Celebrity Endorsement and Brand Credibility in the Carbonated Soft Drink Industry in Sri Lanka

This study aims to assess the degree of celebrity endorsement and brand credibility in the carbonated soft drink industry in Sri Lanka. Questionnaires were used to collect the data from youth educated consumers of Sri Lankan Universities. Unidimensionality, reliability and validity assessments were performed to confirm scale reliability and validity, whereas one sample t-test was conducted to test hypotheses using 338 responses. The results revealed that the degree of celebrity endorsement and brand credibility were at low levels in the carbonated soft drink industry in Sri Lanka, especially among youth educated consumers. The findings of this empirical study can be effectively utilized by marketing managers in the carbonated soft drink industry in Sri Lanka, in their effort to develop and implement successful and strong celebrity endorsement and brand credibility related strategies.


Introduction
The food and beverage industry performs a vital role in expanding economic opportunity and it is the fastest growing sector across the world as well as Sri Lanka (Gehlhar & Regmi, 2005;Peiris, 2014;Vlachvei & Oustapassidis, 1998) since food and beverages are common to human life and health.Indeed, soft drinks sector is one of the fastest growing, most innovative and rapidly changing areas in the food and beverage industry (Gehlhar & Regmi, 2005;Leatherhead Food Research, 2014;United Nations Industrial Development Organization, 2014).Bloomberg (2011) implies that the carbonated soft drink market is the largest segment in the global soft drink market, whereas Bruce (2011) argues that carbonated soft drink industry is one of the significant and growing industries in Sri Lanka, especially among the educated youth generation.Notably, Coca Cola and Pepsi Cola have strong global brand value in the global carbonated soft drink industry (Smithers Pira, 2014), while Elephant House, Coca Cola and Pepsi Cola have higher brand value in the carbonated soft drink industry in Sri Lanka (Mendis, 2012).Therefore, this study considers the carbonated soft drink industry, which includes three key players and brands namely, Elephant House, Coca Cola and Pepsi Cola.Nevertheless, many companies including the companies who are in the carbonated soft drink industry use celebrity endorsement as one of the effective advertising strategies to get competitive advantage (Kamins, Brand, Hoeke, & Moe, 1989;Spry, Pappu, & Cornwell, 2011).One-in-four advertisements use celebrity endorsement (MarketWatch, 2006;Pughazhendi & Ravindran, 2012), since it is a more effective tool in advertising (Muda, Musa & Putit, 2012).In fact, Friedman and Friedman (1979) indicate that celebrities are more effective than other types of endorsers, such as 'the professional expert', 'the company manager', or 'the typical consumer'.To illustrate this effectiveness of celebrity endorsement, Seno and Lukas (2007) cite Thomaselli and Neff (2004), '[i]n 2004, Gillette signed an endorsement deal with soccer celebrity David Beckham worth between $US 30 and $US 50 million'.For this effectiveness of celebrity endorsement, companies who are in the soft drink and carbonated soft drink industry spend huge amounts on celebrities to endorse their product (VinayagaMoorthy & Madevan, 2014), but what is the degree of youth educated consumers' perceived evaluation on celebrity endorsement in the carbonated soft drink industry in Sri Lanka?Extant published research does not answer this question.
According to Spry et al. (2011) celebrity endorsement perhaps associate with a higher level of brand credibility.In fact, firms can keep their promises via the brand, which is known as brand credibility (Erdem & Swait, 2004).This brand credibility is significant to the carbonated soft drink industry (Alam, Arshad & Shabbir, 2012), but what is the degree of youth educated consumers' perceived evaluation on brand credibility in the carbonated soft drink industry in Sri Lanka?Extant published research does not answer this question.Further, these preceding two questions convey the two gaps in the relevant literatures.Thus, this study attempt to answer these two questions and fill these gaps by empirically and comprehensively examining the degree of celebrity endorsement and brand credibility in the carbonated soft drink industry among the educated youth generation in Sri Lanka, since this industry is one of the significant industries in Sri Lanka (Bruce, 2011).
The rest of this study is organized as follows: the next section literature review, followed by method, data analysis and results, and discussion.The final section offers the limitations and further research directions.

