Research on Improving Consumer Purchase Intention by Enhancing Brand Image: Evidence from Field Experiments on Starbucks

. How to improve consumer purchase intention and then corporate performance is a hot issue that enterprises are concerned about. Many researchers have investigated typical factors influencing purchase intention from the dimensions of cultural factors, scarcity effects, and market strategies. However, few types of research are available on how brand image affects the purchase intention of consumers. Therefore, this paper uses Starbucks as an example to identify the relation between brand image and the purchase intention consumers hold by collecting data on consumers’ purchasing behavior, evaluating brand images, and conducting correlation and heterogeneity analysis. Empirical results have shown brand image is significantly correlated to purchase intention, which varies among groups with various characteristics. Correlation is stronger if a consumer is female, younger, earn a lower salary, visits Starbucks less frequently, or spends less in a single visit. This paper can help companies fully understand how the brand image affects consumers’ purchase intention. In that case, they can make clear the effectiveness of their brand marketing when formulating business strategies to obtain more turnover and profit through this method.


Introduction 1.1 Research background
More companies are now choosing to attract customers through brand image, taking brand image as an important means to improve revenue.Kim et al., studying the factors influencing the performance of luxury hotels and chain restaurants, empirically found that brand equity, based on components such as brand image, is positively correlated with corporate performance [1].This is believed that correct understanding and successful brand equity management could produce solid financial performance.Purchase intention can be explained as the subjective tendency of consumers to buy some product and is the premise of their purchase decision and behavior.According to Mullet's study, a consumer's attitude towards a particular product or brand and external factors constitute consumer purchase intention [2].Previous studies have paid little attention to the mechanism of brand image on consumer purchase intention.This study aims to compensate for this deficiency by analyzing how brand image impacts upon purchase intention of consumers and the heterogeneity among different groups with various characteristics.Suppose the company can fully understand how the brand image affects consumers' purchase intention.In that case, they can make clear the effectiveness of their brand marketing when formulating business strategies to obtain more turnover and profit through this method.

Literature review
When consumers begin to learn, memorize, and get used to a particular brand, they simultaneously form their personal opinion, cognition and expectations about the brand.This brand association reflected the subjecive perception held in the memory of cosumers, which is regarded as the brand image according to Keller's study [3].De Chernatony regarded brand image as an intangible asset of a company that is difficult to imitate, unlike a product brand that emphasizes the importance of brand values [4].Based on this, Ailawadi et al. proposed the index of revenue premium [5].This study measured the association between the revenue premium of brands under multiple categories of packaged goods and their brand equity.They empirically found that brand equity is significantly correlated to revenue premium a product can create.This premium is believed to be a reliabe indicator of brand equity, which can manifest actual dynamic changes of the brand image.Burmann believed it is the brand image that propels purchase behavior of consumers [6].The evaluation of the image composes an important asset for many brands, which is intangible but can be evaluated by the identity-based model.According to the rational action theory, behavioral action results from the intention.Behavioral intention of purchasing is determined by the perspective of consumers when choosing to buy commodities of some brand as well as a normative value or subjective norm [7].Purchase intention is one kind of behavioral intention, which is the subjective probability of a person performing a certain behavior based on rational action theory [8].Therefore, purchase intention is defined as the degree to which consumers are willing to buy certain products and services in the future [9].Based on this, Mirabi et al. believed that purchase intention can work as a valid instrument for making puchase predictions but they also pointed that effectiveness of this measurement varies with the product price or perceived quality and value [10].This intention is under influences from various aspects, including perceived value, price discount, buyer knowledge, celebrity endorsement, package design, etc. [11][12][13][14][15].
If consumers have a higher evaluation of a brand product, they are more willing to get a higher utility from buying the product, thus more likely to make a purchase.When examining consumers' reactions to a series of extension concepts referring to six famous brand names, Aaker & Keller found that brands with a positive image can improve consumers' loyalty and trust, enhancing consumers' purchase intention [16].Anatolevena Anisimova studied the influence of brand image on the attitude and behavior of consumer loyalty [17].The results also indicated a significant correlation between consumer-perceived brand image and behavior identification.Wang & Yang investigated how credibility of a particular brand impacts consumers' intention to buy [18].They found that there exists a positive correlation where brand image takes on a moderating role.From five perspectives of packaging, brand, price, advertising, and quality, Mirabi et al. examined the factors in ascendancy over purchase intention.The results showed that quality, advertising, and brand significantly impact consumer purchase intention, while packaging and price don't [10].
Most scholars mainly investigated the relation between corporate brand and purchase intention from the background of automobiles, tile, cosmetics, and other products.These articles mainly explored how brand image, together with other factors including packaging, quality, price, etc. Impacts purchase intention; brand image is just one of the influencing factors.Few scholars focused on the single dimension of brand image.They rarely analyzed how brand image influenced consumers' purchase intention and explored the heterogeneity among groups with different characteristics.

