Social science data harmonization and replication: challenges and solutions for the 21st century
|Convenor||Dr Kristi Winters (GESIS - Leibniz Institute for the Social Sciences )|
This paper will focus on the importance of "output" or "retrospective" data harmonization across both the social and physical sciences with an emphasis on the development and implementation of comprehensive guidelines to create harmonized datasets. Descriptions of these guidelines will be presented and the effect that they have had in studying important health issues. The author will discuss the importance and applicability of a set of general standards for harmonizing data.
The presentation discusses the challenges in preparing data for harmonization of 22 international surveys containing questions on protest behavior covering 142 countries and territories in time span from 1966 to 2013. The project database contains 1720 national samples (project*wave*country) for which interviews were conducted. The endeavor is a part of the international project Democratic Values and Protest Behavior: Data Harmonization, Measurement Comparability, and Multi-Level Modeling, coordinated by the Institute of Philosophy and Sociology, Polish Academy of Sciences, and the Mershon Center for International Security Studies, The Ohio State University.
The paper suggests new fractal approach to methodology of cross-disciplinary research synthesis in social science and liberal arts on the basis of fractal evolutionary theory of natural ordering. In comparison with theories of selection and nomogenesis it has universal explanatory and prognostic potential. It includes not only synthesis of methods but also synthesis of principles. Key issues for more realistic task of forming ‘United Information Space' (UIS) standards are discussed rather than UIS itself to harmonize common research space. Key implementation risk managerial applications for decision makers are discussed in regards to support of such projects.
Statistical analyses often require extensive data preparation and variable harmonization, yet there are no agreed documentation standards in the social sciences to facilitate a transparent replication process. Further, researchers who make contributions to their discipline with original data harmonization work are not credited through citations. CharmStats Pro (CS Pro) addresses these problems. CS Pro is free and open-source software for documenting, organizing and publishing data harmonization, aiming to standardize variable harmonization documentation in the social sciences. This article reviews some of the scientific demands driving the need for accurate replication in social research from the perspective of international relations.
The harmonization of longitudinal data presents interesting challenges to the scientific norms of replication, reliability and transparency. Many researchers turn to spreadsheet programs in order to organize and document conceptually similar variables over time. The paper reviews the challenges, strategies and solutions applied to longitudinal Australian Election Study variables using the harmonization software CharmStats. Individual variable harmonizations are reviewed and the organization of the codebook is detailed.