Brand Awareness Effects on Consumer Decision
Making for a Common, Repeat Purchase Product:
Emma K. Macdonald
Byron M. Sharp
This article is a replication of a study of Hoyer and Brown that used a
keep the brand in the consumer’s consideration set—the set
of brands to which a consumer gives serious attention when
controlled experiment to examine the role of brand awareness in the
consumer choice process. The replication used the same methods, but with
making a purchase decision. Brand awareness has been argued
to have important effects on consumer decision making by
a different (but similar) product category, a larger sample, and a sample
group that included experienced as well as inexperienced consumers. Results
influencing which brands enter the consideration set, and it
also influences which brands are selected from the consider-
support the original study’s findings that brand awareness is a dominant
choice tactic among awareness group subjects. Subjects choosing from a
ation set (Macdonald and Sharp, 1996). Brand awareness af-
fects the latter through its use as a heuristic for choice (e.g.,
set of brands with marked awareness differentials showed an overwhelming
preference for the high awareness brand, despite quality and price differen-
“I’ll choose the brand I know”) and its influence on perceived
quality, (“I’ve heard of the brand, so it must be good”). A
tials. They also made their decisions faster than subjects in the nonaware-
ness condition and sampled fewer brands. In a surprising finding, respon-
study of Hoyer and Brown (1990) carried out pioneering
research at the individual decision level by examining the
dents use of the awareness choice heuristic did not seem to decline steadily
over repeated choice trials, but rather showed something of a U-shaped
effects of brand awareness on consumer choice. It examined
the impact of brand awareness as a heuristic, as well as explor-
pattern, with subjects returning to the high awareness brand in the latter
choice trials. Little support was found for Hoyer and Brown’s finding that
ing its effect on perceived quality.
subjects in the no brand awareness conditions chose the quality brand on
Little research has examined the effect of brand awareness
the final trial more often that those in the awareness differential conditions.
on choice. Consumer behavior theory in both the marketing
In summary, awareness differentials seem to be a powerful influence on
and economic literature has tended to see product choice as
brand choice in a repeat purchase consumer product context. Consumers
a highly involving problem-solving process (Foxall, 1992).
show a strong tendency to use awareness as a heuristic and show a degree
Marketing research has focused upon more elaborate knowl-
of inertia in changing from the habit of using this heuristic.
J BUSN RES
edge structures than awareness, such as attitude and brand
2000. 48.5–15. 2000 Elsevier Science Inc. All rights reserved.
image. However, a study of Hoyer (1984) indicated that, in
many purchase situations, the consumer is a passive recipient
of product information who spends minimal time and cogni-
tive effort choosing brands. In situations involving common,
t has long been held that one of the major goals of market- repeat purchase products, it may be that consumers choose the
ing is to generate and maintain brand awareness, this is brand on the basis of a simple heuristic (e.g., brand awareness,
seen as particularly important in low-involvement situa- package, price). More detailed evaluation, if it happens at all,
tions where consumers may engage in little active search for occurs subsequent to purchase (Ray, Sawyer, Rothschild, Heeler,
information to aid choice. Repetition of advertising is used to Strong, and Reed, 1973; Olshavsky and Granbois, 1979).
The concept of habitual consumer behavior is not new,
and limited problem solving has been acknowledged in several
Address correspondence to: Byron Sharp, Marketing Science Centre, Univer-
cognitive decision-making models of consumer choice (e.g.,
sity of South Australia, Adelaide 5000, Australia; E-mail: Byron.Sharp@
msc.unisa.edu.au and Emma K. Macdonald, Cable and Wireless Optus, Level
Engel, Blackwell, and Miniard, 1993; Engel, Blackwell, and
16, 101 Miller Street, North Sydney NSW 2060, Australia; E-mail: emma_
Kollat, 1978). East points out, however, that there is contradic-
Journal of Business Research 48, 5–15 (2000)
2000 Elsevier Science Inc. All rights reserved. ISSN 0148-2963/00/$–see front matter
655 Avenue of the Americas, New York, NY 10010 PII S0148-2963(98)00070-8
J Busn Res E. K. Macdonald and B. M. Sharp
tion in this inclusion: if choice is habitual, than there is no publication of the original study), in a different place (in
Australia rather than the United States), and the researchersdecision in the sense of conscious cognitive processing before
action, habits, if they account for much of consumption, need a in both studies have had no communication. (Requests were
made to one of the researchers of the original study to providebetter explanation than absence of thought (East, 1996). Behav-
ioral theories are one approach to this gap in consumer behav- details of original methods and results. The first mail request
received no reply. The second request through a colleague,ior theory, brand awareness/salience-based choice theory is
another potential direction that is still cognitivist in nature but received a positive response, but at the time of writing, no
material has been received.)needs not to be predicated upon the assumption of conscious
cognitive processing and attitude formation prior to purchase. Implicit in this discussion is the assumption that the origi-
nal study is worth replicating. As discussed, H&B’s study wasThis article reports on a recent replication of Hoyer and
Brown’s (H&B) (1990) study which used a controlled experi- exploratory and carried out in an area where little research
ment to explore the effects of brand awareness on choice, has been done previously. It is an important study within a
brand sampling, and the frequency with which the highest- new low-involvement habitual repeat purchase paradigm of
quality brand is chosen following a series of trials. consumer behavior. It consisted of a well-constructed experi-
mental design that used actual brands to enhance its external
validity. Because of the exploratory nature of the study, the
Why Replicate Hoyer and
sound experimental design employed, and its interesting re-
sults, it is a study worthy of further examination.
