IN
DEGREE PROJECT INDUSTRIAL MANAGEMENT, SECOND CYCLE, 15 CREDITS
,
STOCKHOLM SWEDEN 2020
Analysis of Customer Behaviours
on Customer Reviews and Ratings
ALKIM YETIS
ERKUT KAVAK
KTH ROYAL INSTITUTE OF TECHNOLOGY
TRITA ITM-EX 2020:191
Analysis of Customer Behaviours on
Customer Reviews and Ratings
Alkım Yetiş
Erkut Kavak
Master of Science Thesis INDEK TRITA-ITM-EX 2020:191 KTH School of Industrial Engineering and Management
Master of Science Thesis TRITA-ITM-EX 2020:191
Analysis of Customer Behaviours on Customer Reviews and Ratings
Alkım Yetiş and Erkut Kavak Approved 2020-06-11 Examiner Kristina Nyström Supervisor Johan Nordensvärd Abstract
With the rapid accretion of online shopping, online customer reviews have become a very essential source. An interpretation of this essential source is not fulfilled with every aspect since online customer reviews consists of unstructured information. This paper enlightens the understanding of behavioural analysis on how customers rate and review gender-based products by focusing on two brands in the fashion industry. Since gender-based products are analysed, gender as a search term is used for diversification in terms of product categories. To understand behavioural differences of customers to rate gender-based products, quantitative statistical methods were used. The two-sample t-test investigates if the means of two samples have statistically significant differences. Additionally, the word frequencies are analysed with the quantitative content analysis methodology to procure an answer for how customers review gender-based products for different product categories. The results show that the product categories which consist of the keyword “Men” in their product description have earned a higher average of total number or reviews and star rating. However, the words that customers type have similarities among different product categories and there are not any consistencies in their frequencies. Therefore, our results show that the customer comments are not significantly dependent for the determined gender-based product categories.
Keywords: Customer reviews, e-commerce, customer behaviour, fashion industry,
Examensarbete TRITA-ITM-EX 2020:191
Analys av Kundbeteenden på Kundrecensioner och Betyg
Alkım Yetiş and Erkut Kavak Godkänt 2020-06-11 Examinator Kristina Nyström Handledare Johan Nordensvärd Sammanfattning
Med den snabba ökningen av online-shopping har kundrecensioner på nätet blivit en viktig källa. Kundrecensioner består av information som är ostrukturerad, vilket betyder att tolkningsprocessen av kundrecensionerna försvåras. I denna uppsats är en beteendeanalys genomförd för att belysa förståelsen av hur kunder betygsätter och recenserar könsinriktade produkter. Fokus har lagts på två varumärken inom modebranschen. Eftersom produkterna är könsinriktade används kön som en sökterm för att differentiera produktkategorierna. För att förstå skillnaden i kundernas beteende vid betygsättning av könsinriktade produkter användes kvantitativa statistiska metoder. Ett ”two-sample” t-test användes för att undersöka om det fanns statistiskt signifikanta skillnader i stickprovernas medelvärden. Med en kvantitativ innehållsanalys undersöktes ordfrekvensen för att få en förståelse av hur kunder granskar könsinriktade produkter, inom olika produktkategorier. Resultatet visar att de produktkategorier vars produktbeskrivning består av nyckelordet ”män”, har fått ett högre genomsnitt av antal recensioner samt fått högre betyg. Orden som kunder använder är lika inom de olika produktkategorierna och ordfrekvensen har ingen fast struktur. Resultatet visar därmed att kundrecensionerna inte är signifikant beroende av de fastställda könsinriktade produktkategorierna.
