A Demonstration of the Influence of Advertising on Brand Choice

J. Edward Russo

Anne-Sophie Chaxel

Cornell University

J. Edward Russo (corresponding author), Johnson Graduate School of Management, Cornell University, Ithaca NY 14853-6201, 607-255-5440 (office), 607-255-5993 (fax), .

Anne-Sophie Chaxel, Johnson Graduate School of Management, Cornell University, Ithaca NY 14853-6201, 607-255-8921 (office), 607-255-5993 (fax), .

Abstract

We propose one way that advertising can influence brand choice. A superior TV commercial creates an initial preference for its brand over a competitor. During a choice this early preference is reinforced by the biased evaluation of new information to support the leading brand. In an experiment, the superior of two award-winning commercials installed its brand as the initial preference for most participants. The subsequent choice process exhibited greater information bias, more final choice of that brand, and greater confidence in that choice. This study shows not only that advertising can influence choice but how it can do so.

Key words: Advertising; biased evaluation; choice process; information distortion; TV commercials


Advertising can influence brand choice (e.g., Mehta, Chen, & Narasimhan, 2008). This is true even for TV commercials that rely less on communicating product information and more on conveying a positive feeling about the brand. Though there is general agreement that such commercials can influence brand choice, there is much less agreement on how they do it. That lack of agreement may reflect the existence of several different ways that commercials succeed, each with its own process through which the viewing of the commercial influences a subsequent brand choice. The purpose of the present work is to demonstrate one such process.

Our proposed process requires two steps. First, an effective commercial must install its brand as the initial leader in a choice against a competing brand. This should be straightforward so long as the commercial for the targeted brand is superior to a commercial for the competitor. Not all viewers may see the selected commercial as superior, but it should install its brand as the initial leader for a majority of them.

The second step relies on a phenomenon that occurs during the subsequent product choice. As new information is received, its evaluation tends to be biased toward supporting whichever alternative is currently leading in overall preference (Russo, Meloy and Medvec, 1998). If the superior commercial makes its brand the initial leader, then the distortion of new information during the choice process should enable it to remain in the lead and emerge as the chosen brand for a majority of consumers. Information distortion (ID) is, by now, a well established “predecision” bias (see Brownstein, 2003, for a review). ID has been frequently observed (e.g., DeKay et al., 2009; Holyoak and Simon, 1999; Russo, Medvec, and Meloy, 1996; Wilks, 2002). It is also systematic, increasing linearly with the strength of preference for one option over the other (Russo et al., 1998). Finally, it seems that it cannot be eliminated -- not by offering a financial incentive (Meloy, Russo and Miller, 2006), not by holding decision makers accountable (Russo, Meloy and Wilks, 2000), and not even by official instructions to prospective jurors to wait until hearing all evidence before forming a tentative judgment (Carlson and Russo, 2001).

Pertinent to the present strategy is the success of a similar approach that did not involve advertising. Carlson, Meloy and Russo (2006) provided six attributes of information in a binary choice. Two attributes were written to favor one brand or the other, while each of the remaining four favored neither (i.e., were neutral). Carlson et al. then targeted one of the brands to be the eventual choice by manipulating the order of attributes. The first one always favored the target and the fourth attribute always favored the competing product, while the remaining positions were filled with the four neutral attributes. This order installed the targeted brand as the initial leader for 94% of the participants. ID helped to retain this leadership position until about 70% chose the targeted brand.

The present work differs from the above in how the targeted brand is installed as the initial leader. Instead of using a first attribute that favors the target brand, we propose to use the more effective of two commercials to place its brand in the leadership position. If the same leader-supporting ID obtains, then the brand in the superior commercial should be chosen by more than half of the participants.

Method

Overview

Imagine having to decide between two Caribbean resorts for your next vacation. You might see a commercial for each resort, possibly on their websites or on a travel-oriented TV channel. This would constitute Step 1 of our process. Then you might use the internet, books or friends to gather information on several attributes of these two resorts, such as their activities and amenities. This would be Step 2.