Some replications are better than others. Replications con-
ducted early in the history of a particular research program
are usually more useful (Monroe, 1992; Rosenthal, 1990).
H&B (1990) employed an experimental procedure in which
The study of H&B is a pioneering study. As its authors noted:
subjects were asked to make a series of decisions regarding
Studies that pinpoint the impact of brand awareness on brand choice for a common, repeat purchase product: peanut
the individual-level choice process are badly needed. This butter, in this replication study: orange cordial (in Australia,
article takes a first step toward filling this void by examining a nonalcoholic soft drink). After each selection, subjects were
the nature of brand awareness effects on the purchase of asked why they chose a particular brand, and they were then
a common household product. (Hoyer and Brown, 1990 permitted to taste it. After five trials, a series of post-task
p.141) questions were asked regarding product usage and experience.
The replication differs in the following respects.
A replication that offers insight into the generalizability of
H&B’s findings will be particularly useful as consumer behav-
ior begins to address this area.
Such replication research helps us establish the scope or H&B selected their sample from U.S. freshman college stu-
boundaries of previous outcomes. In this regard, studies using dents (n⫽173). The replication employed a larger sample
different stimuli, samples of people, methods, procedures, selected from a comparable population of undergraduate Aus-
and analytical techniques, in different situations, over different tralian university students (n⫽472).
occasions serve the advancement of knowledge in consumer
EXPERIENCE WITH THE PRODUCT CATEGORY.
The majority of
behavior (Rosenthal, 1990). The replication reported in this
H&B’s subjects had never purchased peanut butter for them-
paper, employs the same procedures as H&B but uses a differ-
selves, and the rest indicated they had purchased peanut butter
ent product category (still a common, repeat purchase prod-
only a few times at most. In the replication study, more than
uct), a slightly altered (inclusion of non-novices) and larger
half of the 462 included in the sample (56.3%), had purchased
sample, and a more realistic experimental setting.
cordial before. Two levels of experience with the product
Generally, replication research that originates with an inde-
category were identified: (1) inexperienced: never bought or-
pendent researcher is more highly regarded (Monroe, 1992).
ange cordial before or only a few times at most (n⫽226);
Outcomes reproduced by independent researchers reduce
and (2) experienced: buy cordial at least every few months
concerns about the interaction of the researcher and study
being a reasonable explanation for the original results. The
Most Australian university students are employed part time
value of the results of a replication is maximized for each of
and even full time, and do not live with parents, and tend to
the following dimensions that apply: time, physical distance,
shop in supermarkets. In addition, with the increase of both
personal attributes of the experimenters, experimenter’s ex-
parents working, teenagers are increasingly becoming respon-
pectancy, and experimenter’s degree of personal contact with
sible for some or all of the family’s grocery purchases. There-
each other (Rosenthal, 1990). According to these dimensions,
fore the sample used in the current study, with half being
the replication reported in this paper is of high value, because
the study was conducted at a different time (several years after experienced and the other half novices (although still more
Brand Awareness Effects on Consumer Decision Making J Busn Res
experienced than subjects in the original study), is more realis- (77.1% consume cordial often: daily to once per month)
with the flavor orange dominating the category.
tic than H&B’s in terms of the consumption experience of
• The category contains a number of brands. In a pretest
of unprompted recall, 25 different brands of cordial were
The decision to extend the replication through the inclu-
sion of non-novice respondents was in line with the potential
• A few well-known, national brands exist in addition to
theoretical contribution of empirical research on this topic.
a large number of nonadvertised regional brands and
The habitual low involvement paradigm of consumer choice
store brands. Thus, marked awareness differentials exist
is contrasted with the problem-solving models not only in
in the real world, enhancing the external validity of the
terms of degree of cognition and attitude formation prior to
study (at least in terms of generalizing to the tested
purchase but also in the implicit assumption that many choice
product category). These awareness differentials indi-
decisions are first purchases. The empirical fact that most
cated that awareness levels could be easily manipulated
purchases are repeat or replacement purchases (see Wilkie
in the experimental situation. In the pretest, Cottees had
and Dickson, 1985) provided a powerful rationale for the
the highest awareness level, which is consistent with the
inclusion of experienced cordial purchasers in the sample.
high level of national advertising for the brand.
EXPERIENCE WITH THE TEST BRANDS.
To be included in H&B’s • As with peanut butter, cordial can be easily tried and
study, subjects had to have never purchased any of the brands tasted in an experimental situation, permitting an assess-
used in the experiment. ment of the impact of awareness on brand evaluations.
Unlike H&B’s study, many of the students in the replication • As with peanut butter, cordial is not a status/symbolic/
had purchased and/or used one of the test brands. The high fashionable product category.
awareness brand in question, Cottees, is a brand name used
One limitation of cordial as a test category is
across several product categories including jam and desserts.
the need for it to be diluted with water to taste it. To ensure that
Many subjects claimed to have used this brand before (87.2%).
product quality was standardized, the cordial was premixed in
Few subjects (5.6%) said they had used (or thought they might
the ratio recommended by manufacturers of 1:4 parts (cor-
have used) a second brand in the study, Mynor. It was not
dial:water). Spring water was used because of the lamentable
expected that the small degree of familiarity for this brand
variability of the available tap water.
would have an impact on the study.
