The Influence of Social Media Marketing, Hedonic Shopping Motivation and Electronic Word of Mouth towards Impulse Purchases for Shopee’s Customers in Medan

This paper aims to research regarding the impacts that social media marketing, hedonic shopping motivation and electronic word of mouth has to a Shopee customer’s impulse purchases. The method applied in this study is quantitative and data will be collected by providing online to Shopee’s customers located in Medan, Indonesia. Convenience sampling method will be used and the sample will be distributed to whom the writer can easily distribute to. From the questionnaires distributed, 97 respondent’s data were obtained and they will be processed with the SPSS statistical software program. Result shows that only hedonic shopping motivation will significantly impact the impulse purchases made by Shopee customers in Medan while the others will not much impact to impulse purchases.


INTRODUCTION
Living in an evolving world, the use of marketing means has significantly changed throughout the years. Traditional marketing has now shifted to digital marketing, whereby products and services are advertised online, via social medias, television, Youtube, and also their own websites. Nowadays, digital marketing, especially social media marketing plays an important role to a company's marketing actions. Researches have also shown that 92% of marketers said that social media marketing plays a significant role for their business operations (Ghufran et al, 2016).
Social Media Marketing is a crucial part of marketing as social media keeps evolving, more and more users are going online in many different platforms, allowing businesses to target a larger audience and generating more sales when proper social media marketing have been made. This leads to a pathway for many businesses to further develop, shape and create brand awareness to maximize profit. With this, a lot of firms and e-commerce sites have now adapted to the use of social media marketing. Especially during this pandemic, most people are staying at home during their free time, thus, they have more time to browse through the web and be on their social medias. According to a survey made by Rakuten Insights, 55% of Indonesian respondents admitted that they've made more online purchases, and only 9% did not make any purchases online (Wolff, 2021).
Additionally, electronic word of mouth has also played a huge role in the marketing of a product and service, especially to businesses who set up their shops in e-commerce sites. With E-WOM, information and feedback is spread at a quick rate and honest customer can be easily received. This creates a loyal customer base that allows them to engage with the brand, sharing information with others, allowing firms to reach a wider audience. According to researches done by Husnain et al (2016), there is a positive effect between electronic word of mouth on impulse purchases, meaning that e-WOM does effect a person's stimulus to make a sudden unplanned purchase.
Another major predictor for the impulse purchases of customers are found to be hedonic shopping motivation and behavior. Hedonic environment is one of the strongest stimulator that ignite impulse purchases, especially for online purchases as it is easier for these e-commerce websites to target its customers and stimulate the hedonic behavior of the customers, simply by just conducting proper marketing strategies. There are various types of hedonic shopping motivations that an individual falls under and these differs from one and another for each individual and situation. In a journal written by Astuti.S.R.T, et al in 45 2020, she confirmed that hedonic shopping motivations does have a positive effect on impulse buying of Indonesian Instagram users. With this, firms will usually conduct social media marketing means, in support of electronic word of mouth (E-WOM) in order to trigger the hedonic shopping motivations of each customer in order for them to make a purchase.
According to researches, around 40 to 80% of customer's purchases are a result of impulse buying, which contributes to a large percentage of sales. (Khokhar, et al. 2019). Impulse buying via e-commerce sites have significantly increased throughout the year with more products advertised digitally. With more exposure of products through digital and social media advertisement, there's an increase in the likelihood of individuals to make more impulsive purchases, contributing to more sales.
Shopee have created an immaculate social media campaign for each platform however, since they're mainly focused on creating campaigns and mass marketing in different platforms, they might not have the full knowledge regarding a customer's perspective, and also the impact of their social media marketing means has on a customer's impulse buying behavior. Thus, this research aims to to examine whether there is a significant relationship between Shopee's social media marketing measures and customer impulse buying behavior and to analyze whether electronic word of mouth has an impact on Shopee customer's impulse purchases. This research aims to answer the following research questions of: Does Shopee's social media marketing measures leads to the customer's impulse purchase behavior? Does customer's hedonic shopping motivation contribute to customer's impulse purchases? and Does electronic word of mouth (E-WOM) brings an impact to Shopee customer's impulse purchase behavior.

Name of theory here
Explanation of theory here If your work is quantitative, please provide the previous studies concurring or rejecting your proposed hypothesis. H1: Hypothesis one and so on here Name of theory here Explanation of theory here H2: Hypothesis two and so on here

