Associations between Non-Marital Childbearing and Its Correlates over Time and Space in Germany – Does Voting Behaviour Matter?

Stephanie Thiehoff, University of Southampton
Agnese Vitali, University of Southampton
Brienna Perelli-Harris, University of Southampton

There have been considerable changes in family formation patterns and an uptake of non-marital childbearing in Europe since the 1970s. Germany is probably the most extreme case with large spatial variations in births out of wedlock in Europe. Different theoretical frameworks have been developed to understand this phenomenon: Ideational or value change to more liberal values is advocated to be the reason on the one hand and economic uncertainty is seen to be the driver on the other hand. Theories as well as research in politics and political psychology propose that voting behaviour is closely linked to personal values. How liberal voting behaviour helps to understand fertility behaviour will be explored here as proposed by the SDT framework. Other important correlates are economic uncertainty, female labour force participation as well as education. The objective of this paper is to study the spatial associations of non-marital childbearing and especially which correlate contributes the most to it in different points of time (1994, 2004 and 2014). A Geographically Weighted Regression using standardised variables is applied to determine the variable with the highest explanatory power and adding the most to childbearing out of wedlock. When comparing the variables with most explanatory power the usual East-West-divide completely disappears. Liberal and egalitarian family values seem to matter in explaining an increase in non-marital childbearing. As non-marital childbearing becomes more prevalent, male unemployment becomes more important in understanding this behaviour in times of higher economic uncertainty. The analysis suggests as well that when childbearing within cohabitation becomes more prevalent, different variables (values and economic uncertainty) have higher explanatory power than when births to single mothers are more widespread.

Therehave been considerable changes in family formation in Europe since the 1970s(Sobotka 2008; Billari & Liefbroer 2010; Perelli-Harris 2015). Onephenomenon related to this is the increase in non-marital childbearing(Klüsener, Perelli-Harris, & Sánchez Gassen, 2013;Perelli-Harris 2015) which can mostly be attributed to an uptake in birthswithin cohabitation (Perelli-Harris et al. 2010). Most recent research suggeststhat the geographical scale matters as well to understand childbearing withincohabitation (Lappegård, Klüsener & Vignoli 2017): The SDTtheory is important for the across-country level, where the pattern ofdisadvantage hypothesis on the within-country and individual level.

Germanyis probably the most extreme case with large spatial variations in births outof wedlock in Europe (Kreyenfeld & Konietzka 2010) which also showscontinuity over decades (Klüsener & Goldstein 2014). Given thelong-standing regional heterogeneity in non-marital childbearing it seemssurprising that researchers have not questioned that its determinants varyspatially as well. The objective of this paper is to study if the associationsbetween non-marital childbearing and its correlates space and time in thecontext of Germany, especially, which correlate contributes the most tonon-marital childbearing in each cross-section (1994, 2004 and 2014).

Theoriesas well as research in politics and political psychology propose that votingbehaviour is closely linked to personal values (Schwartz 2013; Barnea &Schwartz 1998).[1]In Germany, left-wing parties tend to represent the most liberal values (e.g.strong support of same-sex-marriage, support for different family types).[2] Hence, this analysisuses the share of votes for the two left parties (Green and Left [Bündnis90/Die Grünen and Die Linke]) in the most previous federal electionto each cross-section to measure the association between non-maritalchildbearing and these cultural or political values. This variable is mostlylinked to the SDT framework and supports the ideational change proposition. Maleunemployment is used as an indicator for economic uncertainty and thus for the“pattern of disadvantage” (Perelli-Harris et al. 2010). Male unemployment is oftenassociated with looser partnership forms (Oppenheimer 2003) and simultaneouslywith a rise in childbearing out of wedlock. Female education is considered tobe associated with childbearing within cohabitation (Lappegård,Klüsener, & Vignoli 2017; Vitali, Aassve, & Lappegård 2015). Here,it is measured by the percentage of female school leavers finishing school withhigher education entrance certificate (Abitur). Female labour forceparticipation as an indicator for female economic autonomy which is oftenlinked to having children outside marriage as well (Goldschneider, Bernhardt,& Lappegård 2015; Esping-Andersen & Billari 2015). Bothvariables[3]can be seen from an ideational change perspective (Lesthaeghe 2010) but from aneconomic perspective as well (Oppenheimer 2003).

Geo-referencedGerman data on the district level is used in this analysis. These districts areequivalent to the NUTS-3 units (Nomenclature of Territorial Units forStatistic). The dependent variable is the non-marital birth ratio (NMR) foreach district. A Geographically Weighted Regression (GWR) using standardisedvariables is performed to examine the potential spatially varying associationbetween district-level non-marital fertility (Fotheringham, Brunsdon, &Charlton. 2002). Due to the standardisation of variables it ispossible to determine which of the explanatory variable contributes relativelythe most to non-marital childbearing in each district. In addition to theaforementioned explanatory variables, the GWR controls for population densityand gross-domestic product per capita.

Whencomparing the variables with most explanatory power an East-West-dividecompletely disappears (Fig. 1). Liberal and egalitarian family values seem tomatter in explaining an increase in non-marital childbearing. In 1994, in aroundtwo thirds of the districts this variable has the most explanatory power. Thissupports argument that liberal values are associated with a diffusion of non-maritalfertility in Germany. As non-marital childbearing becomes more prevalent, maleunemployment becomes more important in explaining this behaviour in times ofhigh economic uncertainty (Fig. 1, 2004). When economic uncertainty isdecreasing again, egalitarian values play a larger role. In 1994, in North-Eastmale unemployment and GDP of highest relevance. This situated might suggestthat the value change already took place earlier. Clearly, non-marital childbearingis already widespread in these regions before the first cross-section which isanalysed here. Hence, economic indicators seem to be more important.



[1] Other influential variables on votingbehaviour: group membership, special individual interest andpopularity/character of candidate (Sears 1987).

[2] This is supported by their electionmanifestoes.

[3] Collinearity diagnostics for allexplanatory variables were performed to

Presented in Session 1156: Fertility