Celebrity Endorsement
Nowadays most of the marketers use celebrity endorsement as an effective promotional tool.McCracken (1989) defines celebrity endorsement as 'any individual who enjoys public recognition and who uses this recognition on behalf of a consumer good by appearing with it in an advertisement'.
The three sub dimensions of Ohanian (1990) are used by many recent studies (e.g., Hussain, Mehmood & Khan, 2012;Lord & Putrevu, 2009;Muda, Musa, Mohamed & Borhan, 2014;Spry et al., 2011;Wei & Lu, 2013) to measure the construct of celebrity endorsement and this study also considers attractiveness, expertise and trustworthiness are the sub dimensions of celebrity endorsement as replication to the existing literature (see Figure 1).Furthermore, operational definitions for the dimensions of celebrity endorsement (i.e., attractiveness, expertise and trustworthiness) are given in Table 1.Since, celebrity endorsement in advertisements is important to the carbonated soft drink industry (VinayagaMoorthy & Madevan, 2014), there is a need to assess the degree of celebrity endorsement in this industry.Notably, the companies who are in the carbonated soft drink industry spend around 35 per cent of the total cost for advertising and they allocate millions of dollars for celebrity endorsement in advertisements and therefore, celebrity endorsement is high degree in the carbonated soft drink industry (e.g., Ojha, 2013;Singh, 2012;VinayagaMoorthy & Madevan, 2014).Further, even though there has been no recent statistics available to show the amount spent on celebrity endorsement in the Sri Lankan carbonated soft drink industry, and generally the celebrity endorsement is high degree in the carbonated soft drink industry in Asian countries such as in South Korea (e.g., Money, Shimp & Sakano, 2006) and India (e.g., Goutam, 2013) and thus, the researcher assumed that the celebrity endorsement is high degree in the carbonated soft drink industry in Sri Lanka, especially among the educated youth generation.According to the above argument the first hypothesis is advanced.

H1:
The degree of youth educated consumers' perceived evaluation on celebrity endorsement is at a high level in the carbonated soft drink industry in Sri Lanka.
Indeed, firms make a huge investment on celebrity endorsement, which is an effective advertising strategy (Elberse & Verleun, 2012).These investments for celebrity endorsement perhaps associate with a higher level of brand credibility (Spry et al., 2011).

Brand Credibility
Credibility is one of the significant attributes of a brand (Maathuis, Rodenburg & Sikkel, 2004;Spry et al., 2011).Erdem and Swait (2004) define brand credibility is the believability of a brand, which has the willingness (i.e., trustworthiness) and ability (i.e., expertise) to deliver what has been promised.
Analysis of past scientific literature in consumer marketing, global marketing, marketing and logistics, and retailing and consumer services has found that the topic of brand credibility rest on Erdem and Swait (2004) brand credibility model (e.g., Alam et al., 2012;Kemp & Bui, 2011;Sweeney & Swai, 2008;Wang & Yang, 2010).Indeed landmark study of Hovland et al. (1953) is adapted to the brand context by Erdem and Swait (2004) and they develops brand credibility model.Accordingly, brand credibility comprises two key dimensions: trustworthiness and expertise (Erdem & Swait, 2004).
The two dimensions of brand credibility (Erdem & Swait, 2004) are used by many recent brand credibility related studies (e.g., Alam et al., 2012;Kemp & Bui, 2011;Spry et al., 2011;Sweeney & Swait, 2008;Wang & Yang, 2010) to measure the construct of brand credibility.Thus, this study also considers trustworthiness and expertise as dimensions of brand credibility as replication to existing literature (see Figure 2).In addition, operational definitions for trustworthiness and expertise are given in Table 2.  Expertise means that capable or ability to continuously deliver what has been promised.
Erdem and Swait (2004) Erdem and Swait (2004) Brand credibility performs an important role in the beverage industry, especially it is vital for and high degree in

Expertise
Trustworthiness the carbonated soft drink industry (e.g., Alam et al., 2012).Though there has been no recent statistics available to show the degree of brand credibility in the Sri Lankan carbonated soft drink industry, and generally the brand credibility is high degree in the carbonated soft drink industry in Asian countries such as in Pakistan (e.g., Alam et al., 2012) and thus, the researcher assumed that the brand credibility is high degree in the carbonated soft drink industry in Sri Lanka, especially among the educated youth generation.According to the above argument the second hypothesis is advanced.

H2:
The degree of youth educated consumers' perceived evaluation on brand credibility is at a high level in the carbonated soft drink industry in Sri Lanka.