See Appendix A.
In H&B’s study, the chosen product category was peanut
butter because of the availability of a number of well-known,
easily identifiable brands, and several lesser-known nonadver-
As in H&B’s study, awareness was operationa-
lized as a two-level blocking factor consisting of awareness
PEANUT BUTTER AS A NATIONAL ICON.
Peanut butter is very and no-awareness conditions. In the awareness condition,
much a part of the culture in the United States, where peanut subjects were presented with three brands of peanut butter
butter and jelly sandwiches are almost a staple in the U.S. (in H&B’s study) and three brands of cordial (in the replica-
child’s diet, but peanut butter is of less importance in Austra- tion). In each study, one of these brands was a well-known
lian culture (with the sandwich combination of peanut butter national brand that had been heavily advertised and was highly
and jelly being a specialist taste). recognized, as indicated by a pretest. Two unknown brands
CRUNCHY OR SMOOTH? EXISTENCE OF CUSTOMER SEGMENTS.
from other regions of the country completed the three-brand
One difficulty encountered in replicating this study was the set. These brands elicited very low levels of recognition, and
existence of wide variation in types of peanut butter, such as none was recalled by consumers in a pretest of free recall.
crunchy versus smooth. Different segments within the popula- Thus, it was deemed acceptable to use them in the study.
tion may regard one or other of these product attributes as The high awareness brand, Cottees, elicited 97% (296/306)
more preferable, and respondent preferences could potentially recognition and 86% (38/44) free recall; that is, very large
alter, even during the experiment. The researchers of the differences in levels of awareness were recorded across the
original study do not make it clear whether the existence of test brands.
customer segments with different tastes was considered in In the no-awareness condition, subjects were presented
selecting the products used in the original experiment. with three totally unknown brands. Two brands were the
same brands used in the awareness condition. The third was
SELECTION OF DRINKING CORDIAL.
The selection of orange another unknown brand from another region of the country.
cordial in the replication was based on the following reasons: As discussed previously, unlike the original study, the replica-
tion found some level of awareness, although very low and• It is widely used by those in the sample population
J Busn Res E. K. Macdonald and B. M. Sharp
Table 1. Reported reasons for choice on trial one. Awareness cf no-awareness condition subjects. original study and replication.
Original Study Replication
H&B: H&B: REP: REP:
Not Aware Aware Not Aware Aware
Criteria (n⫽90) (n⫽83) Z (n⫽166) (n⫽296) Z
1. Known brand 0 60.0 11.11** 3.0 48.3 ⫺15.55**
2. Taste 4.3 0 NS 0 0 NS
3. Lowest price 2.2 0 NS 10.8 2.7 3.13**
4. Ingredients 10.8 3.3 2.16* 10.2 1.7 3.446**
5. Package 45.2 4.4 6.95** 34.3 8.1 6.53**
6. Try new brand 1.1 0 NS 0.6 1.4 NS
7. Known brand and taste 0 3.3 NS 0 0.7 NS
8. Known brand and price — — — 0.6 18.9 7.78**
9. Known brand and other 0 18.9 4.41** 0 7.1 ⫺4.756**
10. Price and taste 1.0 0 NS 0 0 NS
11. Price and other 14.0 4.4 2.63* 30.7 8.5 5.65**
12. Taste and other — — — 0 0 NS
13. Other 21.4 5.7 — 9.6 2.7 —
weak, for two of these brands. One brand in particular, Mynor, Although it is not made clear how price was manipulated
had a 17% level of recognition; however, it was noted that across the entire H&B study, the basis for the price manipula-
respondents were far more uncertain of themselves when iden- tion in the replication was as follows: two prices were selected,
tifying this brand as known to them in a prompted recall a high and low price, which were realistic as indicated from a
test. This brand was retained in place of having two totally survey of cordial brands’ pricing in local supermarkets. Using
unknown brands, because it was considered of interest to a counterbalancing procedure for each awareness and quality
determine whether low levels of awareness might be significant condition, price was varied so that, in each condition, one
in terms of choice. brand was marked at the high price, and two brands were
The small impact of this degree of awareness for one of marked at the low price.
the “no awareness” brands can be seen in Table 1, where 3%
of respondents in the no-awareness condition reported the
reason of “known brand” to explain their first choice. The As per H&B (1990, the three dependent variables were the
consequence is that the experiments are a stronger test of nature of choice tactic used in brand selection (obtained by
H&B’s main findings. the open-ended question “Can you tell me why you have
To identify high and low-quality selected the brand you have chosen?”), the number of brands
cordials, seven brands were evaluated by a group of pretest sampled (determined by a count of the number of brands
subjects (n⫽49) in a blind taste test. As in the original study, sampled by subjects across five trials), and whether the highest
these brands ranged from presumably high-quality brands quality brand was finally selected (the analysis involved only
(i.e., national, well-advertised brands) to those of presumably subjects in the quality difference condition: original study n⫽
lower quality (e.g., store and generic brands). Pretest subjects 88; replication n⫽401).
