Norbert Schwarz - Gricean Maxims and Survey Response

Gricean Maxims and Survey Response

Norbert Schwarz is also well known for his research on cognitive processes underlying survey response. This work generally treats the survey interview context as a conversation between the researcher and the respondent. According to this logic, surveys are governed by the cooperative principle advanced by Paul Grice, the late philosopher of language. Put simply, the cooperative principle states that people try to communicate clearly and truthfully, in as much detail as required (but not more so), giving only relevant information. In Schwarz's view, the respondent not only follows the Gricean maxims (Quality, Quantity, Relation, and Manner) when responding to surveys but also assumes that the questions the interviewer asks are guided by the same principles.

Schwarz's research implicates the operation of these maxims during various stages of the survey question and answering process, and highlights how features of the research instrument can significantly impact the answers obtained. For example, when asked about how successful their lives have been, 34% of respondents reported that their lives have been highly successful when the response scale was labeled -5 to 5, whereas only 13% reported high success when the scale was labeled 0 to 10. Presumably this is because the survey respondent assumes that negative integers refer to the presence of negative features, while smaller positive integers refer to the absence of positive features.

Similarly, Schwarz has found that when a question about marital satisfaction precedes a question about general life satisfaction, responses for the two questions are highly correlated because the first question renders information about one's marriage highly accessible, but other studies have found the same correlation when the marital satisfaction question is asked after the marital satisfaction question, presumably because marital satisfaction is chronically accessible. However, this correlation vanishes when the two questions are framed as subordinate parts of a larger question, presumably because the respondent infers that the interviewer does not want redundant information and thus marital satisfaction should be specifically subtracted from general life satisfaction. Similar reasoning has been applied to understanding the relation between people's ratings of social groups' central tendency and variability.