Dissociating explicit and implicit effects of cross-media advertising

Abstract

Knowledge of cross-media advertising effects is mainly based on explicit psychological measures, such as self-reports. To fully understand the mechanisms responsible for the success of cross-media advertising, it is important to also use implicit measures. We used both types of measures to assess whether exposure to different media combinations affects the cognitive and evaluative impact of advertising. Results show that participants performed better on all explicit and implicit measures of memory and brand preference after exposure (versus no exposure) to target ads, which validated the use of these measures. Comparison of cross-media versus single medium exposure showed differences on the explicit level, but not on the implicit level. This suggests (1) that cross-media advantages may be driven by explicit rather than implicit memory mechanisms, and (2) that implicit advertising effects may require more drastic manipulations than context changes, such as varying the combination of media used for exposure to advertising.

Keywords:

• cross-media
• advertising effectiveness
• implicit measures
• implicit memory

Acknowledgements

We would like to thank the Amsterdam School of Communication Research for financing this project from the Research Priority Area Communication. Also, we thank three anonymous reviewers for their valuable comments on an earlier version of the manuscript. We further thank Remziye Temirçi for running a pilot study and Wouter Weeda for thinking along with the statistical analyses. Finally, we thank the members of the Psychology of Persuasion Lab at the University of Amsterdam for their suggestions based on earlier drafts of the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. On an affective level, processing information across different medium types is thought to affect peoples’ attitude towards the advertised brand by appealing to different cosmetic and aesthetic characteristics (see, e.g., the review provided by Stammerjohan et al. 2005).

2. The remaining 8.47% reported some background noise and no distractions; 3.10% reported background noise and some distractions; 1.24% reported noise and/or many distractions. None of these participants would be removed from the sample as long as they satisfied the control conditions described in the result section.

3. This resulted in six counterbalance versions of the ad presentation phase: two in the control condition (two medium orders) and four in the experimental condition (2 target brands × 2 medium orders). We randomly presented the different versions of this phase to different participants. As a result, half of the participants in the experimental condition were exposed to the butter brand across media (and milk within a single medium condition) whereas the other half was exposed to the milk brand across media (and butter within a single medium condition). Of these groups, half of the people were presented with the ads in one medium order (starting with TV: TV–TV for single and TV–web for cross), whereas the other half was presented with the ads in another order (starting with web: web–web for single and web–TV for cross). With this design, any potential stimulus effects or order effects remained present for each individual subject, but were fully counterbalanced across the subjects.

4. A pilot study showed that, for the target milk brand, people used Pauls™ (brand name) and Parmalat™ (company name) interchangeably. We therefore included Parmalat in the task and scored Pauls and Parmalat as being one and the same item.

5. Inclusion of all these participants did not alter the pattern and significance of the results reported below.

6. The variances in all the reported t-tests are not equal across exposure conditions, as was shown by Levene's tests for equality of variances (all Fs > 4.86, all ps < 0.05). The reported degrees of freedom and statistics are corrected for this inequality of variances.

7. One possible alternative explanation for these findings is that we had a higher statistical power to find significant effects in the explicit measures (all participants) compared to the implicit measures (one fourth of the participants per measure; see the design in Figure 1). To exclude this alternative hypothesis, we drew a random sample of one fourth of the participants to analyse their data on the explicit tests (N = 112, equally divided over counterbalancing and implicit measurement lists). These data showed similar result patterns, with a better recall Χ² (1) = 8.40, p < 0.01, ϕ = 0.26, better recognition Χ² (1) = 4.33, p = 0.053, ϕ = 0.15, and higher attitude ratings (t (111) = 3.20, p < 0.005, η² = 0.09) for cross-media versus single-medium brands. This excludes the alternative hypothesis that the higher power for the explicit measures caused the significant p-values.