Documenting cross-national comparability of survey data: the Online - - PDF document

1

Documenting cross-national comparability of survey data: the Online Codebook & Analysis of the Generations & Gender Programme

Arianna Caporali arianna.caporali@ined.fr Institut national d’études démographiques, INED Abstract In international survey projects, metadata are paramount to evaluate cross-national comparability of data. They help identify issues of conceptual or methodological equivalence between country datasets. This paper presents the ways in which metadata are provided in the framework of the Generations and Gender Programme (GGP). This is a longitudinal demographic survey of 18-79 year olds in 19 countries in Europe and beyond. It is managed by a consortium of research institutions on the basis of a decentralized model that is relatively dependent upon post hoc harmonization of data. The challenge of documenting GGP data is the need to combine information from existing surveys with information on the harmonization

• process. To face this challenge, metadata are provided in compliance with the Data

Documentation Initiative (DDI), an international standard for documenting social science data, and are made available in an online codebook and analysis tool through the software package Nesstar. The paper illustrates the DDI elements chosen to document cross-national comparability of GGP surveys and discusses the experience of the GGP team in providing the relevant information. It ends with an outlook on possible future development. Content 1. Introduction: metadata in international survey programs ........................................... 2 2. The Data documentation Initiative and the software Nesstar ...................................... 3 3. The case of the Generations and Gender Surveys ......................................................... 3 3.1. Main characteristics and methodology ....................................................................... 3 3.2. Implications for data documentation ........................................................................... 5 4. The Generations and Gender Programme Online Codebook and Analysis ............... 5 4.1. Description of the workflow steps ............................................................................... 5 4.2. Online GGP surveys data files and their DDI metadata items ................................... 6 4.2.1. The pooled datasets .............................................................................................. 7 4.2.2. The country-specific data files ............................................................................. 8 4.2.3. The variable availability data file ....................................................................... 10 5. Conclusions ...................................................................................................................... 11 6. References ........................................................................................................................ 12

2

• 1. Introduction: metadata in international survey programs

Survey data can be usable only if accompanied by comprehensive metadata. These are ‘data about the data’ necessary to transform the numbers into meaningful knowledge. Without metadata, data sharing and secondary analysis would not be possible, and researchers could not test and replicate each other’s studies (King 1995). In international survey programs, metadata are paramount to document data quality across countries and evaluate issues of cross-national equivalence between country methodologies and datasets (Mohler et al. 2010). This is especially crucial for survey programs based on a decentralized management model. Indeed, in these programs, survey instruments and guidelines can be either adapted to the specific country contexts or ‘incorporated’ into existing surveys. Such surveys require full documentation of country specificities in fieldwork methodologies, as well as of the harmonization procedures that are implemented to ensure data comparability. Metadata are even more important in the framework of longitudinal surveys, where subsequent waves need to be described. Metadata preparation concerns both survey producers and survey data archives (Caporali, Morisset, and Legleye 2015). Data producers create coherent data files and prepare survey documentation. Survey data archives must ensure their reusability even behind national boundaries. To this end, they review quality of data files, ensure that they are sufficiently anonymised to avoid risk of re-identification of respondents, and create exhaustive and structured metadata files following international standards. Today the standard called Data Documentation Initiative (DDI) is widely used in social sciences, both for national and for cross-national comparative surveys (Mohler et al. 2010). It is often used in connection with Nesstar (Networked Social Science Tools and Resources), a user-friendly software (Vardigan, Heus, and Thomas 2008), which can be used to prepare survey metadata according to the DDI standard and to disseminate surveys online. DDI and Nesstar are recommended by the Consortium of European Social Science Data Archives (CESSDA). What is it to document cross-national comparability? What are the documentation challenges posed by longitudinal surveys? How can they be overcome? This paper presents the ways in which metadata are prepared and disseminated in the framework of the cross- national longitudinal survey project called Generations and Gender Programme (GGP; see section 3). GGP has a decentralized management model. It therefore relies on considerable post hoc harmonization of data. The specific challenge that data documentation poses in this project is the need to combine information from already collected surveys with information

