Journal of Consumer Sciences E-ISSN : 2460-8963

2019, Vol. 04, No. 02, 76-89

THE EFFECT OF SOCIAL MEDIA INFLUENCER ON BRAND

IMAGE, SELF-CONCEPT, AND PURCHASE INTENTION

Atika Hermanda*)1, Ujang Sumarwan2, dan Netti Tinaprilla3

1
Management and Business, Business School, IPB University, Indonesia, 16151
2
Department of Family and Consumer Sciences, Faculty of Human Ecology, IPB
University, Indonesia, 16680
3
Department of Agribusiness, Faculty of Economics and Management, IPB University,
Indonesia, 16680

*
Corresponding author: [email protected]

Abstract

The purpose of the research was to discover how social media influencers influenced
the brand image, self-concept and the purchase intention of cosmetic consumers. The
social media influencers became the third party which provided information about
cosmetic products to the social media audience. Those who had the same self-concept
with the influencers often viewed them as role models in consumption. The increase of
self-concept and brand image affected a consumer’s purchase intention. The research
would implement the cross sectional design, with the data gathered through online
questionnaires shared in social media accounts. The samples would be gathered using
the convenience sampling technique, with respondents up to 219 people. The research
implemented the purchase intention as the endogenous variable and social media
influencers as the exogenous variable, as well as brand image and self-concept as the
intervening variable. The analysis method that would be implemented was the
Structural Equation Modeling (SEM), using the SmartPLS software. The research
resulted in the significant negative influence of both social media influencer and self
concept towards the purchase intention, in contrast with the brand image which had a
significant positive effect.

Keywords: brand image, social media, social media influencer, self-concept, purchase
intention

Abstrak

Penelitian ini bertujuan untuk mengetahui pengaruh social media influencer terhadap
brand image, konsep diri dan minat beli konsumen kosmetik. Influencer dari media
sosial adalah pihak ketiga yang memberikan informasi produk kosmetik kepada
audiensi di media sosial. Rekomendasi produk dari social media influencer dapat
memengaruhi opini konsumen. Konsumen yang merasakan konsep diri yang sama
dengan social media influencer akan cenderung menjadikannya sebagai panutan dalam
mengkonsumsi. Peningkatan konsep diri konsumen dan citra merek di benak konsumen
diduga dapat mempengaruhi minat beli konsumen. Desain penelitian ini adalah cross
sectional. Data diperoleh dari kuesioner online yang disebarluaskan melalui media
sosial. Adapun teknik pemilihan sampel yang digunakan penelitian ini adalah

76
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

convenience sampling. Responden yang didapatkan yaitu sebanyak 219 responden.

Variabel endogen yang digunakan dalam penelitian ini adalah minat beli sementara
variabel eksogen yaitu social media influencer. Variabel
intervening yang digunakan adalah citra merek dan konsep diri. Metode analisis dalam
penelitian ini yaitu Structural Equation Modeling (SEM) menggunakan perangkat lunak
SmartPLS. Hasil yang diperoleh adalah social media influencer dan konsep diri tidak
berpengaruh positif signifikan terhadap minat beli. Sementara itu, brand image memiliki
pengaruh positif signifikan terhadap minat beli.

Kata kunci: media sosial, social media influencer, brand image, konsep diri, minat beli

Introduction

Currently, the social media was an online communication tool that made a huge
impact towards the society (Dân & Nam 2018), encouraging companies to expand their
marketing practices in the social media. In the previous research, tradisional marketing
media such as televisions, magazines, and radios were currently unstable and ineffective
due to the change of market brought by the internet (Opreana & Vinerean, 2015; Tiago
& Verissimo, 2014). Moreover, product recommendations from social media
influencers received a higher trust rate compared to products recommended by families
or friends, resulting in 40% of consumers purchased the products used by social media
influencers (Sekhon et al., 2016).
Social media influencers were individuals who actively used their social media
accounts and often involved in a certain topic, as well as providing new information
(Loeper et al., 2014). Social media influencers often used products according to their
lifestyle and shared the reviews in their social media accounts. As stated by Munukka et
al. (2016), an influencer could evaluate a product that consumers would admire,
encouraging them to purchase it. Furthermore, they provided the latest information and
were able to influence consumers’ attitudes and behaviors (Liu et al., 2015). There were
several experts who believed that social media influencers would increase over the
years, since the use of social media influencers as a marketing media was becoming
popular in Indonesia. Until 2018, companies in Indonesia had allocated their marketing
budget for social media influencers around 20-30% higher than last year.
Social media influencers could also act as the third party that recommended and
described the products through the social media contents, which could influence the
consumers’ opinions, behaviors, and attitudes towards a product (Uzunoglu & Klip,
2014). Consumers who searched for information about the product often tought the
social media influencers as a trustworthy source of information (Wang et al., 2012; Park
& Cho, 2012).
Consumers also often received more messages about the product from social
media influencers than from companies (De Veirman et al., 2017). Solomon (2017)
stated that social media influencers influenced the perceptions of consumers towards the
product. It was also in accordance with a research by Ateke (2013), stating that the
higher the perception, the higher the product’s brand images used by influencers as
well. Furthermore, Godey et al. (2016) stated that the marketing practice through social
media influencers would result in a positive connection with the brand image, as the