Sample
Previous carbonated soft drink related studies considered educated youth generation as target population including, especially university undergraduate students (Atilgan, Aksoy & Akinci, 2005;Koordeman, Anschutz, Van Baaren & Engels, 2010;Mathew, Casamassimo & Hayes, 2002;Netemeyer et al., 2004).Therefore, this study too considered educated youth generation studying in Sri Lankan universities (undergraduate students) as target population.Clearly, the population was defined as 'educated youth generation between the age of 18 and 26 who had been studying in Sri Lankan universities (i.e., undergraduate students), and who has consumed carbonated soft drinks'.The final sample was selected using a judgmental sampling method, where number of sample was 381 undergraduates from Sri Lankan universities.

Measures and Measurement
Measurement scales adapted from previous studies are summarized in Table 3. Celebrity endorsement was measured using fifteen item scale devised by Ohanian (1990), which measures celebrity endorsement dimensions of attractiveness, expertise and trustworthiness.Further, expertise and trustworthiness were considered sub-dimensions of brand credibility, and six items were used for measuring brand credibility (Erdem & Swait, 2004).As can be seen in Table 3, the measurement scales for the constructs (21 items) were used to develop a questionnaire in English.This questionnaire is divided into three sections (i.e., Section A-C).The section A was developed to measure celebrity endorsement, followed by section B, which was developed to measure brand credibility.Finally, section C was developed to measure demographics, including, gender, age and parental income.All items were measured on a five-point Likert scale ranging from '1 = Strongly Disagree' to '5 = Strongly Agree'.

Pre-Test
First, the developed 21 items questionnaire was pre-tested in a pilot study by six marketing academics and six marketing managers.Based on these twelve experts' comments, the wording and layout of the questions were rearranged.Second, the modified questionnaire was subject to pre-test in a pilot study with a sample of 38 undergraduate students from universities in Sri Lanka.Based on the pilot study with undergraduates reworded several questions to enhance the clarity.Then, the main study followed with 21 items.

Data Collection
Questionnaires were distributed to 381 samples of undergraduate students and 338 usable samples were obtained after excluding the incomplete ones, yielding an 89 per cent response rate.

Statistical Treatment of Data
This study used both descriptive statistics (i.e., mean and standard deviation) and inferential statistics (i.e., one sample t-test at .05 significance level, by using SPSS 19 software to test the hypotheses namely, H1 and H2), where these are the major two types of statistics (Owen, 2009).Further, the mean scores of the 5-point Likert scale measurement (mean scores for low, moderate and high levels) were adopted from Hair, Black, Babin, Anderson and Tatham (2006) to determine the degree of celebrity endorsement and brand creditability (e.g., Thusyanthy & Senthilnathan, 2013;Thusyanthy & Tharanikaran, 2015;Thusyanthy, 2016).These mean scores as decision criteria were used in the one sample t-test statistical analysis in order to test the hypotheses namely, H1 and H2 (see Table 4).

Sample Profile
The sample (n = 338) was composed by 56.8 percent of male and 43.2 percent of female.With respect to age group, high portion of the respondents was aged between 24 and 26 years (41.72 percent), whereas 27.81 percent and 30.47 percent of respondents age intervals 18-20 and 20-23, respectively.Regarding the parents income level, 13.61 percent of the respondent's parents had monthly income below 30000; 23.96 percent, 30000-99000; 43.79 percent, 100000-169000; and 18.64 percent, above 170000.

