Back to Overview of Sessions

Session Details

Cross-national survey projects exhibit wide variation in data quality, both within and across projects. Some departures from quality standards that the specialized literature has established for data collection, cleaning, and documentation, such as the presence of non-unique records (or duplicates), are unequivocal instances of ‘bad data,’ while others, such as certain types of processing errors are more ambiguous. Between the clearly bad and clearly good survey data there may be a range of ‘decent’ quality surveys, with potentially interesting and important information collected form under-surveyed countries and less well covered time periods. However, to date there is little research that systematically assesses the quality of extant international survey data, or that looks at whether and how the ‘messiness’ in existing surveys can be minimized ex-post, and with what consequences for empirical analyses.

For this session we invite theoretical and empirical papers on evaluating the quality of extant surveys, after the stages of data gathering and documentation are completed. This could include creating new metadata for survey quality. Contributions could focus on different dimensions of survey quality, including (a) the quality of surveys as reflected in the general survey documentation, (b) the degree of consistency between the description of the data with the data records in the computer files, and (c) the quality of the data records in the datasets. We aim to engage with both survey methodologists and with researchers who use survey data primarily for substantive analyses, to discuss data quality and ways of accounting for its variation in empirical analysis within and beyond the Total Survey Error, Total Survey Quality and Total Quality Management frameworks.