Research background
Premised on the identified research gaps, the remainder of this article is structured as demonstrated below.The methodology employed in this paper is illustrated and explained after the introduction, containing research designs, measurement instruments and ethical considerations.What follows next is the result section, which offers the sample profile, descriptive statistics, reliability examination, correlation results and heterogeneity analysis.The subsequent section refers to the discussion and possible implications of the empirical results.Lastly, the study's key findings and limitations are highlighted in the section of conclusion.

Methodology 2.1 Research designs
Quantitative method is adopted in this research.The research object is the consumers of Starbucks, and online questionnaire surveys are conducted on them by random sampling, links of which are all conveyed via online groups.According to the research theory of Cuieford, when the reliability value of Cuieford coefficient α is greater than 0.7, it can be used.After the reliability test, the research continued data analysis in StataSE 15.0 from the perspectives of brand image and customer purchase intention and conducted the regression analysis of the two variables.Based on that, the relationship between the two variables was finally obtained.

Measurement instruments
The questionnaire was mainly composed of 15 questions in three parts.They are demographic information, brand image, and customer purchase intention.The questionnaire is self-designed after referring to relevant literature.The questions in the survey questionnaire were developed in Mandarin because Mandarin is the official language of China.The questionnaire consisted of three sections.All indicators were scored via a Likert scale categoryed into seven degrees, except for the demographic variables.The questionnaire is displayed in Table 1.

Ethical considerations
The norms of conduct recognized by the scientific community are considered in this research.This study runs under this basic principle where the relationship between researcher and participants (including coworkers, funders, objects or the public), researchers, and objects (including test animals, natural environment, etc.) is required to be timely adjusted.According to this principle, the ethical issue is considered in this research from start to end.This is embodied in the fact that all interviewees in this study are adults and are informed of the content before the interview with their consent.No one or anything was harmed in the whole process.

Sample profile of respondents
A total of 5,339 participants were surveyed, and descriptive statistical analysis was displayed.As for gender, women accounted for 51.98%, and the ratio of the two is about 1:1, in line with the gender distribution of the sociological population.In terms of frequency, the high frequency of respondents who go to Starbucks daily accounts for 23.99%.And the low-frequency respondents who go less than once a month account for 33.36%, which also conforms to the realities.In terms of spending at Starbucks each time, roughly 70% of respondents vary from 40 to 99 yuan.It can be seen that the respondents in this survey meet the requirements of this study and can be analyzed in the next step.

