Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
This study investigates the effect of the Brand Equity on the Customer Satisfaction and the Brand Loyalty of Star-
bucks and Excelso coffee shops customers. Starbucks is the strongest competitor of Excelso. Data used in this study was
primary data which were collected through closed questionnaires with 1-5 Likert scale. A sample of this study was 270
respondents; 135 respondents for each brand. The study used Structural Equation Modelling (SEM) operated by LISREL
program to analyze the hypothesis. The results of this study showed that The Physical Quality, the Ideal Self-Congruence,
and the Lifestyle Congruence have a positive and significant impact on the Customer Satisfaction on Excelso and Star-
bucks. Staff behaviour only have a positive and significant impact on the Customer Satisfaction in the Excelso; whereas,
brand identification only have a positive and significant impact on the Customer Satisfaction in the Starbucks. However,
in both of coffee shop brand, the Consumer Satisfaction have a positive and significant impact on the Brand Loyalty.
Keywords: brand equity, consumer satisfaction, brand loyalty, SEM, Excelso, Starbucks
Penelitian bertujuan menguji dampak brand equity terhadap customer satisfaction dan brand loyalty pada konsumen Kedai
Kopi Excelso dan Starbucks. Dalam hal ini, Excelso merupakan kompetitor terkuat dari Starbucks. Penelitian menggunakan data
primer yang dikumpulkan melalui pengisian kuesioner tertutup menggunakan skala Likert 1-5. Total responden yang menjadi
sampel dalam penelitian ini adalah 270 orang; 135 orang diantaranya merupakan konsumen dari Excelso dan 135 orang merupakan
konsumen dari Starbucks. Penelitian ini dilakukan dengan menggunakan Structural Equation Modelling (SEM) dan diolah dengan
bantuan Software LISREL. Hasil penelitian menunjukkan bahwa physical quality, ideal self-congruence, dan lifestyle congruence
merupakan faktor-faktor yang berpengaruh positif secara signifikan terhadap customer satisfaction di Excelso dan Starbucks. Staff
behaviour hanya memiliki pengaruh positif secara signifikan terhadap customer satisfaction di Excelso; adapun brand identification
hanya memiliki pengaruh positif secara signifikan terhadap customer satisfaction di Starbucks. Baik di Excelso maupun di Star-
bucks, customer satisfaction memiliki pengaruh positif secara signifikan terhadap brand loyalty.
Kata Kunci: brand equity, consumer satisfaction, brand loyalty, SEM, Excelso, Starbucks
and colleagues while using a laptop and having lunch in of the respondents are more than 34 years old. Respondents
addition to accessing the internet. In 2004, forty-two per- are most likely to have found out about the coffee shop (Ex-
cent of adults in the United States aged 18-34 purchased their celso and/or Starbucks) from their friends, followed by social
coffee at a coffee shop with 48% consuming the beverage on media, and finally due to the fact that the coffee shop occupies
the premises. These numbers seem to indicate that the coffee a strategic location. Less than one-sixths of the respondents
shops appeal doesnt lie solely with the coffee; and that the learned about the place from the advertising. A large propor-
venue may have significance in itself (Waxman, 2008). tion of respondents for both brands had visited the respective
The items used to measure the variables investigated in coffee shop 3 times in the last 3 months. In this case, Star-
this study are detailed in Table 3. Additionally, this study bucks had more respondents who had visited the coffee shop
uses a 5-point Likert scale for its questions, where 1 in- more than 3 times in the last 3 months than Excelso did. In
dicates that the respondent strongly disagrees and 5 indi- Excelsos case, most respondents had visited the coffee shop
cates that the respondent strongly agrees with the state- exactly 3 times in the last 3 months, followed by 4 times in the
ment. last 3 months as the second-largest category, then more than
5 times in the last 3 months, and 5 times in the last 3 months.
