Journal of Consumer Research, Inc.

Brand Credibility, Brand Consideration, and Choice

Author(s): Tülin Erdem and Joffre Swait
Source: Journal of Consumer Research, Vol. 31, No. 1 (June 2004), pp. 191-198
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/10.1086/383434 .
Accessed: 10/06/2014 13:16

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

The University of Chicago Press and Journal of Consumer Research, Inc. are collaborating with JSTOR to
digitize, preserve and extend access to Journal of Consumer Research.

http://www.jstor.org

This content downloaded from 86.0.148.195 on Tue, 10 Jun 2014 13:16:17 PM

All use subject to JSTOR Terms and Conditions
Brand Credibility, Brand Consideration, and
Choice
TÜLIN ERDEM
JOFFRE SWAIT*

We examine the role of brand credibility (trustworthiness and expertise) on brand

choice and consideration across multiple product categories that vary in regard to
potential uncertainty about attributes and associated information acquisition costs
and perceived risks of consumption. We find that brand credibility increases prob-
ability of inclusion of a brand in the consideration set, as well as brand choice
conditional on consideration. We also find that although credibility impacts brand
choice and consideration set formation more and through more constructs in con-
texts with high uncertainty and sensitivity to such uncertainty, credibility effects are
present in all categories. Finally, our results indicate that trustworthiness, rather
than expertise, affects consumer choices and brand consideration more.

O ne of the most important roles played by brands (un-

derstood to be “a name, term, sign, symbol or design,
or a combination of them which is intended to identify the
tential uncertainty about attributes and associated informa-
tion acquisition costs and perceived risks of consumption.
This work extends previous work on brand credibility by
goods and services of one seller or a group of sellers and explicitly considering the two subdimensions of brand cred-
to differentiate them from those of competitors”; Kotler ibility (trustworthiness and expertise).
1997, p. 443) is their effect on consumer brand choice and Our results indicate that brand credibility affects both
consideration. conditional brand choice and consideration. These effects
In this article, we propose and test that one important are found in a broad array of product classes we tested,
mechanism through which brands’ impact on choice and varying from simple fruit juices to personal computers
consideration materializes is via brand credibility. When im- (PCs). The products were selected to reflect different degrees
perfect and asymmetric information characterize a market, of consumer uncertainty with respect to product attributes
economic agents (i.e., consumers and firms) may use signals and associated information acquisition costs and to per-
(i.e., manipulable attributes or activities) to convey infor- ceived risks of consumption; this uncertainty is confirmed
mation about their characteristics (Spence 1974). To be ef- by this research to underlie the differential role of brand
fective, such signals must be credible (Tirole 1990). Pre- credibility in consideration and choice. In general, it is found
vious literature has studied the credibility of a brand as a that trustworthiness, the subdimension of brand credibility
signal of quality or product positions (Erdem and Swait relating to consumers’ perceptions of firms’ willingness to
1998; Rao and Ruekkert 1994; Wernerfelt 1988). The cred- carry through on promises made, exerts a more important
ibility of a brand as a signal (i.e., brand credibility) has been impact than expertise (firms’ perceived capability to deliver
conceptualized as the believability of the product position on promises) on respondents’ brand consideration and
information contained in a brand. choice. However, there is considerable variation in the rel-
While previous research has explored the impact of cred- ative importance of trustworthiness vis-à-vis expertise
ibility on product utility, we investigate brand credibility’s across product categories. The mechanisms whereby brand
effect on consideration and brand choice conditional on con- credibility works its impacts, that is, via perceived quality,
sideration. We also research the differential mechanisms perceived risk, and information costs saved constructs, are
through which credibility exerts its influence on choice and explored in some detail, but specific results are left for later
consideration across categories that vary in regard to po- sections.

*Tülin Erdem is E.T. Grether Professor of Business Administration and

Marketing, Haas School of Business, University of California, Berkeley, BRAND CREDIBILITY’S IMPACT ON
CA 94720-1900; e-mail: erdem@haas.berkeley.edu. Joffre Swait is partner, BRAND CHOICE AND CONSIDERATION
Advanis, Inc., and adjunct professor, Faculty of Business, University of
Alberta, 12 W. University Avenue, #205, Gainesville, FL 32601; e-mail:
Joffre_Swait@Advanis.ca.
Previous empirical work on how consumers may narrow
attention to a subset of brands out of a bigger set has focused
191