Nyckelord: Kundrecensioner, e-handeln, kundbeteende, modebranschen, könsinriktade
Contents List of Abbreviations ... i 1. Introduction ... 1 1.1 Background ... 1 1.2 Purpose ... 2 1.3 Research Questions ... 3 1.4 Methodological Approach ... 3
1.5 Scope and Delimitations ... 4
1.6 Sustainability ... 4
1.7 Outline... 5
2. Literature Review ... 6
2.1 E-Commerce and Customer’s Behaviour in E-Commerce ... 6
2.2 Customer Review and Rating ... 6
2.3 Word of Mouth (WOM) ... 8
2.4 Customer Behaviour According to Gender ... 8
3. Methodology ... 10
3.1 Explanation of Methodological Approach ... 10
3.2 Description of Methods for Data Collection ... 10
3.3 Description of Method of Analysis ... 11
3.4 Evaluation and Justification of Methodological Choices ... 12
3.5 Limitations ... 13
3.6 Ethical Consideration ... 13
4. Findings and Analysis... 14
4.1 Analysis of the Number of Total Customer Reviews ... 14
4.1.1 Adidas Shoes ... 14
4.1.2 Levi’s Jeans ... 17
4.2 Analysis of Star Ratings ... 20
4.2.1 Adidas Shoes ... 20
4.2.2 Levi’s Jeans ... 23
4.3 Text-Typed Comments and Word Frequencies ... 26
4.3.1 Adidas Shoes Men ... 27
4.3.2 Adidas Shoes Women ... 27
4.3.3 Adidas Shoes Men and Women Comparison ... 28
4.3.4 Levi’s Jeans Men ... 29
4.3.5 Levi’s Jeans Women ... 30
4.3.6 Levi’s Jeans Men and Women Comparison ... 31
5. Conclusion ... 32
6. Future Research ... 35
List of Graphs
Graph 1: Summary Report of Total Number of Reviews for “Adidas Shoes Men” Category ... 14
Graph 2: Summary Report of Total Number of Reviews for “Adidas Shoes Women” Category ... 15
Graph 3: Boxplot of Total Number of Reviews for “Adidas Shoes Women” and "Adidas Shoes Men" Category... 17
Graph 4: Summary Report of Total Number of Reviews for “Levi’s Jeans Men” Category ... 18
Graph 5: Summary Report of Total Number of Reviews for “Levi’s Jeans Women” Category ... 18
Graph 6: Boxplot of Total Number of Reviews for “Levi’s Jeans Men” and “Levi’s Jeans Women” Category... 20
Graph 7: Summary Report of Star Ratings for “Adidas Shoes Men” Category ... 21
Graph 8: Summary Report of Star Ratings for “Adidas Shoes Women” Category ... 21
Graph 9: Boxplot of Star Ratings for “Adidas Shoes Men” and “Adidas Shoes Women” Category .. 23
Graph 10: Summary Report of Star Ratings for “Levi’s Jeans Men” Category... 24
Graph 11: Summary Report of Star Ratings for “Levi’s Jeans Women” Category ... 24
Graph 12: Boxplot of Star Ratings for “Levi’s Jeans Men” and “Levi’s Jeans Women” Category .... 26
List of Figures Figure 1: Word distribution and frequencies for "Adidas Shoes Men" ... 27
Figure 2: Word distribution and frequencies for "Adidas Shoes Women" ... 28
Figure 3: Word Frequency comparison of "Adidas Shoes Men" and "Adidas Shoes Women" ... 29
Figure 4: Word distribution and frequencies for "Levi’s Jeans Men" ... 30
Figure 5: Word distribution and frequencies for “Levi’s Jeans Women” ... 30
i
List of Abbreviations
OCR Online Customer Review
WOM Word of Mouth
eWOM Electronic Word of Mouth
E-commerce Electronic commerce CI Confidence Interval
1
1. Introduction
1.1 Background
Online shopping has been very popular for years because of its convenience and ease of use. Almost all companies try to reach their customers online by selling their products on their websites and/or in any online marketplaces (Ingham, Cadieux, & Berrada, 2015). Understanding the customer’s wants and needs are the key to success in e-commerce (Singh, 2006). Online customer reviews (OCRs) have recently become a very important source of knowledge for customers in the online environment. To understand the aspects and importance of OCRs gained extensive attention for academics and professional communities (Kostyra, Reiner, Ratter, & Klapper, 2016). This developed technology gives the customer the opportunity to share their opinions and comments about a specific product or service in a free and effortless way, which has a significant outcome on the purchase decision of the consumer (Elwalda, Lü, & Ali, 2016).
Word of mouth (WOM) is an important way to share information between person to person. Electronic word of mouth (eWOM) is an online form of WOM. eWOM is very useful for consumers to share their views on a product or a service with other people and also getting information from other’s experiences (Huete-Alcocer, 2017). That is one of the ways that affect how people make a decision whether or not to buy and thus it is related to the customer’s behaviour (Fan and Miao, 2012).