• n post hoc harmonization processes over subsequent survey waves. Then metadata have to

follow international recommendations, be compliant to DDI standard and disseminated online through the Nesstar software. After an explanation of what DDI and Nesstar are (section 2), the paper focuses on the GGP characteristics and methodology and on the issues that are to be overcome to document its surveys in a way that ensures reusability by international researchers (section 3). This is a collaborative work between GGP national data collectors and GGP central coordination team and follows a standardised procedure (section 4.1). The paper also explains the DDI elements chosen to document the surveys, as well as the ways in which Nesstar functionalities are used to present the surveys online. The content of the GGP Online Codebook and Analysis is detailed and illustrated with some examples (section 4.2). The paper ends with some reflexions about the GGP experience in data documentation and the plans for improving the data documentation process in future rounds of GGP data collection.

3

• 2. The Data documentation Initiative and the software Nesstar

DDI was launched in the mid-1990 by researchers affiliated to data archives. It consists of a set of standardised items in eXtensible Markup Language (XML) for the general description of surveys down to the level of each single variable of a dataset. It provides a framework and a format for metadata documentation and for data and metadata online

• publication. Two types of the DDI specification are used by data archives. The first is the

DDI-Codebook (DDI-C or DDI 2), which is designed to document simple survey data and focuses on the elements of a traditional codebook. This type of DDI is broadly used especially in connection with the Nesstar software system (Vardigan, Heus, and Thomas 2008). The second type of DDI is the DDI-Lifecycle (DDI-L or DDI 3) which documents surveys across their entire life cycle, by including all the DDI-C elements and further extending it (Kramer et

• al. 2011). It is particularly suited for longitudinal and cross-national comparative surveys

because it contains features that allow explicit comparisons between items of different datasets (Mohler et al. 2010).1 A third version of DDI (DDI 4) is being developed to accommodate the needs of users outside the social sciences by providing a more flexible model and enhanced technical functionalities. Nesstar was developed in the second half of the 1990s, in the framework of CESSDA seminars (Marker 2013) and it is today managed by the Norwegian Centre for Research Data (NCD). It is a software system for managing documentation in DDI-Codebook format without having to know the XML language. When it is linked to a web server, it allows publishing data and metadata on the Internet. It provides a convenient way for users to search, browse and explore data and metadata online. It offers an online analytical tool that visualizes variable distributions with tables and graphs and performs basic statistical analyses (i.e., cross-tabulations, linear regressions and correlations). It is also possible to download data and metadata in a variety of file format (including DDI), as well as analysis results.2 Thanks to its convenient analytical functionalities, Nesstar and DDI-Codebook are today widely used not only by social science data archives (e.g., the members of the French Réseau Quetelet)3, but also in the framework of international survey programmes (e.g., the European Social Survey - ESS4, European Values Study - EVS, The International Social Survey Programme - ISSP)5.

• 3. The case of the Generations and Gender Surveys

3.1. Main characteristics and methodology The GGP is a panel survey on 18-79 year olds in Europe and beyond, launched in 2000 by the United Nations Economic Commission for Europe (UNECE), to study demographic change (Vikat et al. 2007). It is run by a consortium of research institution. The GGP central coordination team is based at the Netherlands Interdisciplinary Demography Institute (NIDI, The Hague). Among the other main institutions contributing to its development there is the French Institute for Demographic Studies (INED, Paris). The GGP is the follow up of some international fertility survey programmes carried out since the 1960s, such as the Family and Fertility Surveys (FFS). Compared to its predecessors, it presents

1 For further information see DDI website (http://www.ddialliance.org/) and Caporali, Morisset, and Legleye

2015, box 1 (p.541).

2 For further information see Nesstar website (http://www.nesstar.com/) and Caporali, Morisset, and Legleye

2015, box 2 (p.542).

