developing and assessing the reliability and validity of
TRANSCRIPT
Developing and Assessing the Reliability andValidity of an Alternative Scale to Measure Brand Equity
by Arga Hananto
AbstractThis paper attempts to propose an alternative scale to measure brand equity. Using Aaker's conception of brand equity, the scale is constructedby elaborating on each of the four dimensions related with consumerperceptions (i.e. Brand awareness, brand associations, perceived qualityand brand loyalty).The reliability and validity of the proposed scale is assessed using twoways. First, Cronbach's alpha for the proposed scale is computed usingSPSS. It is then compared to Cronbach's alpha computed for scaledeveloped by Yoo and Donthu (2001), which was also administered torespondents. Second, mean and standard deviation for both scales arecomputed. Then means for both scales are correllated in order to assessthe proposed scale's convergent validity.Result on small sample of 20 respondents suggest that the proposedscale exhibit reliability and validity. The Cronbach’s alpha generated bySPSS has proven that items in the author’s scale have relatively goodreliability. The convergent validity has been proven by the existence ofsignificant correlation between his scale and Yoo and Donthu’s scale. Thelatter itself has been validated a couple of times (Atilgan, 2005;Washburn and Plank, 2002).
Introduction
Brand Equity is one of popular topics in marketing. Brand equity is
considered important because it represents the power of brands in the
market (Keller, 2003). Higher brand equity is proposed to increase brand
preference and purchase intention (Cobb-Walgren,1995); increase
interpretation/information processing, use satisfaction, a base to charge
premium price, and source of competitive advantage for firms (Aaker,
1991, quoted in Keller, 2003).
Inspite of agreement amongst researchers about the value of brand
equity, there is still no consensus regarding its conceptualisation and
measurement.
1
This paper aims to propose an alternative scale to measure brand equity
based on conception by Aaker (1991, quoted in Yoo and Donthu, 2001).
The proposed scale is then compared to other scale adapted from Yoo and
Donthu (2001) in order to compare proposed scale’s reliability and
validity.
Before we proceed to discuss the scale development, we need to revisit
some conceptions of brand equity and its components.
Consumer-Based Brand Equity
According to Aaker (1991, quoted by Washburn and Plank, 2002), brand
equity is defined as “a set of brand assets and liabilities linked to a
brand that add to or subtract from the value provided by a product or
service to a firm and/or to that firm’s customers.” Keller (2003), defines
brand equity as “the differential effect that brand knowledge has on
consumer response to the marketing of that brand. Their conceptual
framework of brand equity is widely used throughout academic literature
about brand equity
Based on Aaker’s conceptualisation (1991, quoted in Yoo and Donthu,
2001), brand equity construct consists of five dimensions, namely brand
awareness, brand association, perceived quality, brand loyalty and other
proprietary brand assets. Keller (2003) on the other hand proposes two
dimensions, namely brand awareness and brand image.
In this paper, Aaker’s conceptualisation of brand equity is used to develop
the proposed scale. Therefore we will look at each of the five dimensions
more thoroughly.
1. Brand awareness
Brand awareness deals with the strength of a brand’s presence in
consumer’s mind. Aaker (1991, quoted in Atilgan, et. al ,2005) defines
brand awareness as” the ability of potential buyers to identify and
recall that a brand is a member of a certain product category.” Keller
2
(2003) proposes that brand awareness consists of brand recognition
and recall. Whereas brand recognition is defined as the ability of
consumers to correctly identify that he/she has heard of the brand
before, brand recall deals with consumer’s ability to retrieve the brand
from memory when given certain cues related to product category,
needs, and purchase or usage situation (Keller, 2003).
2. Brand associations
Aaker defines brand associations as “anything linked in memory to a
brand (Aaker 1991, quoted in Washburn and Plank 2002). Brand
associations are believed to contain the meaning of the brand for
consumers (Keller 1993). Strong and favourable brand associations
contribute toward brand equity. Aaker (1991, quoted in Atilgan, 2005)
suggests that brand association creates value for customers as well as
marketers because brand associations help consumer process
information, differentiate an offering, provide reason to buy and a
basis for extension.
3. Perceived quality
Perceived quality is defined as consumer’s subjective evaluation of
overall excellence of a product (Zeithaml, 1988). Aaker (1996)
suggest that perceived quality is one of key dimensions to measure
brand equity. A report by Total Research (quoted in Aaker, 1996)
suggests that it is related with price premiums, price elasticity, brand
usage and stock return.
4. Brand loyalty
Brand loyalty is defined by Aaker (1991, quoted by Pappu, et. al, 2005)
as “the attachment that a customer has to a brand. Aaker (1996)
argues that brand loyalty is the core of brand equity as it provides
barrier to entry, a basis for price premium, gives time to respond to
competitor action, as well a safeguard against price competition.
5. Other proprietary brand assets
According to Aaker (1991, quoted in Keller, 2003), other proprietary
brand assets include trademarks, patents, and channel relationships.
