Evaluation - CiteSeerX

0 downloads 0 Views 117KB Size Report
All rights reserved. Not for commercial use or unauthorized distribution. at PENNSYLVANIA STATE UNIV on April 17, 2008 http://evi.sagepub.com. Downloaded ...
Evaluation http://evi.sagepub.com

An Impact Assessment of Policy Measures to Influence the Gender Division of Work Tinke M. Willemsen, M Jacobs, A P Vossen and G A B Frinking Evaluation 2001; 7; 369 DOI: 10.1177/13563890122209739 The online version of this article can be found at: http://evi.sagepub.com/cgi/content/abstract/7/3/369

Published by: http://www.sagepublications.com

On behalf of:

The Tavistock Institute

Additional services and information for Evaluation can be found at: Email Alerts: http://evi.sagepub.com/cgi/alerts Subscriptions: http://evi.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 369

Evaluation Copyright © 2001 SAGE Publications (London, Thousand Oaks and New Delhi) [1356–3890 (200107)7:3; 369–386; 019814] Vol 7(3): 369–386

An Impact Assessment of Policy Measures to Influence the Gender Division of Work TINEKE M. WILLEMSEN Tilburg University, Netherlands

M E N N O J . G . J AC O B S Tilburg University, Netherlands

A D VO S S E N Tilburg University, Netherlands

GERARD A. B. FRINKING Tilburg University, Netherlands In this study, two different approaches, within one theoretical framework, were used to assess the impact of some public policy measures on the division of paid and unpaid work in the Netherlands. The first (macro-level) approach was based on the views of a panel of national experts; the second (micro-level) approach consisted of the analysis of information provided by members of households in a representative survey. In general, the results of the two studies indicate that the relevant policy measures have little, if any, impact on actual behaviour. In the survey study, the only variable that had a strong impact on both men’s and women’s behaviour was an attitude variable: egalitarianism. However, we conclude that the combination of methods, i.e. of the retrospective Delphi study with the cross-sectional survey, leads to better insight into the impact of the policy measures than each of the separate methods. K E Y WO R D S : triangulation

Delphi study; division of work; gender; policy evaluation;

During the last few decades, a number of public-policy measures have been introduced in the Netherlands with the aim of influencing patterns of paid work and family life, particularly, the division of work and household and care tasks between the sexes. This division of work in families has indeed changed rapidly during the last few decades: the female labour force has more than doubled since 369 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 370

Evaluation 7(3) 1970, while male participation in paid labour has been stable during this same period (Van Praag and Uitterhoeve, 1999). In the domestic sphere, changes have been less drastic, but here too important changes can be noticed: the time spent on parenting in families has increased slightly; the time spent on household tasks has decreased slightly, especially in double-income families (Van Praag and Uitterhoeve, 1999). The main question addressed in this study is: are these changes indeed a consequence of the introduction of policy measures? 1 In order to answer the question whether the policy measures have changed the gender division of unpaid and paid work, we conducted an impact assessment of a number of public policy measures in the field of work and family. Our aim was to find the net effects of these policy interventions, that is, ‘an estimate of the intervention uncontaminated by the influence of other processes and events that also affect the conditions the program attempts to change’ (Rossi et al., 1999: 70). To conduct an impact assessment that is more than a black box evaluation, we needed a research design that includes the relevant variables that may affect the division of paid and unpaid work. Therefore, the development of an appropriate conceptual model that meets these requirements was an important element in the present study. Assessing the impact of policy measures in a country, generally a difficult task, is even more difficult than usual when it involves the division of paid and unpaid work between the sexes. Many changes have taken place, but these are similar to changes that have occurred elsewhere in the Western world where different policies have been implemented, so it is questionable whether they are a result of the specific measures that are part of the national policies under study here. Moreover, there is a lack of clear-cut objectives, which is not unusual for public policies. However, in the case of policies for the re-division of (un)paid work, this creates an additional problem because there is not even an implicit consensus on objectives. For instance, some may think that such policy measures can only be considered successful if there is equality between men and women in paid and unpaid work, while others may think that equality on the labour market is what really counts, and yet others, that a certain amount of difference in the actual situation is inevitable, but it is the equality of opportunities that counts. To avoid these difficulties, no attempt was made to determine whether the policy measures met their objectives, as one would do in an evaluation study. Rather, the goal was to assess whether these measures had any impact at all, according to experts, and whether people who have made use of certain policy measures have a different position on the paid/unpaid labour field than those who have not. We developed one theoretical framework to guide us in asking the relevant questions and, within this framework, used a two-way approach to assess the impact of policies. We conducted a survey study to assess the impact on the micro-level. The main question in the survey was if, and how, the situation regarding the (un)paid work of people who had made use of the policy measures was different from the situation of those who had not used them. The survey study was a cross-sectional study. We also carried out a macro-level Delphi-type study in which a panel of experts assessed the impact of policy measures by estimating the difference between the current situation and the situation that might have 370 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 371