Data Analysis Tools In Starbucks, the largest group of respondents is once again
those who had visited the coffee shop exactly 3 times in last 3
The research was carried out through Structural Equation months, but the second-largest group had visited more than 5
Modelling (SEM) with the aid of the Linear Structural Re- times in the last 3 months, followed by those who had come
lationships (LISREL) Program. The SEM is a second gen- 4 times in the last 3 months, and finally 5 times in the last 3
eration data analysis technique for estimating complex rela- months. For both brands, more than one third of the respond-
tionships among multiple constructs (Byrne, 2010). The SEM ents visited the coffe for the purpose of socializing with others.
differs from traditional statistical methods (e.g., regression, The second most important purpose in Excelsos case is busi-
ANOVA, LOGIT) in several important ways (Gefen, Straub, ness, followed by relaxation, spending some free time, and
and Boudreau, 2000). Where traditional statistic methods can others. For Starbucks, the second most common purpose is
only test pairwise relationships between observed variables, to spend some free time, then relaxation, business, and others.
the SEM can construct latent variables (abstract concepts that
cannot be measured directly) and assess complex (e.g., hier- Test of Model Fit
archical, recursive) causal paths among such variables. There-
fore, the SEM technique has seen increasing use in the social In this study, eleven criteria that belong to the Goodness
sciences, behavioral sciences, and management science, espe- of Fit Indices (GOF Indices) are used to determine the consist-
cially for modelling complex and multivariate relationships. ency of the theoretical model as a whole with the empirical
data. The results for the eleven criteria in the GOF Indices can
Results and Discussion be seen in Table 2.
The first criterion is Normed 2 (normed chi-square),
Respondent Characteristics where the chi-square is adjusted by the degrees of free-
dom to assess model fit (Jreskog, 1969). Models with
The characteristics of the respondents who completed this adequate fit should have a norm chi-square of less than
studys questionnaire are described in Table 1. The majority 2.0 or 3.0 (Carmines and McIver, 1981). A good-fit model
of respondents for both brands (Excelso and Starbucks) are must demonstrate a normed chi-square greater than 1.00
male, and most are 18 to 34 years old. Less than one-fourths but less than 2.00 (Hair, Anderson, Tatham, and Black,
1998). Values close to 1.0 indicate a good fit although The second criterion is RMSEA (root mean square
values up to 3.0 may indicate a reasonable fit (Holmes- error of approximation). According to Browne and Cu-
Smith, 2000). With a normed chi-square of 1.85, the pro- deck (1993), RMSEA values 0.05 can be considered as
posed model for Excelso provides a good-fit representa- a good fit, values between 0.05 and 0.08 as an adequate
tion of the data; meanwhile, with a normed chi-squared fit, and values between 0.08 and 0.10 as a mediocre fit,
of 2.15, the proposed model for Starbucks provides a whereas values > 0.10 are not acceptable. Marsh, Hau,
reasonable fit of the data. and Wen (2004) suggest that 0.08 should be acceptable
in most circumstances (see also Wijanto, 2008). With Reliability is a measure of the internal consistency of
a RMSEA of 0.054 and 0.072, the proposed model for the construct indicators, depicting the degree to which
Excelso and Starbucks provide an adequate-fit represen- they indicate the common latent (unobserved) construct.