3
Scale Development Process
The proposed scale was developed based on Aaker’s conceptualisation of
brand equity, consisting of brand awareness, brand association, perceived
quality and brand loyalty dimensions. Since the fifth dimension proposed
by Aaker is not based on consumer perception, it is not used in the
measurement.
Definitions and discussion of each dimension in various journal articles
were used to develop the scale. Eventhough the author has attempted to
come up with his own scale, but sometimes it is unavoidable to totally
differentiate the question(s) from previously developed scales by other
researchers.
The scale consists of 6 items measuring brand awareness, 16 items
measuring brand association, 6 items measuring perceived quality and 4
items measuring brand loyalty. These items are then compared to a scale
developed by Yoo and Donthu (2001) which was validated by Washburn
and Plank (2002) and Atilgan, et. al (2005).
Both the author’s own scale as well as the scale adapted from Yoo and
Donthu’s are measured using 7 points ordinally interval scale (a modified
Likert scale, see Hair et al, 2003). The scale descriptors are anchored at
1= “strongly disagree” and 7 = “strongly agree”. The original scale
reported in their journal article was measured using 5 items Likert scale
anchored at 1= “strongly disagree” and 5 = “strongly agree”.
A list of items used in the scale as well as scale adapted from Yoo and
Donthu (2001) is presented below.
Table 1
List of items developed by the author
Brand Awareness1. Toothpaste X is the name that comes into my mind when I think of toothpaste brand2. I have heard of Toothpaste X before3. I can easily find Toothpaste X in a supermarket aisle4. I can quickly identify Toothpaste X in a supermarket rack5. When you see a toothpaste tube with blue, italic font, you think of Toothpaste X
4
6. When it comes to purchase toothpaste, Toothpaste X comes up in my mind firstBrand Association1. Toothpaste X gives me performance worth the money I spent to buy it2. Toothpaste X has better performance than other toothpaste brands3. Toothpaste X has better performance than other toothpaste brands4. I have a clear idea about what kind of people who use Toothpaste X5. Toothpaste X is associated with friendliness6. Toothpaste X is associated with sincerity7. Toothpaste X is associated with family values8. Toothpaste X is associated with excitement9. Toothpaste X is very reliable10.Toothpaste X really delivers its promise11. Toothpaste X is associated with sophistication12. Toothpaste X is associated with strength13. I’m proud to buy Toothpaste X 14. I know the company that produce Toothpaste X15. I like the company that produce Toothpaste X16. Toothpaste X comes from a company with good reputationPerceived Quality1. Toothpaste X is high quality toothpaste2. Compared to other brands of toothpaste, Toothpaste X performs better3. For the price I pay for Toothpaste X, I get good quality4. The brand name of Toothpaste X signals good quality5. The price of Toothpaste X reflects its quality6. Judging from the frequency of Toothpaste X advertisement in the media, Toothpaste X
must be of good qualityBrand Loyalty1. I would feel happy to buy other brand of toothpaste next time*2. I feel committed to Toothpaste X3. It is very likely that I will buy Toothpaste X next time I need to buy toothpaste4. I prefer Toothpaste X compared to other brands of toothpaste
* denotes reverse-scored item
Table 2
List of Items Adapted from Yoo and Donthu (2001)
Brand awareness/associationsI can recognize Toothpaste X among other competing brandsI am aware of Toothpaste XSome characteristics of Toothpaste X come to my mind quicklyI can quickly recall the symbol or logo of Toothpaste XI have difficulty in imagining Toothpaste X in my mind*.Perceived QualityThe likely quality of Toothpaste X is extremely highThe likelihood that Toothpaste X would be functional is very highBrand LoyaltyI consider myself to be loyal to Toothpaste XToothpaste X would be my first choiceI will not buy other brands if Toothpaste X is available at the store.
• denotes reverse scored item
•
Object measured is a well-known brand of toothpaste in Indonesia
produced by a multinational company. For reporting purpose, the brand
name is disguised as “Toothpaste X.”
5
Operationalisation of dimensions used to measure brand equity
1. Brand Awareness
Brand awareness is measured by brand recognition only. Aaker (1996)
suggest that measurement of brand recall may not be convenient to
administer in a survey. Other consideration involves the need to use a
scale measure that allows more powerful statistical analysis (interval
or ratio scale) as well as uniform scale so that the score of each item
on each dimension can be averaged, and in turn a mean for the whole
brand equity construct can be derived. Measuring brand recall would
likely result in categorical data as doing it would involve asking
respondent to recall a certain brand based on a certain cue.
Nevertheless, the author attempts to partly emulate brand recall by
structuring the question around product attributes (by asking their
agreement/disagreement of statement “Toothpaste X is the name that
comes into my mind when I think of toothpaste brand”), and purchase
situation (“When it comes to purchase toothpaste, Toothpaste X
comes up in my mind first”). This is done in order to emulate Keller’s
suggestion (2003) to employ such cues in actual recall measurement.
Other questions include pseudo recall of packaging (“When you see a
toothpaste tube with blue, italic font, you think of Toothpaste X”) as
well as more “traditional” question of brand recognition (“I have heard
of Toothpaste X before”).
