# README These files accompany the article by Willem R. J. Vermeulen, Mioara Zoutewelle-Terovan, Niels Kooiman and Aart C. Liefbroer on the effects of couple-level, contextual-level and cross-level religiosity on union dissolution rates in the Netherlands titled "Religion and Union Dissolution: Effects of Couple and Municipal Religiosity on Divorce and Separation". Using the code and data (described) in these files, we show that within the Netherlands, the effect of couple religiosity on union dissolution risks depends on the religious context in which a couple lives, and the relationship type of the couple. Moreover, we show that union dissolution risks are higher for religious couples who live in less religious contexts or choose to cohabit instead of marry. Reference: - Vermeulen, W. R. J., Zoutewelle-Terovan, M., Kooiman, N., & Liefbroer, A. C. (2023). Religion and Union Dissolution: Effects of Couple and Municipal Religiosity on Divorce and Separation. Demographic Research. ## Code The code used to perform the analyses used in this paper has been developed inside the secure environment of Statistics Netherlands (CBS). Running it on the data requires access to the data by Statistics Netherlands, as well as permission to use Stata/MP 16.1. To run the code in the secure environment of Statistics Netherlands, first navigate to the "code" folder and run "do main.do" in Stata/MP 16.1 to perform all data transformations. When all data transformations are done, run "do models.do" to run all models used in the paper. Be aware that some processes can take up quite some time. To dive into the code itself, it is important to know that the data transformation code is structured in multiple phases, which consist of several files which perform mutations or analyses on the data. Every file has an accompanying load and save file to aid the researcher. Software used: - https://www.stata.com/statamp/ ## Data To be able to access the data used in this paper, you need to request access data using the GAP file (the data request format used at Statistics Netherlands) found in the "data/documentation" folder. To do this, you need need to work for an authorised institution. If your institution has not been authorised to access Statistics Netherlands data, you can request such authorisation at the general director of Statistics Netherlands. More information (Dutch): - https://www.cbs.nl/nl-nl/onze-diensten/maatwerk-en-microdata/microdata-zelf-onderzoek-doen/aanvraag-toegang-microdata After providing the GAP, Statistics Netherlands will construct files from microdata accordingly. The data transformation code expects the following files to be present: - BESTAND3REGIO.sav - BURGSTATBUS.sav - GBAMIGRATIEBUS2019.sav - GBAOVERLIJDEN2019TAB.sav - INKOMENANAV1.sav - RELIGIEENSCHEIDING.sav - SAMENWONERBUS2018ANAV1.sav Complete documentation of every variable requested in the GAP can be found in the "documentation" folder, including information on every file and origin of the data. Be aware that all variables and documentation supplied by Statistics Netherlands is in Dutch. ## List of Files The following code and data files (excluding load and save files) are included in this replication package: replication │ readme.txt Contains all descriptive information of this file and all other files. │ ├───code │ │ main.do Run all data preparation. │ │ models.do Run all models as seen in the paper. │ │ │ ├───env_phase1 │ │ env_religion_make.do Generate religious variables on the municipal level. │ │ │ ├───paper │ │ analyse.do │ │ correlation.do Correlation matrix, full sample. │ │ correlation_mar.do Correlation matrix, married sample. │ │ descriptives.do Descriptive table information. │ │ h0_modelbase.do Base model used as reference, married sample. │ │ h0_modelbase_everyone.do Base model used as reference, full sample. │ │ h1_model1.do Control + Religious affiliated, married sample. │ │ h1_model2.do Control + Religious affiliated + Visits to religous gatherings, married sample. │ │ h1_model3.do Control + Religious affiliated + Denomination, married sample. │ │ h1_model3_ref_rk.do Model 3; switch denomination reference group to Roman Catholic. │ │ h2_model4.do Control + Religious affiliated + Visits to religous gatherings + % Religious (municipal), married sample. │ │ h3_model5.do Control + Religious affiliated + Visits to religous gatherings + % Religious (municipal) + % Religious (municipal) x Religious affiliated, married sample. │ │ h3_model6.do Control + Religious affiliated + Visits to religous gatherings + % Religious (municipal) + % Religious (municipal) x Visits to religous gatherings, married sample. │ │ h3_model6_too_complex.do No, it doesn't work to do multiple cross-level interactions. │ │ h4_model7.do Model 5; full sample. │ │ h4_model7_regpar_gone.do Model 7; couples in registered partnerships removed. │ │ h4_model7_regpar_is_mar.do Model 7; couples in registered partnerships added to the married sample. │ │ h4_model7_switched.do Model 7; switch reference group. │ │ h4_model8.do Model 6; full sample. │ │ h4_model8_regpar_gone.do Model 8; couples in registered partnerships removed. │ │ h4_model8_regpar_is_mar.do Model 8; couples in registered partnerships added to the married sample. │ │ h4_model8_switched.do Model 8; switch reference group. │ │ run_all_models.do Run all models as seen in the paper. │ │ │ ├───phase1: Transform CBS SPSS files into new datafiles with variables with the right types to perform analyses. │ │ bsbn_make.do Process civil status information. │ │ ebbn_make.do Process information from the Dutch Labor Force Surveys (LFS). │ │ inkn_make.do Process income data. │ │ swbn_make.do Process cohabitation data. │ │ │ ├───phase2: Link individual survey entries to individual income information and information about parental divorce. │ │ ebbi_make.do Combine LFS and income data in one file. │ │ parents_divorce_make.do Generate parental divorce variable. │ │ │ ├───phase3: Link individuals to their partners' survey data (create couples - filter homosexual couples from the data). │ │ ebbp_make.do Link individuals with partners, filter couples. │ │ │ ├───phase4: Transform files to couple-year files. │ │ bsbym_make.do │ │ bsbyv_make.do │ │ bsby_make.do │ │ ebby_make.do │ │ swby_make.do │ │ │ ├───phase5: Link the survey and civil registration data. │ │ complete_make.do Merge all yearly data together. │ │ │ ├───phase6: Strip away a lot of variables we do not use to speed up loading times of data, and censor. │ │ stripped_make.do Strip a lot variables not used in the analyses. │ │ │ ├───phase7: Generate all couple level variables and link municipal level religious variables, use regressions to fill in missing values. │ │ final_make.do Generate all couple level variables, link municipal level religious variables, fix missings. │ │ │ └───phase8: Filter the final sample to create a married-only sample. │ final_married_make.do Filter for only married couples. │ └───data └───documentation: All files containing full background information on the variables used in the GAP. 200511_GAP_Scheiding_en_Religie.docm Document used for the original data request at Statistics Netherlands. meta.txt Contains all meta information on the data, including construction.