䉷 2004 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 31 ● June 2004

All rights reserved. 0093-5301/2004/3101-0016$10.00

This content downloaded from 86.0.148.195 on Tue, 10 Jun 2014 13:16:17 PM

All use subject to JSTOR Terms and Conditions
192 JOURNAL OF CONSUMER RESEARCH

on modeling consumer choices as the outcome of a two- promises will be kept and (2) demonstrate longer-term com-
stage process of consideration set formation and conditional mitment to brands (Klein and Leffler 1981). Furthermore,
brand choice (e.g., Andrews and Srinivasan 1995; Roberts it has also been shown that the clarity (i.e., lack of ambi-
and Lattin 1991; Swait and Ben-Akiva 1987). The main guity) of the product information contained in a brand is an
approach to conceptualizing consideration sets in the liter- antecedent to brand credibility (Erdem and Swait 1998).
ature has been the cost-benefit approach (Hauser and Wer- As also suggested by Aaker (1991), higher perceived (or
nerfelt 1990). This approach employs the expected utility expected) quality, lower information costs, and risk asso-
maximization framework to advance the notion that con- ciated with credible brands may increase consumer evalu-
sumers weigh the cost of brand evaluation for membership ations of brands. Indeed, Erdem and Swait (1998) have
in this subset against the benefits of adding or dropping the shown, using structural equation models, that expected util-
brand. This implies consumer uncertainty about brands. ity is increasing in perceived quality and decreasing in per-
ceived risk and information costs.
Brand Credibility
When consumers are uncertain about brands and the mar- Brand Credibility, Brand Consideration, and
ket is characterized by asymmetric information (i.e., firms Choice
know more about their products than do consumers), brands
can serve as signals of product positions (Wernerfelt 1988). Such a signaling framework of brand effects on con-
As a signal of product positioning, the most important char- sumer brand utility and choice also implies that when there
acteristic of a brand is its credibility. A firm can use various is consumer uncertainty about brands and information is
marketing mix elements besides the brand to signal product costly to obtain and/or process, the credibility of a brand
quality: for example, charging a high price, offering a certain may be an important factor underlying the formation of
warranty, or distributing via certain channels. Each of these consideration sets. The cost-benefit approach to consid-
actions may or may not be credible, depending on market eration and choice set formation suggests that the higher
conditions including competitive and consumer behavior. perceived value and lower perceived risk associated with
However, what sets brands apart from the individual mar- a higher credibility brand are anticipated to increase ex-
keting mix elements as credible signals is that the former pected benefits (Hauser and Wernerfelt 1990). Addition-
embody the cumulative effect of past marketing mix strat- ally, the lower information costs associated with credible
egies and activities. This historical notion that credibility is brands are likely to decrease expected costs, while the
based on the sum of past behaviors has been referred to as credibility of a brand decreases perceived risk because it
reputation in the information economics literature (see Her- increases consumers’ confidence in a firm’s product
big and Milewicz 1995). claims. Credibility also decreases information costs since
Credibility is broadly defined as the believability of an consumers may use credible brands as a source of knowl-
entity’s intentions at a particular time and is posited to have edge to save information gathering and processing costs
two main components: trustworthiness and expertise. Thus, (e.g., reading Consumer Reports, doing online searches for
brand credibility is defined as the believability of the product product reviews, seeking advice from experts; Erdem and
information contained in a brand, which requires that con- Swait 1998). Indeed, previous work has suggested that high
sumers perceive that the brand have the ability (i.e., exper- levels of cognitive effort needed to evaluate specific brands
tise) and willingness (i.e., trustworthiness) to continuously may induce negative affect and may lead to lower choice
deliver what has been promised (in fact, brands can function probabilities (Garbarino and Edell 1997). Consequently,
as signals since—if and when they do not deliver what is we expect that
promised—their brand equity will erode). Both the expertise H1: Higher credibility will both increase the proba-
and trustworthiness of a brand reflect the cumulative impacts bility of a brand being included in the consid-
of associated past and present marketing strategies and ac- eration set and the probability of its being chosen
tivities. The credibility of a brand has been shown to be from the consideration set.
higher for brands with higher marketing mix consistency
over time and higher brand investments, ceteris paribus However, there is evidence in the literature that different
(Erdem and Swait 1998). Consistency refers to the degree variables may affect consideration and brand evaluation con-
of harmony and convergence among the marketing mix ditional on consideration (Nedungadi 1990). Given that it
elements and the stability of marketing mix strategies and has been shown in previous research that brand credibility
attribute levels over time. The consistency of attribute lev- effects on consumer utility materialize through perceived
els over time—for example, consistency in quality lev- quality, perceived risk, and information costs saved (Erdem
els—implies low “inherent product variability” (Roberts and Swait 1998), the next question would concern the dif-
and Urban 1988), which can be achieved by a dedication ferential effects of these three brand credibility mediator
to quality standardization. However, the consistency to constructs on choice versus consideration.
which we refer is that of the brand positioning in general. More specifically, perceived risk can be used in the initial
Brand investments, on the other hand, are resources that screening before the brand choice stage. Furthermore, the
firms spend on brands to (1) assure consumers that brand impact of perceived risk may be much less, or even neg-