Unidimensionality Assessment
Unidimensionality refers that all items forming an instrument are measuring the same thing in common (Briggs & Cheek, 1986;Cortina, 1993;Hattie, 1985;Lumsden, 1961).Unidimensionality can be established via principal component factor analysis with Varimax rotation to allow for inter-correlations among the dimensions (Brown et al., 1997;Chung & Pennebaker, 2008;Johnson, 2005;Zaichkowsky, 1985).Thus, 338 samples in this study were also subject to factor analysis by using principle component analysis with Varimax rotation procedure in order to explore the unidimensionality of each scale by applying SPSS 19 software.
After determining the sampling adequacy and appropriateness of the factor analysis, every single item for each dimension of celebrity endorsement and brand credibility was individually submitted for a factor analysis by using principle component analysis with Varimax rotation (see Table 5 and Table 6).The factors/components 1, 2 and 3 in Table 5 were interpreted as expertise, trustworthiness and attractiveness, respectively, whilst factors/components 1 and 2 in Table 6 were defined as trustworthiness and expertise, respectively.
It is notable that factors needed eigenvalue greater than one, and number of factors extracted should account for as a minimum 50 per cent or 60 per cent of the total variance interpreted (Hair, Anderson, Tatham, & Black, 1998;Kaiser, 1960;Streiner, 1994;Ledesma & Valero-Mora, 2007).As consistent with these criteria, the results of factor analysis revealed that a three factor with 15 items (Table 5) and a two factor with 6 items (Table 6) representing the eigenvalues greater than one, and a three factor accounted for 58.278 per cent of the accumulated variance, while a two factor explained 64.707 per cent of the total variance.Accordingly, as suggested in eigenvalue rule by Kaiser (1960), the three factors in Table 5 and the two factors in Table 6 were selected.
As can be seen in Table 5 and Table 6, all factor loadings were significant at the 0.05 level and factor loadings of far above 0.40 were presented in bold-face type.Indeed, all individual item factor loadings for their respective components were far above the required value of 0.40 (Bagozzi & Baumgartner, 1994;Clark & Watson, 1995) and thus, there was no necessity for eliminating any items.In fact, these results ultimately lead to an acceptance of the unidimensionality of two constructs, which were used in this study.

Reliability and Validity
Beckstead (2013) suggested that Cronbach's alpha coefficient is most commonly used index of reliability.Cronbach's alpha for each dimension ranged from 0.698 to 0.844.Since, these reliability estimates exceeded the recommended standard of 0.70 (Nunnally, 1978;Thusyanthy & Senthilnathan, 2012cite George & Mallery, 2003) and thus, satisfactory reliability for the dimensions achieved (see Table 7).
Discriminant validity was assessed using the procedure has been suggested by Fornell and Larcker (1981).Fornell and Larcker (1981) suggest that the average variance extracted for a construct should be substantially higher than the squared correlation of two constructs.The results for constructs and dimensions indicated that sufficient discriminant validity because the average variance extracted for each dimension was greater than the squared correlations between dimensions (see Table 8).

Descriptive Statistic Analysis and Results
Using SPSS 19 software, descriptive statistics has been calculated.In fact, the means for celebrity endorsement and brand credibility were 2.18 and 2.33, respectively on a five-point Likert scale and thus, indicating lower level of means (Hair et al., 2006).Further, corresponding standard deviations for celebrity endorsement and brand credibility were 0.63 and 0.61, respectively, which reflected the acceptable variability within the data set.

Inferential Statistic Analysis: Hypothesis Testing and Results
To test H1 and H2, this study used one sample t-test (see Table 9).The decision criteria in Table 4 were adopted to determine the degree of youth educated consumers' perceived evaluation on celebrity endorsement and brand credibility in the carbonated soft drink industry in Sri Lanka.As indicated in the Table 9, significance values of 0.000 for both lower and upper boundaries (i.e., p < 0.05) in celebrity endorsement and brand credibility shows that there are significance differences between assumed means and observed means.Clearly, the observed means 2.18 and 2.33 for celebrity endorsement and brand credibility, respectively have fallen the low level range of 1-2.5, which has represented the lower and upper boundaries of assumed mean.It is thus, the degree of youth educated consumers' perceived evaluation on celebrity endorsement and brand credibility are in low level in the carbonated industry in Sri Lanka.Notably, the one sample-t-test failed to offer support for H1 and H2.

Theoretical Implications
H1 predicted that the degree of youth educated consumers' perceived evaluation on celebrity endorsement is at a high level in the carbonated soft drink industry in Sri Lanka; however, there was no support for this high level, and not consistent with previous studies.In particular, celebrity endorsement is high degree in the carbonated soft drink industry in Asian countries such as in South Korea and India (e.g., Goutam, 2013;Money et al., 2006).Therefore, the researcher assumed that the celebrity endorsement is high degree in the carbonated soft drink industry in Sri Lanka, especially among the educated youth generation.However, the H1 not supported.There is a possible explanation for this result.This study considered educated youth generation as target population, since they have more rational knowledge than typical consumers (Martin, 2005), which may lead to have less credible for the celebrity endorsement, and thus the results suggested that low level.The celebrity endorsement might have had high level, if the target population whole Sri Lankan consumers.Nevertheless, this study offers the empirical support to close the first gap.Consequently, the answer gets for the first research question is that celebrity endorsement at a low level in the carbonated soft drink industry in Sri Lanka, especially among the educated youth generation.
The results indicate that the degree of youth educated consumers' perceived evaluation on brand credibility is at a low level in the carbonated soft drink industry in Sri Lanka, and not provided support for H2, and not consistent with past study.In fact, Alam et al. (2012) argue that brand credibility is at high degree in the carbonated soft drink industry, especially in Pakistan.Accordingly, the researcher assumed that the brand credibility is high degree in the carbonated soft drink industry in Sri Lanka, especially among the educated youth generation.However, the H2 not supported.Nevertheless, the possible explanation for this result is that this study selected educated youth generation as target population.The brand credibility might have had high level, if the target population was the whole Sri Lankan consumers.Moreover, this empirical study assists to close the second gap.Consequently, the answer gets for the second research question is that brand credibility at a low level in the carbonated soft drink industry in Sri Lanka, especially among the educated youth generation.