Willemsen et al.: Impact Assessment of Policy to Influence Gender Division of existed without those policies, i.e. the autonomous trend. In principle, such a study should have a longitudinal design, as the impact of policies over a long period of time is estimated. The present study was restricted to three types of policy measures that are the most widespread in the Netherlands and were introduced or changed after 1985. First, these concern leave arrangements: maternity leave, parental leave (a form of leave from work that both fathers and mothers are entitled to), and calamity leave (a short leave everyone is eligible for in the event of an acute situation). Employed women are entitled to 16 weeks of paid maternity leave (fully paid), with the option of flexible, i.e. part-time, implementation. New fathers get a fully paid two-day delivery leave. According to the current regulation for parental leave, employees, both fathers and mothers, are entitled to unpaid parental leave for a maximum total period of 13 weeks full-time if they care for a child under eight years old. In government and public services, as well as in a few other organizations, parental leave is (partially) paid. Secondly, we studied two types of child-care facilities: daycare centres and after-school care. There is great variety in daycare, which can be categorized in terms of subsidized/non-subsidized, formal/informal, or private/government initiative. The role of the government concerns the provision of subsidies for daycare centres (decentralized to municipalities since 1996); the provision of financial compensation to parents when the costs of professional daycare exceed the standardized parental contributions; and fiscal regulations for companies that contribute to the costs of daycare centres. There are only a few facilities for afterschool care. Finally, the policy measures regarding flexibility in working hours were studied: part-time work, working in shifts, and temporary work. Since 1 January 1996, a new law on working hours has been in force in which matters pertaining to working hours and breaks were arranged in such a way that greater flexibility, for employers and employees, was made possible.

Theoretical Framework Within the social sciences (including sociology, social psychology and economics) there are basically three theoretical approaches to explain the division of paid and unpaid work between men and women on the micro-level of families (Van Doorne-Huiskes and Willemsen, 1997). These are new home economics (Becker, 1991), exchange theory (Cook, 1987) and role theory (Parsons and Bales, 1955; Eagly, 1987). New home economics and exchange theory both have rational choice theory as their starting points, assuming that individuals and households make rational decisions for optimizing profit according to their preferences, restrictions and utility curves. Role theory assumes that behaviour is influenced by norms and values and postulates a strong relationship between gender-role attitudes and behaviour. All these theoretical approaches have, to some extent, been able to explain aspects of the division of paid and unpaid work within households. However, none of the above-mentioned theoretical approaches can fully explain time allocation patterns of women and men (Van Doorne-Huiskes and 371 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 372

