Not All Nutrition Claims Are Perceived Equal: Anchoring Effects and Moderating Mechanisms in Food Advertising

Abstract

Despite the increased use of health claims in food advertising, few studies have investigated how specific nutrition claims have differential effects depending on how they are presented. In this context, the current study tests the anchoring hypothesis. Anchoring refers to a common human tendency to evaluate information differently depending on the presence or absence of a numerical “anchor” or reference point. Two (pilot and main) experimental studies explore anchoring effects on audience response to food advertising both directly and moderated by cognitive, motivational, and message factors. The pilot study finds that food product ads employing nutrition claims with an anchor rather than without an anchor generate two results: First, participants perceive the product to have lower fat/lower calorie contents (anchoring hypothesis); second, they prefer the messages with an anchor over those without an anchor. The main study reports that when anchoring is successfully evoked, it produces favorable attitudes toward the ad, favorable attitudes toward the brand, and purchase intention—but only when moderated by health orientation, claim believability, and nutrition knowledge. Practical implications are provided with respect to regulatory guidelines and effective communication strategies for promoting low-fat and low-calorie products in food advertising.

Notes

¹Our study contains no manipulation check because it focuses on the relationship of a psychological state to a persuasive outcome. Specifically, we were interested in the extent to which anchoring, generated by messages with anchors, affects audience response to food advertising. Relevant to this context is an argument of O'Keefe (2003): “Variation in the psychological state of interest is created by exposure to different message conditions, but the message variations are not themselves of interest; they are simply a methodological device for creating variance in the psychological state. In such a design there is no need for a message manipulation check” (p. 255). This argument is also consistent with that of Sigall and Mills (1998), who maintain that manipulation checks may be unnecessary in social-psychological research, particularly when the independent variables of interest are psychological states.

²To overcome the limitation of the pilot study and to replicate its findings, this follow-up study was conducted among a larger college student sample (n = 369) at a large Midwestern university. The three types of messages (no anchor, within-product-category anchor, across-product-category anchor) were counterbalanced. Participants were also randomly assigned to one of the three food product types—sandwich, cracker, and flavored water. To replicate the pilot study's findings, the first two products were the same as those used in the first study. The third product was added to see whether our findings could be replicated for another type of food product. The anchoring message was also different for the third product as follows: (1) copy without anchor (control)—“Thirsty? Grab Our Flavored Water. Only 2 Grams of Sugar in Each Serving!”; (2) copy with within-product-category anchor—“Thirsty? Grab Our Flavored Water. Only 2 Grams of Sugar in Each Serving, 45% Less Than Other Brands!”; and (3) copy with across-product-category anchor—“Thirsty? Grab Our Flavored Water. Only 2 Grams of Sugar in Each Serving, 45% Less Than Soda!” Each of the cells ranged from 37 to 46, with fairly even distribution (females = 69.6%). Repeated-measures ANOVA via the GLM procedure, same as in the first study, was performed for each of the three product types. Results closely resembled those in the first study. Participants tended to perceive a low fat/low calorie/low sugar claim to be lower for the message with an anchor as opposed to that without an anchor. They also tended more favorably to evaluate the advertising copy of a low fat/low calorie/low sugar claim when they were presented with an anchor as opposed to without an anchor. The difference seems less salient between messages with a within- versus an across-product-category anchor, although anchoring effects seem more salient in ad messages with a within-product-category anchor. Detailed descriptions of design, procedure, and findings of this post hoc study are available upon request.

³One reviewer raised the question of whether there were anchoring condition × product type interactions. Our series of ANOVA and ANCOVA tests (for three dependent variables [DVs]) using product type as a fixed factor and including the interaction term showed no significant interaction effects. We also tested the interaction effect using repeated-measures ANOVA via the GLM procedure (for two DVs: perceived nutrition value and copy preference) in our post hoc experiment for the pilot study and found no significant interaction effects. Thus, the results provide no evidence that product type may have influenced how the anchoring conditions impacted our findings.