77
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

information shared through the social media were more effective in affecting the
behaviors and purchase intention of consumers (Labrecque et al., 2013).
The result of the research stated that social media users indirectly viewed social
media influencers as role models in terms of lifestyle, as consumers could view
similarities between them and the influencers through the social media (Choi & Rifon,
2012). They also viewed the influencers as the desired ideal concept. According to
McCracken (1989), a consumer obtained a self-concept of satisfaction when using the
same product as their idols, while the self-concept of perception was reflected throught
the consumer’s behavior (Sumarwan, 2015). The similarities of self-concepts between
consumers and social media influencers could create the purchase intention (Choi &
Rifon, 2012; Ilicic & Webster, 2015).
Through the explanations about the result of the resaerch and theories about the
social media marketing, brand image, self-concept, and purchase intention, there were
no current researches that combined such variables to study the local cosmetic products.
The majority of local cosmetic companies in Indonesia were still dependent with the
traditional marketing strategy, such as televisions, newspapers, or radios, which was
outdated by the growing promotions of imported beauty products through online media,
that gathered a wider range of consumers. Imported beauty products also began to enter
in large shopping centers that can enhance the company’s image. According to Yunita
& Rosa (2016), there are several factors that make consumers tend to choose imported
beauty products, one of which is consumers feel unfamiliar to local beauty products
because the advertisements and promotions do not reach consumers. It was believed that
the promotion strategy through social media influencers could increase the sales of local
cosmetic products in Indonesia, as well as broadening the reach of the promotion in
order to compete with imported cosmetic products. Based on the research background,
the purpose of the research was to furtherly understand the concept of social media
influencers towards the brand image, self-concept and purchase intention of local
cosmetic consumers and formulate the implemented managerial implications. The
conceptual framework of the research, along with the hypotheses, could be seen in
Image 1.
H1 : Social media influencer has a significant positive effect on purchase intention
H2 : Social media influencer has a significant positive effect on brand image
H3 : Social media influencer has a significant positive effect on self-concept
H4 : Brand image has a significant positive effect on purchase intention
H5 : Self-concept has a significant positive effect on purchase intention

Figure 1 Conceptual Framework

78
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

Methods

The reasearch period was around a three-month period or during October until
December of 2018, using the quantitative approach. The research implemented the cross
sectional design, in order to discover the connection between the independent and
dependent variables. The primary data was obtained through online questionnaires in
the Google Docs link, shared in the social media applications such as WhatsApp,
Instagram, and Twitter. The questionnaires consisted of (closed) structural questions,
with the non-probability sampling technique.
Randomization is not important in nonprobability sampling when selecting
samples from the interest population. Subjective methods are instead used to determine
which elements are included in the sample. Therefore, non-probability sampling is a
sampling technique in which samples are collected in a process that does not give all
participants or units in the population equal opportunities to be collected. Determining
the nonprobability sampling technique that applicable to the study is very important for
a researcher. In this study, convenience sampling was chosen because the subject was
very close to the researcher where it was easier for researchers to access.
Convenience sampling is known as haphazard sampling or accidental sampling
which is a type of nonprobability or nonrandom sampling in which members of the
target population meet certain practical criteria, such as easy accessibility, geographical
proximity, availability at a certain time or desire to participate in the reseach.
Convenience sampling is a technique that makes it easy for researchers because it’s
pretty affordable, easy and the subjects are available. Researcher must essentially
explain how the sample will differ from the one selected randomly. It is also important t
describe the topic that could be excluded during the selection process or the topic
represented in the sample.
The criteria of respondents were at the age range of older than 17 years old and
had purchased products recommended by social media influencers for the last six
months. The chosen age range was in the assumption that consumers had a tendency to
make product purchase intentions during such age. The respondents of the research
were 219 respondents, based on the fact that the indicator was implemented around 5-10
times more than the assumed amount of the parameter.
The research implemented the dependent variable or the endogenous variable (Y),
which was the purchase intention, the independent variable or the exogenous variable
(X) (social media influencers); and the intervening variables which were the brand
image (Z1) and the self-concept (Z2). The operational definitions of the research could
be seen in Table 1. The data process in this research implemented the use of softtwares
such as Microsoft Excel 2007, SPSS and SmartPLS.

