How Entrants Affect Multiple Brands: a Dual Attraction Mechanism

Citation:

Timothy B. Heath and Subimal Chatterjee (1991) ,"How Entrants Affect Multiple Brands: a Dual Attraction Mechanism", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 768-772.

Advances in Consumer Research Volume 18, 1991      Pages 768-772

HOW ENTRANTS AFFECT MULTIPLE BRANDS: A DUAL ATTRACTION MECHANISM

Timothy B. Heath, University of Pittsburgh

Subimal Chatterjee, University of Pittsburgh

[The authors wish to thank Mike Rich and Bob Soergel for their help in data collection.]

Past research has demonstrated that adding an asymmetrically dominated alternative to choice sets generally increases the share of the dominating (target) brand. The present paper demonstrates that such attraction effects do not always occur. In particular, it is argued that entrants can enhance the attractiveness of nontarget as well as target brands. In such situations, the dual attraction effects nullify each other. Thus, although asymmetrically dominated, the entrant has little effect on market share.

Understanding the effect of new entrants on existing markets is important for understanding consumer choice processes and devising marketing strategies (for some strategic issues see Hauser and Shugan 1983). For example, one fundamental assumption underlying most mathematical choice models is regularity (Luce 1959): Introducing a new brand cannot increase the market share of an original brand. Yet violations of regularity are well documented. Entrants often increase the share of superior brands to which they are similar (e.g., Huber, Payne, and Puto 1982). This has been dubbed the attraction effect since an entrant similar to a target brand seems to attract consumers to the target.

The current paper addresses the possibility that asymmetrically dominated alternatives will increase the share of the target brand in some instances but not others. The lack of market share effects is predicted and empirically demonstrated in those instances where a single entrant is expected to enhance the attractiveness of multiple brands simultaneously. It will be argued that such dual attraction effects offset one another such that the entrant has little effect on market shares.

Prior Research and Theory on Attraction

The attraction effect is best understood by referring to Brands A, B, and E summarized in Table 1 and Figure A. Neither A nor B dominates the other since each is superior on at least one attribute. Entrant E, however, is dominated by B but not by A. Thus, E is said to be asymmetrically dominated. In Figure A, the region of such asymmetrically dominated entrants is shown by the rectangular shaded area. Relative to the market consisting of only Brands A and B, adding asymmetrically dominated entrants like C and E generally increases the probability of choosing the dominating brand B (Huber, Payne, and Puto 1982; Huber and Puto 1983; Ratneshwar, Shocker, and Stewart 1987; Simonson 1989).

The middle shaded region of Figure A consists of relatively inferior alternatives. Entrants from this region (e.g., Brand D) are always more similar (closer) to B than to A, irrespective of the attribute weights or the distance metric used. Huber and Puto (1983) found that some entrants from this region reduced B's share by stealing from B (substitution effect), but that other entrants increased B's share by moving consumers from A to B (attraction effect).

Several explanations have been put forth for the attraction effect (see Ratneshwar et al. 1987 for a review). These can be broken down into (1) perceptual mechanisms, (2) choice heuristics, and (3) moderating influences.

Perceptual Mechanisms. Huber et al. (1982) suggested that range and frequency effects (Helson 1964) may be involved in attraction. Entrant D (Table 1 and Figure A) increases the quality range from 10 to 15 and thereby makes A's 10-point quality advantage over B seem smaller. On the other hand, increasing the number (frequency) of brands along the dimension on which the target (dominating brand) dominates may draw more attention to that dimension (e.g., price). Thus, Entrant E may enhance the salience of B's price advantage since B now dominates two brands on price. Although such perceptual mechanisms may be involved in attraction effects, attraction effects have been found that cannot be explained in these terms (Huber et al. 1982; Huber and Puto 1983).

Choice Heuristics. Choice heuristics have been suggested as explanations for attraction effects (Huber et al. 1982; Huber and Puto 1983). One heuristic consists of comparing all brands on all attributes and counting the "number of wins." Another related process is referred to as relative attribute comparisons (Huber and Puto 1983). Consumers use one brand, for example Entrant D in Figure A, as an anchor. They then evaluate the attribute tradeoffs when switching from the anchor to the other brands. Switching from D to A costs 75 cents but gains 15 quality points. Each added quality point costs 5 cents. But switching from D to B costs 5 cents and gains 5 quality points, such that each added quality point costs only 1 cent. Thus, B looks better than A when comparing relative attributes. While one of two studies using concurrent verbalizations reported some evidence of such heuristic processing (Simonson 1989), the other did not (Ratneshwar et al. 1987).