2. Brand Associations
Measurement of brand association in the scale is based on Aaker’s
concept to measure brand equity across products and markets (1996),
which suggests the inclusion of value, brand personality and
organisational associations.
In value subdimension, a value indicator is measured. It indicates to
what extent a brand is successful in creating a value proposition
(Aaker,1996). Aaker further suggests that it can be measured by
6
asking whether the brand provides value for money and/or reason to
buy over competitor brand. Eventhough evidence show that perceived
value is related with perceived quality, but further research by Young
and Rubicam (quoted in Aaker, 1996) indicate that they represent
different dimensions, thus deserves to be separated. In this scale, the
author measures value dimension as proposed by Aaker (1996).
(Questions 1-2)
Brand personality is defined as “the set of human characteristics
associated with a brand” (Aaker, 1997). Aaker further suggests that it
consists of five dimensions: sincerity, excitement, competence,
sophistication, and ruggedness. In this scale, the author attempted to
measure brand personality using items suggested by Aaker (1996)
with some adaptation (Question 4). Other items attempt to measure
brand personality dimensions based on concepts proposed by Aaker
(1997). This is featured in questions 5-13.
The third subdimension, organisational associations is based on the
premise that an organization can be used to differentiate an offering,
especially when brands in the market have similar attributes (as in the
case of toothpaste market). Questions regarding this subdimension are
featured in questions 14 – 16)
The author also adds one more item to measure overall brand
association (Question 3). This is done in accordance to Aaker’s
suggestion that we can supplement the three brand-associations
measure with an overall indicator measuring a brand’s ability to
achieve differentiation.
3. Perceived Quality
Perceived quality is operationalised using framework proposed by
Zeithaml (1988). She suggests that perceived quality is an overall
subjective judgment of a product quality, and that perceived quality is
7
also a form of an attitude. Another point she posits is that a consumer
judges perceived quality of a product relative against other
competitive brands in the market. She also proposes that consumers
use extrinsic attributes such as price, brand name, and level of
advertising frequency to judge a product’s perceived quality. Using
these postulates, the author generated four items (questions 3-6).
There are also two more questions that directly assess respondents’
attitude toward Toothpaste X’s perceived quality relative to its
competitors (questions 1-2)
4. Brand Loyalty
There is no consensus of the definition of brand loyalty as well as its
measurement (Bennet and Rundle-Thiele, 2002; Rundle-Thiele and
Mackay 2001). Mellens (1996, quoted in Rundle-Thiele and Mackay,
2001) classifies brand loyalty measurement as behavioural and
attitudinal. Whereas behavioural loyalty is measured based on
observed actual purchases over time period, attitudinal loyalty is
measured based on stated preferences, commitment or purchase
intentions (Mellens 1996, quoted by Rundle-Thiele and Mackay, 2001).
In this scale, the author adopts attitudinal loyalty measurement in
order to create consistent measurement based on consumer
perception.
The attitudinal loyalty construct is operationalised based on various
key measures proposed by a number of researchers summarised in
table from Rundle-Thiele and Mackay (2001) below:
Table 3Classification of attitudinal loyalty measurement
Attitudinal loyalty measurement Question no.Attitude toward the loyal/disloyal act (Sharp, et al,
1997)
1
Brand Preference (Guest, 1944, 1955) 4Commitment (Hawkes, 1994) or attitude toward the
brand measures (Sharp, et al, 1997)
3
Probability of purchase (Danenberg and Sharp, 1996;
Jacoby and Chestnut, 1978)
2
8
Measures of Brand Equity Used in Yoo and Donthu’s (2001) scale
Yoo and Donthu based their scale development on Aaker’s conception of
brand equity, consisting of dimensions mentioned in the previous sections.
They initially generated 48 candidate items to measure the four dimensions;
then the items were screened based on conformity with theoretical
definitions and the work of previous researchers. As a result, they
maintained 22 items (5 for brand loyalty, 4 for brand awareness, 7 for
perceived quality, and 6 for brand associations) for psychometric testing.
For brand loyalty, they used attitudinal loyalty approach, and adapted items
from Beatty and Kahle (1988, quoted in Yoo and Donthu, 2001).
They measured brand awareness based on simple brand recognition instead
of recall, but they did not mention the rationale behind this decision. They
based the items on works of previous researchers as mentioned in Srull
(1984), Alba and Hutchinson (1987), and Rossiter and Percy (1987), all
quoted in Yoo and Donthu’s article (2001).
Perceived quality items were designed to capture quality as “consumers’
subjective judgment about a brand’s overall excellence as postulated by
Zeithaml (1988). They adopted seven items from Dodd et al ’s work (1991,
quoted in Yoo and Donthu, 2001).
The last dimension, brand association is defined following Keller’s (1993)
definition (“the strength of connection to a brand node as a function of both
the amount or quantity of processing the information received at encoding
and the nature or quality of the processing of the information received at
encoding”). They developed their own items based on this definition.
