0.17***

0.14***

Partner

 

0.04

0.06*

Ref: no children

Child < 6

 

0.09**

0.01

Child 6–17

 

0.05*

0.00

Lone parent

 

0.10*

0.08~

H/work hours/week

 

0.00~

0.00

Paid week hours/week

 

0.02***

0.01***

Start/finish time (1–4)

  

−0.01~

Unsocial hours (1–5)

  

0.15***

Job pressure (1–5)

  

0.26***

H/work stressful (1–5)

  

0.08***

Egalitarian attitudes (1–5)

  

0.05**

Adjusted R square

0.39

0.106

0.253

N

5,496

5,496

5,496

Work-life conflict index varies from 1 (low) to 5 (high), see text for further details

*** p ≤ 0.001; ** p ≤ 0.01; * p ≤ 0.05; ~p ≤ 0.1

What happens to these class differences when we account for time availability (Model 2, hypothesis 2)? We find a significant effect of the presence of children on work-life conflict, particularly children under six. The effect of being an (employed) lone parent also increases work-life conflict. Increasing paid work hours has a significant positive effect on work life conflict, though weekly housework hours has no effect, at least after accounting for paid work. All of these findings are consistent with previous research (Dex and Bond 2005; Crompton and Lyonette 2006; Van der Lippe et al. 2006; Scherer and Steiber 2007) and hypothesis 2. However, while there is a significant effect for paid work hours and caring responsibilities, we find that class differences are reduced but are still significant.

Model 3 adds variables capturing both how paid working time is allocated, and the stress of both paid and unpaid work. Working unsocial hours (weekend, overtime etc.) increases work-life conflict, although the effect of flexibility of schedule is negligible, possibly due to opposite/countervailing effects (Fagan 2003). Overall we find that while working unsocial hours increases work-life conflict, this has little impact on class differences. We find that job pressure substantially increases work-life conflict. Finding housework stressful also increases work-life conflict, and those with more egalitarian attitudes report more work-life conflict. As professionals, on average, report more work pressure, find housework more stressful and have more egalitarian attitudes, taking account of these factors reduces class differentials further. This is an important part of why professionals experience more work-life conflict than non-professionals: they have more stressful jobs.5 These findings, particularly the results for job pressure, support hypothesis 3. However, note that the difference between professionals and non-professionals does not disappear (Model 3).6

Do overall class differences vary for men and women? We estimate models 1 and 3 for men and women separately (see Tables 4 and 5 in the Appendix). Here we find the class difference in work-life conflict, without any other covariates, is greater for women, and this gender difference is significant. When we account for family composition, hours of paid and unpaid work, job pressure and other covariates, this gender difference disappears (although the class difference for both men and women remains). What is interesting is that work hours have a greater impact on work-life conflict for women than for men. In addition, these potential explanatory factors explain a greater portion of the difference between professionals and non-professionals for women than for men.

7.2 Work-Life Conflict by Social Class: Country Variation

Next, we explore whether the relationship between work-life conflict and social class varies by country. As Fig. 1 shows, in all of the countries under study, professionals report higher levels of work-life conflict. The differentials in mean work-life conflict score between the professionals and non-professionals group is particularly marked in the UK, Ireland and France, though overall the pattern is remarkably consistent, given policy and labour market variation in these countries. The difference in means is lowest in Sweden.
https://static-content.springer.com/image/art%3A10.1007%2Fs11205-008-9433-2/MediaObjects/11205_2008_9433_Fig1_HTML.gif
Fig. 1

Country variation in work-life conflict (mean scores) by class

Model 4 in Table 3, which simply estimates the mean effect of being a professional for the pooled sample, is a variant of model 1 (Table 2), and is presented for comparison with the other models.7 In models 5, 6 and 7 the effect of class by country is introduced using interaction terms. Germany is the reference category, and here professionals experience more work-life conflict than non-professionals. The ‘main effects’ for country indicate that country variation is significant and maintained even after controlling for a range of covariates (Model 7). Compared to non-professionals in Germany, work-life conflict is lower among non-professionals in Britain, The Netherlands, Sweden, Denmark, Spain and particularly in Ireland.
Table 3

Linear regression models—work-life conflict all employees

 