This content downloaded from 86.0.148.195 on Tue, 10 Jun 2014 13:16:17 PM

All use subject to JSTOR Terms and Conditions
BRAND CREDIBILITY AND BRAND CHOICE 193

TABLE 1

MEASUREMENT MODEL ITEMS AND ESTIMATED COEFFICIENTS

Construct Item Cronbach’s a

Expertise This brand reminds me of someone who’s competent and knows what .77
he/she is doing (+)
This brand has the ability to deliver what it promises (+)
Trustworthiness This brand delivers what it promises (+) .89
This brand’s product claims are believable (+)
Over time, my experiences with this brand have led me to expect it to
keep its promises, no more and no less (+)
This brand has a name you can trust (+)
This brand doesn’t pretend to be something it isn’t (+)
Perceived Quality The quality of this brand is very high (+) .77
In terms of overall quality, I’d rate this brand as a . . . (+)
Perceived Risk I’d have to try it several times to figure out what this brand is like (+) .64
I never know how good this brand will be before I buy it (+)
Information Costs Saved I need lots more information about this brand before I’d buy it (⫺) .75
I know what I’m going to get from this brand, which saves time shop-
ping around (+)
I know I can count on this brand being there in the future (+)
This brand gives me what I want, which saves me time & effort trying
to do better (+)
NOTES.—All scales formed as simple averages of component items, reverse scored for negative items. All items measured on nine-point agree/disagree scales
except second quality item, which was measured on nine-point scales with 1 p low quality, 9 p high quality. Signs (+/⫺) indicate a priori sign expectation.

ligibly, important at the conditional brand choice stage due Brand Credibility’s Impact on Choice Processes
to the exclusion of high-risk brands from the consideration and Product-Category Specific Factors
set during the earlier screening stage of the choice process.
Therefore, we expect In contexts where uncertainty levels and sensitivity to
uncertainty are higher (e.g., PCs, as compared to orange
H2: Brand credibility’s impact through perceived risk juice), one would expect credibility to have a greater impact
to be found mainly at the consideration set for- on consumer choice processes; additionally, credibility’s im-
mation stage, rather than at the brand choice con- pact through perceived risk and information costs saved,
ditional on consideration stage. over and above its impact through perceived quality, should
be more pronounced.
Furthermore, quality and price trade-offs are more likely
to be made explicitly at the brand choice stage rather than EMPIRICAL TEST
at the choice set formation stage (Nedungadi 1990). There- In the data collection for this study we have used six
fore, in the context of brand choice conditional on consid- product classes: athletic shoes, cellular telecommunications
eration set, we expect services, headache medication, juice, personal computers,
and hair shampoo. The survey we fielded at a major North
H3: Brand credibility’s impact through perceived American university had eight different versions, each cov-
quality to be more pronounced on brand choice ering three of the six product classes just enumerated. In
conditional on consideration set, rather than on each product class, five brands were presented, selected so
consideration set formation. that the brand set represents a wide range of market share.1
For each brand, respondents were asked to provide responses
The third construct following brand credibility, infor- to the items listed in table 1 for the Trustworthiness, Ex-
mation costs saved, can have the same screening function pertise, QUAL (Perceived Quality), RISK (Perceived Risk),
as perceived risk at the consideration set formation stage. and ICS (Information Costs Saved) constructs. These items
However, in most product categories in which potential in- are directly taken from Erdem and Swait (1998), who val-
formation costs are above a certain threshold, or there is idated the desired scales; we employed the same nine-point
relative brand uncertainty (Moorthy, Ratchford, and Taluk- agree/disagree scales used by them. In addition, three out-
dar 1997), information costs saved are likely to continue to
1
be important at the brand choice stage. Hence, one needs Brands utilized were for athletic shoes: Adidas, Asics, New Balance,
to test empirically whether brand credibility’s impact Nike, Reebok; cellular service providers: AT&T, Cingular, Nextel, Sprint
PCS, Verizon; headache medication: Advil, Bayer, Excedrin, Motrin IB,
through information costs saved can be found at both the Tylenol; juice: Dole, Minute Maid, Sunkist, Tropicana, Welch’s; personal
consideration set formation stage and brand choice condi- computers: Apple, Compaq, Dell, Gateway, IBM; hair shampoo: Alberto
tional on consideration set. VO5, Herbal Essence, Finesse, Pantene Pro V, Suave.