ranked these brands in order of preference. The full methodol-
ogy and results to identify high- and low-quality brands is
presented in Appendix B. Over the 401 subjects in the quality Based on the original study, an experiment was used to test
difference condition, each brand (i.e., bottle) contained the the three hypotheses. Subjects were run individually and made
high-quality product (cordial) an approximately equal number a series of brand choices for orange cordial. After each selec-
of times. tion, they were asked why they chose a particular brand and
A third independent variable, not discussed in detail were then permitted to taste it. As in the original study, to
by H&B, is that of price. H&B did not present their pricing complete the session, the subject was asked a series of post-
results in their findings other than to state that the price task questions to check the purity of the sample. The only
manipulation: major procedural differences between the original study and
the replication was the use of a more natural setting in the
had little effect on subjects’ choices and was orthogonal to
replication (a university cafeteria). This research approach is
the other manipulations. Hence, it is not discussed further.
(p143). gaining support (e.g., see Lutz, 1991) and added more realism
Brand Awareness Effects on Consumer Decision Making J Busn Res
to the choice task, because consumers usually make buying The results in Table 1 show substantial use of awareness
as a choice tactic by consumers in the awareness conditionchoices for this type of product in a supermarket, where
distractions are present. on the first trial with 60 and 48.3% reporting use of this tactic
in the original study and the replication, respectively. A further
22% (original study) and 27% (replication) in both studies
reported using a combination of awareness and some other
tactic as a basis for their decision.
The questions of interest in this study are how awareness affects
Contrasting these results with subjects in the no-awareness
choice probability and sampling of a common, repeat-purchase
condition in the absence of awareness, consumers used a
product under varying conditions of brand quality and awareness
number of other criteria upon which to base their choice:
(Hoyer and Brown 1990, p. 142): To investigate the extent
package (45 H&B, 34% replication), and then price and ingre-
to which a simple heuristic based on awareness, such as “buy
dients. In the replication (cf. the original), a much larger
the best known brand,” was utilized in a simple choice task.
proportion of subjects reported choice based on price either
H1: Brand awareness serves as a dominant choice tactic alone (10%) or in combination with some other factor, such
among inexperienced consumers presented with a as ingredients (31%).
brand selection task. Both H&B’s study and the replication indicate that consum-
ers may rely on awareness as a cue for choosing a brand, whenThe hypothesis was examined by observing actual choice and
by eliciting subjects’ free responses about their choice strategies. a clear distinction between brands exists on this dimension.
However, when no brands are known, other criteria, such as
packaging, ingredients, and price, are likely to be employed.
H&B found that subjects’ initial choices provided strong sup-
TIME TAKEN TO MAKE THE INITIAL DECISION.
port for the hypothesis, and this finding is backed up by the ploy heuristics to simplify decision-making tasks. The results
replication. Specifically 93.5 and 85.5% of subjects in the so far indicate that consumers have used awareness as a simpli-
original study and the replication, respectively, chose the fa- fying heuristic. A further hypothesis developed in the replica-
miliar brand on the first trial (Analysis of chi-square difference tion that builds on this is that:
in the replication:
⫽1560.1, df ⫽2, p⬍0.005). From
H1a: Consumers choosing from a set of brands which
the results of both studies, it seems that, when faced with
includes one known brand will make a decision more
a choice situation in which a known brand competes with
quickly than consumers choosing from among a set
unknown brands, consumers are considerably more likely to
of three unknown brands.
choose the known brand.
This hypothesis was supported. Consumers in the awareness
Reported Reason for Choice
condition made their initial decisions more quickly than con-
To assess the use of awareness directly as a choice heuristic, sumers in the no-awareness condition, mean of 9.8 seconds
subjects were asked to give their reasons for choosing a partic- cf. 15.1 seconds (t ⫽2.61, df ⫽227, p⫽0.01, critical
ular brand after the first and last trial. Table 1 presents the interval (⫺9.252, ⫺1.299)). Therefore, it seems that in the
results of this open-ended question. absence of brand awareness as a simplifying choice heuristic,
Of the twelve criteria listed in Table 1, agreement between consumers exert more decision effort by evaluating other
the replication and the original study occurs on nine of the brand attributes. Because the only difference between the two
criteria (Z test scores for criteria 1 and 9 were reported without conditions was the presence or absence of one known brand,
the negative sign in the original study). Of the remaining three it can be assumed that the difference in decision-making time
criteria, two (8 and 12) were not listed in the original study. indicates that brand awareness is important as a heuristic that
The remaining difference between the two studies occurred in simplifies decision making for a common, repeat purchase
the reported use of price as a reason for choice. The replication product.
found a significant price effect difference between consumers
DECREASE IN IMPORTANCE OF BRAND AWARENESS AS A CHOICE
in the awareness and no-awareness conditions. That is, it
TACTIC OVER TIME.