Descriptive statistics and reliability examination
As for the brand image, the average score is 5.5311 points, which is in the upper middle level.Decoration of the store scored highest with 5.8932.Similarly, in terms of purchase intention, the average score is 5.6073, which is at the upper middle level and higher than the brand image.The environment of Starbucks scored highest, with 5.8739.The standard deviation of each indicator in both sections is less than 1, showing the concentrated distribution of the collected data.
Cronbach α trust coefficient serves as the test index to examine the reliability of the questionnaire.When conducting reliability analysis, the brand image and purchase intention of this study was used to conduct reliability analysis, which is above 0.85, indicating the reliability of the measurement instruments.Moreover, the average variance extracted from the two constructs was 0.6739 and 0.7890, respectively, exceeding the minimum criterion of 0.5, varying from 0.6243 to 0.8483, which has assured the convergent validity.Also, the discriminant validity is tested by estimating correlations among the dimensions to determine whether they differed significantly from 1.All correlation coefficients among all dimensions were significant at the 0.01 level, suggesting that the measurement instruments satisfied discriminant validity.The correlation matrix for the dimensions is displayed in Table 4. Table 4. Correlation Matrix of measurement dimensions.b1 b2 b3 b4 b5 p1 p2 p3 p4 p5 b1 1.000 b2 0.595 1.000 b3 0.654 0.542 1.000 b4 0.578 0.726 0.556 1.000 b5 0.624 0.681 0.530 0.705 1.000 p1 0.540 0.529 0.529 0.548 0.610 1.000 p2 0.514 0.513 0.486 0.535 0.549 0.466 1.000 p3 0.548 0.674 0.529 0.697 0.688 0.542 0.525 1.000 p4 0.604 0.660 0.541 0.680 0.784 0.603 0.511 0.692 1.000 p5 0.559 0.523 0.503 0.538 0.630 0.509 0.420 0.512 0.636 1.000 Note: b means brand image; p means purchase intention.

Correlation results and heterogeneity analysis
Using the structural model to explore path regression coefficients indicates the direct influences of the indicators upon the predicted latent constructs, all path regression coefficients are significant at the 99% confidence intervals, and the results are shown in Figure 1.Also, the Pearson correlation analysis method was implemented to study the overall correlation.The study results told that the correlation coefficient was 0.875, and the value of significance was 0.000, less than 0.01, indicating a very significant positive correlation.Simultaneously, the study is also interested in whether various characteristics will have an impact on the correlation coefficient, like whether there is a difference in the correlation between female and male groups.The analysis focuses on whether there are significant differences in the two group correlation coefficients between the two variables.This study first groups the samples according to different features to explore whether heterogeneity exists in the correlation among groups with particular characteristics.The grouping results are displayed in Table 5.Then, based on the above binary sequence, the OLS regression of purchase intention and brand image is carried out under different groups such as gender, age, monthly income, frequency, and expenditure.For example, as can be seen from the results, the coefficients for the female and male groups are 1.008092 and .9399603,respectively.But can this result indicate that the correlation coefficient of the females is greater so that the purchase intention and brand image are more strongly correlated among female consumers?
Therefore, the third step to determine whether different characteristics will significantly impact the correlation is to test for differences in coefficients between groups.This paper used Fisher's Permutation test to obtain the corresponding P value by boot-strap 1000 times.For example, the P value of the correlation coefficient between the female and male groups is 0.020, indicating that the correlation coefficientvis significantly different at the 5% level.The regression coefficients and between-group difference results of all groups are displayed in Table 6.The difference test of the correlation coefficient of all groups was significant, and the significance level of groups in terms of frequency and expenditure reached 1%.connotations of Starbucks, for instance, their mission of inspiring and nurturing the human spirit.In contrast, the older generation in China is more accustomed to drinking tea, which is somewhat of a substitute for coffee.Even if they buy products from Starbucks stores, they may not actively perceive the warm and belonging culture that Starbucks is endeavoring to create.Therefore, the correlation between the purchase intention of the elderly and their brand image evaluation of Starbucks is weaker than that of the younger generations.