This content downloaded from 86.0.148.195 on Tue, 10 Jun 2014 13:16:17 PM

All use subject to JSTOR Terms and Conditions
194 JOURNAL OF CONSUMER RESEARCH

TABLE 2

BRAND CONSIDERATION BINARY LOGISTIC MODELS

Athletic Cellular Headache Personal

shoes provider medication Juice computer Shampoo

A. Trustworthiness/Expertise models:
Brand 1 1.44** 2.50** 2.89** 2.38** .41 .94**
Brand 2 1.03** 1.09** .95** 2.19** 1.18** 1.35**
Brand 3 2.09** .48* 1.10** .71** 2.50** .86**
Brand 4 2.68** .77** 1.45** 1.52** 1.25** 1.73**
Brand 5 .24 1.21** 2.53** 1.41** 1.56** .77**
Trustworthinessa 1.07** 1.12** .96** .83** 1.56** .73**
Expertiseb .63** .70** .72** .77** .38* .62**
Trust # Uncerc ⫺.48 ⫺.06 ⫺1.02 ⫺.06 ⫺1.14 .22
Exper # Uncer .66 ⫺.52 1.38** ⫺.35 1.14** ⫺.12
Trust # Uncer # Famd .55 .12 ⫺1.23 .38 1.88** ⫺.94
Exper # Uncer # Fam ⫺.42 .13 ⫺.46 .94 ⫺1.71 ⫺.24
McFadden’s r 2 .47 .45 .49 .42 .48 .38
B. Perceived Quality/Risk/Information
Cost models:
Brand 1 1.37** 2.28** 2.39** 2.23** .53* .97**
Brand 2 1.35** 1.38** 1.06** 2.28** 1.41** 1.09**
Brand 3 2.16** .49 1.26** 1.07** 2.45** 1.02**
Brand 4 2.09** .77** 1.72** 1.68** 1.04** 1.35**
Brand 5 .03 1.19** 1.72** 1.53** 1.35** 1.07**
Perceived Qualitye .77** 1.17** 1.56** 1.44** 1.09** .88**
Perceived Riskf ⫺.23* .08 ⫺.18 .06 ⫺.28** ⫺.09
Information Costs Savedg .88** 1.02** .26 ⫺.2 .62** .62**
McFadden’s r 2 .45 .47 .47 .46 .44 .36
a
Trustworthiness score calculated as the average on nine-point agree/disagree scales of five Trustworthiness items from table 1. This is then centered across
brands in the product category, within each subject.
b
Expertise score calculated as the average on nine-point agree/disagree scales of two Expertise items from table 1. This is then centered across brands in the
product category, within each subject.
c
Uncertainty score is calculated as the sum of “Whenever one chooses in this product category, one never really knows whether that is the one that should have
been bought” and “When one makes purchases in this product category, one is never certain of one’s choice,” where association of the statement with the product
category results in a score of one, and zero otherwise. This is then centered across product categories, within each subject.
d
Familiarity score is defined as one if “I consider myself familiar with this product category” is associated with the product class, and zero otherwise.
e
Perceived Quality is calculated as the average of two Perceived Quality items in table 1, centered within subject.
f
Perceived Risk is calculated as the average of two Perceived Risk items in table 1, centered within subject.
g
Information Costs Saved is the average of four Information Costs Saved items in table 1, centered within subject.
*p ! .05.
**p ! .01.