Evaluation 7(3) Willemsen, 1997). Moreover, for the present study, which will be part of a larger European study, we needed a theoretical approach that could also incorporate the division of work on the macro-level and the cultural differences between countries. Therefore, we have made an adaptation of Bourdieu’s theory of practices (1977, 1984, 1994). 2 Stated very briefly, Bourdieu argues that all practices result from a relationship between habitus and capital in a given field (Bourdieu, 1984: 101). Habitus is a system of dispositions that function as durable generative classificatory schemes of perception, evaluation and action. In the habitus, the basics of culture are encoded during childhood, as ways of doing things. Habitus causes social reality to be perceived as self-evident. Particularly important for the present research project is the fact that habitus refers to dispositions to behave in a certain way (rather than to think in a certain way). In our view, habitus also includes gender expectations, the expected ‘normal’ division of tasks between men and women, the gendered customs that are taken for granted. Capital refers to both material and symbolic goods that have value in a certain situation. The three basic forms of capital are economic capital (e.g. income, assets), cultural capital (e.g. knowledge, value orientations) and social capital (e.g. networks). Practices refer to frequently repeated actions, customs, or duties. Practices are more often than not accomplished without conscious deliberation; otherwise, the routine of everyday life would not be possible. Social interaction is a mixture of freedom and constraints; people use adaptive strategies to reach their goals. Male domination, which is still obvious in our culture, is also part of this routine. It is a form of exercising power in which the actual power is concealed and only works because of authority that is recognized by both sexes (Bourdieu, 1994: 195–6). These three concepts, habitus, capital and practices, can be used on an individual level but also aggregated on, for instance, the macro-level of a country, where they can be used to describe a country’s common opinion, welfare and (in our case) division of work. On the macro-level, social reality consists of fields, configurations of social positions occupied by individuals or institutions. These are relatively autonomous networks that function according to their own logic. The structure of a field is determined by the power relations between its occupants. Bourdieu distinguished a number of fields, of which the fields of politics, paid labour and family relations are the most important for our study. Politics is regarded as the dominant field in any society: it is the source of the hierarchical power relations that structure all other fields. Therefore, although both the labour market and politics can be considered fields, in this theoretical model we assume that politics and policies influence the structure of the labour market and aspects of it, such as the level of participation of women. This theoretical framework allows us to combine aspects of the macro-, structural level with aspects of the micro-, individual level. We are interested in the influence of policy measures on the actual behaviour of men and women, especially with regard to paid work, child care and household work. On the individual level, we will study whether people make use of the relevant available socialpolicy measures. We expect that habitus and capital influence the use of policy 372 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 373

Willemsen et al.: Impact Assessment of Policy to Influence Gender Division of measures, and that the use of policy measures influences practices. At the same time, it is also possible that habitus and capital variables have a direct impact on practices, as is postulated in the above-mentioned social-role theory and rationalchoice theories. For instance, capital, especially cultural capital, if operationalized as educational attainments, allows one to earn well on the labour market. According to rational-choice theories, the more one can earn on the labour market, the more attractive paid work is in comparison to unpaid work. As for habitus, it is to be expected that the more egalitarian the attitude, the more egalitarian the division of work will be. On the micro-level, we can test both direct and indirect influences. The macro-study has, by its design, necessarily put greater emphasis on direct influences, as will be explained in the next section.

The Delphi Study The approach we used on the macro-level is a so-called policy-oriented Delphi study, a method that is particularly suitable to policy-oriented analyses (Dunn, 1994). One purpose of this study was to determine how experts view the impact of government policy on the division of paid and unpaid work in households. The study combined quantitative and qualitative methods of information gathering, i.e. the experts were requested to give quantifications of the perceived impact of policy measures as well as the arguments for these quantifications.

Methodological aspects The method we used, a policy Delphi, can be traced back to the Delphi technique that was developed by the Rand Corporation in 1948 for military applications. It is a ‘judgmental forecasting procedure for obtaining, exchanging, and developing informed opinion about future events’ (Dunn, 1994: 242). The technique has been applied to numerous fields of research, and various versions have been developed. In particular, a policy-oriented version, generally indicated as ‘policy Delphi’, has been added to the conventional, prospective Delphi (Turoff, 1970; Linstone and Turoff, 1975). With this technique, one seeks ‘to generate the strongest possible opposing views on the potential resolution of a major policy issue’ (Turoff, 1970: 149), in order to estimate the impact of various options. Although, in general, this technique is used as an ex-ante evaluation of some intended policy (e.g. Keuzenkamp [1997] used it that way in a study on the possible effects of various versions of equal-opportunity policies), in the present study we use it as a technique for evaluating an existing set of policy measures.