Managerial Implications
Marketing managers in the carbonated soft drink industry in Sri Lanka can effectively use the finding in this empirical study to develop and implement successful celebrity endorsement and brand credibility related strategies.Present study indicates that the degree of youth educated consumers' perceived evaluation of celebrity endorsement and brand credibility are in the low level in the carbonated soft drink industry in Sri Lanka.In fact, the celebrity endorsement is an effective advertising tool in the carbonated soft drink industry (VinayagaMoorthy & Madevan, 2014), and brand credibility performs a vital role in this same industry (Alam et al., 2012), however the degree of both celebrity endorsement and brand credibility in the carbonated soft drink industry are in low level among the educated youth generation in Sri Lanka.Therefore, the marketing managers in this industry should concentrate these findings and try to use more credible celebrity endorsers in advertisements based on their attractiveness, expertise and trustworthiness, and try to build strong brand credibility to get the sustainable competitive advantage among the educated youth generation in Sri Lanka.

Limitations and Future Research Directions
The limitation of this study was mainly related to the generalization of findings due to the usage of student sample, since this sample has unique characteristics and non-representativeness of the population and thus, threatens the external validity (Zdravkovic, Magnusson & Stanley, 2010).However, many past studies used this same sample and therefore, this study also used student sample as replication to the existing literature (Junco & Cotton, 2012;Yoo & Donthu, 2001;Yoshida & Gordon, 2012).In addition, the data for this study was cross-sectional in nature.However, longitudinal data could be revealed an interesting result.
There are several avenues for future research.Previous studies have shown that celebrity endorsement impact on many variables such as, customer based brand equity, brand attitude, customers' brand perception and purchase intention, brand loyalty and word of mouth, consumer risk perception, brand image, advertising effectiveness, brand recall and brand recognition (Biswas, Biswas & Das, 2006;Bush, Martin & Bush, 2004;Hakimi, Abedniya & Zaeim, 2011;Kamins, 1989;Jain, 2011;Sharma & Kumar, 2013;Spry et al., 2011).Therefore, in future research, the impact of celebrity endorsement and its dimensions, namely attractiveness, expertise and trustworthiness on the abovementioned variables could be examined in the carbonated soft drink industry.Further, Kemp and Bui (2011) imply that brand credibility partially mediates the relationship between health motivation and self-brand connection; Alam et al. (2012) affirm that brand credibility mediates the association between trustworthiness, perceived quality and customer loyalty; Shamim and Butt (2013) assert that brand credibility mediates and strengthen the relationship between consumer based brand equity and brand attitude; and Hur, Kim and Woo (2014) clime that corporate brand credibility mediates the association between corporate social responsibility and corporate reputation.Therefore, the brand credibility could be considered as mediating variable with the preceding variables in the same carbonated soft drink industry, and it is another avenue for possible further research.

Figure 1 .
Figure 1.Ohanian source model for celebrity endorsement and dimensions

Table 1 .
Definition for celebrity endorsement dimensions

Table 2 .
Definition for brand credibility dimensions

Table 4 .
Decision criteria

Table 4 .
Principal component analysis results for celebrity endorsement Notes.Factor loadings after Varimax rotation of three components extracted by principal components extraction by using SPSS 19 software; factor loadings of far above 0.40 are presented in bold-face type.

Table 5 .
Principal component analysis results for brand credibility Notes.Factor loadings after Varimax rotation of two components extracted by principal components extraction by using SPSS 19 software; factor loadings of far above 0.40 are presented in bold-face type.

Table 6 .
Reliability and convergent validity assessment results

Table 7 .
Discriminant validity assessment results

Table 8 .
One sample T-test result