3 http://www.reseau-quetelet.cnrs.fr/spip/ 4 http://nesstar.ess.nsd.uib.no/webview/ 5 http://zacat.gesis.org/webview/

4

innovative aspects (Vikat et al. 2007): The GGP is a longitudinal panel survey (three year interval between each wave) aimed at studying socio-demographic and economic challenges, such as low fertility, changes in family structures and population ageing, with a multidisciplinary approach including demography, sociology, economics and psychology. It provides survey data with the so-called Generations and Gender Surveys (GGS) on the relationships between generations and gender along individuals’ life-courses, and it complements the micro-level survey data with a macro-level contextual database (Caporali et

• al. 2016; Vikat et al. 2007). Among the research subjects that it allows to study there are, for

instance: the impact of early events on life trajectories, the dynamics of family relations, the relationship between fertility intentions and the decision to have a child, the impact of the role

• f family carers on their lives, the interactions between individual behaviours and the socio-

economic and political contexts of where they live, the reconciliation of family and working life.6 The GGP is based on a relatively decentralized management model. The survey instruments and the fieldwork guidelines are developed by the central coordination team (UNECE 2005, 2007). They include a number of recommendations7 and a ‘standard’ questionnaire provided in English in .pdf file format. These recommendations can either be adapted by the national teams to the different national contexts in ‘bespoke’ surveys (e.g., in the case of Bulgaria, France and Germany), or partly incorporated into existing surveys (e.g., in the case of Australia, Hungary, and Italy). This leads to country differences in fieldwork calendars and methodologies (Fokkema et al. 2016), as well as in the level of ‘compliance’ to the standard questionnaire (Emery and Caporali in prep.) For example, there are deviations in the starting years for the first wave (varying between 2004 in Bulgaria and 2012 in Sweden) and in the minimum number of contact attempts. The level of compliance to the questionnaire varies between 84% (Bulgaria) and 23% (the Netherlands) of the questions asked. Country- specific variables and response categories are often maintained in the data files available to secondary users. In spite of weak central fieldwork coordination, the GGP implements centralized data harmonization (see documentation available from the website) and documentation procedures to ensure cross-national comparability and extensive metadata about country deviations (see section 4.1). After the EU funding within the 7th Framework Programme (2008-2012), a new era for GGP has started in 2016, when it was granted the status of Emerging Project by the European Strategic Forum for Research Infrastructures (ESFRI). The GGP plans to expand its databases through the addition of new and existing surveys. It is also preparing a new round

• f data collection for 2019 (Gauthier and Emery 2016; see also the GGP website). This new

round of data collection is planned to be based on a renewed methodology. This includes the implementation of survey fieldworks under the supervision of the GGP central coordination team at NIDI, with the possibility to monitor the fieldwork and data as they are collected. This new methodology may allow for a higher degree of compliance with the standard questionnaire and, as a result, for a considerable reduction of the efforts needed for post hoc

• harmonization. The new round of data collection is also planned to use a revised

questionnaire comparable with the FFS and the previous GGP round.

6 For further information on the GGP data and coverage see its website (http://www.ggp-i.org/). 7 The methodological recommendations include: to carry out three waves (one wave every three years), to

sample a panel of about 10,000 individuals at the first wave living in households, to have at least 8,000 respondents at the third wave, to implement random sampling, to have comparable number of men and women, to limit losses between waves through oversampling and follow-up of people, and to carry out CAPI (Computer- assisted personal interviewing) or PAPI (Paper and pencil personal interviewing).

5

3.2. Implications for data documentation The GGP methodology poses a number of challenges for data documentation. First, it is necessary to provide comprehensive description of country specificities. This implies detailed metadata on data collection methodology including the description of the data collection mode, contact protocols, sampling and weighing procedures. It also implies documentation of the levels of compliance to the standard questionnaire. For this, it is necessary to describe for each variable country deviations in question text and response categories, as well as to provide information on its availability across countries and waves. The country-specific metadata should also be enriched with the provision of national questionnaires, methodology reports, and any additional documents that may be of relevance for the understanding of country-specificities. All the country-specific information needs to be collected from the national team carrying out the fieldwork. Second, the country-specific metadata needs to be complemented with comprehensive metadata that are the same for all GGP surveys, including the details on the post hoc harmonization process. This concerns metadata at the level of each variable which should include the description of the question texts and response categories as phrased in the standard questionnaire, as well as the calculation method for derived variables. It also concerns a detailed description of the harmonization process, including label and routing checks and consolidated variables. All this information, which is equal for any GGP survey, needs to be regularly updated by the central coordination team, easily comprehensible and appraisable by the data users. Third, all this metadata needs to be provided for each wave and in a way that data users can compare country-specificities and harmonization processes across subsequent