Table 1 Variables and Indicators

Variables Operational Definition Dimension
Social media A communication tool that can provide product Visibility (VS)
influencer (X) information to consumers Credibility (CR)
Attractiveness (ATR)
Power (PWR)

Brand Image (Z1) Consumer perceptions of brands are reflected Company image (CPR)
through brand associations in consumer Product image (CPU)
memory Service image (CPE)

79
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

Variables Operational Definition Dimension

Self-Concept (Z2) Perceptual (PER)
Character or personality of a person who shapes Conceptual (CEP)
perceptions of oneself so that it influences their Attitudinal (ATD)
consumption behavior
Purchase Intention (Y)
Consumer statement to oneself which reflects
the plan to purchase a number of products with
a certain brand

Validity Test
The validity test measured the accuracy between the research data and the data
occurred as the actual objects of the research. This stage was carried out using
Statistical Package Social Science (SPSS) from the Windows version. The measuring
instrument was the questionnaires, which would be valid if the questionnaires would
interpret something that was measured. The questionnaire can be said to be valid if the
question in the questionnaire is able to reveal something that will be measured by a a
questionnaire. Validity test is done by correlating item scores with total scores.
Spearman rank correlation if the data obtained is ordinal, whereas if the data obtained
by interval we can use product moment correlation. From the results, it can be seen also
the validity of each question item representing the research indicators.
Determaining the validity of an instrument is measured by comparing the r-
statistic with r-table. The determination is presented as follow:
 r-statistic > r-table or sig value r < 0.05 : valid
 r-statistic < r-table or sig value r > 0.005 : invalid
If there are invalid items, then that invalid items should be discarded and the analysis
process is repeated for only valid items. From that statement it can be concluded that if
there is an invalid item, it should be retest with valid item.
Reliability Test
Reliability is a test that shows if a measuring instrument used in the study has
reliability as a measuring instrument, including measuring the consistency of measuring
result from time to time or can be said that reliability is a test that show how consistent
the measurement results are when measured multiple times with the same measuring
instrument. A data was declared as ‘reliable’ if the same data was generated in the same
object and produced the same data of the research at different times. Meanwhile, the
questionnaires could be considered reliable if the respondents answered the
questionnaire consistently.
The reliability test would be conducted using SPSS for Windows. Research
require data that is truly valid and reliable. In the framework of this urgency, the
questionnaire was first tested into a sample of research trial before being used as a
primary research data. The trial was conducted to obtain evidence of the extent to which
the accuracy of measuring instrument in carrying out their measuring function. Question
are said to be reliable if an answer to the question was consistent or stable over time.
Measurement of reliability was done by one shot or just one measured using the SPSS
program on scale analysis. Where the reliability was measured by Cronbach Alpha
statistical tes, with a limitation of a variable said to be reliable if the value of Cronbach
Alpha > 60.

80
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

SEM Analysis
The SEM (Structural Equation Model) was a powerful statistical technique that
determined the measurement and structural models, as it was considered as the second
generation of multivariate analysis. There were two types of SEM, the SEM-based
covariance or better known as covariance-based structural equation modeling (CB-
SEM) and the partial least square path modeling (PLS-SEM). The CB-SEM was
basically a parameter-oriented, with an approach based on the covariance-based
relationship between the latent variables and reflective indicators. Meanwhile, the PLS
was more oriented to predictions with a variance-based approach, concerning the
relationship between latent variables and the reflective and formative indicators. PLS
was also known as a soft-modeling analysis method because it did not add the
assumption of the regression of the ordinary least square (OLS). The PLS could also be
used to explain the possible relationship between latent variables or predictions, as well
as confirming the theory.
Data that used in PLS-SEM does not have to meet the requirement for the
assumption of normality, thus PLS-SEM give concession to data that are not normally
distributed. PLS-SEM consist of three components, namely structural model,
measurement model and weighting scheme. This third part is a special feature of PLS-
SEM and does not exist in covariant-based SEM. PLS-SEM only allows the relationship
model between recursive variable only. This is the same as the path analysis model is
not the same as covariant based SEM that allows the occurrence of non-recursive
relationship. In the structural model, which is also called th einner model, all latent
variables are connected to one another based on the theory of substance. Laten variables
are divided into two, namely exogenous and endogenous. Exogenous latent variables
are cause or variables without being proceded by other variables with arrow leading to
other varables. Meanwhile, the measurement model, also known as the outer model,
connects all manifest variables or indicators with their latent variables. In the SEM-PLS
framework, one manifest varaible can only be connected with one latent variable. All
manifest variable that are associated with one latent variable are referred to as a block.
Thus each latent variable has a manifest variable block. A block must contain at least
one indicator.