METHODOLOGY
In this paper, a quantitative research method will be used, whereby quantitative researches collects their information by conducting sampling methods such as sending our questionnaires, polls, surveys and etc to a sample group. Quantitative sampling method also deals with analysis of numerical data using different statistical techniques required (Apuke.O.D, 2017). This data will be analyzed numerically with the help of the SPSS software. The main problems and variables we would like to analyze how the independent variable, which is social media marketing (X1), electronic word of mouth (X2), and hedonic motivation (X3) will lead to impulse buying, which is the dependent variable Y.
In this research, as it is conducted in Medan, Indonesia, our population in this research will be the citizens of Medan. Citizens within all age groups are included within our population group. Only Medan region is included to ease the data collection. For the sampling method, non probability sampling in concentration to convenience sampling method will be used as it is inferred in being a sampling method that draws the sample from a group of individuals that met certain criteria such as they are easier to reach and contact, and those who are more willing to participate (Etikan.I, 2017). Data are conveniently collected by respondents who are easier to reach by the writer.
As the entire population is unknown and not defined, the number of samples taken for this research will be calculated by implementing the Lemeshow formula (Sugianto.A, Rahman.S, 2019). After calculating, we can acknowledge that the sample size required for this research adds up to 97 respondents.
Both primary and secondary data will be collected via observations and questionnaires for primary data, and also retrieving information from the internet, articles, books and journals for the secondary data.
There are a few data analysis methods that are used in calculating the result, including the descriptive statistical analysis to arrange and summarize the data obtained which comprises of mean, median, mode, variance and standard deviation (Kaur.P, 2018), research instrument tests to ensure that the data is valid by conducting reliability and validity test, the classical assumption test that consists of the multicollinearity, heteroskedasticity, normality, and linearity test. A multiple linear regression analysis is also conducted to predict the value of the dependent variable based on the independent variables and a coefficient of determination test is conducted to show the proportion of variance of the dependent variable that can be predicted by the independent variable. Lastly, hypothesis testing including the f test, p value and partial t test are required to realize which hypothesis will be accepted.

Descriptive analysis
Upon collecting all of the respondent's responses, we've grouped them into different categories, starting off with the genders.  (2021) From the table above, it shows the distribution of genders whom answered the questionnaires. There is a higher amount of female respondents as compared to male. The result obtained is as following maybe because of the higher readiness of female in comparison to male when these questionnaires are distributed.  (2021) From the data obtained, it shows that most respondents whom answered this questionnaire are between the age of 18-29 years old. This is probably due to the convenience sampling that the writer used, meaning that this questionnaire is distributed to respondents who are easily reached by the writer, thus, it results in respondents who are within the same age group.  (2021) From the data obtained, it shows that most students answered this questionnaire. The reason behind so might be similar to the ones I've mentioned above as convenience sampling is used, thus the questionnaire is distributed to respondents who are easily reached by the writer, whom are also students.  (2021) With this table provided as above, we can tell the mean, mode, median, variance and standard deviation of the data we obtained. From the mean, we can tell that the average of most question ranges between 3 to 4, meaning that in average, people do agree to the questions. With the mode, it shows what most people agree to, and as a whole, most people agree to the questions too as "4"s are most dominant answer from the given questionnaires. Standard deviation tells us the dispersion of the data, and thus, the higher the number, the more dispersed it will be. For the variables, we can tell that impulse purchases as the Y variable has the highest standard deviation, more so than the independent variables. This shows that the data is more dispersed than the others.

Validity test
A validity test is conducted before hand to analyze whether the questions made for the questionnaires are valid to ensure the indicator of the research is good before distributing it to the respondents (Yenty, 2020). In order to conduct this test, the questionnaire was distributed to 30 respondents whom are Shopee customers that are located in Medan, Indonesia. The degree of freedom for this test is 28, and the value for the r table is 0.361, thus, those questions who has R count above 0.361 will then be accepted.   (2021) From the validity test, there are 6 questions that are eliminated before distributing it to the 97 respondents as they are invalid. After these questions are eliminated, a reliability test will then be conducted afterwards. Cronbach Alpha test where by the result of this test should be greater than 0.6 (Yenty. 2020). From the table above, all of the variables are proven reliable, thus, the research will then be proceeded.  (2021) After the reliability test is conducted, to ensure that the data is normally distributed, the Kolmogorov-Smirnov Test will also be done additionally. If the significant value of these data are over 0.05, it shows that the data is normally distributed, thus, data below 0.05 refers to abnormal data (As'ari.R, 2018). From the data above, it shows that the significant value is 0.200, which is above 0.05, meaning that the data is normally distributed. Table 11. Linearity Test Variables X1 X2 X3 Linearity 0.000 0.005 0.000 Source: Prepared by writer (2021) Linearity test is conducted test and all of the values are way below 0.05, meaning that data is well accepted as data above 0.05 will be rejected and considered not linear.  (2021) The heteroskedacity test result is provided as following and since all values are above 0.05 significant value, it means that the variables have passed this test.  (2021) From all of the values stated in the table above, VIF values are within the range of 1.4 to 1.7, which means that the correlation is still very slight, resulting in an acceptable data.  (2021) Equation= Y = 6.796 + 0.052X1+0.937X2-0.164X3.