Based on the hierarchy of effects model,
seems that in the presence of awareness, consumers are less
H&B suggested that:
likely to use price as a heuristic. Although this might be seen
as an exciting finding for owners of high-awareness, high- H1b: Over a number of trials, awareness will decrease in
price brands, it should be noted that consumers can use price importance as a choice tactic. Choice will instead be
as a heuristic in two alternative ways: seek lowest price to based on previous trials and evaluations.
avoid financial risk; or seek highest price in an effort to gain high
This would imply that where a known brand exists in a choice
product quality. The original study produced a nonstatistically
set, consumers are more likely to choose the known brand
significant result in the same direction, the lack of statistical
significance presumably caused by the small sample size. on the first trial, but given the opportunity to try other brands
J Busn Res E. K. Macdonald and B. M. Sharp
Table 2. Self-reported reasons for choice: first and last trials
after the initial trial. This somewhat U-shaped curve was also
observed in the awareness–no-quality difference condition
Awareness Condition Subjects
(Figure 2), but did not occur in the condition where no-
Trial 1 Trial 5
awareness existed (Figure 3).
This finding seems to be consistent with the brand-switch-
Awareness H&B 82% H&B 33%
Rep 75% Rep 51%
ing literature, which indicates that consumers are loyal not
Taste H&B 0% H&B 41%
only to a single brand but to a repertoire of brands (Ehrenberg,
Rep 0% Rep 24%
1988). Consumers will select from among these brands when
in a purchase situation. In addition, from time to time, con-
sumers will try a brand from outside the repertoire, most
likely because of some situational event.
over a number of trials, they are more likely to base their The results of this study indicate that consumers demon-
choice on their evaluation of the brand, so such factors as strated curiosity about the other brands available to them, but
taste will become more important. still preferred to choose the well-known brand. This could
To investigate this, we first examined the differences in the be a result of some of the effects attributed to high-awareness
self-reported reasons for choice on the first and last trials brands, such as reassurance of popularity and quality (Tybout
(Table 2). and Artz, 1994) or that they preferred to stick to their aware-
H&B found that reported use of awareness as a choice
tactic dropped from 82% down to 33% between the first and
The assumption has been that brand awareness is an impor-
last trials. This strongly supports their hypothesis that other
tant simplifying heuristic for choice. Therefore H&B (1990)
factors become important in making a decision over a number
hypothesized that, in its absence, consumers would exert more
of trials. For example, reported use of a taste as a tactic
effort in selecting among brands. Therefore, their second hy-
increased from 0 to 41% by the final trial.
pothesis was that:
It should be noted, however, that fairly strong experimental
effects would also encourage the same result. Respondents H2: Consumers choosing among a set of unknown brands
might make their first choice based on awareness, but once are likely to sample more brands across product trials
asked to repeat the choice, might infer that they are required than consumers who choose among a set of brands
to do something different (e.g., choose a different brand). that includes one well-known one.
Results of the replication show the same trends as in the
H&B predicted that this hypothesis would be supported by
original study, although they are substantially weaker. Al-
a significant main effect of awareness in a two-way awareness-
though 75% of subjects reported use of brand awareness as
by-quality analysis of variance (ANOVA). In fact, a significant
a choice tactic on the first trial, this dropped down to 51%
main effect for awareness did occur in the analysis for both
by the final trial. Thus, the use of brand awareness did decline,
studies (H&B F[3,169] ⫽7.87, p⬍0.01; replication
but it continued to be important, because over half the respon-
F[1,458] ⫽6.040, p⬍0.05). Neither the main effect of
dents reported using it as a choice criterion, even after having
quality, nor the two-way interaction approached significance
the chance to sample other brands over a number of trials.
in either study. These results support H2.
This result is not exactly supportive of the hierarchy of effects
Further confirmation of H&B’s results is found in the com-
reasoning that lead to the hypothesis, but rather could be
parison of the mean number of brands sampled between
interpreted as being more supportive of the simpler habitual
awareness and no-awareness conditions. The results indicate
theories of choice.
that subjects in the no-awareness condition tended to sample
Supporting the original study, again, it was found that such
more brands than did subjects in the awareness condition
experience factors as taste increased in importance over the
(H&B: 2.67 cf. 2.29, replication: 1.86 cf. 1.58 (t ⫽3.94, p⬍
five trials, but the increase in the use of taste over the five
0.001). Interestingly, the mean number of brands chosen was
trials was less pronounced in the replication (0% on the first
much lower in the replication than in the original study,
trial, as compared with 24% on the final trial).
despite there being some prior awareness for one of the no-
Moving to subjects’ actual behavior, the findings from the
awareness brands, which should have lead to its also enjoying
replication are clearly different from those predicted by H&B.
the greater chance of trial effect. This lower mean could be
Instead of observing a decline in choice of the high-awareness
caused by some variation in the procedure used or the differ-
brand over the five trials, a U-shaped curve was observed
ence in product category.
