Approaching quality in survey research: towards a comprehensive perspective

The article has two goals: (1) bring attention to the problem of inappropriate treatment of survey data quality issues in the social sciences, and (2) introduce the basic principles of contemporary approaches to survey quality. If quality evaluation focuses solely on sampling error, most aspects of data quality are ignored and surveys are assumed to have 'ideal' statistical characteristics that are rarely attainable in the pragmatic world of survey fieldwork. A complex overview of the entire process of data collection provides a more solid foundation for evaluating data quality. Under this approach, quality is ensured by controlling the whole survey process. Accuracy, which is commonly elaborated using the concept of survey error, ceases to be the only dimension of quality. Nevertheless, this data quality component is crucial for data analysis and statistical testing. A comprehensive approach to survey data quality requires us to take account of complex sample designs when evaluating sampling error and to identify and distinguish between different dimensions of nonsampling error. Analysts who are not directly involved in data collection have limited ability to obtain information necessary for data quality evaluation. There are two types of quality standards: administratively imposed standards (ISO20252:2006) and the technical and ethical criteria of professional associations (e.g. ICC/ESOMAR, AAPOR/WAPOR, SIMAR). These help break this information barrier between data producers and data users.