Multiple linear regression
The regression equation can be described as follows: Constant (a) = 6.796, meaning that if social media marketing, hedonic motivation and EWOM is zero, the impulse purchases will be 6.796.
Since the coefficient for social media marketing is 0.52, this means that if there is once increase in social media marketing, there will be also be an increase in impulse purchases by 0.52. Similar to hedonic motivations, which meant that if there is an increase of one for hedonic motivations, there will be an increase in 0.937 for impulse purchases. However, for electronic word of mouth (EWOM), if there is one increase in EWOM, there will be a decrease of 0.164 for impulse purchases.
With this result, only the hedonic motivation is statistically significant as p is lesser than 0.05. Meanwhile, the other variables are way over 0.05, meaning that it is not statistically significant.  (2021) This test shows that 41.3% (0.413 of the r square) of the factors that influence impulse purchases are ewom, hedonic shopping motivations, and also social media marketing, meaning the remaining 58.7% composes of external factors that are not mentioned.  (2021) From the Anova table seen above, it shows that the significant value is 0.000, which is lesser than 0.05, thus, we can tell that the model is a great fit for the data.

Hypothesis test F test
From the F table, we can tell that the amount will be 2.70 as we have 4 variables and 97 respondents. With this, the F table will be significantly lower than the Anova F count, which is 23.514. With this, we will accept that Ha hypothesis.  (2021) According to T-table, the t value will be 1.964 as we take 100 as degree of freedom, closest to 97.

Partial t test
If the t count stated above is larger than the T-table, it means that the Ha is accepted and H0 is rejected. From this, we can tell that only the Ha for hedonic motivation is accepted while the EWOM and social media marketing will be rejected. Thus, we can conclude that hedonic motivation does influence impulse purchases while social media marketing and electronic word of mouth does not.

DISCUSSION
This part allows you to elaborate on your results findings academically. You must not put numbers related to your statistical tests here; instead, you have to explain that numbers here. You have to compile your discussion with academic supports to your study and a good explanation according to the specific area you are investigating.
To summarize this research, starting off with the descriptive analysis results, it shows that there are more female respondents in comparison to male, having almost a 7:3 ratios for those. The majority age group for this questionnaire is also within an age group of 18-29 years old.A dditionally, A little over 50% are students who are currently pursuing different level of education. Looking at the descriptive statistics, overall, question 22 has the highest mean of 4.598, with a high median and mode of 5, on the other hand, question 36 will have a lowest mean of 2.639, with a low median and mode of 2.
For the classical assumption test, the Kolmogorov-Smirnov test is conducted and the value shows that its 0.200, which is the above the required 0.05, showing that the data is normally distributed. From the linearity test, none of the values for the independent variables are above 0.05, and they are around the range of 0.000 to 0.005, which then shows that there's a linear relationship between the independent and dependent variables. From the heteroscedacity glesjer test, all of the significant values are above 0.05, showing that the glesjer test are passed.Lastly the multicollinearity test VIF statistics are around the range of 1.4 to 1.7, meaning that they are slightly correlated, however, it is not problematic Multiple linear regression analysis is done in this research as there are more than 2 variables. After calculating, the equation will be Y=6.796+0.052X1+0.937X2-0.164X3. Constant (a) will be 6.796, meaning that if social media marketing, hedonic motivation and EWOM is zero, the impulse purchases will be 6.796. Since the coefficient for social media marketing is 0.052, this means that if there is once increase in social media marketing, there will be also be an increase in impulse purchases by 0.052. Similar to hedonic motivations, which meant that if there is an increase of one for hedonic motivations, there will be an increase in 0.937 for impulse purchases. However, for electronic word of mouth (EWOM), if there is one increase in EWOM, there will be a decrease of 0.164 for impulse purchases.
Lastly, there will be the hypothesis test. The determination test result shows that 41.3% of the influence impulse purchases are electronic word of mouth, hedonic shopping motivations, and also social media marketing, meaning the remaining 58.7% composes of external factors. From the F test, results show that the Ha hypothesis will be accepted as significant value is 0.000, which is below the required 0.05, which means that simultaneously, social media marketing, hedonic shopping motivation and electronic word of mouth will have an impact in impulse purchases. . For the T test, only hedonic shopping motivation has a t count of 6.432 above the value in the t table, of 1.984984, and from the significant figures, only the significant value of hedonic shopping motivation is below 0.05, which then concludes that only hedonic shopping motivation Ha's will be accepted .

CONCLUSIONS AND RECOMMENDATIONS
From this research, we can conclude that only hedonic shopping motivation will significantly impact impulse purchases made by Shopee customers in Medan. The purpose of this study is to find whether there is any relation for social media marketing, hedonic shopping motivations and electronic word of mouth towards the impulse purchases made by Shopee customers in Medan, thus, we can see that only the X2 variable, hedonic shopping motivation will significantly impact the Y variable.
In the future, an in depth research regarding other e-commerce sites can also be done to analyze whether the results are similar or not. The writer would recommend using similar variables to this research when researching other ecommerce sites in order to result in better outcomes for comparison. Additionally, this research can be further developed too by providing other variables not written in this research that might have an impact to the impulse purchases made by Shopee customers.

FURTHER STUDY
This research is only limited to the indicators mentioned for each variable, and the sample is limited to Medan City. For further studies, a more in depth research could be made that covers a larger sample group globally.

ACKNOWLEDGMENT
I would like to thank my advisor, Mr Genesis for guiding me towards the completion of this journal. I would also like to express my gratitude to the UPH family, as they've provided me with the knowledge and ability to create this journal. Lastly, I would like to thank my family and friends whom supported me towards the completion of this paper.