Page 55 - IJPS-2-2

Mark Lyons-Amos

starting school is also included as a measure of how easy it is to combine childbearing with work,
due to rising female employment and increased difficulty in fulfilling multiple roles influencing fer-
tility outcomes (Mason, 1998; McDonald, 2000; Mills, 2010). The effect of the interaction between
leave provision (as well as other indicators of the integration of women into the labour market post
childbearing) and fertility and partnership behaviour is widely documented (Adserà, 2004; Ahn and
Mira, 2002; Chesnais, 1996; Neyer, 2003) as well as the broader social contexts within which both
demographic behaviour and policy formation occur (McDonald, 2000).
Measures on the legal status of cohabitation is also included (Perelli-Harris and Sánchez-Gassen,
2012). Specifically, the proportion of policies in which cohabitation was mentioned and the propor-
tion of policies in which marriage and cohabitation were afforded equal status are used. Finally, a
binary indicator of whether the country in question has a legally recognised cohabiting state distinct
from marriage is included (e.g., PACS).
Individual level data for this analysis are drawn from the Harmonized Histories (Perelli-Harris,
Kreyenfeld, and Kubisch, 2010, and see www.nonmarital.org where the data can also be requested).
The Harmonized Histories is a dataset containing consistent retrospective demographic histories
from 16 countries across Europe based largely on GGP surveys (as well as some other national surveys).
A subset of 10 countries from the full Harmonized Histories dataset is extracted to ensure that
each country in the dataset has at least some contextual level information available. Therefore, res-
pondents from Austria, Bulgaria, Estonia, France, the Netherlands, Norway, Romania, Russia, Spain,
and the United Kingdom are selected since these countries have at least some information available
either from the GGP contextual database (or comparable source) or the study by Perelli-Harris and
Sánchez-Gassen (2012). The sample size within each country is presented in Table 1. Only women
within the sample are retained consistent with the study by Perelli-Harris and Lyons-Amos (2015)
and Perelli-Harris and Lyons-Amos (2016).
Three individual level processes are used in this analysis, capturing the processes of marriage
formation, the formation of non-marital cohabitating unions and becoming a mother (first birth). The
processes are modelled accurate to the nearest year consistent with Perelli-Harris and Lyons-Amos
(2015) and Perelli-Harris and Lyons-Amos (2016). I model these processes between the ages of 16
and 45, delimiting the effective exposure to the three processes modelled (women cannot marry before
the age of 16 in the selected countries, and are generally speaking post-menopausal by the age of 45).
C
M
Marriage and cohabitation processes are represented by response variables y and y take
tj
tj
the value 0 for years t when the respondent is not in a marital or cohabitating relationship respec-
tively, and 1 where they are. These are not cumulative rates: women can exit both marriage and co-
habiting relationships reflected in the falling probability of being in either of these states reflecting

Table 1. Sample size within each country
Country and original survey Number of women
Austria GGS 1505
Bulgaria GGS 5236
Estonia GGS 4252
France GGS 4556
Netherlands FFS 3476
Norway GGS 5922
Romania GGS 5176
Russia GGS 5836
Spain SFS 7300
UK BHPS 5890

International Journal of Population Studies | 2016, Volume 2, Issue 2 49