• waves. Finally, all the information has to be organized in Nesstar in a way to allow useful
• nline browsing and analysing of data and metadata, and according to the DDI specification

supported by Nesstar. This is the so-called DDI-Codebook, which is designed to document simple surveys.

• 4. The Generations and Gender Programme Online Codebook and Analysis

4.1. Description of the workflow steps GGP metadata preparation involves INED, acting as GGP data archive, NIDI, which is in charge of post hoc data harmonization, and the national GGP teams that are in charge of national data collection and of providing NIDI and INED with pre-harmonized and documented data files. In order to face the challenges that GGP poses for data documentation, the work at INED is structured in two procedures, one concerning the documentation of variables, and the other one the documentation of fieldwork methodologies. As to the variables documentation, once new data files are received by NIDI (it can be either unreleased data files or revisions of already published ones), all the variables are tabulated to check whether there are any undocumented values and whether country deviations are correctly labelled and taken into account, especially in the calculation of consolidated variables. Missing codes are also revised to ensure that they are correctly

• assigned. Checks are done both in SPSS and in STATA format (the two formats in which

GGP data files are made available), to make sure that codes and labels are correctly assigned regardless of the software used for the analysis. At this stage, constant contacts are maintained with the team at NIDI to report errors and, if necessary, ask for further information about country deviations. This may take a lot of time depending on the number of issues to be discussed and their complexity (e.g., some issues may require further clarification to be asked to the national GGP teams). Once the datasets are finalised, they are imported in Nesstar,

6

along with a DDI file already pre-filled with metadata that are the same for all GGP surveys. This allows to automatically enter this information in the DDI fields where it is stored, thus allowing for considerable time saving. Each single variable is then further browsed and revised and, if necessary, variable metadata are refined ‘by hand’ to take into account and explain country specificities. This is a detailed, meticulous, and time demanding work. Indeed it needs to be done for more than 2000 variables for Wave 1 (version 4.3), and for more than 2000 variables also for Wave 2 (version 1.3). Additionally, at this stage, a separate data file is created in SPSS to document the variable availability across countries and waves. This data file is also imported and documented into Nesstar. As to the fieldwork metadata, first country-specific fieldwork metadata are collected from national teams based on a template structured according to DDI items chosen to document GGP surveys. INED then revise the provided information and, if necessary ask for further clarifications. Additional useful documents and resources are also identified, such as national methodology reports and survey instruments. This often implies a lot of documentation work both from the side of the national teams and from INED. This information is then imported into Nesstar and the links to relevant national documentation are

• added. When the whole documentation process is completed, the data files and the metadata

are published online in the GGP Nesstar webview, called GGP Online Codebook and Analysis. 4.2. Online GGP surveys data files and their DDI metadata items In the GGP Online Codebook and Analysis8, there are three types of data files (left side frame of Figure 1). First, to provide users with cross-national overviews of data, pooled datasets are published for each available GGP survey wave. Second, in order to fully account for national deviations in fieldwork methodologies, country-specific data files are also

• provided. Third, the so-called ‘Variable Availability’ data file is aimed to document