Findings

Respondent Characteristic
There were 219 respondents, with the demographic aspect used to explain the
characteristics of respondents, including age, education level, types of work, and total
income per month. Based on Table 2, the respondents of the research were mostly at the
age between 20 to 30 years old, with the percentage of 74.9 percent. For the education
level, the respondents were dominantly bachelor degree graduates with the percentage
of 49.8 percent. The types of work of the most respondents were high school / college
student with the percentage of 50.7 percent. Lastly, the total income of respondents per
month was dominantly under Rp. 2,000,000 per month, with the percentage of 50.7
percent.

81
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

Table 2 Respondents’ characteristics

Category Profile Total (n) Percentage (%)
<20 46 21
Age 20-30 164 74.9
31-50 9 4.1
Senior High School 87 39.7
Diploma 13 5.9
Education
Bachelor 109 49.8
Master 10 4.6
Student 111 50.7
Employee 58 26.5
Civil Servant 14 6.4
Entrepreneur 6 2.7
Job Doctor 2 0.9
Housewife 14 6.4
Teacher 4 1.8
BUMN 2 0.9
Not Employee 8 3.7
< IDR2.000.000 111 50.7
IDR 2.000.000–IDR 81 37
5.000.000 25 11.4
Salary per Month
IDR 5.000.001–IDR 2 0.9
10.000.000
>IDR 10.000.000

Analysis of the Relationship between Social Media Influencer to Brand Image,

Self-Concept and Purchase Intention
The Structual Equation Modeling - Partial Least Square (SEM-PLS) analyzed
the relationship between variables, using the external model assessment criteria, or the
measurement model, and the assessment of the internal or structural model. The
measurement model connected all of the manifested variables or indicators with their
latent variables, by assessing the reliability and validity aspects. The structural model,
which connected all the latent variables based on the existing theories, could be
measured by analyzing the values of R2 and Q2. Furthermore, after the data had
underwent the measurement requirements, a hypothesis would concluded using the
bootstrapping method.
The social media influencers, as the endogenous variable of the research, had
four dimensions: Visibility (VS), Credibility (CR), Attraction (ATR), and Power
(PWR). The same also applied with the Purchase intention, as the exogenous variable,
which also had four indicators: transactional interest (MB1), referential interest (MB2),
preferential interest (MB3), and explorative interest (MB4). Furthermore, thee two
intervening variables, brand image and self-concept, had three dimensions respectively.
The brand image variable consisted of company image (CPU), product image (CPR)
and service image (CPE), while the self-concept variable had perceptual (PER),
conceptual (CEP), and attitudinal (ATD).

Measurement Model
The reliability test of the PLS SEM was interpreted with the value of Cronbach's
Alpha, in addition to the composite reliability value. This value reflects the reliability of
all indicators in the model. The minimum value wass 0.7, with the ideal value around

82
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

0.8 or 0.9 (Sarwono & Narimawati, 2015). Each latent variable must be able to explain
the variance of each indicator at least 50%. Therefore, the absolute correlation between
latent variable and the indicator must be more than 0.7. Reflective indicators should be
removed from th emeasurement model if the outside standard loadings value is below
0.4. As seen in Table 3, all the variables met the requirements of the reliability value,
reflecting the reliability of each indicator in the model.

Table 3 Reliability test results

Variables Composite Reliability Cronbach’s Rule of Conclusio
alpha thumb n
Social Media Influencer 0.904 0.884 >0.7 Reliable
Brand Image 0.883 0.851 >0.7 Reliable
Self-Concept 0.891 0.862 >0.7 Reliable
Purchase Intention 0.905 0.860 >0.7 Reliable

The validity test of the PLS SEM could be explained through the values of the
Average Variance Extracted (AVE) and loading factors, as well as the latent variables
of the indicators. The indicators were considered as valid if the values of the loading
factors were above 0.7 and could still be tolerated up to 0.5. This value describes
adequate validity which means that a latent variable is able to explain more than half of
the variants of the indicator in the average. As seen in Table 4, all the indicators had the
values of loading factors that met the requirements, which were above 0.5. These
indicators could later be used for further testings.