There has been a great number of researches to understand the buying behaviours of people. According to Ling and Yazdanifard (2014) many external elements affect the customer behaviour such as gender, social, culture and some internal ones as emotion, motivation and personality. Though there are some articles in the far past that state gender differences do not impact shopping behaviour, recent researchers strongly advise the opposite (Szymkowiak and Garczarek-Bąk, 2018). Therefore, to understand if the reviews vary according to different gender-based product categories plays an important role in online shopping.
2
the video game industry. Those studies have analysed the impact of OCRs on the sale amount but did not focus on how customer behaviour changes to review and rate those products when the product description includes gender keywords. Another subject is negative OCRs that were analysed in many pieces of research. For instance Bradley, Sparks and Weber (2016) have examined the personal impact of negative online reviews and the perceived prevalence of those reviews. The respondence to negative consumer-generated online reviews and specific forms of managerial responses were analysed by Sparks and Bradley (2014). The impact of the negative reviews and the managerial responses to those reviews were analysed, however those were not categorized according to product categories and did not focus on the retail industry. Online shopping behaviour depending on gender has been worked vastly. However, main focus on this researches are mostly about the psychology of buying behaviour. Lin, Featherman, Brooks and Hajli (2018) concentrated on the purchase decisions of the online customers. Ling and Yazdanifard (2014) investigated how different elements which are perception, motivation, preferences affect the customer behaviour. On the other hand, eWOM is another important form of online shopping in order to understand the online shopping behaviour. Some of the studies about eWOM include the cultural effect of gender and trust such as in the paper of Awad and Ragowsky (2014). Ratings in e-commerce is also a significant feature that needs to be understood. Flanagin, Metzger, Pure, Markov and Hartsell (2014) looked into how a person assesses the trueness of the rating, whether she or he finds that information useful and the relation of the product rating and rating volumes. The integration and relationship of eWOM and rating in e-commerce is not well defined yet. Elli and Wang (2016) investigated the reviews in Amazon for mobile phones, however they did not take into consideration if there is a difference in the construction of the world clouds according to different product categories. This relationship is not studied with all aspects of e-commerce. Therefore, the effects of the word cloud and word frequency on e-commerce is another subject that needs to be investigated due to lack of information.
This paper investigates the analysis of customer behaviour to review and rate a product that includes gender as a keyword in their product description on Amazon.com. These products are called gender-based and the categories shoes and jeans in the retail fashion industry are the focus of this paper which are explained with details in the further parts. Nevertheless, the gender of the customer who rates and reviews the product is not taken into consideration. The aim is just to figure out the variety of different gender-based product categories in terms of reviews and rates. The total number of reviews and star-ratings are collected to understand the relation of how these factors are differentiated for gender-based products.Additionally, word clouds which consist of text-typed comments are analysed to investigate the relationship between word frequency and wording for these product categories.
1.2 Purpose
3
but how the reviewers act to review and rate the products which have gender in their search term is. The consideration of products that include the keywords men and women in their product description strengthens the understanding of the behaviour varies to review and rate a product that includes those keywords. Therefore, the gender of the customer who reviews and rates is not taken into consideration as mentioned before. Additionally, the behavioural analysis of customers in terms of which wording is used and in which frequencies the words in the text-typed comments are used, is accomplished in this paper. To build up the understanding, different product categories are used for comparison and consideration of behavioural analysis. Therefore, the purpose of this paper is to understand the behaviour of the customer in the fashion retail industry in detail by exploring how customers behave to rate and comment on gender-based products. To understand the scale of the word frequency and the wording that customers use as a comment is also an aim of this paper. The results also show whether there are different approaches from customers to different gender-based product categories without taking gender of the reviewer into consideration.
1.3 Research Questions
In order to figure out how customer behaviour differs in specific product categories and to understand if this behaviour varies to comment and to rate gender-based product categories, quantitative research methods are used. To procure the outputs of the method, this research paper comes up with the following two questions:
1) How does the total number of reviews and star ratings of a gender-based product vary according to different product categories?
2) How does the word frequency and wording of text-typed comments change for different gender-based product categories?
1.4 Methodological Approach
4
of the interpretations, the research question “How does the word frequency and wording of text-typed comments change for different gender-based product categories? ” is answered.