availability of variables across countries and waves, so to have insights into country compliance with the GGP standard questionnaires. As of September 2017, there are 19 Wave 1 data files (i.e., Australia, Austria, Belgium, Bulgaria, Czech Republic, Estonia, France, Georgia, Germany, Germany Turkish- Subsample, Hungary, Italy, Lithuania, Netherlands, Norway, Poland, Romania, Russian Federation and Sweden), and 12 Wave 2 data files (i.e., Australia, Austria, Bulgaria, Czech Republic, France, Georgia, Germany, Hungary, Italy, Lithuania, the Netherlands and Russia). From the left side frame of the Online Codebook and Analysis, it is possible to open a data file and start browsing its metadata and data, based on the DDI fields chosen by the GGP coordination team. The metadata content will appear on the right side frame (Figure 1). Metadata are organised along two main fields: 1) ‘Metadata’, which contains information on the survey in general, 2) ‘Variable Description’, which contains detailed description of each variable. The field Metadata is further organised in three subfields, offering information that can be either the same for all GGP surveys or differ based on country-specific characteristics: a) ‘Document Description’, containing information about the DDI-Nesstar document (the document which is being created, such as its version date and explanation of its content), that is the same for all the GGP surveys; b) ‘Study Description’, including both general information about GGP surveys that it the same for all of them (such as bibliographic citation, study scope, and data access) and country-specific fieldwork information with metadata

8 http://www.ggp-i.org/data/browse-the-data/

7

• n methodology and processing, as well as other study description

materials; c) ‘Data Files Description’, comprising both metadata that describe the content

• f the file containing the variables and that are the same for all GGP

datasets (e.g., information on the harmonization process, description of how missing data codes are assigned), and country-specific descriptions of dataset versions. Figure 1: Main DDI fields documenting W1 consolidated dataset and available by clicking on it (left side frame), and the dataset abstract (right side frame) In the field Variable Description, for each dataset there is a list of variable organised according to the modules of the GGP survey questionnaires. For each variable, there are detailed metadata necessary to understand how to use the data and their cross-national

• comparability. This information is the same for all GGP surveys.

Below it follows a description of the main content of each of the three types of data files available in the Online Codebook and Analysis. Examples of some analytical functionalities of Nesstar are given. 4.2.1. The pooled datasets The pooled data files contain in each of the DDI fields information that is the same for all GGP surveys. In the subfield Study Description of the Metadata field, there is for instance: information about how the study should be cited, abstract and keywords about the content of the GGP data, and data access procedure. The subfield Data Files Description contains for example: explanations of the harmonization process and of the variable names. The field Variable Description includes information on the GGP standard questionnaires, and the possibilities to visualise country deviations and variable distributions

8

for several countries at the same time (Figure 2). There is for instance: the text introducing the question, the text of the actual question asked, a descriptive text including country-deviations from the question and/or from the response categories of the standard questionnaire, the variable distribution, and the summary statistics. If applicable, there is also the universe (i.e., the subset of respondents to whom the question was asked), and, in case of derived variables, the explanations of the calculation method. Figure 2: Extract of the documentation of the variable a405a_a ‘Who makes decisions about household: Routine purchases for the household’ in the GGS Wave 1 pooled dataset 4.2.2. The country-specific data files The country-specific data files contain Metadata and Variable Descriptions that are the same for all GGP surveys. In addition to that, the sub-field Study Description of the Metadata field provides country-specific information on fieldwork methodology and processing, such as sampling procedures, mode of data collection, characteristics of the data collection situation (e.g., information on the interviewers, contact protocol, questionnaire testing and translating), actions to minimise losses between waves, weighting, response rates (Figure 3). There are also links to relevant country-specific study description materials. This includes for example: website of the national GGP project, methodology and/or technical reports and documents, national questionnaires (in mother tongue and, if available, also in English). For each country data file, the country-specific metadata are available for all the available waves. Users can therefore compare information concerning different waves. However, the availability of this country-specific information can vary across countries: due

9

to changes in the national GGP teams for some countries it was not possible to collect all the required metadata. Figure 3: Extract of metadata about methodology and processing for French GGP survey Wave 1 and Wave 2 There is country-specific information also in the subfield Data Files Description of the Metadata field. This includes information on the differences between subsequent versions of GGP survey country data files (e.g., derived variables that were added or revised). The Variable Description field provides for each variable the same information that is available for the corresponding variable in the pooled dataset of the corresponding wave. However, this does not apply to the variable distribution which only includes the country- specific cases. Additionally, the Variable Description field of the country-specific data files gives access to two main groups of variables: one for the country Wave 1 dataset, and the