To track the commercials’ influence on each attribute to the eventual brand choice, we used a method known as the stepwise evolution of preference (SEP; Meloy and Russo, 2004).

This method requires participants to provide three responses after they have read each product attribute. The first is an evaluation of the attribute. This response is the basis of the measurement of ID (as explained below). Second, participants are asked which brand they prefer at this point in the process, knowing that more information is coming. This response identifies the current leader. Finally, participants are asked for their confidence in this current leader. This third response captures the strength of their preference for the identified leader.

If the superior commercial succeeds in installing its brand for a substantial majority of viewers and if ID is sufficient to preserve enough of that majority in the face of neutral information (which should dilute that early preference), then the targeted brand should be chosen by a majority of participants. If it is, then we have identified one mechanism through which advertising can influence choice.

Materials

Commercials. We sought commercials for vacation resorts that met two criteria. First, one had to be somewhat superior so that it could install its resort as the initial leader in the subsequent choice. Second, we wanted both to be judged as fully professional and effective so that the executional quality of the inferior commercial would in no way suggest low quality of its resort. Because vacation resorts are a service that is more affective than utilitarian, we chose commercials that differed on how much positive affect they elicited.

We selected two commercials for beach resorts that were not being broadcast in the United States at the time of the study. Both had been recognized for their quality. One received a silver prize at the Asia Pacific Advertising Festival in 2007, the other a silver award at the European Advertising Festival Eurobest in 2005. Both were 1-minute long, although one was originally 45 sec and was lengthened to 1 min to avoid any effects of exposure duration. We expected that one would be superior because the feeling it conveyed was more uniformly positive. Pending a confirmation via a pretest, this was designated the superior commercial. The real brand names were replaced by fictitious resort names: Club Zephyr (superior commercial) and Club Helios (control commercial). This substitution was facilitated by the appearance of the resorts’ names only once, at the very end of each of the original commercials.

Club Zephyr’s (superior) commercial depicted a varied series of vacation scenes, with relaxing background music. In each scene, a smile was embedded in a part of the visual environment (e.g., in the clouds above or in the water in a lagoon) as shown in figure 1. The commercial ended visually with the slogan “It’s not where you go, it’s who you meet – Club Zephyr”.

Club Helios’ (control) commercial began with a computer desktop completely filled with work icons and accompanied by the noxious sounds of a ringing telephone and a pneumatic drill. The desktop was progressively cleared of all its icons to reveal a beach scene as its wallpaper (see figure 2). At the same time the telephone and drill sounds ceased and were replaced by the soothing slosh of waves reaching a beach, like the one now revealed on the desktop. “Escape – Club Helios” were the last frames of the commercial.

Pretest. To verify that the commercials differed significantly in the positive affect that they evoked but not in their perceived effectiveness, a pre-test was conducted. Twelve participants viewed the two commercials. They then responded on a 5-point unipolar scale (1 = I did not feel this feeling at all and 5 = I felt this feeling very strongly) to ten items that reflected affect: calm, angry, peaceful, hopeful, sad, warmhearted, repulsed, emotional, depressed, and regretful. This was followed immediately by responses on a seven-point bipolar scale from -3 (negative) to 3 (positive) to ten items that captured a commercial’s overall effectiveness: untrustworthy/ trustworthy, not at all persuasive/ persuasive, unbelievable/ believable, not at all effective/ effective, unimpressive/ impressive, not clear/ clear, unconvincing/ convincing, imprecise/ clear, badly structured/ well structured, meaningless/ meaningful. Measures of affect and effectiveness were computed by averaging their ten individual ratings (with reverse scoring of the five negative feelings).

Results confirmed that Zephyr’s affect (M = 1.22) was significantly higher than that of Helios (M = .47; t (10) = 1.87, one-sided p = .05), and that the two commercials did not differ significantly in effectiveness (for Zephyr, M = -.03; for Helios, M = .82; t(10) = -.87, two-sided p = .40. Thus, this pair of commercials met the design criteria.

