Panel Abstract

Error sources and quality in interviewer-administered surveys

While alternate survey modes such as online surveys are gaining in popularity, face-to-face interviewer-administered surveys are still a common mode in survey research in Asia today due to limited penetration of other modes and/or the lack of high quality sampling frames in many countries. Interviewers can affect the quality of the data (e.g., introduce interviewer effects), which can occur because of the interactional dynamics between the respondent and the interviewer caused by interviewer characteristics, lack of standardization in adherence to study protocols, or falsification. Question characteristics can also lead to measurement error in the context of an interviewer-administered survey. Cultural factors such as privacy concerns, social desirability concerns, and wariness of strangers can make it difficult to implement more rigorous quality assurance and quality control procedures and can impact the availability of important information about data quality. Drawing on the Total Survey Error (TSE) framework and focusing on the measurement aspect, we highlight those error sources particularly vulnerable in interviewer-administered surveys. We then explore further and more specifically interviewer error using examples from several case studies from non-Western settings. We conclude with discussion of current and future approaches to minimizing this and other important sources of error in comparative surveys.