come measures for each brand were elicited: consider- classes, our final sample sizes are 83 for athletic shoes, 84
ation—“Assuming prices are the same, I would seriously for cellular services, 83 for headache medication, 83 for
consider buying this brand. (Check all that apply.)”; juice, 82 for personal computers, and 83 for shampoo. All
choice—“Assuming prices are the same, this is the one brand respondents received course credit as incentive for partici-
I would be most likely to buy. (Check one only.)”; allo- pation.
cation—“Assuming prices are the same, if you were to buy Construct estimates were obtained as follows for each
10 [units of . . .] at one time, how many of each brand brand, product class, and individual: an average score over
would you buy?” (Both choice and allocation measures were the relevant items (as given in table 1) was calculated for
included for convergent validity purposes. We will report each product class, brand, and individual, and then centered
only on the choice results, since the allocation models mirror about the mean brand score of the individual in the product
them closely.) Note the ceteris paribus equal price condition: class. For completeness we report in table 1 the Cronbach’s
this was done to permit the omission of price (unknown to alpha measures for the constructs thus defined; all constructs
us and the participants) from subsequent analyses. Respon- have alpha measures over .6. The definitions of Familiarity
dents also indicated whether they considered themselves fa- and Uncertainty, as well as the other constructs, are detailed
miliar or not with each of the six product categories, as well in the notes of table 2. Centering of constructs is done to
as whether or not they viewed their choices in each of the eliminate ambiguity of interpretation: a significant construct
product categories as being uncertain or not (i.e., the degree estimate can only mean that within-respondent construct var-
of uncertainty about having made a good selection they iation underpins the result.
associate with a choice in a product category). Table 2 presents two sets of models, per figure 1 (and
With each respondent providing data on three product Erdem and Swait 1998), wherein Trustworthiness and Ex-

This content downloaded from 86.0.148.195 on Tue, 10 Jun 2014 13:16:17 PM

All use subject to JSTOR Terms and Conditions
BRAND CREDIBILITY AND BRAND CHOICE 195

FIGURE 1 Uncertainty # Familiarity). These last results reflect the

STRUCTURAL RELATIONSHIPS BETWEEN BRAND
intuition that individual difference variables (Uncertainty
CREDIBILITY, PERCEIVED QUALITY, PERCEIVED RISK, AND and Familiarity, among many others, we are sure) are likely
INFORMATION COSTS SAVED to help establish boundary conditions for the applicability
of the theory presented.
It is noteworthy that in all product categories the main
effects of the centered Trustworthiness and Expertise con-
structs are statistically significant, indicating that the prin-
cipal hypothesis 1 holds generally but becomes stronger or
weaker under certain conditions. Note again that the results
hold due to brand credibility differences at the individual
respondent level.
Table 2B also presents the logistic regression results for
brand consideration as a function of QUAL, RISK, and ICS.
These models are interesting to examine because they in-
dicate how brand credibility may work in impacting brand
consideration. (These models do not include individual dif-
ference variables since their main intent is to explore the
means whereby credibility impacts are generated.) We find
that QUAL consistently shows up as a statistically signifi-
cant explanator of brand consideration and is the most im-
pactful variable of the three constructs in all product cate-
pertise are posited as antecedents to QUAL, RISK, and ICS. gories save athletic shoes. The RISK model is statistically
Thus, in turn, the binary logistic regressions relate brand significant in athletic shoes and PCs, marginally significant
consideration (yes/no) to the two sets of constructs in each in headache medication, and not significant in juice, cellular
of the six product classes: (1) Trustworthiness and Expertise, services, or hair shampoo. Finally, ICS is statistically sig-
the subcomponents of brand credibility, interacted with Fa- nificant in athletic shoes, cellular services, PCs, and hair
miliarity and Uncertainty, and (2) QUAL, RISK, and ICS,
shampoo; thus, increased brand credibility allows these sub-
the downstream constructs reflecting brand credibility. Our
jects to simplify decision making by saving time and only
motivation in interacting Familiarity and Uncertainty (about
considering brands with high ICS (i.e., implying high Trust-
choices) with the core Trustworthiness and Expertise con-
worthiness and/or Expertise as well as low RISK). These
structs is to aid our understanding about limits that may hold
combined results seem to suggest that overall perceived risk
to the main effects being considered here. We will explain
these in more detail subsequently. and information costs saved play a bigger role in categories
Let us now examine the Trustworthiness/Expertise results with higher uncertainty and sensitivity to uncertainty.
in table 2A. Clearly, across all product classes, as each of Table 3 presents estimation results for multinomial logit
these credibility components increases, so does the probability models of the choice response variable. Note that all these
that the brand is considered by the respondent. In all product models have choice sets that include only those brands that
classes, estimated coefficients for the main effects of these the individual respondents indicated they would consider
two constructs are significantly different from zero at the 95% buying, hence they are conditional on stated consideration.
confidence level. In this population of respondents, under- This seemed to us a reasonable procedure to help isolate
graduate university students, it seems that Trustworthiness is the brand credibility effects that are found to those that occur
generally more impactful than Expertise perceptions; this is during product evaluation and not brand consideration (al-
particularly notable in PCs, where the main effect of Trust- ready captured in the models discussed in table 2). This
worthiness is about three times more important than that of does, however, imply that the choice sets defined for the
Expertise in terms of increasing brand consideration proba- choice models in table 3 contain far fewer observations than
bilities (using log odds ratios as the comparison metric). In were available for the consideration analyses presented in
two product classes, Headache Medication and Personal Com- table 2 (each respondent provided one choice observation but
puters, the impact of Expertise on consideration probabilities five consideration observations in each product category).
is moderated by Uncertainty: the more uncertain a respondent Table 3A generally demonstrates that brand credibility
perceived him/herself to be about choices made in these prod- (via Trustworthiness and Expertise) significantly impacts the
uct categories, the greater role their brand Expertise percep- stated choice behavior of our respondents across all product
tion had in determining consideration. No systematic inter- classes used in the study, with the exception of the shampoo
action between Uncertainty and Trustworthiness was found product class. As mentioned above, due to smaller sample
in any of the product classes. However, the role of Trust- sizes in the choice models compared to the consideration
worthiness is moderated by Uncertainty among respondents models of table 2, several of the Trustworthiness and Ex-
that were familiar with the PC category, as captured through pertise main effect coefficients are not significantly different
the significant third-order interaction (Trustworthiness # from zero, but all are directionally correct. Almost none of