Table 4 Loading factor values

Indicators Value Conclusion Indicators Value Conclusion
VS1 0.727 Valid CPU3 0.710 Valid
VS2 0.872 Valid CPE1 0.809 Valid
VS3 0.804 Valid CPE2 0.863 Valid
CR1 0.856 Valid CPE3 0.790 Valid
CR2 0.874 Valid PER1 0.835 Valid
CR3 0.685 Valid PER2 0.690 Valid
ATR1 0.755 Valid PER3 0.822 Valid
ATR2 0.840 Valid CEP1 0.656 Valid
ATR3 0.815 Valid CEP2 0.895 Valid
PWR1 0.778 Valid CEP3 0.877 Valid
PWR2 0.716 Valid ATD1 0.846 Valid
PWR3 0.840 Valid ATD2 0.837 Valid
CPR1 0.771 Valid ATD3 0.741 Valid
CPR2 0.786 Valid MB1 0.867 Valid
CPR3 0.830 Valid MB2 0.876 Valid
CPU1 0.864 Valid MB3 0.808 Valid
CPU2 0.837 Valid MB4 0.805 Valid

The AVE value of around 0.5 stated that one latent variable represented more
than half of the variants of the average indicators (Sarwono and Narimawati, 2015). The
AVE value, as seen in Table 5, showed that the value of the loading factors in all
indicators has met the requirements above 0.5, therefore that the indicator could be
considered as ‘valid’.

83
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

Table 5 Result of Average Variance Extracted (AVE) values

Variable AVE Rule of Thumb Conclusion
Attitudinal 0.6551 >0.5 Valid
Attraction 0.6465 >0.5 Valid
Service Image 0.6749 >0.5 Valid
Company Image 0.6339 >0.5 Valid
Product Image 0.6508 >0.5 Valid
Conceptual 0.6668 >0.5 Valid
Credibility 0.6553 >0.5 Valid
Purchase Intention 0.7053 >0.5 Valid
Perceptual 0.6164 >0.5 Valid
Power 0.6075 >0.5 Valid
Visibility 0.6450 >0.5 Valid

Structural Model
The evaluation of the structural model could be conducted by analyzing the
values of R2 and Q2. Based on data processing using SmartPLS 3.0, the R2 value was
generated, as seen in Table 5. According to Sarwono and Narimawati (2015), the value
of R2 measured the level of variation in each variable. As seen in Table, the latent
variable of the brand image was around 36.3%, indicating that the brand image variable
could be explained by the influencer variable of 36.3% while the remaining 63.7% was
explained by other factors. Furthermore, a consumer’s self-concept could be explained
by an influencer of 6.9%. Meanwhile, the purchase intention variable had the value of
R2 around 49.2%, indicating that the variables of influencers, brand image and self-
concept could affect the purchase intention around 49.2%. On the other hand, the rest
was explained by other factors not included in the research model.
The acquisition of the R2 value calculated the value of Q2 or the relevance of
predictions. The relevance value was used to analyze the relative influence of structural
models on the measurement of the endogenous variable observation (Sarwono and
Narimawati, 2015). The value of Q2 above 0 indicated that the model had a good
predictive relevance, while the value of Q2 below 0 would lack a good predictive
relevance. The calculation of the Q2 rate could be seen below.

Q2 = 1 – (1-R12)(1-R22) … (1-Rp2)
Q2 = 1 – (1-0.363) x (1-0.069) x (1-0.492)
= 0.699
The result of the Q2 acquisition was based on the calculation, which was 0.699. This
indicated that the magnitude of the data diversity from the research could be explained
by the structural model of 69%, while the remaining 31% showed that other variables
not included in the model could affect the purchase intention of local cosmetic
consumers.
After fulfilling the measurement conditions, the next stage was the hypothesis
testing, using the bootsrapping method on SmartPLS. According to Wong (2013), the
bootstrapping method was performed by repeatedly taking new N samples from data of
size n, in this case by obtaining the sample points from the data one by one until n
times. The hypothesis would be decided by comparing the values of t count with t table,
with the level of confidence of α = 0.05 and a t table value of 1.96. The results of the t-
test could be seen in Table 6.

