Raising the BAR: Bias Adjustment of Recognition Tests in Advertising

Abstract

Advertising recognition tests use advertisements as visual retrieval cues; they require consumers to report which advertisements they remember having seen earlier and whether they noticed the advertised brand and read most of the text at the time. Using a heterogeneous randomly stopped sum model, the authors establish the relationship between consumers' actual attention to print advertisements, as measured through eye tracking, and subsequent ad recognition measures. They find that ad recognition measures are systematically biased because consumers infer prior attention from the ad layout and their familiarity with the brands in the advertisements. Such biases undermine the validity of recognition tests for advertising practice and theory development. The authors quantify the positive and negative diagnostic value of ad recognition for prior attention and demonstrate how these diagnostic values can be used to develop bias-adjusted recognition (BAR) scores that more accurately reflect prior attention. Finally, the authors show that differences in the scores from ad recognition tests based on in-home versus lab exposure attenuate when the bias-adjustment procedure is applied.