This content downloaded from 86.0.148.195 on Tue, 10 Jun 2014 13:16:17 PM

All use subject to JSTOR Terms and Conditions
196 JOURNAL OF CONSUMER RESEARCH

TABLE 3

BRAND CHOICE MULTINOMIAL LOGIT MODELS (CONDITIONAL ON STATED CONSIDERATION)

Athletic Cellular Headache Personal

shoes provider medication Juice computer Shampoo

A. Trustworthiness/Expertise models:
Brand 1 1.47** .64* .12 ⫺.27 ⫺3.86** .24
Brand 2 .32 ⫺.15 ⫺1.12 ⫺.09 ⫺1.07 .97
Brand 3 1.44** ⫺.42 0 ⫺1.36 .73* .23
Brand 4 1.36** ⫺.38 .38 ⫺.79 ⫺1.55* 1.14
Brand 5 0 0 0 0 0 0
Trustworthinessa .54 .33 1.37** 2.48** 2.14** .64
Expertiseb .72** 1.15** .73 1.02 .17 .96
Trust # uncerc ⫺.69 .66 ⫺.14 ⫺.94 ⫺3.67* .46
Expert # uncer .58 ⫺1.06** .5 ⫺.24 6.31 .02
Trust # uncer # famd 1.18 ⫺1.18* .16 2.11 2.29 ⫺1.07
Exper # uncer # fam .73 1.07** ⫺.05 1.1 ⫺4.76 ⫺.07
McFadden’s r 2 .27 .27 .39 .37 .55 .29
B. Perceived Quality/Risk/Information
Cost models:
Brand 1 1.95** .60 .08 ⫺.14 ⫺2.03 .65
Brand 2 .80 .25 ⫺.70 .23 ⫺.27 .64
Brand 3 1.92** ⫺.60 .17 ⫺.77 1.17** .04
Brand 4 1.66** ⫺.20 .43 ⫺.12 ⫺.76 .80
Brand 5 0 0 0 0 0 0
Perceived Qualitye 1.19** 1.29** .93** 3.09** 1.53** 1.55**
Perceived Riskf ⫺.07 .07 ⫺.29 .19 ⫺.53 ⫺.35
Information Costs Savedg .69** ⫺.20 1.09** .29 1.56** .65
McFadden’s r 2 .34 .36 .36 .58 .59 .41
a
Trustworthiness score calculated as the average on nine-point agree/disagree scales of five Trustworthiness items from table 1. This is then centered across
brands in the product category, within each subject.
b
Expertise score calculated as the average on nine-point agree/disagree scales of two Expertise items from table 1. This is then centered across brands in the
product category, within each subject.
c
Uncertainty score is calculated as the sum of “Whenever one chooses in this product category, one never really knows whether that is the one that should have
been bought” and “When one makes purchases in this product category, one is never certain of one’s choice,” where association of the statement with the product
category results in a score of one, and zero otherwise. This is then centered across product categories, within each subject.
d
Familiarity score is defined as one if “I consider myself familiar with this product category” is associated with the product class, and zero otherwise.
e
Perceived Quality is average brand scores calculated using items in table 1, which were then centered at the subject-brand level.
f
Perceived Risk, average brand scores calculated using items in table 1, which were then centered at the subject-brand level.
g
Information Costs Saved is average brand scores calculated using items in table 1, which were then centered at the subject-brand level.
*p ! .05.
**p ! .01.