1.5 Scope and Delimitations
The scope of this research paper is to demonstrate what the attitude of customers is to review and rate a gender-based product in e-commerce. Comparison is used to strengthen the understanding of the structure and organization of the different product categories which lead to developing ideas. The product categories shoes and jeans were chosen because of the following common reasons: The size and design of these categories vary according to gender. The brand “Adidas” is used for the shoe analysis and the brand “Levi’s” for jeans. According to Levi’s online shopping website, the jean waist 23-34 and length 26-34 sizes for women and 26-50 waist size and 30-38 length sizes are sold ("Men's Jeans", n.d.; "Women Jeans" , n.d.). In terms of shoe size differences between men and women; according to Hong, Wang, Xu and Li (2011), the shoe size differs in gender for Chinese adults as they state that the common foot size is 230 mm and 250 mm for women and men respectively (Hong et al., 2011). Therefore, using these gender-based products strengthen the understanding of how customers behave to rate and comment on a product which includes keywords “men” and “women” in their description. Another reason why the focus is on these two product categories and brands is that studying one brand for reviews may not give a correct outcome since this result can be affected by the brand profile (East, Hammond & Wright (2007). Therefore, it is better to investigate more than one brand.
For this paper, the offered products are narrowed down to the product categories “Jeans” and “Shoes” which was mentioned before. For the procurement of the data, the following headings were typed sequentially: The brand name, category, gender (e.g. Adidas Shoes Men). Afterwards, the needed data, which is mentioned for every product category in the analysis part, is collected from the website Amazon.com.
Customer reviews differ according to their functionalities. There are star ratings and text-typed comments where the choice is mainly given to the reviewer to fulfill either one or both of them. Star ratings indicate a negative sign with low star rates and text-typed comments indicate it by negative structured sentences. This research paper is narrowed down by just focusing on the wordings, word frequencies, number of reviews and star ratings that the customer has given. The helpfulness rating of text-typed comments is not included in this paper.
The date where the product was first available on amazon.com was not taken into consideration. This is the date where the products actually could have started to obtain customer reviews and rates. It was not taken into consideration since the variation between the first available dates is high which affects the accuracy of the analysis.
1.6 Sustainability
5
generations to meet their own needs.”. Sustainable development includes three core elements; economical growth, social inclusion and environmental protection which are interconnected to each other ("About the Sustainable Development Goals – United Nations Sustainable Development", n.d).
Understanding the customer behaviour is important for the companies since the companies adapt themselves according to their customers. They can use this kind of information to increase the efficiency in their businesses. Increment of the efficiency in the operation means a reduction in waste which means less usage of their resources. Responsible Consumption and Production, which suits with economical growth element, is one of the Sustainable Development Goals of the United Nations ("#Envision2030 Goal 12: Responsible Consumption and Production Enable", n.d.). This is also related to the customer side of e-commerce. The customers who check the reviews before they purchase a product may give their decision depending on the reviews rather than ordering it and testing it by themselves. Without checking any reviews of a physical product may cause a return after the parcel is reached to the customer. Since a return requires more transportation processing, this has a negative effect on the environment and aligns with the 13th Sustainable Development Goals of the United Nations which is Climate Action ("#Envision2030 Goal 13: Climate Action Enable", n.d.). This goal relates to the environmental protection of the three core elements of Sustainable development goals.
1.7 Outline
6
2. Literature Review
2.1 E-Commerce and Customer’s Behaviour in E-Commerce
Over the past decade, Internet technology has impressively modified people’s way of life. In terms of online shopping, e-commerce has experienced fast improvements and online shopping has ended up as a well-known strategy for purchasing goods. Amazon provides a variety of products and services from its platforms which one of them is in e-commerce through its website called Amazon.com. In the year 2013, Amazon reached a sales amount of US$74.4 billion (Yan, et al., 2016). E-commerce is as often as possible likened with the utilize of the Web to execute business. Whereas the Web has verifiably been at the cutting edge of the quick development of e-commerce. Since it provides low-cost, it is a prevalent tool to facilitate business transactions (Boyd & Bilegan, 2003). Even though the popularity of e-commerce was not broad in the early 90s, it became very popular in the late 90s with the boom of the World Wide Web (Rosenbloom, 2003). From that year until nowadays, e-commerce evolved in many aspects and utilization of e-commerce expanded dramatically. The expansion of e-commerce can be described with an exponential rate (Gefen, 2000).