9
They concluded their article by suggesting that brand awareness and
associations need to be combined because discriminant validity of the two
were not proven in their model.
Assessing Scale Reliability and Validity in Comparison with Yoo and
Donthu’s Brand Equity Scale
Both scale was administered through two informal mailing lists of teaching
staffs at Department of Management of Faculty of Economics, University of
Indonesia as well as some friends in Indonesia around April-May 2006.
There are 20 questionnaire filled out and returned via email. The result were
inputted into SPSS. Prior to processing, data were screened for missing
values as well as data input errors. Then several statistics below were
computed:
• Mean and standard deviation of the two scales (including mean scales and
standard deviation on each dimensions on both scales)
• Cronbach’s Alpha for both scales
• Pearson correlation between the two scales
Complete SPSS output is available in Appendix 1-3. In this section we will
highlight only some of the important statistics.
● Cronbach’s Alpha
Cronbach’s alpha result from the author’s own scale:
a. Brand Awareness
Cronbach’s alpha for the overall brand awareness construct is
0.682. It is slightly under minimum recommended alpha value of
0.7 (see Pappu, et al, 2005; Yoo and Donthu, 2001).
An evaluation of item-total statistics for each item in the
construct reveals that items 5 (“When you see a toothpaste tube
10
with blue, italic font, you think of Toothpaste X”) and 6 (“When it
comes to purchase toothpaste, Toothpaste X comes up in my mind
first)” have low corrected item-total correlation (0.2380 and 0.2630
respectively). “Cronbach’s alpha if item deleted” for both items
indicate that if both were deleted (one at a time) from the scale, then
the overall brand awareness reliability would improve significantly (to
0.720 if item 5 were deleted, 0.732 if item 6 were deleted). If both
items were deleted altogether, then the total brand awareness
reliability would improve even more significantly to 0.820.
Cronbach’s alpha result for brand awareness thus suggests
that item 5 and/or 6 (or both) may need to be deleted in order to
obtain more reliable awareness scale.
b. Brand Associations
Cronbach’s alpha for brand associations scale is 0.840,
indicating good level of construct reliability.
The Alpha if item deleted” column suggests that item 4 (“I
have a clear idea about what kind of people who use Toothpaste X”),
item 7 (“Toothpaste X is associated with family values”), and item 14
(“I know the company that produce Toothpaste X”) may be deleted.
If item 4 were deleted, the cronbach’s alpha would increase
fractionally to 0.8490, deleting item 7 would increase the alpha to
0.8450, while deleting item 14 would even increase the alpha more
significantly to 0.853. If all of them were deleted, reliability would
increase to 0.874.
Based on values above, we may conclude that although the
brand association scale already yield good reliability, eliminating
items 4, 7, and 14 may increase it further.
Despite Aaker’s (1996) suggestion in his article to use question
similar to item 4 to measure brand personality subdimension of
brand association, it seems that this kind of question may not
measure brand personality subdimension. Could this possibly caused
by ambiguous nature of the question, the wording used, or the
11
nature of object under study? Further research is needed to examine
this further.
Item 7 was intended as part of sincerity subdimension of brand
personality, but apparently “family values association” may not
associate with brand association.
Item 14 (“I know the company that produce Toothpaste X”)
also probably unrelated with brand association. It might be more
appropriate to include it in the brand awareness dimension as part of
new subdimension named corporate awareness.
c. Perceived Quality
Cronbach’s alpha coefficient for perceived quality of this
dimension is 0.930. This value indicates a highly reliable scale.
Examining the “alpha if item deleted” column, we may say
that no items need to be eliminated, since all of them indicate high
corrected item-total correlation as well as small differences in “alpha
if item deleted” column, indicating good reliability.
Five out of six items indicate lower alpha if item were deleted. Only
item 6 (Judging from the frequency of Toothpaste X advertisement in
the media, Toothpaste X must be of good quality) would increase the
total cronbach’s alpha coefficient of the scale slightly to 0.939 if it
were deleted.
d. Brand Loyalty
Cronbach’s alpha for this dimension indicates high reliability
(0.922). All items indicate high item-total correlation. Item 2 (I feel
committed to Toothpaste X) is the only item that might improve
loyalty scale reliability very slightly if it were deleted (reliability
would increase slightly to 0.928).
Based on statistics above, we may conclude that brand loyalty
scale has very good reliability and no items need to be removed.
Cronbach’s alpha result of scale adapted from Yoo and Donthu (2001)
12
a. Brand awareness/associations
Cronbach’s alpha indicates relatively high reliability of 0.793
for 5 items scale of combined awareness/associations.
Examination of “alpha if item deleted” column reveals that
deleting item 2 (I am aware of Toothpaste X) may improve scale
reliability marginally (to 0.812).
Since cronbach’s alpha scale of 0.7 (Pappu et al, 2005; Yoo
and Donthu, 2001) is regarded sufficient to indicate acceptable
reliability, thus no need to eliminate any items.
b. Perceived Quality
For perceived quality, the 2 items scale yield cronbach’s alpha
coefficient of 0.922, indicating high reliability. Both items indicate
similar corrected item-total correlation, with no alpha if item deleted
values available.