Participants and Method The panel members, all experts in matters such as work–family relations or equalopportunity policies, were selected from employers’ organizations, trade unions and universities to guarantee multiple viewpoints. From the employers’ organizations and trade unions the participants were in general employees responsible for developing or executing its equal-opportunity policy, or work–family policy; from academia we chose colleagues from other universities or research institutes we knew had expertise in this field and who were not in any way involved in the 373 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 374

Evaluation 7(3) research project. By choosing panel members who hold strongly diverse positions in the field of equal-opportunity policies and the like, we hoped to guarantee that they would generate the ‘strongest possible opposing views’ mentioned above. Fifteen experts participated in the written research round. After they had agreed to participate, the respondents received a questionnaire in which they were asked to estimate the impact of public-policy measures regarding leave arrangements, child-care facilities and flexible working arrangements on the division of paid and unpaid work and to indicate on which arguments these estimates were based. The questionnaire was accompanied by relevant background information on the policies and on developments in the actual division of work. The questionnaire consisted of four parts: the impact of leave arrangements; the impact of child-care facilities; the impact of flexible work options; and the joint impact of these measures. Each part contained quantitative questions (estimates of the impact on the division of work in specific types of households) and questions about the underlying arguments. The estimates were made for households with children, because most of the measures concerned are meant to affect these households, with ages of the youngest child specified (younger than 4 and between 4 and 18),3 along with the paid-work situation in the household (dual-income, one-and-a-half-income, single-income, and so-called half–halfincome households where both partners work part-time). The paid-work situation is not really a good operationalization of income, which is a capital variable, but it gives some indication. A combination of these two specifications gives eight types of households. For each age of youngest child, the respondents estimated whether the policy measures had increased or decreased the percentage of families in certain working situations and whether, in each type of household, the time spent on household work and child care had increased or decreased, for men and women separately. Additionally, respondents were asked to estimate the importance of the influence of education (specified in three levels) on households. This specification of educational level was meant as an operationalization of cultural capital. No habitus variables were included in the design of the Delphi study. A meeting of nine panel members and the research team was held to present and discuss the results of the written round. In accordance with one of the goals of a Delphi study, i.e. to elicit opposing views, the focus of the discussion was on controversial estimates and arguments. Arguments from the written round were addressed in greater depth, and points of view were explained. After the discussion, the participants completed a short questionnaire to examine whether their points of view had converged.

Results of the Written Round Paid Work On the whole, the impact of policies on the division of paid work, expressed as shifts between types of households, was considered marginal. Nevertheless, the experts believed that both leave arrangements and child-care facilities had resulted in an increase in the number of dual-income households and one-and-a-half-income households, and in a decrease in the proportion of singleincome households. That is, they considered these policy measures somewhat 374 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 375

Willemsen et al.: Impact Assessment of Policy to Influence Gender Division of beneficial for women’s labour-force participation. The impact of child-care facilities was considered stronger than the impact of leave arrangements. For both groups of arrangements, the impact was considered greater for households with young children than for households with older children, and households with a higher level of education were assumed to derive greater benefit from government policy measures than households with a lower level of education. The panel members argued that the leave arrangements were implemented too late and that the duration of the various leaves was too short. Moreover, they felt that these measures had not encouraged a re-division of paid labour among men. The small impact of child-care arrangements was attributed to the fact that it is mostly arranged informally and that the government policy on these issues is reactive rather than proactive. Unpaid Work According to the experts, neither leave arrangements nor childcare facilities resulted in a substantial increase in the contribution of men to household work. Leave arrangements have only relieved the family responsibilities of women with young children while child-care facilities had no impact on these responsibilities. Although both leave arrangements and child-care facilities have resulted in a slight increase in the amount of time fathers spend on child care, these same leave arrangements have resulted in an increase of the time women spend on child care in dual-income households with young children. The most significant impact of child-care facilities was perceived in dual-income and oneand-a-half-income households with young children. The arguments offered for the marginal impact of leave arrangements were, first, that parental leave is usually taken by mothers, and second, that fathers who make use of parentalleave provisions spend most of this time caring for the children rather than on household work, whereas mothers in the same situation perform both types of tasks. The effect on men was also considered to be merely incidental because parental leave is so short. The marginal impact of child-care facilities was attributed to the lack of a logical relationship between child care and household work. Panel members also believed that men are only prepared to do housework if their partners have a full-time job. For flexible work arrangements, only open questions were asked as this policy had not been long in use and could be implemented in a variety of ways. Responses often concerned the increase in part-time work. The positive effect of flexibility on the continued labour-force participation of women with children, or on their re-entry into the labour force, was often mentioned. To assess the joint impact of these policy measures, the questionnaire included an overview of the actual amount of time men and women in couples, either married or cohabiting, and specified by the age of the youngest child,4 spent on various activities in 1985 and in 1995. The figures were based on the results of the most recent time-budget surveys conducted by the Social and Cultural Planning Office (SCP, 1997). To express the differences between men and women, the ratio of hours spent by women on a certain task to the hours spent by men on these tasks was calculated. The experts were requested to estimate a hypothetical ratio, representing 375 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 376

