Informed Consent: Consequences for Response Rate and Response Quality in Social Surveys

Abstract

The study described in this paper was designed to measure the effects of variations in (1) the amount of information provided to respondents about the content of a survey ahead of time, (2) the assurance of confidentiality given to respondents, and (3) the request for and timing of a signature to document consent on three aspects of social surveys: (a) overall response rate, (b) response rates to individual questions, and (c) response quality. That is, the study was designed to measure the impact of informed consent procedures on response rates and response quality in social survey research. For this purpose, a fully crossed 2 x 3 x 3 factorial design was imposed on a national probability sample of 2,084 potential respondents. The questionnaire consisted of a large number of items in such presumably sensitive areas as drinking, marijuana use, sexual behavior, and mental health, in addition to more conventional questions about leisure activities. Majorfindings can be summarized as follows: 1. The overall response rate to the survey was 67%; of the three variables investigated, only the request for a signature had a significant effect on the probability of responding. Seventy-one percent of those not askedfor a signature were interviewed, compared with 64% and 65% of those asked to sign before and those asked to sign afterwards. 2. Only the assurance of confidentiality had a significant effect on item nonresponse. Despite the sensitive nature of the interview, nonresponse to individual questions was very low. On those questions to which the nonresponse rate totaled more than 3%-all of them questions about behavior rather than attitudes-respondents given an assurance of absolute confidentiality had a lower nonresponse rate than those in two other experimental groups, in some cases by a statistically significant margin. 3. None of the three independent variables had either consistent or large effects on the quality of response. However, there are suggestions in the data that asking for a signature before the interview has a sensitization effect, so that better data are obtained if the respondent is asked to sign a consent form afterwards.