(Figure 1). The majority (86%) of subjects chose the well-
H&B reasoned that when no known brand is available in
known brand on the first trial, but after selecting this brand
a choice set, and consumers are given an opportunity to sample
one or two times on subsequent trials, some were included
brands, their perceptions of quality are likely to be unaffected
to try one of the brands they didn’t know (31% chose an
by the biases and distortions that awareness of the brand can
unknown brand on trial 3). However, most (75%) chose the
create. They, therefore, hypothesized that consumers in this
high awareness brand on the final trial (see Figure 1), indicat-
ing that the use of awareness as a heuristic continued long situation were more likely to detect “objective” quality differ-
Brand Awareness Effects on Consumer Decision Making J Busn Res
Figure 1. Brand choice: Awareness-quality condition (n⫽265).
ences between brands following a number of trials. Therefore, not statistically significant, difference between the awareness
and no-awareness groups (41% cf. 59%). Their sample sizetheir third hypothesis was that:
was very small with only 44 subjects in each condition, in
H3: After a series of product choices, consumers choosing the replication the sample size was much larger (n⫽401)
among a set of three totally unknown brands are more but the differences were still not statistically significant (Pear-
likely to choose the high-quality brand than are con- son’s chi
⫽0.63150, p⫽0.42681). The lack of statistical
sumers who choose among a set of brands that in- significance in either study makes it impossible to say if the
cludes one well-known and two unknown brands, results varied meaningfully between the studies but if they
especially when the well-known brand is not the high- did one factor that may account for a difference is that in the
quality brand. replication study subjects selected significantly fewer brands
over the five trials than they did in the original study. This
This hypothesis received the weakest support in both the
may have affected their ability to choose the ‘high quality’
original study and in the replication.
brand as they may not have tried the high quality brand at
An analysis of the percentage of subjects who chose the
any stage during the experiment.
quality brand on the final trial in the replication in both
awareness and no-awareness conditions shows a small differ-
Awareness Effect Is Mediated by Price
ence between the two groups (32% awareness condition cf
36% no-awareness condition). Both groups chose the high Price effects were not detailed in H&B (1990), in the replica-
quality brand around a third of the time which is the expected tion, high price had a (realistic) consistent effect to decrease
the chance of a brand being chosen (purchased). When thelevel owing to chance alone. H&B found a stronger, but again
Figure 2. Brand choice: Awareness no-quality condition. (n⫽31).
J Busn Res E. K. Macdonald and B. M. Sharp
Figure 3. Brand choice: No-awareness quality condition (n⫽136).
high-awareness brand was marked at a high price relative to The U-shaped curve observed in the over-all awareness-
quality condition is still present in Table 3. It is observedthe other two brands, the proportion of subjects choosing it
on each trial dropped markedly. Table 3 shows a comparison when subjects in this condition are split into those who saw
the high awareness brand marked at a high price and thosebetween over-all choice of the well-known brand in the aware-
ness-quality condition and the choice of a well-known brand that saw it marked at a low price.
in this condition when it was marked at a high price relative
to the other brands and when it was marked at a low price
relative to the other brands. Even when marked at a higher
relative price, the well-known brand was still chosen more A replication of Hoyer and Brown’s (1990 study exploring
the effects of brand awareness on consumer decision makingfrequently than either of the other two brands. This indicates
strong support for the brand awareness effect already observed. was carried out using a much larger sample (n⫽462 cf.
n⫽173). Not an exact replication, the study used the sameWhen the well-known brand was marked as one of the
two low-price brands, the number of subjects who chose it procedure but with a broader sample definition, a more realis-
tic test situation, a different product category, and a largeron each trial was very high (ranging from 93% on the first
trial to 82%). This indicates that brand awareness has a very sample.
The results showed strong support for two of the originalstrong effect on consumer choice for a common, repeat pur-
chase product; whereas, price has a moderate effect on choice. study’s findings: (1) that brand awareness is an important
choice tactic for consumers facing a new decision task; andIf the price of the well-known brand is high relative to compet-
ing brands, then, although a large number of consumers will (2) that subjects who are aware of one brand in a choice set
tend to sample fewer brands across a series of product trials.choose the well-known brand, some consumers will also be
inclined to try other cheaper brands. Thus, the awareness Only extremely weak support was found for H&B’s third
finding that, in the absence of awareness differentials amongeffect is stronger than the price effect.
Table 3. Choice of High-Awareness Brand: Awareness-Quality Condition
Trial Trial Trial Trial Trial
(n⫽265) 86% 77% 69% 73% 75%
(n⫽88) 72% 59% 50% 59% 60%
(n⫽177) 93% 85% 79% 80% 82%
Where Cottees is the “high-awareness” brand.
Brand Awareness Effects on Consumer Decision Making J Busn Res
brands, consumers are more likely to choose the highest qual-
Number of Trials
ity product; H&B also only found weak support. Five trials were used in the original study. A pretest carried
H&B’s hierarchy of effects derived hypothesis, that aware- out before the replication found that over five trials, 86.5%
ness would reduce in prevalence as a choice tactic over re- of subjects selected each brand at least once (n⫽37), and
peated trials, received mixed support with the choice of high the average number of brands chosen over the five trials was
awareness brand typically declining over the first few trials 2.811 (from a possible 3). Therefore, five trials was deemed
but then increasing over the latter trials. adequate for consumers to make a choice between three
Additional findings from this replication include: brands and to determine a preference. A not unreasonable
expectation would be that the majority of subjects would
1. that brand awareness seems to be an important choice experiment somewhat and then settle on the brand for which
tactic for consumers, even when facing a familiar or they had highest preference.
repeat choice task; and
2. although some consumers can be enticed to break their
habit using an awareness heuristic many show a ten-
dency to return to this habit.