84
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

Tabel 6 Statistical hypotheses test

Hypothesis Variables Original T-value Conclusion
Sample
H1 Social Media Influencer -> Purchase 0.067 0.977 Rejected
Intention
H2 Social Media Influencer->Brand Image 0.603 9.346 Accepted
H3 Social Media Influencer ->Self-Concept 0.263 3.163 Accepted
H4 Brand Image -> Purchase Intention 0.674 11.545 Accepted
H5 Self-Concept -> Purchase Intention -0.0482 0.951 Rejected

The result of the hypothesis test indicated that social media influencers did not
had significant influences on the purchase intention (t-value = 0.977), but had
significant positive influences in the brand image (t-value = 9.346) and self-concept (t-
value = 3.163). Furthermore, the brand image had a significant positive influence
towards the purchase intention (t-value = 11,545), while the self-concept did not had a
significant influence (t-value = 0.951). The result stated that H1 and H5 were rejected,
while H2, H3 and H4 were accepted.
In addition to conducting a hypothesis test, the SmartPLS calculation was also
obtained through the calculation of indirect influences, in order to measure the strength
of the connection between the mediator variable and other variables. Zhou et al. (2010)
stated that the mediation was occured when a variable affected the relationship between
the endogenous and exogenous variables. In this research, there were two variables used
as the mediation or referred as the intervening variables, brand image and self-concept
variables, that mediated the social media influencer variable with the purchase intention.
The indirect effect value was also obtained through the bootstrapping test.
The magnitude rate of the indirect influence of social media influencers towards
the purchase intention through a brand image was 0.407, This rate was higher than the
value of the direct influence of social media influencers towards the purchase intention
(0.0677), stating that social media influencers could influence indirectly on the purchase
intention through the brand image, in a positive and significant relationship. Meanwhile,
the influence of social media influencers towrads the purchase intention through self-
concept had a negative relationship of -0.0127, lower than the direct influence of social
media influencers towards the purchase intention (0.0677), stating that social media
influencers could not influence the purchase intention through the self-concept variable.

Discussions

The first result of the hypothesis test showed that there were no significant
relationships between the influencers towards the purchase intention (H1 was rejected).
The result was similar with a research by Johansen and Guldvik (2017), stating that
there were no significant relationships between the influencers towards the purchase
intention, but there would be an indirect influence through the attitudes toward
behavioral variables. The influencers with firm opinions about their product reviews,
along with equal negative and positive reviews could obtain positive results to
consumers’ attitudes towards influencers (Johansen & Guldvik, 2017). Based on the
indirect effect test, the influencers' social media accounts could influence the purcahse
decision indirectly through the brand image variable, as analyzed by Hassan and Jamil
(2014). Their research suggested that influencers could not attract consumers directly to

85
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

commence purchases, but after watching good reviews from influencers, consumers
would relate their perceptions with their own on the brand image, encouraging
consumers to purchase the products.
The second result showed that there was a significant positive relationship
between social media influencers and the brand image (H2 was accepted). Godey et al.
(2016) stated that the marketing practice through social media influencers had a
significant influence on the brand equity of a product, considering that the brand image
was an important aspect in the brand equity. Meanwhile, according to Hariyanti and
Wirapraja (2018), social media influencers could increase the sales significantly and
display a good brand image for consumers. In other words, they acted as intermediaries
between companies and consumers in order to convey the product information to the
consumers well. A research conducted by Barreda et al. (2015) showed that a word of
mouth on social media could improve the company's brand image. In addition, the
presence of influencer social media created a two-way communication that encouraged
consumers to be indirectly connected with the company (Howard et al, 2014).
The third result showed that there was a significant positive relationship between
social media influencers and the self-concept (H3 was accepted). This result was similar
with the research conducted by Choi and Rifon (2012), stating that influencers had a
significant relationship with consumers’ self-concepts. Consumers would find
similarities with social media influencers that could strengthen their own self-concepts.
A strong self-concept of a consumer would influence the consumers wore clothes and
used services and products (Khan, 2016).
The fourth result showed a significant positive relationship between brand image
and the purchase intention (H4 was accepted). According to a resaerch by Anggraeni
and Asnawati (2017), a brand image could be implemented as an intervening variable
between the endorsement and purchase intention because of the significant positive
relationship between the brand image and consumers’ purchase intention. Another
research by Vien et al. (2017) stated that the brand image was one of the factors of
brand attitude with a significant positive relationship to the purchase intention.
The fifth results furtherly indicated that there was no significant relationship
between the self-concept and purchase intention (H5 was rejected). In accordance with
Octaviani and Kartasasmita (2017), the self-concept did not significantly influence a
consumption behavior. Another research by Gumulya and Widiastuti (2013) stated that
the self-concept only represented the 0.6% of the consumers’ behaviors. Moreover,
Khan (2016) stated that consumers did not only match products with the self-concept in
determining a purchase, but also matched products with the perception of brand image
before having the purchase intention.