Examination of the “alpha if item deleted” column of the SPSS
output identifies something that the author could not explain. The
column is left blank by SPSS and there’s a short notification
underneath, saying that there is negative correlation between the
two items and that the author should check the data coding. The
data coding were checked, and there was no mistake or coding error,
so the problem is remained unsolved to the date this report was
written.
c. Brand Loyalty
The three items scale measuring brand loyalty yield cronbach’s
alpha coefficient of 0.946, indicating high reliability. All items in the
scale indicate high item-total correlation. The only item that can be
deleted is item 1(“the likely quality of Toothpaste X is extremely
high”). But deleting this item would only yield small increase in the
alpha coefficient (0.948).
13
Conclusion based on cronbach’s alpha values of both scales
The two scales appear to exhibit high reliability. In general, the
author’s own scale exhibit somewhat comparable reliability values in
terms of cronbach’s alpha. The author’s scale for perceived quality
dimension is slightly higher than Yoo and Donthu’s, whereas for
brand loyalty it is slightly lower than Yoo and Donthu’s. Since Yoo
and Donthu merged brand awareness and associations dimensions,
we cannot directly compare the values, but if the author’s brand
awareness and associations dimensions were combined and simply
averaged, Yoo and Donthu’s scale is still slightly better in reliability
(0.7608 compared to 0.7935).
Table 4 below summarise cronbach’s alpha values of both scales in
each dimension they measure.
Table 4
Summary of Cronbach’s alpha values of both scales
Brand
Awarenes
s
Brand
Associations
Perceived
Quality
Brand
Loyalty
Author’s scale 0.682 0.840 0.930 0.922Yoo and Donthu’s
scale
0.793* 0.922 0.946
*Combined dimensions (awareness/associations)
Table below presents a part of Cronbach’s alpha of Yoo and Donthu’s
original scale as reported in their article. In their study, they
administered their scale to three groups of students with different
background (Americans, Koreans, and Korean Americans).
14
Table 5
Cronbach’s alpha values from Yoo and Donthu’s original study
Americans Koreans Korean
AmericanBrand Loyalty 0.88 0.86 0.87Perceived Quality 0.92 0.90 0.84Awareness/associations 0.92 0.89 0.89
These values suggest that both Yoo and Donthu’s adapted scales as
well as the author’s scale used in this study exhibit higher reliability
in brand loyalty and perceived quality dimensions. On the other
hand, for brand awareness/associations dimension, the original scale
exhibits higher reliability.
● Mean, standard deviation and correlation of both scales
After examining the reliability of the proposed scale, we need to
assess its convergent validity. Hair et al (2003) defines convergent
validity as “how well the construct’s measurement positively
correlates with different measurements of the same construct. To
assess this, we shall compare the mean total brand equity of both
scales as well as their correlation with each other.
The mean value for total brand equity is derived by averaging all
mean values of each dimension. Mean value of each dimension is
derived by averaging all items in a dimension.
The mean value of each dimension as well as total brand equity is
summarised in tables 4 and 5 below.
Table 6
Mean and Standard Deviation of Author’s Scale
Total
Awarenes
s
Total
Associatio
n
Total
Perceived
Quality
Total
Loyalt
y
Total
Brand
EquityN 20 20 20 20 20
15
Mean 5.7167 4.344 4.5833 3.612
5
4.5617
Standard
deviation
0.8552 0.7124 1.2085 1.706
2
0.9100
Source: SPSS data output
Table 7
Mean and Standard Deviation of Scale adapted from Yoo and Donthu
Total
Awareness/
Association
Total
Perceived
Quality
Total
Loyalty
Total
Brand
EquityN 20 20 20 20Mean 5.24 4.750 3.7167 4.5689Standard
deviation
1.0889 1.2825 1.8266 1.0984
Source: SPSS data output
Examination of the two tables above show us that the mean value of
total brand equity of the author’s own scale is not much different
from the mean value of brand equity of Yoo and Donthu’s scale
(4.5617 and 4.5689 respectively). In terms of standard deviation,
there is not much difference between the two scales as well (0.9100
and 1.0984 respectively). Therefore, we may predict that both scales
are measuring the same thing.
In order to confirm that both scales are relatively convergent, we
shall look at the correlation between the two means of total brand
equity.
Table 6 below displays correlation result between total brand equity
of the author’s scale (TOTALBE) and Yoo and Donthu’s total brand
equity (TOTYBE).
Table 8
16
Correlations
1.000 .898**
. .00020 20
.898** 1.000
.000 .
20 20
Pearson CorrelationSig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
TOTALBE
TOTYBE
TOTALBE TOTYBE
Correlation is significant at the 0.01 level (2-tailed).**.
