Read me file for the replicability of Demgrapihc Research paper 4317 
“Parental socio-economic status and the timing of first marriage: What is the role of unmarried cohabitation? Results from a cross-national comparison”
by M.D. (Anne) Brons, A.C. Liefbroer & H.B.G. Ganzeboom


Statistical Program used:
All analyses in this paper were conducted using STATA 12.

Data used:
The data used in this study are the Generations and Gender Survey Wave 1 and the Harmonized Histories dataset, which both can be retrieved from the GGP website (https://www.ggp-i.org/), 
after permission has been granted by the Generations and Gender Programme. 
We only included countries for which sufficiently detailed information was available on the partnership history and parental and individual educational attainment, 
namely, Austria, Belgium, Bulgaria, Czech Republic, Estonia, France, Georgia, Hungary, Italy, Lithuania, Netherlands, Norway, Poland, Romania, Russia, and Sweden. 
For the United Kingdom and the United States, we use the Harmonized Histories dataset created by the Non-Marital Childbearing network and made publicly available to the Gender and Generations Programme research community. 
For the Netherlands and Canada we used separate datasets, namely the Dutch Survey on Family Formation (2008) and the Canadian General Social Survey Cycle 20. 
The Dutch Survey on Family formation (2008) is available from Data Archiving and Networked Services (https://dans.knaw.nl/nl) with the permission of Centraal Bureau voor de Statistiek (Statistics Netherlands). 
The Canadian General Social Survey Cycle 20 is available via Statistics Canada with the permission of Statistics Canada. 

Variables used:
Variables							Coding

Ever married?							Dummy (1= yes, 0 = no)
Year first marriage						Continuous
Ever lived together with partner?				Dummy (1= yes, 0 = no)
Year of first co-residential union				Continuous
Cohabitation history						Time varying binary variable whether cohabiting (1) or not (0) at certain age
Highest educational level of father and mother			Continuous (ISLED scale)
Highest educational level respondent				Continuous (ISLED scale) & time-varying
Educational enrolment						Dummy (1= yes, 0 = no), also time-varying
Age respondent							Continuous & time varying
Gender								Dummy (1=male, 0=female)
Birth year							Continuous
Country								String variable (AT, BG, BE etc..)
Prevalence of cohabitation					Continuous (aggregated from data)
Percentage who got first child within cohabitation		Continuous (aggregated from data)
Percentage who married within 2 years after cohabitation	Continuous (aggregated from data)
Percentage who separated within 2 years after cohabitation	Continuous (aggregated from data)

Note: More information about construction of variables in the paper.

Data preparation & analyses:
Within the CONOPP project (ERC grant 324178), we harmonized separate parts of the GGP and Harmonized Histories datasets, including all partnership and fertility histories and all childhood indicators, 
before we started with the data preparation and analyses. 

The supplemental .zip archive for this publication contains the following syntax files:
1. Syntax data preparation & logistic regressions: 
	Syntax parental edu_first marriage_data GGP & Harmonized Histories_final.do

By using this syntax we combined the harmonized partnership and fertility histories dataset with the childhood indicators dataset into one dataset, 
constructed all variables and ran all the logistic regressions needed to estimate the parameters that were included in the meta-analyses and meta-regressions

2. Syntax data preparation &logistic regressions (Netherlands):
	syntax parental edu_first marriage_dataset netherlands_final.do

For the Netherlands, we used a separate dataset and a separate syntax. 
In this syntax, we constructed all variables for the Netherlands and ran all the logistic regressions needed to estimate the parameters that were included in  the meta-analyses and meta regressions

3. Syntax data preparation &logistic regressions (Canada):
	syntax parental edu_first marriage_dataset canada_final.do

For Canada, we had a separate dataset and a separate syntax. 
In this syntax, we constructed all variables for Canada and ran all the logistic regressions to estimate the parameters that were included in  the meta-analyses and meta regressions

4. Syntax meta analyses & meta regressions: 
	syntax meta analysis parental edu_first marriage_final.do

In this syntax, we performed all the meta-analyses and meta-regressions based on the estimates and standard errors of the country-specific logistic regression analyses.