In summary, brand awareness seems to play an important
part in explaining habitual choice patterns. The research re- To identify high-quality and low-quality brands, seven brands
sults fit with the observed empirical regularity that consumers were evaluated by a group of pretest subjects (n⫽49) in a
tend to maintain brand repertoires from which choice is made. blind taste test. As in the original study, these brands ranged
It would seem that the habit or inertia theories of consumer from presumably high-quality brands (i.e., national, well-
choice, based upon an understanding of the use of awareness advertised brands) to those of presumably lower quality (e.g.,
as a choice heuristic, offer considerable potential in explaining store and generic brands). Pretest subjects ranked these brands
low-involvement, repeat purchase patterns. For example, Eh- in order of preference.
renberg et al. (1990) use awareness differentials to explain
the widely observed double jeopardy pattern. Such explana-
Ranking as a Sensory Evaluation Technique
tions may prove superior to competing attitude or hierarchy Ranking samples as a sensory evaluation technique is accept-
of effect-based theories, which seem to be more appropriate able when aiming to compare samples according to a single
for a small subset of buying behavior, that is, high-involve- attribute, such as preference (see Meilgaard, Civille, and Carr,
ment, first purchase choice scenarios. 1991). It has the advantage of being a very simple technique,
but it does have these disadvantages: (1) the resultant data
are merely ordinal, and no measure of the degree of difference
Appendix A. Experimental Design
is obtained from each respondent; and (2) consecutive samples
that differ widely, as well as those that differ slightly, will be
separated by one rank unit.
Consideration Set Size
The purpose of this study was to examine consumer’s decision
Pretest for Quality
making under varying conditions of awareness and quality. For the purpose of this study, ranking of the cordial brands
Consumers usually make a decision from a set of brands they was considered acceptable. Each cordial brand was identified
recall from memory, and/or they recognize in their environ- by a letter from A–G. For the purposes of effective sensory
ment, known as their consideration set. In this experiment evaluation, samples should be presented to subjects in a bal-
the consideration set has been decided for the consumer so anced, random order (Meilgaard, Civille, and Carr, 1991).
that all the subject is required to do is choose from this limited In this study, to ensure there was no ordering effect, three
set of brands. In H&B’s study, the choice set included three conditions were designed, and in each condition, the seven
brands, and in the replication it was decided not to vary the brands were labeled differently.
number of brands. This is partly to allow for a more complete The use of three conditions also ensured that no profes-
replication of the original study, partly because a consideration sional letter effect could bias the results. A study by Ehrenberg
set of three was not unrealistic for cordial choice, and partly and Charlton (1973) found markedly different consumer re-
because with three independent variables employed in this peat buying for four brand names, even when the brand names
study (i.e., awareness, quality, and price) and three brands, were only letters of the alphabet: M, J, C, and V respectively.
the study requires 24 different variations in these factors. The By varying which brands were labeled by particular letters,
addition of one or more brands would quickly increase the we were able to avoid this problem. The results indicated no
complexity of the study. The opportunity exists for further significant preference for any one of the letters used.
The use of ranking data instead of rating data means thatresearch using larger consideration set sizes.
J Busn Res E. K. Macdonald and B. M. Sharp
Table 4. Testing for Low-Quality Brand
Lowest Rating Brand vs. Others Lowest Two Rating Brands vs. Others
Compare subjects who rated brand as Compare subjects who rated brand as
lowest in quality with other subjects) one of the two lowest with other subjects)
Brand (1 vs. 2–7) (1–2 vs. 2–7)
Cottees Country Blend Not significant Not significant
⫽0.2301, df ⫽2, ␣⫽0.89133) (
⫽0.1798, df ⫽2, ␣⫽0.91405)
(Low ⫽21, high ⫽26) (Low ⫽29, High ⫽18)
Coles Savings Cannot test (L ⫽8,H⫽39) Not significant
⫽0.7929, df ⫽2, ␣⫽0.67272)
(L ⫽21, H ⫽26)
Cottees Cannot test (L ⫽1,H⫽46) Cannot test (L ⫽9,H⫽38)
Farmland Cannot test (L ⫽2,H⫽45) Cannot test (L ⫽8,H⫽39)
Cottees All Natural Cannot test (L ⫽8,H⫽39) Cannot test (L ⫽14, H ⫽33)
Berri Cannot test (L ⫽2,H⫽45) Cannot test (L ⫽4,H⫽43)
Golden Circle Cannot test (L ⫽5,H⫽42) Cannot test (L ⫽9,H⫽38)
1 is the lowest quality ranking and 7 is the highest ranking.
a parametric statistical test, such as ANOVA, cannot be un- as the lowest quality brand overall). Coles Savings brand was
testable as one of the two lowest ranking brands with 21employed, because it would be violating the test’s underlying
assumptions: (1) normality; (2) independence; and (3) con- responses (45%).
stant variance. Therefore a nonparametric test, the chi-square
Golden Circle was ranked as the top
analysis, has been employed. The underlying assumption of quality brand by far, with 20 subjects rating it as the highest
chi-square analysis is that for a contingency table where ror quality brand, although this result is not significant (Table 5).
care greater than 2, then the sample size must be large enough Thirty subjects (64%) ranked it as one of the two top quality
so that there is a minimum sample size of 5 in each cell. brands. Berri was ranked as one of the top two brands by 24
Because of the small sample size in this study, it was necessary (51%) of subjects. The other five brands were ranked as one
to combine cells to carry out the test. This is acceptable, of the top two brands with much lower frequency.
because the purpose of this test is only to identify two extreme
brands—a high-quality brand and a low-quality brand.
