ABSTRACT
It has often been argued that Southern European countries are more familialistic in their culture than Western and Northern European countries. In this paper, we examine this notion by testing the hypothesis that adult children are more responsive to the needs of their elderly parents in countries with more familialistic attitudes. To test this hypothesis, we analyse the Survey of Health, Ageing and Retirement in Europe (SHARE). We focus on three indicators of need: (a) the partner status of the parent, (b) the health status of the parent, and (c) the education of the parent. Using Heckman probit models, we examine the effects of these variables on whether or not the parent receives instrumental support from children, thereby controlling for whether or not children live independently from their parents. We estimate effects of need on support and we compare these effects across 10 European countries, using both graphic devices and a multilevel probit model where individuals are nested in countries. We find significant cross-level interactions of need variables and the degree of familialism in a country. Our analyses, thereby provide more positive evidence for the hypothesis than earlier studies, which have focused largely on comparing aggregate levels of support among countries.
Notes
Earlier versions of this paper were presented at the session on Intergenerational Relations of the FAMNET/EQUALSOC Workshops in Turin, January 27, 2006 and in Barcelona, September 8, 2006. The authors gratefully acknowledge the comments received in both occasions as well as those by the anonymous reviewers.
2If there were more than four children, the children who live closest to the parent were selected.
3The age of the child may be associated with support giving, but this will largely be due to the correlated age of the parent or to the life course stage of the child. Since the substantive equation includes the age of the parent as well as information on the life course stage of the child, there will be no net remaining effect of the child's age on support. Additional analyses also show that there is no net effect of child age on support (see later).
4We use STATA to estimate all the regression models, including the multilevel models.
5These models are not controlled for clustering of dyads within families since this was not possible in the multilevel probit model. A three-level probit model would be an alternative, but this is not available in STATA.
Additional information
Notes on contributors
Matthijs Kalmijn
Matthijs Kalmijn is Professor in the Department of Sociology of Tilburg University
Chiara Saraceno
Chiara Saraceno is Professor at the Faculty of Political Sciences, University of Torino, Italy, and Research Professor on Demographic Development, Social Change, and Social Capital at the Wisseschaftszentrum Berlin für Sozialforschung