A Call for a Broader Range of Dependent Variables in Advertising Research

Using experiments to evaluate the effectiveness of various advertising strategies has a long tradition in academic research in marketing (see e.g., Strong, 1912, for an early example). These experiments typically evaluate the effects of using a certain strategy (e.g., humor or celebrity endorsement) or executional style or elements (e.g., color or images) on a small set of dependent variables. The dependent variables in most cases are brand attitude and brand purchase intention, and to a limited extent attitude toward the ad. The dominance of these variables is demonstrated by the fact that they are the most common dependent variables in meta-analyses of advertising research (e.g., Brown and Stayman, 1992; Eisend, 2006, 2009). This practice is problematic for at least two reasons. Firstly, it suggests a belief that advertising works by having a positive effect on either brand attitude or brand purchase intention, or both, but not by influencing other variables (Figure 1).