Abstract
When developing household surveys, much emphasis is understandably placed on developing survey instruments that can elicit accurate and comparable responses. In order to ensure that carefully crafted questions are not undermined by 'interviewer effects', standardised interviewing tends to be utilised in preference to conversational techniques. However, by drawing on a behaviour coding analysis of survey paradata arising from the 2012 UK Poverty and Social Exclusion Survey we show that in practice standardised survey interviewing often involves extensive unscripted conversation between the interviewer and the respondent. Whilst these interactions can enhance response accuracy, cooperation and ethicality, unscripted conversations can also be problematic in terms of survey reliability and the ethical conduct of survey interviews, as well as raising more basic epistemological questions concerning the degree of standardisation typically assumed within survey research. We conclude that better training in conversational techniques is necessary, even when applying standardised interviewing methodologies. We also draw out some theoretical implications regarding the usefulness of the qualitative-quantitative dichotomy.