Moderating Influences. Three variables have been suggested as moderators of attraction effects: (1) familiarity, (2) meaningfulness, and (3) accountability. Ratneshwar et al. (1987) argued that the attraction effect could be due to the lack of familiarity and meaningfulness of stimuli typically used in experimental choice tasks. They found only directional evidence of familiarity effects. However, using elaborated attribute descriptions to increase the meaningfulness of the stimuli reduced the size of attraction effects. While meaningfulness may moderate entrant effects, it cannot explain all attraction effects since attraction occurred even with the more meaningful stimuli.

TABLE 1

EXAMPLE CHOICE SETS

FIGURE A

POSITIONING OF ASYMMETRICALLY DOMINATED AND RELATIVELY INFERIOR ENTRANTS

Simonson (1989) hypothesized that people seek reasons for their choices, particularly when expecting to justify their selections to others. He found that attraction effects were stronger when subjects believed they would be held accountable for their choices. Although it may be easier to justify decisions when asymmetric dominance exists, the mechanisms underlying this ease are probably some combination of those discussed earlier (e.g., relative attribute comparisons).

An Added Mechanism for Asymmetric Dominance

Failure of Relative Attribute Comparisons. As noted above, the relative-attribute-comparison heuristic can account for attraction effects when entrants are relatively inferior (e.g., Brand D in Figure A). However, it does not always work with asymmetrically dominated entrants. For example, if the consumer anchors on E in Table 1 (also Figure A), switching to B saves 5 cents and has no effect on quality. Switching to A costs 65 cents and gains 10 quality points such that each quality point costs 6.5 cents. Whether A or B looks more appealing according to relative attribute comparisons is not clear. Yet attraction effects are commonly reported with such entrants (e.g., Simonson 1989).

Considering Entrant J in Table 1 further demonstrates the difficulty that the relative-attribute-comparison heuristic has predicting the effects of some asymmetrically dominated entrants. Switching to B saves 60 cents and loses no quality, whereas switching to A costs 10 cents but gains 10 quality. Again, relative attribute comparisons do not clearly favor A or B. Further, and more importantly, the relative-attribute-comparison heuristic does not predict different effects from Entrants E and J. But as discussed next, there is ample reason to expect that Entrant E will increase B's share more than will Entrant J.

Salience of Value Differences. Relative attribute comparisons based on anchor brands are likely whenever entrants differ from both brands on both attributes. However, for the reasons just discussed, attraction effects due to entrants that differ from existing brands on only one attribute are difficult to explain in tern s of relative attribute comparisons. An additional process seems to be at work in such instances. Instead of treating one brand as an anchor, we propose that consumers focus on any obvious differences in value (quality/price). [Although the term value connotes quality per price, in the present context it is not restricted to situations where price is an attribute. For example, Simonson (1989) examined cars differing in their gas mileage and ride quality. In this instance, value could be defined as either mileage/ride quality or as ride quality/mileage.] Given the mental arithmetic necessary to calculate and compare values, value differences that are not obvious are expected to be ignored.

The differential salience of value comparisons can be illustrated with Brands A, B, and E in Table 1 (also Figure A). If consumers were to calculate the values of each brand, then Entrant E would make both A and B look better since its value of 15.12 is lower than that of A (15.15) and B (15.29). But since such calculations can be mentally taxing, consumers may recognize only the obvious value difference. The B-E value difference is obvious since B has the same quality as E but a lower price. B's advantage can then be discerned simply by noting its lower price. On the other hand, the A-E value difference is not so obvious since A costs more but offers more quality than E. To appreciate A's advantage over E, consumers would have to calculate and then compare the values of A and E (i.e., price and quality would have to be traded off carefully). Hence, this heuristic suggests that Entrant E will increase B's market share because consumers focus on the salient B-E value advantage and essentially ignore the less salient A-E advantage.

In contrast to the relative-attribute comparison heuristic, the salience heuristic predicts different effects of Entrants E and J in Table 1. As just noted, Entrant E is expected to increase B's market share since B's value advantage over E is much more salient than A's. But the salience hypothesis also predicts that A's advantage over an entrant will become more salient as the entrant's price increases and approaches that of A (i.e., the entrant's value decreases). For example, A's value advantage over Entrant J should be fairly obvious since A has about the same price as J but higher quality. Unlike Entrant E, Entrant J is expected to enhance the attractiveness of both B and A. This leads us to what on the surface is a counter-intuitive proposition: B's market share should improve more when the entrant is stronger (e.g., E) than when it is weaker (e.g.,i J). Vs nonmonotonic relationship between the entrant's appeal and the target's (B's) market share is captured in the following hypothesis:

H1: Adding an asymmetrically dominated entrant similar to the target will increase the target's market share (attraction effect). But as the entrant becomes weaker and increasingly similar to the nontarget brand, the improvement in the target's market share should become less.