The output indicate the correlation between TOTALBE and TOTYBE is
significant with r = 0.898 and p<0.01. Thus we may conclude that
there is almost 100% confidence that TOTALBE and TOTYBE values
are positively associated. The strength of the association (indicated
by Pearson correlation coefficient value of 0.898) is also very high
(nearing the value of +1 which indicates very strong positive
correlation, see Hair et al, 2003).
This finding suggests that it is highly likely that the author’s scale
and Yoo and Donthu’s scale measure the same thing, since a positive
value in one scale is also reflected with positive value in the other
scale. We may say that both scales have convergent validity.
Table on the next page presents the mean value of the original scale
as reported in Yoo and Donthu’s article (2001). By comparing the
mean values of each dimension as well as the total brand equity
values, we may conclude that the author’s scale and scale adapted
from Yoo and Donthu exhibit higher mean values. But this result is
not comparable because the author used 7 points modified Likert
scale, while Yoo and Donthu used 5 points Likert scale. Other than
that, another source of difference might be in the object of the study
used in both studies.
Table 9Mean of original scale from Yoo and Donthu’s original study
17
Americans Koreans Korean
AmericanBrand Loyalty 2.29 2.00 2.26Perceived Quality 3.48 3.05 3.3Awareness/associations 3.42 2.94 3.42Total Brand Equity 3.06 2.66 2.99
B. Critiques and Reservations of the Study and Suggestion forFuture Research
The main weakness of the author’s scale probably lies in the
methodology used to test the scale. To test a scale that measures
complex multidimensional construct like brand equity require more
rigorous methodology and more powerful statistical analysis such as
confirmatory factor analysis. Real researchers use confirmatory factor
analysis to prove the relationship between the variables (this is
observed in many journal articles cited in this report).
In this small research, the interrelationships between the
variables/dimensions are taken for granted, for it is assumed to exist
based on the work of researchers. For instance, the author just
assumes that brand awareness and brand association are separate
dimensions based on Aaker’s conceptual framework. The author could
not assess for himself whether or not the two dimensions are really
separated and whether or not his scale can really be used to
discriminate the two dimensions.
The second weakness comes from the number of respondents (here
the author only used 20 respondents) as well as the sampling method
used to collect the sample.
Despite those weaknesses above, the author believes that his scale is
quite reliable and valid. The Cronbach’s alpha generated by SPSS has
proven that items in the author’s scale have relatively good reliability.
The convergent validity has been proven by the existence of significant
18
correlation between his scale and Yoo and Donthu’s scale. The latter
itself has been validated a couple of times (Atilgan, 2005; Washburn
and Plank, 2002).
In addition to the facts above, the author’s scale were developed
based on existing theories about brand equity and its dimensions.
Further research may be needed to further assess the author’s scale
validity. More advanced statistical tools such as confirmatory factor
analysis need to be used in order to further test the existing brand
equity model using the author’s scale. Further research also may need
to involve larger number of respondent from more representative
sample in order to gain generalizable result.
Cronbach’s alpha if items deleted for brand awareness dimension
suggest us to remove items 5 and 6 (or at least one of them) in order
to gain acceptable level of reliability. Thus another attempt to validate
this scale might need to exclude items 5 and 6 in order to improve the
scale’s reliability.
Three other items in brand associations dimension may also need to
be excluded or reassessed. Item 4 may need re-examination to
confirm whether its reported low correlation with brand associations
dimension resulted from improper wording, nature of the object used
in the study or could it be just not appropriate item to measure brand
associations? Item 14 might be incorporated into brand awareness
dimension. Although it may not have been proposed in previous
studies, but a new subdimension to brand awareness, namely
company awareness may be proposed. Awareness toward a brand
may be affected by awareness toward company. Aaker (2004)
suggests that brands are somewhat affected by the reputation of the
company behind it. He implied in his book that the manufacturer or
marketer of a brand might affect the way a brand is perceived as well
19
as its credibility and acceptance by consumers. Thus the possibility of
including a new subdimension to brand awareness might need to be
explored further.
For perceived quality and loyalty, there seems to be no items that
need elimination, but items measuring them as well as items
measuring other dimension need to be further validated in another
study.
20
REFERENCES
Aaker, D. A. (1996). Measuring brand equity across products andmarkets. California Management Review, 38(3), 102-120.
Aaker, D. A. (2004). Brand portfolio strategy. New York: Free Press.
Aaker, J. L. (1997). Dimensions of brand personality. Journal ofMarketing Research, 36(3), 345-355.
Atilgan, E., Aksoy, S., & Akinci, S. (2005). Determinants of brandequity: A verification approach in the beverage industry inTurkey. Marketing Intelligence and Planning, 23(2), 237.
Bennett, R., & Rundle-Thiele, S. (2002). A comparison of attitudinalloyalty measurement approaches. Journal of Brand Management,9(3).
Cobb-Walgren, C. J., Beal, C., & Donthu, N. (1995). Brand equity, brandpreference and purchase intent. Journal of Advertising, 24(3), 25-40.
Hair, J. F., Bush, R. P., & Ortinau, D. J. (2003). Marketing researchwithin a changing information environment (International Editioned.). New York: McGraw-Hill/Irwin.