Evaluation 7(3) the ratio in 1995 if there had been no government involvement in these three areas. Table 1 presents the data used by the participants, as well as the estimated ratios. The latter indicate that, according to the experts, government policy has had a greater impact on the division of paid work than on the division of unpaid work, irrespective of the age of the youngest child. Also, that the division of household and family tasks has been influenced more strongly in households with children aged between 6 and 14 than in households with younger children. The estimated differences in terms of paid work were very small while the impact of publicpolicy measures on gender roles in households with older children was considered of little importance.

The Panel Meeting Nine of the 15 experts who took part in the first round also participated in the plenary meeting. The programme included a presentation of the results of the first round, a discussion of the results, a short questionnaire and an evaluation. Since there was consensus in the written round regarding the extent of the effects but a lack of clarity about the reasons, the discussion focussed on clarifying these reasons. Paid Work The panel felt that the increased labour-force participation of women was mainly triggered by autonomous trends, both sociocultural (women’s emancipation) and economic (labour-market situation). Initially, women (and their partners) tried to find their own solutions for combining a paid job and child care. Only later did government begin to play a facilitating role, mainly by creating more favourable conditions. These governmental efforts may then have encouraged other women to look for a paid job since, in a way, combining work and a family had been legitimized by the government. The panel concluded that the availability of child care is not a decisive factor in keeping or taking on a job. The panel also judged that, rather than removing actual barriers, the government Table 1. Time spent (in hours per week) on unpaid and paid work by male and female partners in the Netherlands in 1985 and 1995; ratios of female/male participation and the panel’s estimates of female/male ratios Women 1985 1995

Men 1985 1995

Ratio f/m Estimated f/m ratio 1995 1985 1995

Age of youngest child 5 Housework, care Paid work

53.9 4.3

49.2 8.4

20.0 32.6

20.7 39.4

2.70 0.13

2.38 0.21

2.46 0.18

Age of youngest child 6–14 Housework, care Paid work

45.8 5.8

39.9 11.0

14.4 37.4

17.2 36.2

3.18 0.16

2.32 0.30

2.50 0.26

Age of youngest child 14 Housework, care Paid work

43.9 6.6

37.5 9.0

14.9 34.2

13.4 39.4

2.95 0.19

2.80 0.23

2.82 0.22

376 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 377

Willemsen et al.: Impact Assessment of Policy to Influence Gender Division of had made a (modest) contribution to the change in attitudes towards working women, but employers’ organizations and trade unions played a more important role. The main contribution of the public sector was to bring about a shift from informal to formal solutions. Unpaid Work Regarding housework and family responsibilities, the panel judged that the government had made only a minimal effort to bring about a change of mentality among men. The relationship between the increased labourforce participation of women and the division of unpaid work was considered weak. There was consensus that the influence of education is mediated by labourmarket participation and earning potential: because highly educated women work more and earn more than less educated women, it is only logical that taking parental leave and using child-care arrangements by the former group is greater. Opinions varied considerably on the question whether men with a low level of education are more traditional and therefore less inclined to share family responsibilities than those with a high level of education. In the discussion on flexible work arrangements, most of the arguments concentrated on part-time work. The fact that the increased labour-force participation of women has not resulted in a substantial re-division of paid and unpaid work was attributed to the fact that women with part-time jobs tend to work fewer hours than in the past, while part-time work among men is still very rare. This inequality in paid work was considered the main reason why the division of unpaid work is still far from egalitarian. After the discussion, the panel members gave new estimates of the joint impact of the policy measures in terms of the fictitious ratio introduced above, and it turned out that the group discussion had hardly changed the views of the participants. The differences between these new estimates and the results of the first round were negligible.