Managerial Implication

Demographically, the segmentation for local cosmetic products was mainly

women around 20 to 30 years, with the total income of less than IDR 2.000.000 per
month and with higher education levels. The psycho-physical segmentation was the
study environment that was interested in the cosmetics world and were active in social
media. Nowadays, cosmetic users were increased and were not limited to career women,
since high school and college students also started on using cosmetics a lot. This
showed that the target consumers from teenage girls to adult women, who were active
and had careers, were the right targets for the local cosmetics market. Cosmetics with

86
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

premium quality, various color choices, elegant packagings, and affordable prices were
the types of cosmetics that consumers tend to be interested, making the concept applied
in product positioning.
Conclusion and Recommendation

Conclusion
Based on the results of the research, there were various conclusions to be
formulated. One of which was social media influencers had a significant positive effect
on the brand image of local cosmetic products and consumers’ self-concept. Another
conclusion was the brand image had a significant positive effect on the consumers’
purchase intention, in contrast with the effect of the self-concept. In addition, the results
of the following researches stated that the influencers did not significantly influence
consumers on their purchase intention, but rather produced a significant positive effect
indirectly through the brand image variable. Demographically, the segmentation of local
cosmetic products was mainly women around 20 to 30 years, with the total income of
less than IDR 2.000000 per month and with higher education levels, with the target
consumers of teenage girs to adult women, who were active and had careers. Lastly, the
suitable positioning was premium quality cosmetic products, with various color choices,
elegant packagings, and affordable prices.

Recommendation
Based on the results of the research, there were several suggestions that could be
implemented in further researches. Firstly, it was to recognize the name of one of social
media influencers specifically in order to learn more about the influencers' characters,
such as on improving the product's brand image. Since the brand image was a part of
brand equity, the research model could be expanded by researching more about social
media influencers with brand equity and purchase intention. Further researches were
also recommended to add other compatible variables and produce better results.
References

Anggraeni, M., Asnawati, A. (2017). Pengaruh Celebrity Endorsement terhadap

Purchase Intention dengan Brand Image sebagai Variabel Intervening (Studi
Kasus Iklan Produk Perawatan Kecantikan Wardah). Prosiding Seminar Nasional
Manajemen dan Ekonomi Bisnis, 1: 303-312.
Ateke, B.W. (2013). Marketing message effectiveness: The relevance of celebrity
endorsement. The University Advanced Research Journal, 11, 14-22.
Barreda, A.A., Bilgihan A., Nusair K., Okumus F. (2015). Generating Brand Awareness
n Online Social Networks. Computers in Human Behaviour, 50, 600-609.
doi: 10.1016/j.chb.2015.03.023
Choi S.M., Rifon N.,J. (2012). It Is A Match: The Impact Of Congruence Between
Celebrity Image And Consumer Ideal Self On Endorsement Effectiveness.
Psychology and Marketing, 29(9), 639-650. doi: 10.1002/mar.20550.
Dân H.T., Nam L.,G. (2018). Impact of social media influencer marketing on consumer
at Ho Chi Minh City. The International Journal of Social Sciences and
Humanities Invention, 5(05), 4710-4714. doi: 10.18535/ijsshi/v5i5.10