Class differences

Class differences & country

Class differences with country & time demands

Class differences with country, time, flex. & job pressure

Model 4

Model 5

Model 6

Model 7

Coefficient

Coefficient

Coefficient

Coefficient

Constant

2.50***

2.70***

1.76***

0.84***

Professional

0.34***

0.32***

0.25***

0.21***

Ref: Germany

Denmark

 

−0.26***

−0.26***

−0.27***

Sweden

 

−0.10*

−0.11*

−0.11**

UK

 

−0.23***

−0.18***

−0.22***

Netherlands

 

−0.30***

−0.20***

−0.10*

France

 

−0.02

0.01

0.09~

Spain

 

−0.17**

−0.23***

−0.15**

Ireland

 

−0.62***

−0.62***

−0.58***

Professional * Denmark

 

0.02

0.03

0.01

Professional * Sweden

 

−0.07

−0.09

−0.15*

Professional * UK

 

0.21**

0.17*

0.09

Professional * NL

 

−0.01

−0.04

−0.11~

Professional * France

 

0.05

0.07

0.06

Professional * Spain

 

−0.10

−0.04

−0.10

Professional * Ireland

 

0.08

0.12~

0.10

Female

  

0.18***

0.18***

Partner

  

0.01

0.04~

Ref: no children

Child < 6

  

0.11**

0.07**

Child 6–17

  

0.06*

0.03

Lone parent

  

0.07

0.09~

H/work hours/week

  

0.00***

0.00***

Paid week hours/week

  

0.02***

0.01***

Start/finish time (1–4)

   

−0.03***

Unsocial hours (1–5)

   

0.16***

Job pressure (1–5)

   

0.29***

Egalitarian attitudes (1–5)

   

0.06***

Adjusted R square

0.04

0.09

0.15

0.27

N

5,982

5,982

5,982

5,982

Work-life conflict index varies from 1 (low) to 5 (high), see text for further details. Housework stress excluded because of missing cases for Spain

*** p ≤ 0.001; ** p ≤ 0.01; * p ≤ 0.05; ~p ≤ 0.1

Table 4

Work-life conflict, males and females (pooled model)

 

Males

Females

Difference between coefficients

t-Test for differences

Coefficients

Coefficients

Constant

2.53**

2.51***

  

Professional

0.29**

0.36***

0.07

−2.22

Constant

0.74***

0.57***

  

Professional

0.19***

0.16***

−0.03

0.89

Partner

0.05

0.08*

0.02

−0.60

Ref: no children

Child < 6

−0.00

0.03

0.03

−0.68

Child 6–17

0.00

0.01

0.01

−0.20

Lone parent

−0.01

0.11~

0.12

−1.08

H/work hours/week

0.00

−0.00

−0.00

1.74

Paid week hours/week

0.01**

0.01***

0.01

−5.00

Start/finish time (1–4)

−0.01

−0.01

0.00

0.02

Unsocial hours (1–5)

0.15***

0.16***

0.01

−0.90

Job pressure (1–5)

0.27***

0.25***

−0.01

0.77

H/work stressful (1–5)

0.08***

0.09***

0.01

−0.57

Egalitarian attitudes (1–5)

0.04~

0.04~

−0.00

0.08

Adjusted R square (final model)

0.25

0.261

  

N

2,592

2,919

  

Work-life conflict index varies from 1 (low) to 5 (high), see text for further details

The t-statistic is calculated using the following formula: \( t = \hat{\beta }_{\text{M}} - \hat{\beta }_{\rm F} /\sqrt {(\hat{\sigma }_{{\hat{\beta }{\text{M}}}} )^{2} } + (\hat{\sigma }_{{\hat{\beta }{\text{F}}}} )^{2} \), where \( \hat{\beta }_{\text{G}} \) is the coefficient for males, \( \hat{\beta }_{\text{B}} \) the coefficient for females, and \( \hat{\sigma } \) the relevant standard error

*** p ≤ 001; ** p ≤ 0.01; * p ≤ 0.05; ~ p ≤ 0.1

Table 5

Work-life conflict, males and females with country interactions

 

Males

Females

Difference between coefficients

t-Test for difference

Coefficients

Coefficients

Constant

1.08***

0.84***

−0.24

 