Survey Study On the micro-level of the household, we studied the same research question as in the Delphi study: whether policy measures aimed at reconciling paid and unpaid work have had any effect in the Netherlands. The same groups of variables were used (capital, practices, use of policies) but were operationalized in a different way and supplemented by habitus variables.

Respondents and Questionnaire The sample we report on here5 consisted of 790 heterosexual couples (married or cohabiting) in which at least one of the partners was between 19 and 50 years of age. The respondents regularly participated in a survey panel that is representative of the population of the Netherlands. To reduce non-response and bias, and to avoid negotiation between partners about certain questions, part of the questionnaire was administered by post, the other part by telephone. The response rate for the written questionnaire was 69.4 percent. Out of those respondents, 377 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 378

Evaluation 7(3) 78.5 percent completed the telephone interview, resulting in a total response rate of 54.5 percent. For the purpose of the present study, we only used the data from couples with at least one child under 18 years of age, of whom the man had paid work, and without missing data. This resulted in a data set of 395 couples. The questionnaire was based on the one proposed in Willemsen (1997) and it was developed to test an extended version of the model presented here. The written questionnaire contained questions on time used for work, household and care tasks, on education, income, housing, social networks, composition of the household, and on use of policy measures. Questions asked by telephone concerned the division of paid and unpaid work among members of the household, one’s share of child-care tasks, and opinions about the roles of men and women in paid and unpaid work. The variables of the model were operationalized as follows. The practices variables were based on the time-budget part of the survey. Respondents estimated the time spent on several activities during: a typical working day; a typical non-working day; a typical Saturday; and a typical Sunday. The reported hours were summed over three categories, comprising virtually all usual activities: income-generating work; child-care tasks; and household work. Income-generating work includes paid work, study, and time to travel to and from the place of work or study. Household tasks include cooking, washing the dishes, laundry, cleaning, shopping and gardening. Care tasks include taking care of elderly or sick members of the household, and clothing, bathing, feeding, accompanying or transporting children and playing with them. Time-budget surveys are notoriously difficult methodologically – accurate knowledge of how people spend their time would require large-scale 24-hour observation, which is not feasible for obvious reasons. Activities that have a positive social connotation, e.g. doing paid work, tend to be systematically overestimated in methods using time estimation. The diary method leads to overestimation of the many small tasks that are performed during a day (compare Niemi, 1993). In the present study, we dealt with the reliability problems of time estimation by using the relative share of men and women in paid work and domestic tasks, assuming that men and women over or underestimate to the same extent and in the same activity category. As the respective share in a certain task carried out by the man and the woman are variables that are not independent of each other, we had to use either the man’s or the woman’s share in our analyses. To simplify interpretation, we chose the shares that policy measures intended to increase: the woman’s share in paid work and the man’s share in household duties and care tasks. The use of policy measures was assessed by questions about the incidence of current or former use of seven policy measures: two forms of child care (afterschool child care and daycare centres); paid and unpaid parental leave;6 calamity leave; flexible working hours; and working-hours reduction. Some policy measures were seldom used (the use of unpaid parental leave among men was less than one percent, for example), so we aggregated these variables into three categories similar to those of the Delphi study: use of child-care facilities; leave arrangements; and adaptation of working hours. Since our unit of analysis was the household, we counted the use of either of these policy categories by each 378 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 379

Willemsen et al.: Impact Assessment of Policy to Influence Gender Division of partner of a couple. The maximum score per category, therefore, is four for each of the categories of child-care facilities and adaptations of working hours, and six for the category of leave arrangements. Habitus was defined by a single variable: egalitarianism, measured with the Dutch translation of the Attitudes towards Occupational and Family Roles of Men and Women Scale (Abele and Andrae, 1997; see also Jacobs et al., 2000). This instrument contains four subscales, concerning working women, the traditional division of tasks, the participation of men in family work, and the compatibility of job and family for both partners. As the internal consistency of the total scale was satisfactory (Cronbach’s alpha 0.73), we used the average item score on the total scale. A high score on this variable indicates an egalitarian attitude. The economic capital of the couple was operationalized by the man’s hourly wage.7 Cultural capital was defined by educational level of both the man and the woman. Social capital was defined as the average of the size of each partner’s social network. A social network was measured as the number of people one thinks one can rely on for help in case of some problem.

