Population studies strongly rely on survey data. In order to face with the increasing complexity and rigor of current research studies, the structure of databases has become in the last decades more and more complex: longitudinal data, network data, spatial data, etc. The ever growing size of structured data complicate the task of both documenting, manipulating and sharing data. The Rsocialdata project puts forward two tools: an R package and a web application. First, the Rsocialdata package provides researchers in social sciences with high-level tools for storing, documenting, exploring and recoding survey data in R. It also provides front-ends to classical statistical methods able to print analysis results in a "ready-to-publish" formatting and save them in a PDF or DOCX file. Specific tools for efficiently handling panel data and network data are provided, as for example the possibility to manipulate trajectories as a whole and to export them as sequences ready to be analysed with the TraMineR package. Second, the web application allows (1) to explore survey databases that are available in the Rsocialdata file format and (2) to easily retrieve them in a R environment. By providing a full database search engine, the platform allows researchers to easily find databases which are relevant for their analyses.
As an important part of survey databases are lost because it would have required to much effort to render them accessible, our platform encourages the dissemination of survey databases by facilitating the process of making them available to the scientific community. Databases are securely stored (access restricted to authorized users, data transfers encrypted in SSL) and their authors can manage data access via a license agreement process. The Rsocialdata initiative is free and open to everybody who would like to participate, either by helping to carefully document and prepare existing databases or by sharing their own data with the scientific community on the Rsocialdata platform.

The Rsocialdata.network package

Specific tools for dealing with network survey data are supplied in the Rsocialdata.network package. The extension provides two new classes. First, the NetworkVariable class is designed to store network survey data. Thanks to this object you can store each network variable (emotional support, financial support, conflict, etc.) in one single variable instead of size(network)$^2$ variables. You also can apply most common network measures (centrality, density, betweenness, etc.) directly on your network variables without any additional recoding or data preparation. Naturally, all Rsocialdata features for survey data (distinction between valid cases and missing values, handling several different missing value types) are available. Although the class is designed to store egocentric networks, it is also possible to store community networks. NetworkVariable objects can be created with the netvar function. Second, the toolbox provides the NetworkMetadata class. This new class is designed to store covariates describing individuals cited in the networks. These covariates generally include the family tie with the respondent as well as demographic variables: for instance gender, age, marital status, or education level. Then, you have the possibility to link newtorks variables with corresponding network covariates. Especially, as a main feature of the package, the net.extract function offers facilities to easily extract networks based on both network and demographic criteria. For example you can select your respondent based on the following network properties: (1) the father is cited and (2) has a low education, and (3) the density of the network is lower than 0.2" in one single step.

Authors: Emmanuel Rousseaux and Gilbert Ritschard.
Availability:
  • Rsocialdata Release candidate available from the Rsocialdata R-Forge repository.
       Installation: install.packages("Rsocialdata", repos="http://R-Forge.R-project.org", type = "source")
  • Rsocialdata.network alpha version available from the Rsocialdata R-Forge repository.
       Installation: install.packages("Rsocialdata.network", repos="http://R-Forge.R-project.org", type = "source")
Website: www.rsocialdata.org
Documentation:
  • Series of tutorials from the Rsocialdata website.
  • Seminar's booklet (in French) of the Statistical analysis of categorical data course given at the University of Geneva.
Development: The development version is hosted on the R-Forge platform.



comments powered by Disqus