Professional

0.17**

0.24***

0.08

1.05

Denmark

−0.35***

−0.22**

0.13

1.68

Sweden

−0.20***

−0.04

0.16

2.38

UK

−0.21**

−0.21**

0.00

−0.04

Netherlands

−0.21**

0.04

0.25

2.98

France

−0.03

0.18**

0.22

2.94

Spain

−0.23***

−0.09

0.14

1.75

Ireland

−0.68***

−0.48***

0.20

2.65

Professional * Denmark

0.06

−0.04

−0.10

−0.93

Professional * Sweden

−0.06

−0.24**

−0.18

−1.77

Professional * UK

0.11

0.07

−0.04

−0.37

Professional * NL

−0.04

−0.19~

−0.16

−1.37

Professional * France

0.16~

−0.01

−0.18

−1.58

Professional * Spain

−0.05

−0.13

−0.08

−0.64

Professional * Ireland

0.18~

0.01

−0.17

−1.53

Partner

0.05~

0.04

−0.02

−0.42

Ref: no children

Child < 6

0.06~

0.10*

0.04

0.87

Child 6–17

0.03

0.04

0.02

0.42

Lone parent

0.04

0.10~

0.06

0.73

H/work hours/week

0.00***

0.00***

0.00

−0.34

Paid week hours/week

0.00**

0.01***

0.01

6.00

Start/finish time (1–4)

−0.02~

−0.04***

−0.02

−1.44

Unsocial hours (1–5)

0.16***

0.15***

−0.01

−0.91

Job pressure (1–5)

0.28***

0.30***

0.02

1.05

Egalitarian attitudes (1–5)

0.05*

0.05*

0.00

0.07

Adjusted R square

0.27

0.28

  

N

2,990

2,992

  

Work-life conflict index varies from 1 (low) to 5 (high), see text for further details

The t-statistic is calculated using the following formula: \( t = \hat{\beta }_{\text{M}} - \hat{\beta }_{\text{F}} /\sqrt {(\hat{\sigma }_{{\hat{\beta }{\text{M}}}} )^{2} } + (\hat{\sigma }_{{\hat{\beta }{\text{F}}}} )^{2} \), where \( \hat{\beta }_{\text{G}} \) is the coefficient for males, \( \hat{\beta }_{\text{B}} \) the coefficient for females, and \( \hat{\sigma } \) the relevant standard error

*** p ≤ 001; ** p ≤ 0.01; * p ≤ 0.05; ~ p ≤ 0.1

More interesting for us is whether the difference between professionals and non-professionals varies by country. Here the significant interaction coefficient for Britain in model 5 indicates that the gap between professionals and other classes is greater in Britain than in other countries. Other countries do not differ significantly from Germany in this regard, lending general support to the ‘no country variation in class difference’ hypothesis (4a).

What happens when we account for differences in time demands, home and work (Model 6)? As in the pooled model, we find that having children, particularly young children, and paid working hours and unsocial hours all increase work-life conflict, and this reduces the difference between professionals and non-professionals. In the final model (7), once we account for job pressure, which has a very strong impact on work-life conflict, we find that the professional/non-professional gap in work-life conflict is reduced even further, and country variation in this gap is now negligible, with the exception of Sweden.8

Two points to note on specific country differences. Firstly, investigating country differences before adding covariates (Model 5), we find that in Britain, differences between professionals and non-professionals are greater than in Germany, as indicated by the significant interaction term for professionals in Britain. When we account for working hours (Model 6), scheduling and work pressure (Model 7), this difference is no longer significant. Britain does not now differ from other countries in terms of class differences in work-life conflict. In fact, professionals in Britain have particularly high job pressure scores, which helps explain this result. Secondly, the difference between professionals and non-professionals is significantly lower in Sweden than in Germany.

Figure 2 details the country variations in the net effect of the professional class compared to other classes, calculated from models 5 and 7 in Table 3. The dark bar is the net effect of professional versus non-professionals by country; the light bar shows how this effect changes once we account for time demands and job pressure.
https://static-content.springer.com/image/art%3A10.1007%2Fs11205-008-9433-2/MediaObjects/11205_2008_9433_Fig2_HTML.gif
Fig. 2

Net effect of professional class versus other classes by country

In all countries, the difference between professionals and non-professionals is greatly reduced when we account for time demands, family commitments and job pressure. However, it is also very clear from Fig. 2 that differences between professionals and non-professionals remain. These are statistically significant in all countries but Sweden—a country with comparatively low levels of class inequality. Concerning hypothesis 5, class differences tend to be highest in the UK and Ireland, followed by France and Germany, with Sweden the lowest, as we might expect. Yet class differences are modest in Denmark, and low in The Netherlands. The fact that overall country variation is modest gives, on balance, more support to the ‘no country variation in class differences’ (hypothesis 4), than hypothesis 59. This is not to deny differences between countries in the overall level of work-life conflict.