the Uncertainty and Familiarity interactions with the main and 3, by product class. For instance, table 2A indicates that
constructs are significant, again likely due to sample size. consideration probabilities increase in Expertise for juice;
Table 3B also yields some information about the likely table 3A, on the other hand, shows that conditional choice
routes whereby brand credibility works its impact on final probabilities in this category are essentially unaffected by
choice via QUAL, RISK, and ICS. Across the six product Expertise. This may be a sign that in this population seg-
classes we examined, QUAL is always a statistically sig- ment, and for this product class, perceptions of Expertise
nificant determinant of choice; ICS is found to be statisti- are helpful in getting the brand to be considered, but Trust-
cally significant in athletic shoes, headache medication, and worthiness is likely the sine qua non of final choice. How-
PCs; and RISK is not found to be statistically significant in ever, the simple models underlying tables 2 and 3 make such
any of the product classes. These results seem again to sug- interpretations something of a guess, mainly because there
gest that information costs saved affects the brand choice is no explicit relationship between consideration, choice set
stage more in categories with relatively higher levels of formation, and final choice.
uncertainty and sensitivity to such uncertainty. That RISK We have presented above strong statistical evidence,
is not influential in the conditional brand choice stage in- based on simple models of consideration and overall choice,
dicates that this construct’s impact is likely limited to the that brand credibility works its impact onto these stages of
consideration set formation stage, consistent with hypothesis the choice process through quality and risk perceptions, as
2. However, it must be noted that the results from this table well as through perceptions of information costs saved (i.e.,
are not very reliable due to small sample sizes, as we said decision-making costs). The results from the empirical study
above. indicate that these results are likely to be found in a relatively
Interesting comparisons can be made between tables 2 wide range of product classes; they also show that there is

This content downloaded from 86.0.148.195 on Tue, 10 Jun 2014 13:16:17 PM

All use subject to JSTOR Terms and Conditions
BRAND CREDIBILITY AND BRAND CHOICE 197