A very important characteristic that creates a difference between online and offline shopping behaviour is the low cost of transportation that is required to visit a virtual store. It costs less to visit an online store which has several effects on the behaviour of the customer. To begin with, from the customer’s perspective it has a lower cost and customers who surf online may visit a store without any purpose of buying. In the offline perspective, where the customer takes his/her time and exertion to visit a store, the probability that he/she will not purchase a product is lower (Moe & Fader, 2004). Therefore, nearly everybody has adopted the online purchasing phenomenon (Abdul-Muhmin, 2010). To distinguish and understand the potential drivers of online purchase behaviour has ended up as an interesting research area with multiple research domains and an expanding number of publications each year (Grant, Clarke, & Kyriazis, 2007). However, a systematic review has recognized that the scope of online customer behaviour studies is wide, incoherent and contradictory. This may have come about in a call for the improvement of new theories, an exertion that has not however been embraced cohesively. Researchers ought to attempt to investigate and construct their own theories in online customer behaviour rather than applying existing ones (Chan, Cheung, Kwong, Limayem, & Zhu, 2003). Ganesh, Reynolds, Luckett, and Pomirleanu (2010) states that the fundamentals of online shopping have not been changed through time. Even though past research papers show differences between online and offline shopping such as online shoppers are paying more attention to save time even if they pay more money and to have more information about the product and its varieties, they believe that online shoppers are behaving very similar to regular shoppers. According to their research, they believe that the distinctness of e-commerce stores is interactivity, availability for day and night and ability to edit their content easily (Ganesh et al., 2010). So, understanding the customer’s wants and needs can be described as the key to success in e-commerce (Singh, 2006).
2.2 Customer Review and Rating
7
Mudambi and Schuff (2010) have defined customer reviews in the following way: The definition of OCRs is a peer-generated evaluation of products that are uploaded to a company or third-party websites. Additionally, retail websites give customers the opportunity to upload a review of the product with content in the form of star ratings which are numerical (usually between 1 to 5) and open-ended comments. Customer reviews that are created by consumers provide additional knowledge with the description of the provided product, reviews created by professionals and individual advice which are generated by an automatic recommendation system. Companies like Amazon.com give the opportunity to their customers to upload product reviews for many years. Other smaller firms prefer to buy customer reviews from Amazon.com and then upload those reviews on their electronic stores. So, the review provides a further revenue stream for Amazon.com (Mudambi & Schuff , 2010).
With the increase of OCRs, the decision to which of the reviews should be read among the various reviews posted online has become a very important strategic issue for the customer, who wants to shop online. As an example, in July 2012, more than 1,300 reviews were uploaded for Apple’s product category iPod Nano 16GB 6th Generation media player on Amazon.com. Therefore, a customer who has decided to buy an iPod has to consider which of the reviews to read among the great number of reviews uploaded because to read every single review is not possible in practice (Lee, 2013).
Dellarocas (2003) argues that OCR frameworks are one of the foremost capable channels to create online WOM. With the help of the Internet, not just professional organizations who can access their audience, also individuals can share their personal opinions, reaction which are effortlessly available to the worldwide community of Web customers (Dellarocas, 2003). To explain the development of new channels which are developed by customer reviews; Bounie, Bourreau, Gensollen and Waelbroeck (2008) argue that OCRs led to a new channel to gather knowledge. Leading companies like Amazon, give their customers the opportunity to read and/or write reviews on the product they have purchased and obtain knowledge and advice of the product they plan to purchase. This helps other customers to access comprehensive knowledge on the preferences of the reviewers. So, customers can contrast their enjoyed products with the reviewer’s perspective and gather knowledge on the horizontal and vertical aspects of the goods that they will enjoy (Bounie et al, 2008).
8
There are a lot of publications and a huge number of issues which are utilizing customer review data, but an organized investigation of this research stream is still missing (Trenz & Berger, 2013).
2.3 Word of Mouth (WOM)
WOM communication is an informal communication way for people to share information and their opinion with others about how they feel (Jalilvand, Esfahani, & Samiei, 2011; Buttle, 1998). Hennig-Thurau, Gwinner, Walsh and Gremler (2004) states WOM as a customer's purchasing decision is highly affected by it (Hennig-Thurau et al., 2004). In addition to this, it is believed that it influences people more than traditional commercial types (Laczniak, Decarlo, & Ramaswami, 2001). Therefore, WOM is a very important marketing communication way (Kemper, 2017).