To investigate hypothesis 6 in more depth, we estimate models 5 and 7 separately for men and women. The net effects for women are presented in Fig. 3 (detailed findings are available from the authors).
https://static-content.springer.com/image/art%3A10.1007%2Fs11205-008-9433-2/MediaObjects/11205_2008_9433_Fig3_HTML.gif
Fig. 3

Net effect of professional class versus other classes by country (females only)

The results from Fig. 3 lend some support to hypothesis 6. Professional women in Ireland and particularly in Britain, countries with low state support for caring, show high work-life conflict compared to other working women in these countries. Conversely, in The Netherlands and Sweden, countries with high support for reconciling work and family life, there is a smaller difference between professional and non-professional women: these disappear once we account for time demands and job pressure. Once again the hypothesis is not fully supported: we would have expected women in Denmark to be more like women in Sweden, and women in Spain, a country with low support for combining work and family life, to look more like the UK. It is clear though that the overall low class differences in Sweden and The Netherlands (presented in Fig. 2) are driven by the results for women.

8 Discussion

We find that in the pooled model, work-life conflict is highest among professional workers, often thought of as the ‘privileged’ in the occupational distribution. Part of this is explained by the fact that professionals work longer hours and experience more work pressure than other social classes. Long working hours and work pressure, both strongly associated with work-life conflict, are often an integral part of professional jobs. A key part of the professional/non-professional difference in work-life conflict then concerns the nature of the jobs. However, the class differential in the pooled model is maintained: even after accounting for differences in time availability and differences in the pressure of job and home life, professionals report higher work-life conflict, while routine non-manual workers and both skilled and unskilled manual workers report less work-life conflict.

While levels of work-life conflict vary cross-nationally, the class findings are remarkably consistent across countries; although in Sweden, class differences are no longer significant when we account for family situation, working time and work pressure, and in The Netherlands the difference becomes very small. Separate models by gender reveal that the overall low class differences in Sweden and The Netherlands are driven by females: in these countries with high support for reconciling work and family life, there are less differences between professional and non-professional females.

Hammermesh and Lee (2007) have argued that high-earning (professional) workers suffer more from time pressure (in this case, work-life conflict) because they choose to work more hours (they value money more than time) and also have more money to spend in their leisure time, and thus value it more. With this data it is difficult to adjudicate whether higher and lower professionals place a higher value on leisure, relative to others, as they have more money to spend in their leisure time, which makes it more interesting and desirable. We can however, consider, at least to some extent, whether professionals choose to work more hours. Evidence from the European Social Survey suggests this is not the case. We find that preferred working hours (constrained by earnings) and the respondent’s actual working hours varies by social class. Over 60% of higher professionals and 50% of lower professionals report wanting to work less than their current hours; compared to 35% of routine non-manual workers and 45% of skilled and unskilled manual workers. This contradicts the idea that higher professionals choose to work more—the majority, though not all, would like to work less.

So why do professionals not simply work less? One possibility is that jobs are a “package”. Workers might welcome the opportunity to vary their working hours in response to changes in their preferred time allocations, but in practice working hours are more often fixed than flexible, part of a “package deal” associated with a particular job (McGinnity and McManus 2007). Even those who can formally reduce hours—like parents of young children in all of the countries studied—may find that the nature of their job means this is much more difficult in practice. Schneider and Waite (2005) in their large study of professional dual-earning couples in the US argue that many working parents in their study find themselves conforming to the image of the ‘unconditional worker’, believing that if they do not put in the additional effort, they will be overlooked for promotion. These parents are the ones more likely to arrive home emotionally drained, stressed and resenting the intrusion of work into family life.