Lower-income consumers purchase out of vanity.
In respect of income, the purchase intention of people whose monthly income is less than ¥5000 is more closely related to their cognition of brand image, with a significant difference of about 0.06 at the 5% level.Starbucks is a relatively expensive coffee brand in China, and people tend to consume Starbucks as a symbol of affluence.In fact, vanity might make lower-income people consume Starbucks and pay for the fine and high-end brand image to get psychological satisfaction.However, higher-income people do not need Starbucks to prove their social status and are more willing to pay for the coffee itself rather than the brand image.Under such circumstances, a good brand image evaluation does not necessarily play a positive role in enhancing purchase intention.Therefore, brand image is more strongly correlated with purchase intention in the low-income group.

Correlation diminishes as frequency goes up.
The frequency correlation also significantly differs at the 1% level.The correlation difference between low-frequency and high-frequency consumers could be explained from the perspective of the marginal diminishing effect in economics theory.Brand image may play an important role initially when forming purchase intentions.In the early stage of low consumption frequency, customers know little about Starbucks and will go through the information searching process, including celebrity endorsement, store profile, peer recommendation, etc. Consumers, when forming their purchase intention, will think more about the image the brand conveys through the information available.As time goes by, brand image will have a boundary effect.Its influence on people's purchase intention will diminish until it is almost zero.Customers with high consumption frequency tend to regard Starbucks as a habit without considering their underlying brand image recognition.

Buyers with petty expenditures are more loyal.
Seen from expenditure each time, brand image is more closely correlated with purchasing intention if consumers spend less in a single visit to Starbucks.A possible explanation could be that petty consumers are more loyal.Consumers with a consumption amount greater than ¥70 each time might be first-time purchasers of Starbucks who are just trying something new and dear.That is to say, the driver for their purchase decisions and behaviors is not the purchase intention based on their product's brand image but curiosity.On the contrary, consumers with expenditures not exceeding ¥70 each time emphasize concept consciousness and brand salience in purchasing decisions, as discussed above.They buy products from Starbucks due to their credit for brand image and do this as is their wont.They possess an in-depth understanding of Starbucks' core concepts, appearance designs, and product characteristics and hold high recognition of the brand image composed of the sequences of factors.Brand image plays a significant part in constituting purchase intention and then incentivizing purchase decisions, making them loyal customers of Starbucks.

Key findings
This study first evaluated the correlation coefficient between the corporate brand of Starbucks and the purchase tendency of consumers.The empirical results show a significant and strong correlation between brand image and purchase intention, consistent with previous research results.This correlation can be explained from the perspectives of self-concept and brand salience.Brand images that conform to consumers' self-concept or have distinct visual cognition will affect consumers' evaluation of products and then their purchase intention.Then, this study subdivides sample groups to study the impact of different characteristics on the degree of correlation.The results indicate that the coefficients of consumers with different characteristics are significantly different, which is stronger if a consumer is a female, younger, earn a lower salary, visits Starbucks less frequently, or spends less in a single visit.Lastly, this study explains the differences in coefficients among different groups based on Starbucks' corporate culture and market strategy.

Limitations
In this part, the limitations of this research are demonstrated from the perspectives of questionnaire design, sample bias, and data analysis.Firstly, there are only 10 questions in the questionnaire for measuring consumers' brand image and purchase intention.However, these two variables are relatively abstract concepts, and the dimensions involved are far from enough to measure them comprehensively, resulting in incompleteness.Also, some subjective questions are lacking to capture the real thoughts of consumers.Secondly, the author only selected approximately five thousand samples in this study, which is not enough for the total sample number of China's population.This will make the results unrepresentative, which means biased experimental results.The last limitation is the process of data analysis.The author chooses Stata for sample analysis, which might be weak at pre-processing data, leading to the deviation of the results.

Figure 1 .
Figure 1.Standardized regression coefficients of structural models.

Table 1 .
Indicators in the survey questionnaire.

Table 2 .
Demographic characteristics of participants.

Table 3 .
Descriptive statistics and reliability of each instrument.

Table 5 .
Samples grouped according to characteristics.