In prior research, changes in a brand's market share were used to infer changes in the brand's attractiveness. However, if an entrant can influence multiple brands, then market share may not measure an entrant's effect on a brand's attractiveness. Market. shares may remain constant when an entrant improves the attractiveness of A and B comparably. Readers should be alerted to the fact that the phrase dual attraction does not imply increased market shares of multiple brands. Instead, the phrase dual attraction means that an entrant makes both brands appear more attractive which, due to offsetting effects, has little impact on market share.

STUDY OVERVIEW

Hypothesis 1 was tested in an experiment. The entrant's price was held constant while its quality was varied. For theoretical reasons, the current study used various entrant versions. To keep the size of the study manageable, a single product class was used. Fortunately, this does not pose a serious problem for the current study since the generalizability of attraction effects is well established. Attraction has been demonstrated in such varied product categories as beer, cars, light bulbs, barbecue grills, restaurants, lotteries, film, televisions, calculators, and mouthwashes. To enhance generalizability as much as possible, we selected the one product category from Ratneshwar et al.'s (1987) study that was completely unaffected by stimulus meaningfulness (i.e., beer).

TABLE 2

THE EFFECT OF AN ENTRANT'S QUALITY ON BRAND A'S MARKET SHARE

EXPERIMENT

Subjects

One-hundred-and-ninety-six undergraduate and masters level business students from three large eastern universities voluntarily served as subjects.

Design and Procedure

A one-way between-subjects experiment consisting of four conditions was used. Subjects were told to imagine being in the market for a six-pack of beer. The control group was asked to choose from a market consisting of two brands: Brand A was priced at $4.95 and had a quality rating of 75 (1-100 scale). Brand B was priced at $4.25 and had a quality rating of 65. The experimental groups chose between the same two brands, although an out-of-stock entrant was added.

The entrant's price was held constant across groups at $4.95 (A's price). Its quality was then varied systematically to assess attraction effects on Brand A. Three levels of the entrant's quality were used: 72, 67, and 65. An entrant with quality of 72 was expected to enhance the attractiveness of only Brand A and thereby increase Brand A's share. However, this improvement was expected to be less when the entrant's quality was at 67 since the entrant was then somewhat similar to B and expected to simultaneously enhance the attractiveness of A and B. When the entrant's quality was further lowered to 65, the entrant was expected to have little effect on market shares since it was expected to enhance the attractiveness of A and B comparably.

Subjects in the experimental groups were told that the entrant was out of stock and that they had to confine their choices to Brands A and B. Excluding the entrant from choice consideration eliminated any possibility of substituting the entrant for A or B. Substitution effects are relatively unlikely when entrants are clearly dominated. However, unavailable brands were used in this study in part to make it comparable to a parallel study currently underway.

Results

In keeping with previous attraction studies (e.g., Huber and Puto 1983), between-group differences were assessed with Fisher's Exact Test of Independence (Cox 1989). The Exact Test is used in place of Chi2 tests when sample sizes are small and the normal approximation to the binomial distribution is not valid. The Exact Test restores the power that would otherwise be lost. A separate 2 X 2 contingency table was formed for each pairwise comparison between control and experimental group.

Brand A's share in the control group was 49%. It was hypothesized that adding an entrant at $4.95 and 72 quality would increase A's share, but that this increase would become smaller as the entrant's quality was reduced to 67 and then to 65. Table 2 summarizes the results.

Consistent with Hypothesis 1, adding an entrant with a quality rating of 72 significantly increased A's share from 49% to 80.8% (g = .001). This replicates the attraction effects reported in other studies.

Making the entrant even weaker by decreasing its quality to 67 decreased A's share to 72.3%, and further decreasing the entrant's quality to 65 further decreased A's share to 61.5%. The difference between the quality-72 and quality-67 conditions was not statistically significant, nor was that between the quality-67 and quality-65 conditions. However, both the quality-72 and quality-67 groups differed significantly from the control group while the quality-65 group did not. Further, the quality-72 group differed significantly from the quality-65 group (; = .029).

