Microsoft Word Socialized Choices 31-12 pod docx

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. 
 

---