significant heterogeneity in the overall mechanisms whereby pirical study indicates that neither perceived risk nor infor-
the impact of brand credibility ultimately works itself out. mation costs seem to matter in consumer choice processes.
For example, we found that brand consideration in the athletic However, we also found that credibility affects consumer
shoes category is affected by all three constructs (QUAL, choices through perceived risk, information costs saved, and
RISK, and ICS), whereas RISK does not seem to have an perceived quality in most categories, even those with only
impact on the consideration of different cellular providers. moderate levels of uncertainty. This result is found to hold
Except for juice, a very low uncertainty and low sensitivity at the individual respondent level, indicating that it is brand
to uncertainty (low-risk) category, brand credibility affects credibility differences that are driving consumer behavior.
consideration set formation (as well as brand choice); RISK Finally, we also found some evidence for stronger credibility
seems to be a factor only at the consideration set stage; if impacts for individuals who perceive higher uncertainty
ICS plays a role, it does so in both decision stages; and QUAL when choosing in a given product category.
effects are relatively more pronounced at the brand choice The simple analysis methods employed in this article
stage. Finally, RISK determines more strongly the consid- are unable to cleanly isolate and attribute brand credibility
eration set stage in high-risk categories (PC and athletic impacts between choice process stages. Future research
shoes); ICS affects all stages of choice processes in a broader should examine brand credibility impacts using structural
range of products but affects choice processes less in low- choice set formation models to allow a purer attribution of
risk categories (juice and to lesser extent headache medica- impacts, leading to a better understanding of the mechanisms
tion). These results are consistent with our expectations in whereby brand credibility impacts choice stages. Future re-
regard to how potential uncertainty and sensitivity to it across search should also extend our analyses to explore choice dy-
categories may moderate credibility impacts. namics and thus explain the processes by which credibility
While table 2A supports somewhat the contention that and consideration set formation evolve over time. Addition-
the differentiation of credibility impacts is likely to be a ally, and of great practical interest, more detailed analysis of
function of individual differences (herein, familiarity with individual level and product category specific moderators of
and perceptions of uncertainty with respect to making good credibility effects should be conducted.
choices in the product category), the most notable insight
afforded by that analysis is that across all product classes [Dawn Iacobucci and David Glen Mick served as editors
examined the main hypothesis holds: brand credibility and Joel Huber served as associate editor for this
(Trustworthiness and Expertise) is an important determinant article.]
of brand consideration. The interactions of the credibility
components with Familiarity and Uncertainty indicate that
brand credibility will play an even greater role in deter- REFERENCES
mining consideration for certain product categories (here, Aaker, David A. (1991), Managing Brand Equity, New York: Free
headache medication and PCs) and for certain individuals Press.
(e.g. high Uncertainty and Familiarity). This greater role is Andrews, Rick and T. C. Srinivasan (1995), “Studying Consid-
achieved not only by making brand credibility overall a more eration Effects in Empirical Choice Models Using Scanner
important determinant of consideration but also by shifting and Panel Data,” Journal of Marketing Research, 32 (1),
the relative importance of Trustworthiness and Expertise 30–41.
within the overall credibility impact. For example, in PCs Erdem, Tülin and Joffre Swait (1998), “Brand Equity as a Signaling
the main effects of these component constructs indicate that Phenomenon,” Journal of Consumer Psychology, 7 (April),
on average Trustworthiness is about three times more im- 131–57.
Garbarino, Ellen C. and Julie A. Edell (1997), “Cognitive Effort,
portant than Expertise in determining brand consideration; Affect, and Choice,” Journal of Consumer Research, 24 (Sep-
however, for individuals that associate low Uncertainty and tember), 147–58.
have Familiarity with the product class, Trustworthiness is Hauser, John and Birger Wernerfelt (1990), “An Evaluation Cost
only half as important as Expertise perceptions in deter- Model of Consideration Sets,” Journal of Consumer Re-
mining consideration probabilities. search, 16 (December), 393–408.
Herbig, Paul and John Milewicz (1995), “The Relationship of Rep-
DISCUSSION utation and Credibility to Brand Success,” Journal of Con-
sumer Marketing, 14 (Winter), 5–10.
In this article we present evidence for brand credibility’s Klein, Benjamin and Keith B. Leffler (1981), “The Role of Market
effect on the formation of consideration sets over and above Forces in Assuring Contractual Performance,” Journal of Po-
its effect on brand choice. We also shed light on the mech- litical Economy, 89 (Fall), 615–39.
anisms through which brand credibility effects materialize Kotler, Philip (1997), Marketing Management, 7th ed., Englewood
Cliffs, NJ: Prentice-Hall.
in the consideration set formation and brand choice stages. Moorthy, Sridhar, Brain T. Ratchford, and Debebrata Talukdar
In regard to the two subdimensions of credibility, we also (1997), “Consumer Information Search Revisited: Theory and
found that trustworthiness, rather than expertise, has the Empirical Analysis,” Journal of Consumer Research, 23
bigger impact on consumer choices. (March), 263–77.
Furthermore, we established some boundary conditions Nedungadi, Prakesh (1990), “Recall and Consumer Consideration
to our results. In the juice category, for example, our em- Sets: Influencing Choice without Altering Brand Evalua-

This content downloaded from 86.0.148.195 on Tue, 10 Jun 2014 13:16:17 PM

All use subject to JSTOR Terms and Conditions
198 JOURNAL OF CONSUMER RESEARCH

tions,” Journal of Consumer Research, 17 (December), Spence, Michael (1974), Market Signaling: Informational Transfer
263–76. in Hiring and Related Screening Processes, Cambridge, MA:
Rao, Akshay and Robert W. Ruekkert (1994), “Brand Alliances as Harvard University Press.
Signals of Product Quality,” Sloan Management Review, 36 Swait, Joffre and Moshe Ben-Akiva (1987), “Incorporating Random
(Fall), 87–97. Constraints in Discrete Choice Models of Choice Set Gener-
Roberts, John and Jim Lattin (1991), “Development and Testing ation,” Transportation Research, 21B (April), 91–102.
of a Model of Consideration Set Composition,” Journal of Tirole, J. (1990), The Theory of Industrial Organization, Cam-
Marketing Research, 28 (November), 429–40. bridge, MA: MIT Press.
Roberts, John and Glenn Urban (1988), “Modeling Multiattribute Wernerfelt, Birger (1988), “Umbrella Branding as a Signal of New
Utility, Risk and Belief Dynamics for New Consumer Durable Product Quality: An Example of Signalling by Posting a
Brand Choice,” Management Science, 34 (February), 167–85. Bond,” Rand Journal of Economics, 19 (Autumn), 458–66.

This content downloaded from 86.0.148.195 on Tue, 10 Jun 2014 13:16:17 PM

All use subject to JSTOR Terms and Conditions