Product information. Five attributes provided the information about both resorts: activities, amenities, atmosphere, location, and travel guide descriptions. They were chosen to be clearly relevant to the decision but largely independent of each other. All five were described by short paragraphs like the following for the general atmosphere of both resorts:

The next day you investigate and find out that patrons describe Club Zephyr as trendy, and you can imagine that people come here to be “seen”. It has 235 luxurious hotel rooms and suites, and a choice of four stylish restaurants. The architecture of the hotel gives it a unique look. You learn that Club Helios has an elegant atmosphere, with a polished and refined feel. It is composed of 150 cottage-style rooms, giving a house-like feeling. The five restaurants are recognized for their excellent cuisine. The architecture of the hotel gives the patrons a feeling of relaxation.

This attribute was written to be neutral, that is, to favor neither Club Zephyr nor Club Helios on average.

Design

A between-participant design included one experimental and one control condition. In the experimental condition, participants were shown the two commercials before they saw the five attributes and made their choice of resort. In the control group, participants saw no commercials. Their brand choice provided a baseline level of both choice proportion and ID against which the levels observed in the commercial condition could be compared. Because all five attributes were neutral, the proportion choosing the targeted brand (Club Zephyr) in the no-commercial (control) group should not have differed from .50. In the experimental condition this proportion was predicted, with the help of ID, to significantly exceed .50. Similarly, because the commercials in the experimental condition should have provided a kind of head start to Club Zephyr, the mean ID there should have exceeded the corresponding value in the control condition.

The design was completed by two counterbalancing factors. First the order of the two commercials was reversed for half of the participants (in the experimental condition only). Second, the order of the five attributes that described the resorts in the choice task was chosen randomly and then reversed for half of the participants. Neither factor exhibited any significant impact on any dependent measure.

Participants

Ninety-six undergraduate students participated in “a study of consumer preferences” in exchange for either $5 or one-half extra course credit. Two-thirds were randomly assigned to the advertising condition, with the remaining one-third to the control group.

Procedures

Upon arrival at the experimental laboratory, participants were randomly assigned to a condition and seated at individual computer stations. The cover story explained that they had won a trip to Bermuda in a movie trivia contest organized by a radio station. Furthermore, they got to choose their preferred resort from one of the two available to contest winners. Participants in the experimental condition were also told that they had visited the Travel Channel on television and encountered a commercial for each of the two resorts (which they immediately viewed).

Then all participants received the five attributes, with the three progress questions after each one. First, the attribute was rated on a scale from 1 (“strongly favors Club Helios”) to 9 (“strongly favors Club Zephyr”), with 5 denoting neutrality (“favors neither resort”). After circling the leading resort “based on all the information seen so far”, participants indicated their confidence on a scale from 50 (“50-50, a complete toss-up”) to 100 (“absolutely certain”).

After the final choice of one resort and a confidence rating on the 50-to-100 scale, age, gender, and ethnicity were recorded in order to control for possible confounding effects.

Calculation of Distortion by the SEP method

The five attributes that comprised the core of the choice stimulus were pretested and rewritten as necessary until they favored neither alternative. This neutrality meant that their average rating on the 1-to-9 scale fell between 4 and 6. As a result, the true unbiased diagnosticity of each attribute was approximated by the scale midpoint of 5.

To compute ID, we first subtracted 5 from the participant’s diagnosticity rating of each attribute. Then, if the difference favored the alternative that was leading after the prior attribute, the absolute difference was signed positively. If the difference favored the trailing resort, the absolute difference was signed negatively. For instance, if an attribute’s rating was 7 and Club Zephyr was leading, then ID was calculated as +2. If Club Helios was leading, ID was -2. Note that ID cannot be calculated for the first attribute, because there was no measure of the leading resort prior to that attribute. Finally, the ID values for each of the four attributes were averaged within a participant. This yielded a single measure of ID for each participant.