A entrevista de survey como interação social : fatores explicativos da susceptibilidade aos efeitos nas respostas"

Abstract

This study draws on the contributions to survey methodology of the study of the cognitive aspects involved in the production of answers to a standardized questionnaire, the so called response effects. The notion of responses effects posits that part of what we measure with a survey questionnaire can be the effect of the way we design the questions, leading to different response results. This field of study has generated a significant production of data based on split-ballots experiments, which use probability samples divided into two subsamples, ensuring identical conditions, and allowing for inferential analysis. The response effects are produced by variations in question wording, format, and question order among others factors. This study seeks to analyze the social conditions related to how questionnaires respondents become susceptible to response effects. The point of departure is the understanding - from a sociological perspective - of the survey interview as a social interaction. The data used in this investigation came from the GLOBAL ENVIRONMENTAL SURVEY (GOES, 2001) and the BELO HORIZONTE AREA SURVEY (PRMBH, 2002). The types of experiments analyzed were: 1) responses options – offering or not the “don‟t know” alternative (as a semi-filter) - and 2) question order or the question context. The relevant social traits concern the respondents‟ specific attitudes in relation to the object of study, and their biographical/background situation or position in the social structure such as social capital, experience/social behavior and socialization, which turned out to be more significant when compared to other dimensions such as the environmental context and economic capital. The identification of these social traits and their relations to types of response effects are relevant for questionnaire design and data interpretation as well. Knowing the research target audience and the possible interpretative schemes used by the respondents, and drawing on the relationship between the questions‟ contents and situations (formats, writing, context), this study raises issues of validity and reliability of measures used in survey questionnaires. The knowledge of theses conditions should give rise to a continuous epistemological vigilance, and the search for the various factors producing response effects.