87
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

De Veirman M., Cauberghe V., Hudders L. (2017). Marketing through Instagram

influencers: the impact of number of followers and product divergence on brand
attitude. International Journal of Advertising, 36(5), 798-828. doi:
10.1080/02650487.2017.1348035
Godey B.N., Manthiou A., Pederzoli D., Rokka J., Aiello G., Donvito R., Singh R.
(2016). Social Media Marketing Efforts of Luxury Brands: Influence on Brand
Equity and Consumer Behaviour. Journal of Business Research, 69(12), 5833-
5841. doi: 10.1016/j.jbusres.2016.04.181
Gumulya J., Widiastuti M. (2013). Pengaruh Konsep Diri Terhadap Perilaku Konsumtif
Mahasiswa Universitas Esa Unggul. Jurnal Psikologi, 11(1),50-65.
Hassan S., Jamil R.A. (2014). Influencer of Celebrity Endorsement on Consumer
Purchase Intention for Existing Product: A Comparative Study. Journal of
Management Info, 4(1), 1-23.
Hariyanti N.T., Wirapraja A. (2018). Pengaruh Influencer Marketing sebagai Strategi
Pemasaran Digital Era Modern (Sebuah Studi Literatur). Jurnal EKSEKUTIF,
(15), 133-146.
Howard T., Mangold W.G., Johnston T. (2014). Managing your social campaign
strategy using Facebook, Twitter, Instagram, YouTube & Pinterest: An interview
with Dana Howard, social media marketing manager. Business Horizons, 57, 657-
665.
Ilicic J., Webster C. (2015). Consumer Values Of Corporate And Celebrity Brand
Associations. Qualitative Market Research: An International Journal, 18(2), 164-
187. doi: 10.1108/QMR-06-2013-0037
Johansen I.K., Guldvik C.S. (2017). Influencer Marketing and Purchase Intentions
(thesis). Retrieved from https://brage.bibsys.no/xmlui/handle/11250/2453218
Khan M. (2006). Consumer Behaviour and Advertising Management. New Delhi, India:
New Age International.
Labrecque L.I., Esche J., Mathwick C., Novak T.P., Hofacker C.F. (2013). Consumer
Power: Evolution in the Digital Age. Journal of Interactive Marketing, 27, 257-
269. http://dx.doi.org/10.1016/j.intmar.2013.09.002.
Liu S., Jiang C., Lin Z., Ding Y., Duan R., Xu Z. (2015). Identifying effective
influencers based on trust for electronic word-of-mouth marketing: A domain-
aware approach. Information Sciences, 306, 34-52. doi:10.1016/j.ins.2015.01.034
Loeper A., Steiner J., Stewart C. (2014). Influential Opinion Leaders. Economic
Journal, 124 (581), 1147–1167. doi: 10.1111/ecoj.12100.
McCracken G. (1989). Who Is The Celebrity Endorser? Cultural Foundations Of The
Endorsement Process. Journal of Consumer Research, 16, 310–321. doi:
10.1086/209217.
Munukka J., Uusitalo O., Toivonen H. (2016). Credibility of a Peer Endorser and
Advertising Effectiveness. Journal of Consumer Marketing, 33(3), 182-192. doi:
10.1108/JCM-11-2014-1221.
Opreana A., Vinerean S. (2015). A new development in online marketing: Introducing
digital inbound marketing. Expert Journal of Marketing, 3(1), 29-34.
Park H., Cho H. (2012). Social Network Online Communities; Information Sources for
Apparel Shopping. Journal of Consumer Marketing, 29(6), 400-411. doi:
10.1108/07363761211259214
Sarwono J., Narimawati S. (2015). Membuat Skripsi, Tesis dan Disertasi dengan
Partial Least Square SEM (PLS-SEM). Yogyakarta, Indonesia: Andi.

88
Hermanda, Sumarwan, & Tinaprilla / Journal of Consumer Sciences, 2019, Vol. 04, No. 02

Sekhon T., Bickart B., Trudel R., Fournier S. (2016). Being a likable braggart: How
consumers use brand mentions for self-presentation on social media. Consumer
Psychology in a Social Media World, 23-39.
Solomon M.R. (2017). Consumer Behavior: Buying, Having, and Being 12th Edition.
United States: Pearson Education.
Sumarwan U. (2015). Perilaku Konsumen: Teori dan Penerapannya dalam Pemasaran.
Bogor, Indonesia: IPB Press.
Octaviani C., Kartasasmita S. (2017). Pengaruh Konsep Diri Terhadap Perilaku
Konsumtif Pembelian Produk Kosmetik Pada Wanita Dewasa Awal. Jurnal
Muara Ilmu Sosial, Humaniora, dan Seni, 1(2),126-133.
Tiago M., Veríssimo J. (2014). Digital Marketing and Social Media: Why Bother?.
Business Horizons, 57(6), 703-708. doi: 10.1016/j.bushor.2014.07.002
Uzunoğlu E., Klip M. (2014). Brand Communication Through Digital Influencers.
International Journal of Information Management, 34 (5), 592–602.
doi: 10.1016/j.ijinfomgt.2014.04.007.
Vien C.V., Yun C.T., Fai P.L. (2017). The Effect of Celebrity Endorsement on Brand
Attitude and Purchase Intention. Journal of Global Business and Social
Entrepreneurship (GBSE), 1(4), 141–150.
Wang X., Yu C., Wei Y. (2012). Social Media Peer Communication and Impacts on
Purchase Intentions: A Consumer Socialization Framework. Journal of Interactive
Marketing, 26(4), 198-208. doi: 10.1016/j.intmar.2011.11.004.
Wong K.K. (2013). Partial Least Square Structural Equation Modeling (PLS-SEM)
Techniques Using SmartPLS. Marketing Bulletin, 24(1), 1-32.
Yunita D., Rosa A. (2016). Pengaruh Mencari Variasi, Ketidakpuasan dan Ketidak
Tersediaan Produk terhadap Perpindahan Merk. Jurnal Manajemen dna Bisnis
Sriwijaya, 14(4), 539-558.
Zhou X., Lynch Jr. JG., Chen Q. (2010). Reconsidering Baron and Kenny : Myths and
Truths about Mediation Analysis. Journal of Consumer Research, 37, 197-206.
doi: 10.1086/651257.

89