tation of the data. More reliable measures provide the researcher with
The third criterion is the expected cross-validation greater confidence that the individual indicators are all
index (ECVI). ECVI is a measure of the discrepancy be- consistent in their measurements. Reliability was evalu-
tween the model-implied covariance matrix in the ana- ated by examining the Composite Reliability (CR) and
lyzed sample and the covariance matrix that would be average Variance Extracted (VE) from the measures. A
expected in another sample of the same size (Jreskog commonly used threshold value for CR is 0.70 (Wijanto,
and Srbom, 1993). Thus, ECVI evaluates how well a 2008). Table 2 shows the CR scores for each of the con-
model fitted to the calibration sample would perform structs are well above 0.70, ranging from 0.815 to 0.913,
in comparable validation samples (Kaplan, 2000). The which demonstrates an acceptable level of internal con-
model has a good fit if its ECVI is smaller than the ECVI sistency for the construct indicators. Another measure of
for the saturated model and the ECVI for the indepen- reliability is the variance extracted measure. This mea-
dence model (Ghozali and Fuad, 2008). Thus, the model sure reflects the overall amount of variance in the indi-
for Excelso and Starbucks provide a good fit representa- cators accounted for by the latent construct. Higher VE
tion of the data since the ECVI for this model is smaller values occur when the indicators are truly representative
than the ECVI for the saturated model and the ECVI for of the latent constructs. Recommendations typically sug-
the independence model. gest that the VE value for a construct should exceed 0.50
The fourth criterion is the Comparative Fit Index (Wijanto, 2008). Table 2 shows that the average VE for
(CFI). It ranges between 0 and 1, with values above 0.90 our measures range from 0.562 to 0.721, consistently
being associated with a good fit (Hair, Black, Babin, and above the accepted value.
Anderson, 2010; Gerbing & Anderson 1992). With a CFI
of 0.98, the proposed model for Excelso and Starbucks Hypothesis Test
provides a good fit representation of the data. The last
criterion is Goodness-of-Fit Index (GFI). Values for GFI The results of hypothesis testing are presented in the
range from 0 to 1 (>0.8) with higher values being bet- Table 3.
ter (Hair, Anderson, Babin, Tatham, and Black, 2007). A Physical quality had a significant positive effect on
value of GFI 0.8 is presently recognized as indicative customer satisfaction (PQ CS) for Excelso (SLF
of marginal fit (Handley and Benton, 2009). A GFI value = 0.35, t-value >1.96) and Starbucks (SLF= 0.33, t-
of 0. 95 is presently recognized as indicative of good fit value>1.96); so hypothesis 1 was supported for both
(Browne and Cudeck, 1993; Verschuren, 1991; Bentler, Excelso and Starbucks.
1990; Hu and Bentler, 1999), although the conventional Staff behavior had a significant positive effect on cus-
cut-off seen in the literature is about 0.90 (Russell, 2002). tomer satisfaction (SBCS) for Excelso (SLF = 0.36, t-
With a GFI of 0.88 and 0.80, the proposed model for Ex- value >1.96); staff behavior also had a positive effect
celso and Starbucks provide a marginal fit representation on customer satisfaction for Starbucks, but not signifi-
of the data. However, according to Table 2, the overall cant (SLF = 0.33, -1,96< t-value <1.96), so hypothesis
structural fit results of these analyses showed that the 2 was only supported for Excelso.
model provides a reasonable fit. Ideal self-congruence had a significant positive effect
on customer satisfaction (ISC CS) for Excelso (SLF
Validity and Reliability Test = 0.14, t-value >1.96) and Starbucks (SLF= 0.21, t-
value>1.96); so hypothesis 3 was supported for both
Linn and Stewart stated that validity is the measure Excelso and Starbucks.
of the accuracy of the data used (questionnaire items) in Brand identification had a significant positive effect on
a study (Said, Badru, and Shahid, 2011). To test the va- customer satisfaction (BI CS) for Starbucks (SLF =
lidity of each item in the questionnaire, this study used 0.26, t-value >1.96); brand identification also had posi-
a standardized loading factor of 0.70 (Wijanto, 2008). tive effect on customer satisfaction for Excelso, but not
Standardized factor loadings showed that all the items significant (SLF = -0.01, -1,96< t-value <1.96), so hy-
in the questionnaire used in this study are well loaded pothesis 4 was only supported for Starbucks.
on each latent construct and that the correlation between Lifestyle congruence had a significant positive effect
constructs ranges from 0.71- 0.94. on customer satisfaction (LC CS) for Excelso (SLF