Keller, K. L. (1993). Conceptualising, measuring and managingcustomer-based brand equity. Journal of Marketing, 57(January),1-22.
Keller, K. L. (2003). Strategic brand management: Building, measuringand managing brand equity (second ed.). New Jersey: PearsonEducation International.
Pappu, R., Quester, P. G., & Cooksey, R. W. (2005). Consumer-basedbrand equity: improving the measurement -empirical evidence.The Journal of Product and Brand Management, 14(2/3), 143-154.
Rundle-Thiele, S., & Mackay, M. M. (2001). Assessing the performanceof brand loyalty measures. Journal of Service Marketing, 15(6/7),529-545.
21
Washburn, J. H., & Plank, R. E. (2002). Measuring brand equity: anevaluation of a consumer-based brand equity scale. Journal ofMarketing Theory and Practice, 10(1), 46-62.
Yoo, B., & Donthu, N. (2001). Developing and validating amultidimensional consumer-based brand equity scale. Journal ofBusiness Research, 42, 1-14.
Zeithaml, V. A. (1988). Consumer perceptions of price, quality andvalue: A means-end model and synthesis of evidence. Journal ofMarketing, 52(July), 2-22.
22
APPENDIX 1SPSS OUTPUT: RELIABILITY ANALYSIS
OF THE AUTHOR’S OWN SCALE
1. Brand Awareness Dimension
Reliability Statistics
.682 .768 6
Cronbach'sAlpha
Cronbach'sAlpha Based
onStandardized
Items N of Items
Item-Total Statistics
28.35 16.766 .667 .555 .548
27.75 20.724 .548 .427 .618
27.90 21.042 .556 .488 .621
28.10 20.516 .618 .546 .606
30.00 19.579 .238 .118 .720
29.40 18.042 .263 .135 .732
Comes Into Mind
Heard of before
Easy to find
Quicky identified
Logo characteristic recall
Comes into mind whenneed to buy
Scale Mean ifItem Deleted
ScaleVariance if
Item Deleted
CorrectedItem-TotalCorrelation
SquaredMultiple
Correlation
Cronbach'sAlpha if Item
Deleted
2. Brand Awareness Dimension if Items 5 and 6 Were Deleted
Reliability Statistics
.820 .835 4
Cronbach'sAlpha
Cronbach'sAlpha Based
onStandardized
Items N of Items
Item-Total Statistics
19.15 5.082 .706 .529 .771
18.55 7.418 .613 .419 .788
18.70 7.484 .663 .488 .771
18.90 7.358 .681 .529 .763
Comes Into Mind
Heard of before
Easy to find
Quicky identified
Scale Mean ifItem Deleted
ScaleVariance if
Item Deleted
CorrectedItem-TotalCorrelation
SquaredMultiple
Correlation
Cronbach'sAlpha if Item
Deleted
23
3. Brand Association DimensionReliability Statistics
.840 .848 16
Cronbach'sAlpha
Cronbach'sAlpha Based
onStandardized
Items N of Items
Item-Total Statistics
65.05 114.787 .482 .930 .829
65.60 114.779 .486 .933 .829
65.05 113.313 .484 .943 .829
65.75 119.776 .202 .711 .849
65.60 110.884 .594 .955 .822
65.60 114.989 .478 .956 .829
64.15 118.871 .249 .902 .845
65.65 114.134 .488 .767 .829
64.95 111.839 .768 .945 .817
65.25 110.408 .656 .936 .819
65.65 112.871 .628 .824 .822
64.85 108.661 .737 .920 .815
65.85 111.713 .614 .831 .822
63.25 129.566 -.035 .737 .853
64.55 118.155 .311 .868 .839
63.45 121.839 .334 .892 .836
Value
Better performance
Unique characteristics
Clear image of user
Friendly
Sincere
Family values
Excitement
Reliable
Delivers promise
Sophisticated
Strength
Proud to buy
Know the company
Like the company
Good reputation
Scale Mean ifItem Deleted
ScaleVariance if
Item Deleted
CorrectedItem-TotalCorrelation
SquaredMultiple
Correlation
Cronbach'sAlpha if Item
Deleted
4. Brand Associations Dimension Without Items 4, 7 dan 14Reliability Statistics
.874 .878 13
Cronbach'sAlpha
Cronbach'sAlpha Based
onStandardized
Items N of Items
Item-Total Statistics
50.15 92.555 .614 .886 .86250.70 92.853 .606 .930 .862
50.15 93.397 .519 .919 .867
50.70 93.905 .521 .948 .867
50.70 100.011 .307 .915 .87950.75 93.882 .536 .713 .86650.05 91.839 .832 .933 .853
50.35 89.397 .759 .909 .85350.75 93.145 .666 .772 .859
49.95 91.524 .674 .906 .85850.95 92.050 .651 .796 .860
49.65 100.029 .262 .812 .883
