Chapter 4 - The vital and stabilising role of work preferences
111
Children of parents with ‘modern’ values appear to have a more egalitarian
perspective on work and family roles themselves compared to children of parents
with more traditional values (Barret and White, 2002; Cunningham, 2001; Moen
et al., 1997; Trent and South, 1992; Van Wel and Knijn, 2006; De Valk, 2008).
Having religious parents correlates with more traditional preferences among girls
and boys (De Valk, 2008; Thompson, 1991, p.382). Adolescents tend to have a
more egalitarian gender attitude when they had a working mother and grew up in
a non-standard family arrangement (single parent or foster families) (De Valk
2008; also Marks and Houston 2002b, p.333). Weinshenker (2006) has shown,
with a study among 194 middle class North American families, that the
expectations of female adolescents’ (aged 12 to 18) about their future
employment as a mother were associated with their own mothers’ employment
histories and her support for gender egalitarianism. Several studies have also
demonstrated that having a working mother has a significant and stimulating
effect on the employment behaviour of their daughters (Cloïn, 2010; Sanders,
1997; Van Putten et al., 2008).
Socialization theory has a distinct view compared to the literature on
stratification or intergenerational social mobility. Stratification theory in essence
points to resource transfers from parents to children. What parents transmit is
social status, by their educational level and occupation, and subsequent
similarities in social structural position may generate attitudinal correspondence
between parents and their kin (Bourdieu 1984; Glass, Bengston and Dunham,
1986, p.686; Kraaykamp, 2009; Liefbroer and Dijkstra, 2007; Van Putten et al.,
2008, p.438).
Childhood background characteristics that are included in this study are the
more objective features: the educational level of both parents, the parental
division of paid and unpaid work (when the respondent was twelve years old or
under), and whether the mother was in paid work. The third hypothesis of this
chapter is:
A mother’s preferred number of work hours, her general gender values and
her personal gender and work attitudes are influenced by objective parental
characteristics during childhood.
The three hypotheses are illustrated in figure 3.
Socialized Choices - Labour Market Behaviour of Dutch Mothers
112
Figure 3. Theoretical model and hypotheses
4.5 Research
method
The three hypotheses are tested by analysing data from the LISS (Longitudinal
Internet Studies for the Social Sciences) panel survey, administered by
CentERdata of Tilburg University, the Netherlands. The LISS panel consists of a
representative sample of the Dutch population who participate in monthly internet
surveys. A longitudinal survey is conducted among the panel every year, covering
a large variety of domains, including work, education, income, housing, time use,
political views, values and personality. Apart from this annual survey, the
respondents receive a different questionnaire each month which focuses on a
particular topic.
For the analysis a special questionnaire for this study was conducted in
November 2010 for mothers with at least one child of twelve years old or
younger, living at home.
37
In addition, several questions are used from the
questionnaires ‘Politics and Values’ and ‘Work and Schooling’, also answered in
November 2010. The questionnaire included 40 questions and was sent to a
random selection of 1,374 mothers, of whom 948 returned a completed form
(response rate 69%).
The composition of the sample of mothers with respect to age, number of
children, education and work hours differs only slightly from the composition of
the full population, as registered by Statistics Netherlands. The sample is
therefore representative of the Dutch population of mothers with at least one child
below the age of 13 living at home.
37
Women and their Social Environment, Liss Panel, Centerdata, University of Tilburg, November
2010.
Chapter 4 - The vital and stabilising role of work preferences
113
Analysis
As shown in figure 3, the theoretical model includes three dependent variables,
the labour market decisions, labour market behaviour and work preference, which
are simultaneously analysed with a structural path model. The advantage of a
structural path model is that it enables to examine in one regression analysis the
causal relationship between a number of independent variables and more than one
dependent variable. Moreover, the analysis estimates the direct and the indirect
impact of several independent variables, while controlling for their co-variances.
In this way the relative importance of the total effect of various attitudes and
personal characteristics on work preferences and on labour market behaviour can
be compared. To perform the structural path analysis the study uses the software
package Amos™ 19 (IBM SPSS
®
).
For a well-functioning path analysis, the number of variables included in the
analysis must be limited. For this reason, a number of logistic and linear (OLS)
regressions were performed with the dependent variables separately, in order to
determine which independent variables have a significant effect on the dependent
variables (see results regression analyses in appendix 2.)
38
. Next, the most non-
significant variables were removed from the analysis, until only the significant
variables were left. Finally, all dependent variables and the significant
independent variables were included in a structural path analysis. Within the
regression analyses I also tested for multicollinearity, and based on the values
(Vif) I could accept the variables. The path analysis is based on 935 cases. Based
on the Bollen-stine bootstrap, a measure for the goodness-of-fit in case of non
normal data for a path model, the model is accepted (Arbuckle, 2010).
39
Dependent variables
Table 3 gives an overview of the descriptive statistics of the dependent variables
and the background characteristics used in the analysis.
38
See also for appendix bilateral correlations of all dependent and independent variables.
39
Before I could perform a Bollen-Stine bootstrap I had to recode all the missing values into the
mean values. After recoding the missing values, the model fitted better in 935 bootstrap samples -
testing the null hypothesis that the model was correct – (Bollen-Stine bootstrap p = .001).
However, for the missing values Amos computes maximum likelihood estimates. This is preferred
to regular regression methods handling with missing values (listwise deletion, pairwise deletion or
data imputation (Arkbuckle, 2010, p.270). Therefore the model that is presented includes the
missing values.
Dostları ilə paylaş: |