Some 15 years ago, investigators Clarke and Kline proposed message discrimination as an alternative open-ended measure of media exposure. This study examined its use in a multicity health campaign project. It addressed three issues: message recall patterns as a function of instrument length and sociodemographic variables; types of messages recalled; and interviewer effects. Analysis suggested that three open-ended “question cycles” were sufficient to capture the majority of respondents' message recall and that respondents did not recall different types of messages earlier or later in the instrument's administration. The major argument that message discrimination as a technique is less biased against those with less formal education received qualified support. The study also lent qualified support to the conclusion that interviewers were a source of excess but controllable variability in this open-ended survey method.
Copyright 2016 Elsevier B.V., All rights reserved.