Lateralized goal framing: How health messages are influenced by valence and contextual/analytic processing

Abstract

The effectiveness of health messages has been shown to vary due to the positive or negative framing of information, often known as goal framing. In two experiments we altered, the strength of the goal framing manipulation by selectively activating the processing style of the left or right hemisphere (RH). In Experiment 1, we found support for the contextual/analytic perspective; a significant goal framing effect was observed when the contextual processing style of the RH – but not the analytic processing style of the left hemisphere (LH) – was initially activated. In Experiment 2, support for the valence hypothesis was found when a message that had a higher level of personal involvement was used than that in Experiment 1. When the LH was initially activated, there was an advantage for the gain- vs. loss-framed message; however, an opposite pattern – an advantage for the loss-framed message – was obtained when the RH was activated. These are the first framing results that support the valence hypothesis. We discuss the theoretical and applied implications of these experiments.

Keywords:

• framing
• decision making
• health
• lateralization

Acknowledgments

Special thanks goes to Cathy Seta for her many helpful comments and suggestions on numerous drafts of this paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Only right-handed females were included in Experiment 1 for several reasons: females comprised the bulk of the subject pool to which we had access; this increases the homogeneity of the sample which should allow hemispheric differences to be observed more easily; and we suspected that future mothers may be more engaged in a study about MMR vaccination than males given that mothers are still more likely to be the primary caregiver in a family. Males were included in Experiment 2 because the HPV vaccine had recently become available to males and issues related to HPV vaccination were thought to be highly relevant to both males and females in our college-aged sample.

2. Following previous work (e.g. Gallagher & Dagenbach, 2007; Seta et al., 2010), the relatively high frequencies were amplified 12Dbs more than the low frequencies so that there was no difference in the perceived volume of the messages between conditions.

3. Although the Frame X Hemisphere interaction did not reach conventional levels of significance without controlling for message comprehensibility (i.e. F(1, 72) = 3.38, p = .07), a pointed contrast that examined the difference in the size of the framing effect between hemispheres was consistent with our predictions, F(1, 72) = 6.78, p < .02.

4. Initial analyses were consistent with our predictions regarding personal involvement in that no differences were observed between conditions for individuals who had already received the HPV vaccine (F’s < 1.4). Presumably, because these individuals have already received the shot, their involvement with our message was low and this reduced framing effects.

5. A similar analysis that was conducted with the first item omitted (the one item that asked about getting more information only) revealed no significant differences in the results.

6. Two non-significant interaction effects involving the sex factor were observed; a relationship status X hemisphere X sex interaction, and a four-way interaction that also included the frame factor, F(1, 102) = 2.25, p > .13 and F(1, 102) = 2.30, p > .13, respectively. All other F’s < 1. We feel that these results have little impact on the findings of Experiment 2, however, as most of the difference was due to males reporting highly discrepant likelihood ratings when currently engaged vs. not engaged in a sexual relationship; a pattern of means descriptively consistent with the valence hypothesis was found for males who were currently in a relationship but a reverse pattern was observed in the RH condition for those not engaged in a relationship. No significance tests were conducted on these difference scores given the low number of observations in each cell.