• ther one for the country Wave 2 dataset. Through the functionalities offered by Nesstar tool,

it is possible to execute tabulations and simple statistical analyses by merging datasets of different waves (Figure 4). Among the other functionalities of the Online Codebook and Analysis, there is the possibility to apply weights, execute analyses on subsets of respondents, download the documentation (in .pdf format or .xml format), download analysis results (in .pdf format and .xls). The user guide, available from the welcome page and the green icon on the top of the Online Codebook and Analysis9, also provides useful illustrations and explanations of Nesstar

9 http://ggpsurvey.ined.fr/documents/guide/GGPNesstarUG.pdf

10

analytical functionalities. A complete explanation of Nesstar functionalities is available from the Nesstar webpage10. Figure 4: Bar chart showing the distribution of the variable about feeling depressed at W2 according to answers at W1 for French respondents11 4.2.3. The variable availability data file The pooled data files and the country-specific data files provide an overview of data and metadata either for all the countries and one wave, or for one country and more than one

• wave. To have an overview of several countries and several waves at the same times, there is

the ‘GGS Variable Availability’ data file. This is produced at INED and disseminated only through the Online Codebook and Analysis. This data file contains all the variables included in all released waves and country datasets (including country-specific variables). Variable names have the prefix ‘x’, where the letter ‘x’ stands for any available GGP survey wave. In order to increase transparency, for country-specific variables we indicate in the variable label the name of the country to which

10 http://www.nesstar.com/help/4.0/webview/index.html 11 In order to create tabulations and charts, one has to select the ‘Tabulation’ tab available on the top bar of the

Online Codebook and Analysis. Regressions and correlations can be executed from the ‘Analysis’ tab. W1 W2

11

the variable refers to. Variables are grouped by subject. The grouping follows the sections and subsections of the GGP Wave 1 codebook. Each variable has observations equal to the GGP codes of each country dataset where it is present. For example, if a variable is available in the French Wave 1 dataset, that variable will have an observation equal to ‘15.1 - France W1’, where ‘15’ is the GGP code for France and ‘1’ stands for Wave 1. Similarly, if a variable is available in the French Wave 2 Dataset, that variable will have an observation equal to ‘15.2 France W2’ (Figure 6). Figure 6: Distribution of the variable ‘x802: Is the respondent on maternity, parental leave or childcare leave’ in the GGS Variable Availability data file The information on the compliance of countries to the GGP standard questionnaire that is provided by this data file is particularly interesting for explorative purposes. For example, it is possible to have quick information about the countries in which it is possible to use GGP data to study a given subject matter. More in general, it provides information on the geographical and temporal/longitudinal coverage of each GGP survey variable.

• 5. Conclusions

To properly document GGP surveys, a great amount of metadata is required. This includes information on the GGP surveys in general and on the harmonization process. This is the same for all GGP national data files and is prepared by the central coordination team.

12

Metadata also include country-specific fieldwork information that is prepared in collaboration with the national GGP teams. Especially the preparation of the country-specific metadata may be time-demanding and may require a lot of resources and efforts both from the central coordination team and from the GGP national teams. This is especially the case for GGP surveys that are incorporated into existing national surveys. In these cases, the documentation

• f deviations from the GGP standard questionnaires is crucial and may require the provision
• f very detailed and accurate information on a number of variables, such as the explanations
• f country-specific questions and response codes.

In order to provide this great amount of metadata in a convenient way, the Online Codebook and Analysis tool takes advantage of the functionalities of the Nesstar software. Data documentation is organized in three different data files, based on the different profiles of GGP users. This allows taking into account all the necessary information for cross-national and longitudinal social science research based on GGP data. First, there is the possibility to have cross-national comparisons of data and metadata for each available GGP wave. Second, for each country, it is possible to obtain overview of data and metadata across different GGP

• waves. Third, it is possible to have insights on the availability of each variable in all the

available GGP country data files and waves. The future 2019 round of GGP is planned to be based on a greater monitoring of the data collection by the central coordination team. This may allow for a reduction of country deviations from the standard questions and, thus, also of the efforts to prepare country- specific metadata. In addition to that, the preparation of country-specific metadata would start at the beginning of the fieldwork and follow the data collection process. This involves data check, data corrections, and variable documentation while the fieldwork is still on-going. Greater automation would replace ‘by hand’ operations. The aim is to optimize the collection

• f country-specific metadata, as well as the management of the entire data documentation

procedure.