Results of the Survey The average age of the men in our sample was 41.2 years; the women were on average 2.6 years younger. The men had a slightly higher level of education than the women and the labour force participation was 100 percent for the men and 59.2 percent for the women. A relatively large percentage of the couples was married (97.7%)8 which is a consequence of the fact that we considered only couples that had at least one child in their household: these couples tend to be married. The average number of children in the selected households was 2.2, and 20.8 percent of the households had pre-school children (younger than four years of age). In our sample, 69 percent of the couples made use of, or had made use of, one or more of the three groups of policy measures. About one in four couples (23%) had used child-care facilities, and about one in five couples (19%) had utilized leave arrangements. Adaptation of working hours was the most popular facility, with 60 percent of the couples utilizing this. Adaptation of working hours was quite commonly used among women, with 49 percent of them using it now or in the past and among men, it was also the most frequently used policy measure (28%). However, for men this referred mainly to the use of flexible working hours; few men worked part-time. In general, women made more use of the policy measures than men, with the exception of leave arrangements. Note that maternity (or pregnancy) leave was not taken into account in the leave arrangements variable, and that only people with jobs can take these leaves. The average total number of hours spent on ‘income-generating work’ (i.e. paid work plus study and travelling time) was 15.0 hours per week for women and 48.6 for men – the wife’s share in paid work averaged 21 percent. With regard to household work, women did the majority of the work: 39.1 hours per week as opposed to 13.1 hours for the men – men carried out 25 percent of the total amount of the household work. The division of care tasks was less skewed however with 7.6 hours per week by men versus 14.9 hours per week by their wives; men performed 36 percent of the care work. 379 Downloaded from http://evi.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2001 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

05 Willemsen et al. (bc/d)

5/9/01

4:10 pm

Page 380

Evaluation 7(3) The direct effects of the habitus, capital and use-of-policies variables upon the wife’s share in paid work, the husband’s share in household work and his share in care tasks were established through three multiple regression analyses. Table 2 presents the results of the first two models; the model for care tasks was not significant (F = 1.85) and has therefore been omitted here. Table 2 demonstrates that there were hardly any effects of the use of policy measures on the division of paid and unpaid work in households. Only adaptation of working hours affected the husband’s share of household work but as we mentioned, adaptation of working hours is mostly used by women. The effect of the use of this policy measure on the husband’s share in household work, therefore, may partially be explained by the fact that men do more household tasks if their wives also work; women without paid work have no working hours to adapt. The habitus variables had the largest effects. An egalitarian attitude of the wife had the strongest effect on her own share in paid work, and also increased the husband’s share in household work. The egalitarianism of the husband also had a stronger impact on his wife’s share in paid work than on his own share in household work. The analyses demonstrate that egalitarian attitudes went together with a more egalitarian division of paid and unpaid work. However, the presumed cause–effect relationship is not proven by this analysis: attitudes may follow the division of tasks within the household as well as cause it. In the Delphi study, the experts attached much importance to educational level. They suggested that relatively highly educated people make more use of policies, that less-educated men do more household work, and that there is a strong relationship between attitude and educational level. These suggestions can now be empirically investigated on the individual level, and in this case educational level of both men and women was related to their use of policies and to

Table 2. Standardized regression weights of use of policy measures, capital and habitus upon the wife’s share in paid work and the husband’s share in household work (N = 395 couples) Predictor variable

Dependent variable Wife’s share in paid Husband’s share in household work (beta coefficients) work (beta coefficients)

Use of child-care facilities Use of leave arrangements Use of working hours adaptations Educational level of husband Educational level of wife Social network Hourly wage of husband Egalitarianism of wife Egalitarianism of husband R Square F

.075 .071 .075 .071 .040 .028 .003 .297** .227** .257 14.770**

.074 .003 .103* .062 .036 .073 .040 .190** .120* .158 8.045**

*p