48.55 102.471 .317 .641 .876
Value
Better performance
Unique characteristicsFriendly
Sincere
ExcitementReliable
Delivers promiseSophisticated
Strength
Proud to buy
Like the companyGood reputation
Scale Mean ifItem Deleted
ScaleVariance if
Item Deleted
CorrectedItem-TotalCorrelation
SquaredMultiple
Correlation
Cronbach'sAlpha if Item
Deleted
24
5. Perceived QualityReliability Statistics
.930 .931 6
Cronbach'sAlpha
Cronbach'sAlpha Based
onStandardized
Items N of Items
Item-Total Statistics
22.75 35.671 .831 .816 .913
23.30 35.168 .846 .856 .910
22.90 38.411 .795 .808 .91822.70 34.642 .970 .947 .894
22.70 39.695 .726 .786 .926
23.15 38.555 .634 .720 .939
High quality
Performs better thanothersGood quality for price
Signals good qualityPrice reflects quality
Quality perceivedbased on ad
Scale Mean ifItem Deleted
ScaleVariance if
Item Deleted
CorrectedItem-TotalCorrelation
SquaredMultiple
Correlation
Cronbach'sAlpha if Item
Deleted
6. Brand Loyalty
Reliability Statistics
.922 .921 4
Cronbach'sAlpha
Cronbach'sAlpha Based
onStandardized
Items N of Items
Item-Total Statistics
11.15 28.029 .797 .725 .906
11.15 28.555 .729 .659 .928
10.55 24.155 .939 .882 .855
10.50 26.789 .819 .718 .898
Feel happy to buyother brands (r)Committed
Likelihood to buyPepsodent next time
More preffered thanother brands
Scale Mean ifItem Deleted
ScaleVariance if
Item Deleted
CorrectedItem-TotalCorrelation
SquaredMultiple
Correlation
Cronbach'sAlpha if Item
Deleted
25
APPENDIX 2SPSS OUTPUT: RELIABILITY ANALYSIS OF SCALE
ADAPTED FROM YOO AND DONTHU
1. Brand Awareness/AssociationReliability Statistics
.793 .804 5
Cronbach'sAlpha
Cronbach'sAlpha Based
onStandardized
Items N of Items
Item-Total Statistics
20.80 19.853 .720 .628 .715
20.15 24.871 .350 .248 .812
21.40 18.779 .613 .626 .741
20.90 19.779 .705 .624 .718
21.55 16.471 .575 .537 .775
Can recognizeamong other brands
Aware of Pepsodent
Characteristicsquickly recalled
Logo quickly recalled
Diffficult to imagine (r)
Scale Mean ifItem Deleted
ScaleVariance if
Item Deleted
CorrectedItem-TotalCorrelation
SquaredMultiple
Correlation
Cronbach'sAlpha if Item
Deleted
2. Perceived QualityReliability Statistics
.922 .938 2
Cronbach'sAlpha
Cronbach'sAlpha Based
onStandardized
Items N of Items
Item-Total Statistics
4.80 1.326 .883 .780 .a
4.70 2.221 .883 .780 .a
High likely qualityHigh functionalitylikelyhood
Scale Mean ifItem Deleted
ScaleVariance if
Item Deleted
CorrectedItem-TotalCorrelation
SquaredMultiple
Correlation
Cronbach'sAlpha if Item
Deleted
The value is negative due to a negative average covariance among items. This violatesreliability model assumptions. You may want to check item codings.
a.
3. Brand LoyaltyReliability Statistics
.946 .947 3
Cronbach'sAlpha
Cronbach'sAlpha Based
onStandardized
Items N of Items
26
Item-Total Statistics
7.40 15.095 .856 .743 .948
7.15 12.976 .925 .859 .892
7.75 13.039 .892 .819 .920
Feeling loyal
First choiceWill not buy other brands
Scale Mean ifItem Deleted
ScaleVariance if
Item Deleted
CorrectedItem-TotalCorrelation
SquaredMultiple
Correlation
Cronbach'sAlpha if Item
Deleted
27
APPENDIX 3SPSS OUTPUT: MEAN, STANDARD DEVIATION,
AND CORRELATIONS
1. Mean and Standard Deviation
Statistics
20 20 20 20 20 20 20 20 20
0 0 0 0 0 0 0 0 0
5.7167 4.3344 4.5833 3.6125 4.5617 5.2400 4.7500 3.7167 4.5689
.8552 .7124 1.2085 1.7062 .9100 1.0889 1.2825 1.8266 1.0984
Valid
Missing
N
Mean
Std. Deviation
TOTALAWR TOTASSOC TOTALPQ TOTALYLT TOTALBE TOTYAWAS TOTYPQ TOTYLYLT TOTYBE
2. Correlations
Correlations
1 .898**
.000
20 20.898** 1
.000
20 20
Pearson Correlation
Sig. (2-tailed)N
Pearson Correlation
Sig. (2-tailed)
N
totalbe
totybe
totalbe totybe
Correlation is significant at the 0.01 level(2-tailed).
**.
28
APPENDIX 4LIST OF ORIGINAL QUESTIONS FROM YOO AND DONTHU’S ARTICLE (2001)
29