No significant difference was found Although the quality ranking for none of the brands was
in terms of the low-quality rating for each brand versus the significant, by observation Golden Circle was ranked as highest-
other ratings it received (Table 4). Only Cottees Country Blend quality more frequently than any other brand. For this reason,
was ranked as the lowest-quality brand with enough frequency it was used as the high-quality cordial in the quality manipula-
tion in the experiment. Although, Cottees Country Blend wasto be testable (i.e., 21 (45%) (consumers ranked this brand
Table 5. Testing for High-Quality Brand
Highest Rating vs. Others Highest Two Ratings vs. Others
(Compare subjects who rated brand as (Compare subjects who rated brand as
highest in quality with other subjects) one of the two highest with other subjects)
Brand (1–6 vs. 7) (1–5 vs. 6–7)
Golden Circle Not significant Cannot test
⫽3.586, df ⫽2, ␣⫽0.16647 (Low ⫽17, high ⫽30)
(Low ⫽27, high ⫽20)
Berri Cannot test not significant
(Low ⫽38, high ⫽9)
⫽2.438, df ⫽2, ␣⫽0.29546
(Low ⫽24, high ⫽23)
Cottees Country Blend Cannot test (L ⫽45, H ⫽2) Cannot test (L ⫽39, H ⫽8)
Coles Savings Cannot test (L ⫽47, H ⫽0) Cannot test (L ⫽44, H ⫽3)
Cottees Cannot test (L ⫽44, H ⫽3) Cannot test (L ⫽37, H ⫽10)
Farmland Cannot test (L ⫽44, H ⫽3) Cannot test (L ⫽40, H ⫽7)
Cottees All Natural Cannot test (L ⫽37, H ⫽10) Cannot test (L ⫽34, H ⫽13)
1 is the lowest quality ranking, and 7 is the highest ranking.
Brand Awareness Effects on Consumer Decision Making J Busn Res
Hoyer, W. D.: An Examination of Consumer Decision Making for a
ranked as the lowest-quality brand overall, it was a much paler
Common Repeat Purchase Product. Journal of Consumer Research
orange color than many of the other brands and for this
11 (1984): 822–829.
reason, was not selected for the main experiment to avoid
Hoyer, W. D., and Brown, S. P.: Effects of Brand Awareness on
adding another distracting cue to the choice situation. There-
Choice for a Common, Repeat Purchase Product. Journal of Con-
fore, the brand ranked as one of the two lowest-quality brands
sumer Research 17 (1990): 141–148.
was selected, Coles Savings brand. Over the 401 subjects in
Lutz, R. J.: Editorial. Journal of Consumer Research 17 (1991): i–vii.
the quality-difference condition, each brand contained the
Macdonald, E., and Sharp, B.: Management Perceptions of the Impor-
high-quality product an approximately equal number of times.
tance of Brand Awareness as an Indication of Advertising Effective-
ness. Marketing Research On-Line 1 (1996): 1–15.
Meilgaard, M., Civille, G., and Carr, B.: Sensory Evaluation Techniques,
CRC Press, Boca Raton. 1991.
East, R: Consumer Behavior: Advances and Applications in Marketing,
Monroe, K.: On Replications in Consumer Research: Part 1. Journal
Prentice Hall, London. 1996.
of Consumer Research 19 (1992): i–ii.
Ehrenberg, A. S. C.: Repeat Buying Facts, Theory and Applications,Olshavsky, R. W., and Granbois, D. H.: Consumer Decision Mak-
Oxford University Press, New York. 1988. ing—Fact or Fiction? JCR 6 (1979): 93–100.
Ehrenberg, A. S. C., and Charlton, P.: An Analysis of Simulated Ray, M. L., Sawyer, A. G., Rothschild, M. L., Heeler, R. M., Strong,
Brand Choice. Journal of Advertising Research 13 (1973): 21–33. E. C., and Reed, J. B.: Marketing Communication and the Hierarchy
Ehrenberg, A. S. C., Goodhardt, G. J., and Barwise, T. P.: Double of Effects in New Models for Mass Communication Research, Sage,
Jeopardy Revisited. Journal of Marketing 54 (1990): 82–91. Newbury Park, CA. 1973, pp. 147–176.
Engel, J. F., Blackwell, R. D., and Kollat, D. T.: Consumer Behavior,Rosenthal, R.: in Replication Research in the Social Sciences, Neuliep,
Dryden, Hinsdale, IL. 1978. J. W., ed. Sage Publications, London. 1990, pp. 1–30.
Engel, J. F., Blackwell, R. D., and Miniard, P. W.: Consumer Behavior,Tybout, A. M., and Artz, N.: Consumer Psychology. Annual Review
Dryden Press, Fort Worth. 1993. of Psychology 45 (1994): 131–169.
Foxall, G. R.: The Behavioral Perspective Model of Purchase and Wilkie, W. L., and Dickson, P. R.: Shopping for Appliances: Consum-
Consumption: From Consumer Theory to Marketing Practice. ers’ Strategies and Patterns of Information Search. Marketing Sci-
ence Institute, Report No. 85–108, Cambridge, MA. 1985.Journal of the Academy of Marketing Science 29 (1992): 189–198.