The steady reduction in A's share as the entrant became weaker cannot be accounted for by relative attribute comparisons, range effects, or frequency effects. However, the trend is consistent with Hypothesis l's salience-based dual-attraction mechanism. As the entrant's quality approached that of B, the entrant increasingly enhanced the attractiveness of B which then offset the simultaneous attraction effect on A.

GENERAL DISCUSSION

The critical finding in the present study is that decreasing an entrant's attractiveness increases a target's share only to a point, beyond which decreasing entrant attractiveness actually reduces the target's share. Decreasing entrant attractiveness eventually leads to an attraction effect toward the nontarget brand which then nullifies the simultaneous attraction effect on the target. This apparent dual attraction can be accounted for only by appealing to the salience of value differences between different brands. Neither relative attribute comparisons, nor range effects, nor frequency effects can account for these findings.

It is important to note, however, that the choice of a heuristic may depend on the configuration of brands in the market. For example, value salience should play a role when Entrant X below is introduced since X is easily compared with A on value. However, Entrant Y below is not as easily compared with A or B. Therefore, consumers may not focus on value salience when Y is introduced, and instead may use relative attribute comparisons.

TABLE

Two recent studies using concurrent verbalizations are consistent with the expectation that market configurations can lead to different heuristics. Using entrants that differed from targets on only one attribute, Simonson (1989, Experiment 3) found evidence of pairwise comparisons. In contrast, Ratneshwar et al. (1987, Experiment 4) used entrants that differed from targets on both attributes and found no evidence of pairwise comparisons.

Although different heuristics may be used in different markets, scenarios can be devised where either heuristic might be used. For example, a model based on relative attribute comparisons predicts that Entrant X above will enhance B's attractiveness and thereby increase B's share. In contrast, since A's value advantage over X is fairly salient, the salience hypothesis predicts that Entrant X will additionally increase the attractiveness of A and potentially nullify the attraction effect on B. So called "strong tests" between competing mechanisms should help researchers hone in on consumer choice processes.

The current study suffered from shortcomings that future research needs to address. First, only one product class was used which may threaten the generalizability of the findings. However, this concern is partially mitigated by the fact that attraction effects have been found repeatedly in many product classes. Second, choice sets consisted of only three brands. Future research needs to address scenarios that mare closely approximate the multiple-brand markets of the real world.

Third, entrants in the current study differed from targets on only one attribute. Although we have noted that relative attribute comparisons cannot explain the effects of such entrants, relative attribute comparisons probably exist when entrants differ from other brands on multiple attributes as noted earlier.

Fourth, the current study inferred dual attraction effects based on choice data that were consistent with an underlying theory. However, a more direct test of dual attraction would be to test the effects of various entrants on the liking for the two brands. For example, if the dual attraction mechanism is at work, Entrant E in Table 1 should elevate only liking for Brand B, whereas Entrant J should elevate liking for both A and B.

REFERENCES

Cox, David R., and E. J. Snell (1989), Analysis of Binary Data, Monographs on Statistics and Applied Probability, No. 32, New York: Chapman and Hall.

Hauser, John R., and Steven M. Shugan (1983), "Defensive Marketing Strategies," Marketing Science, 4 (Fall), 319-360.

Helson, Harry (1964), Adaptation-Level Theory: An Experimental and Systematic Approach to Behavior," New York: Harper and Row.

Huber, Joel, John W. Payne, and Christopher Puto (1982), "Adding Asymmetrically Dominated Alternative: Violations of Regularity and the Similarity Hypothesis," Journal of Consumer Research, 9 (June), 90-98).

Huber, Joel and Christopher Puto (1983), "Market Boundaries and Product Choice: Illustrating Attraction and Substitution Effects," Journal of Consumer Research, 10 (June), 3144.

Luce, R. Duncan (1959), Individual Choice Behavior, New York: Wiley.

Ratneshwar, Srinivasan, Allan D. Shocker, and David W. Stewart (1987), "Toward Understanding the Attraction Effect: The Implications of Product Stimulus Meaningfulness and Familiarity," Journal of Consumer Research, 13 (March), 520-533.

Simonson, Itamar (1989), "Choice Based on Reasons: The Case of Attraction and Compromise Effects," Journal of Consumer Research, 16 (September), 158-174.

Tversky, Amos (1972), "Elimination by Aspects: A Theory of Choice," Psychological Review, 79 (July), 281 -299.

----------------------------------------