Divorce in Flanders

Summary

Strengths: An important strength of the Divorce in Flanders survey relates to the research design. The multi-actor design has several advantages. First, it is possible to relate the life-courses and wellbeing of three generations. The data contain a researchable number of intergenerational dyads and triads of (grand)parents, children/parents, and (grand)children. Secondly, the perspective of different actors on the same relationships and events, and their causes and consequences, can be used to enhance the reliability and validity of the measurement, or to construct more objective or intersubjective information on family processes. In addition, it can eliminate the problem of shared method variance, overestimating the association between indicators measured with the same person. Another strength of the survey is that it allows to study changes in the organization of the life-course. As the main sample consists of reference marriages conducted between 1970 and 2008, the data allow an exploration of evolutions through time. The sample is proportionally drawn by marriage cohort, but also contains variation in terms of birth and divorce cohort. With a time frame of almost 40 years, there is substantial variation in the combination of age, period and cohort. Weaknesses: Unfortunately, the Divorce in Flanders data are limited to the Flemish Region. No similar surveys exist for the Brussels and Walloon Region. Another weakness relates to the selectivity resulting from the sampling criteria. First, the sample is limited to ever-married people and the parents, children and new parents from ever-married people. Subsequently, we have no information on never-married people. For people above age 65, however, this only concerns 5% of the population. Secondly, the main sample contains no men and women who were divorced two or more times, or whose partner from their first marriage is divorced two or more times. Subsequently, we also have no parents, children or partners of men and women who divorced two or more times. Calculations on National Register data indicate that in this way, 10% of all men and women who ever married between 1970 and 2008 are excluded. A second type of selectivity results from the multi-actor design. Participation of the (ex-)partner of the reference marriage was a condition for participation of a parent, child and new partner. As a result, depending on the selection strategy, there are different mechanisms by which the partners from the reference marriages are important gatekeepers in the composition of the samples of children, parents and new partners.

Type of data

Data Source
Survey

Type of Study
Crosssection occasional

Data gathering method
Telephone
Face-to-face
Self administered questionnaire

Access to data

Conditions of access
The data are freely available. Instructions will soon be available on the website: www.scheidinginvlaanderen.be

Type of available data (e.g. anonymised microdata, aggregated tables, etc.)

Formats available

Coverage

Coverage Years of collection, reference years, and sample sizes
The data were collected in 2009-2010. The sample of Divorce in Flanders consists of so-called reference marriages, drawn from the National Register in 2009. From the selected reference marriages, different actors were questioned, using a so-called multi-actor design: The partners from the reference marriages were questioned with a CAPI-interview. In total, 1,811 partners from intact reference marriages were questioned, including 786 double-interviews. 4,659 ex-partners from non-intact reference marriages were questioned, including 1,134 double interviews. If the partners from the reference marriage have a common child, a random target child was selected, with a preference for children at least 10-years-old living in the parental home (residential children). In total, 1,257 residential children were interviewed using the CAPI-method, 320 non-residential children completed a mail or web survey. If the (ex-)partners from the reference marriage indicated during their interview to have at least one biological parent alive, a random parent was selected. In total, 2,157 parents completed a mail or web survey. If the ex-partners from non-intact reference marriages indicated during their interview that they live together with a partner, this new partner was also asked to participate in the study. 1,837 new partners completed a mail or web survey. 31% (n=2,009) of the (ex-)partners from the reference marriages are 50 years or older, 3% (n=163) are 60 years or older. 96% (n=2,119) of the parents that were questioned are 50 years or older, 81% (n=1,803) are 60 years or older, 48% are 70 years or older (n=1,056) and 12% are 80 years or older (n=271). 29% (n=586) of the new partners were 50 years or older at the time of the interview, 7% (n=131) were 60 years or older.

First year of collection
2009

Stratification if applicable
The sample is disproportionally stratified with regard to the status of marriage (1/3rd intact; 2/3rd dissolved) and proportionally with regard to year of marriage.

Base used for sampling

Geographical coverage and breakdowns
Flanders

Age range
No age limitations apply.

Statistical representativeness
Special group, please specify

Coverage of main and cross-cutting topics
The Divorce in Flanders survey is an important survey to study wellbeing in later life: all actors are questioned in detail about their psychological, emotional and physical wellbeing. The survey further contains several questions relating to intergenerational relationships. A battery of questions relates to the attitudes towards intergenerational support (e.g. Grandparents should look after their grandchildren if the parents of these grandchildren are unable to do so;…). Other questions relate to relationship quality, communication and mutual support between (grand)parents and (grand)children (cf. what is the frequency and importance of mutual support relationships between generations;…). The Divorce in Flanders survey also allows for a study of health and performance (e.g. it contains information on health-related service use and other service use) and work and productivity (e.g. how is work distributed across the life-course). The dataset is especially suitable to study the above mentioned issues from a life-course perspective as the questionnaire contains detailed information on most family and labour market transitions.

Linkage

Standardisation
Where applicable, international validated questions and scales are used in the different questionnaires.

Possibility of linkage among databases
As the sample was drawn from the National Register, the (confidential) data contain the National Register number for each respondent. This could allow for a link with other national databases containing the National Register number.

Data quality

Entry errors if applicable
The data quality is good. Response rates differ between the actors: 38% of the (ex-) partners, 57% of the residential target children, 36% of the non-residential target children, 42% of the parents and 68% of the new partners that were contacted, participated in the study. To correct for selectivity as a consequence of non-response, the project provides weighting procedures. Working with CAPI-interviews ensures a large data quality in terms of item non-response, as the interviewer guides the respondent through the complete interview, following the instructions programmed by the researchers. The item-non response on the mail and web-surveys are overall very small.

Breaks

Consistency of terminology or coding used during collection

Governance

Contact information
Inge Pasteels
Centrum voor Longitudinaal en Levenslooponderzoek (CELLO), University of Antwerp
St. Jacobstraat 2
2000 Antwerp Belgium Phone: +32 47 448 24 17
Email: info(at)scheidinginvlaanderen.be
Url: www.scheidinginvlaanderen.be

Timeliness, transparency
The data were collected in 2009-2010. In the period 2010-2012, the data were only available for members of the Divorce in Flanders research team (datacleaning, exploratory analyses, …). Since 2013, the data are also available for external users upon request.