Some may feel a reduction in working hours could disadvantage their career prospects, and many professional jobs are part of a well-defined career path. Previous research indicates that there may be a penalty in terms of promotion prospects for part-time work, so those reducing hours may face a trade-off between a reduction in current work-life conflict and the quality of future employment. In addition, as many professional jobs are task-based, there are fluctuations in workload which may require overtime. Job autonomy over time and tasks, which makes these jobs count as privileged in social class theory (Goldthorpe 2000) makes them more likely to be associated with long hours and spillover.

Taking a longer-term perspective, there may be a lifecycle effect. Highly educated individuals secure challenging and demanding jobs, and it is only at the family formation stage, when already on a career track, that the time squeeze really comes. Professionals may not only have high expectations of leisure, like Hammermesh and Lee (2007) argue, they might also have higher expectations of parenting. Bianchi et al. (2006), in their study of parenting and time pressure in the US, point to the role of subjective expectations in adding to time pressure. Working parents, particularly mothers, feel a time squeeze because they feel they should be spending more time with children—even though mothers are spending as much time interacting with children as they were 40 years ago (Bianchi et al. 2006). If this pressure is felt more keenly by professionals than other mothers/parents, this will contribute to class differences in perceived work-life conflict, though could hardly be described as ‘yuppie kvetch’.

What are the policy implications of our findings? Hammermesh and Lee (2007) argue that less public concern should be given to being time poor—and by implication given our findings, work-life conflict—if it is experienced by the privileged and is a ‘yuppie’ condition. However, arguably welfare states have recently shifted focus from targeted subsidies to reduce inequality to ‘welfare as social investment’ as the new solution to the problem of balancing economic growth and social justice (Taylor-Gooby 2008). Modern welfare states require an increasing proportion of the population in paid work to both enhance competitiveness and reduce poverty, so the state must facilitate female labour market participation. If European economies/welfare states need highly skilled labour, they cannot afford to lose highly-skilled female labour; the problem of work-life balance and how it is achieved among professionals is therefore particularly important. There may indeed be a coincidence of interest between the preoccupations of the professional class and the needs of the welfare state.

Footnotes
1

Since the late 1980s the Irish path has somewhat diverged from the British model: note the corporatist-style solidaristic agreements between the social partners and government (O’Connell et al., 2003).

 
2

For details on the data, see http://www.europeansocialsurvey.org. Tests were conducted on the results with an index which uses all four questions and also an index which uses three questions but excludes missing cases on any of the items. Results are reported in the results section.

 
3

The Cronbach’s alpha for individual countries varies from 0.63 for France to 0.74 for Ireland.

 
4

See White et al. (2003) for an alternative measure using principal components analysis.

 
5

Note that work pressure, like work-life conflict, is a subjective indicator. We may be slightly overestimating the effect of work pressure if we are picking up an unmeasured underlying anxiety, however the extent of this is not possible to quantify with this data.

 
6

When we replicate the analysis using an alternative 4-item index which includes the question ‘How often do you find that your partner or family gets fed up with the pressure of your job’, we find no difference in the pattern of class effects.

 
7

The only difference being the exclusion of the housework stress variable as there were some problems with missing cases, particularly for Spain. The sample size is now greater than in Table 3.

 
8

We also estimated these models on a sample of couples only. The findings are almost identical to those reported, with three exceptions. In Denmark, Spain and Ireland the difference between professionals and non-professionals is less for couples than for all employees.

 
9

We check whether estimating separate country models instead of a pooled model with interaction terms has any impact on our findings. Patterns of country variation are very similar indeed. The only difference of note is the findings for Ireland. In Ireland, the effect of having children has a greater positive impact on work-life conflict in the separate country model, and the difference between professionals and non-professionals is lower, once we properly account for the presence of children.

 

Acknowledgments

This paper was produced as part of the economic change, quality of life and social cohesion (EQUALSOC) network of excellence, funded by the European Commission (DG Research) as part of the sixth framework programme. See editors’ introduction for further details. The authors wish to thank all the project participants for their helpful input, and in particular, Vanessa Gash, for her comments on an early draft. Thanks also to Helen Russell, for sharing her programmes for manipulating ESS data. This paper received constructive comments when presented at the RC28 Spring Meeting in Florence, May 2008: the authors are also grateful to the referees of Social Indicators Research for their helpful comments and criticisms. Special thanks to Chris Whelan for his insightful comments, and his engagement as editor of the paper.

Copyright information

© Springer Science+Business Media B.V. 2009