• 6. References

Caporali, A., Morisset, A., and Legleye S. (2015). Providing access to quantitative surveys for social research: The example of INED. Population-E 70(3): 537-566. DOI: 10.3917/pope.1503.0537. Caporali, A., Klüsener, S., Neyer, G., Krapf, S., Grigorieva, O., and Kostova, D. (2016). The contextual database of the Generations and Gender Programme: Concept, content and research examples. Demographic Research 35(9): 229-252. DOI: 10.4054/DemRes.2016.35.9. Kramer, S., Banks, R., Chang, V., Sieber, I., Vardigan, M., and Zenk-Möltgen, W. (2011). Presenting longitudinal studies to end users effectively using DDI metadata. DDI Working Paper Series – Longitudinal Best Practice, 4. DOI: dx.doi.org/10.3886/DDILongitudinal04. King, G. (1995). Replication, Replication. PS: Political Science & Politics 28(3): 444-452. Emery, T. and Caporali, A. (in prep.). The added value of cross-national studies: Compliance and usage in the Generations and Gender Programme. Fokkema, T., Kveder, A., Hiekel, N., Emery, T., and Liefbroer, A. C. (2016). Generations and Gender Programme Wave 1 data collection: An overview and assessment of sampling and fieldwork methods, weighting procedures, and cross-sectional representativeness. Demographic Research 34(18): 499-524. DOI: 10.4054/DemRes.2016.34.18.

13

Gauthier, A.H. and Emery, T. (2016), The Generations and Gender Programme: Past, present and future. Demos: bulletin on population and society 32 (7), special issue: 7. http://www.nidi.nl/shared/content/demos/2016/demos-32-07-gauthier.pdf. Marker, H. J. (2013). Strengthening cooperation between European social science data archives: The evolving role of CESSDA. In: Kleiner, B., Renschler, I., Wernli, B., Farago, P., Joye, D. (eds.), Understanding Research Infrastructures in the Social

• Sciences. Zurich, Seismo Press: 39-46.

Mohler, P.P., Hansen, S.E., Pennell, B.E., Thomas, W., Wackerow, J., and Hubbard, F. (2010). A survey process quality perspective on documentation. In: Harkness, J.A., Braun, M., Edwards, B., Johnson, T.P., Lyberg, L.E., Mohler, P.P., Pennell, B.-E., and Smith, T.W. (eds.). Survey Methods in Multinational, Multiregional, and Multicultural Contexts. Hoboken, N.J.: Wiley & Sons: 299–314. DOI: 10.1002/9780470609927.ch15. UNECE – United Nations Economic Commission for Europe. (2005). Generations and Gender Programme - Survey Instruments. New York/ Geneva: United Nations. http://www.unece.org/pau/pub/ggp_survey_instruments.html. UNECE - United Nations Economic Commission for Europe. (2007). Generations and Gender Programme - Concepts and Guidelines. New York/ Geneva: United Nations. https://www.unece.org/pau/pub/ggp_concepts_guidelines.html. Vardigan, M., Heus, P., and Thomas, W. (2008). Data Documentation Initiative: Toward a standard for the social sciences. The International Journal of Digital Curation 3 (1): 107-113. DOI: 10.2218/ijdc.v3i1.45. Vikat, A., Spéder, Z., Beets, G., Billari, F., Bühler, C., Desesquelles, A., Fokkema, T., Hoem, J.M., MacDonald, A., Neyer, G., Pailhé, A., Pinnelli, A., and Solaz, A. (2007). Generations and Gender Survey (GGS): Towards a better understanding of relationships and processes in the life course. Demographic Research 17(14): 389–

• 440. DOI: 10.4054/DemRes.2007.17.14.