Digital form of WOM is called eWOM (Kemper, 2017). One of the key differences between WOM and eWOM is that since the eWOM comments are online reviews, they stay for a long time after they are posted unless it is taken back. Therefore, accessing this information is easier and more people can see this opinion (Dellarocas, Zhang, & Awad, 2007). On the other hand, since there may not be any information about this reviewer, the credibility of this comment is not higher than the credibility of a traditional WOM since the latter one comes from a person that the receiver most probably knows (Cheng & Ho, 2015).
There are different places for people to share their opinions on eWOM which can be stated as Web-based platforms, online forums, blogs, chatrooms, sending an email and so on (Hennig-Thurau et al., 2004; Litvin, Goldsmith, & Pan, 2008). Writing OCRs is one of the most important eWOM ways in the fashion industry (Lohse, Kemper, & Brettel, 2017). According to Lohse et al. (2017) and Kemper (2017), OCRs plays a substantial role on buying decision of customers and one of the first web platforms that provided an opportunity to write OCRs is Amazon (Lohse et al., 2017; Kemper, 2017).
Alexandrov, Lilly and Babakus (2013) state that the volume of the comment plays an important role. According to East et al. (2007), it is hard to understand the effect of WOM easily with low numbers of samples. Addition to this, Cheng and Ho (2015) mention in their paper that if the volume of the OCRs is higher, the attitude of the customers becomes more positive. They also state that the word count of the review also affects the customer’s attitude similar to the volume of OCRs (Cheng & Ho, 2015).
2.4 Customer Behaviour According to Gender
9
the search terms include the gender words, men and women, which indicates that the result of the search is formed for that gender by the provider.
Early research is written by Slyke, Comunale and Belanger (2002) mention that Internet usage of men was more than the women before the year 2000 but this gap diminished significantly from 2000 onwards (Slyke et al., 2002). Sebastianelli, Tamimi and Rajan (2008) also mentioned the same progress towards the use of the internet by gender. They added that the gender gap related to online shopping also reduced a lot and they stated as it is virtually non-existent (Sebastianelli et al. 2008). However, Slyke et al. (2002) believe the opposite in terms of the behaviour. They mentioned that even though women visit websites more, men are the ones who are more into buying products online (Slyke et al., 2002). Dittmar, Long and Meek (2004) explain this difference in their paper by comparing the online and conventional buying motivations, and they stated that though women buy more in a conventional way than men do, men have a more positive posture to online shopping. The reason for this is that while women are looking at shopping as a leisure and social dimension, men are more interested in the functional factors (Dittmar et al., 2004). Another reason is that since women are more requested for emotional and social interaction, they cannot find a face to face communication through online shopping. Therefore, they are less interested in shopping online in comparison to men (Hasan, 2010; Szymkowiak and Garczarek-Bąk, 2018).
Characteristics whilst shopping online also differs from women to men. When men shop online, they tend to see this action as a task (Hasan, 2010). Martín and Jiménez (2011) explain this as “males are more pragmatic and are primarily guided by societal norms that require control, mastery and self-efficacy to pursue self-centered goals” (p.270). In addition to men being task-oriented, they also pay attention to save the time on online shopping and ease of online shopping (Lim & Yazdanifar, 2014). Therefore, as it is mentioned earlier, men find online shopping more suitable than women (Slyke et al., 2002).
Another main difference between men and women towards online shopping is trust in e-commerce transactions and on personal information (Rahim & Rosly, 2014). According to Bae and Lee (2010), the perceived risk of women buying online is higher than men’s. They mentioned that women hesitate to give their personal information such as credit card details, phone numbers and so on. On the other hand, men are more friendly to this case and they trust more to shop online (Bae & Lee, 2010). Kolsaker and Payne (2002) also believe that women are more anxious about trust, security, and confidentiality (Kolsaker & Payne, 2002).
10
3. Methodology
3.1 Explanation of Methodological Approach
The overall approach of this research is to find customers’ behavioral differences towards reviewing and rating gender-based products. To figure out if there are such differences and to accomplish an analysis of the behaviour to review and rate a product, quantitative analysis is used as a research method. According to Nallaperumal (2014), quantitative research is suitable for the phenomena which is possible to be expressed in terms of quantity (Nallaperumal, 2014, p.8). The outputs that are achieved in this research are visualized with numbers, graphs and figures. Denscombe (2010) claims that the unit of analysis for quantitative research are numbers and this analysis is associated with the researchers detachment (Denscombe, 2010, p.237). The total number of reviews that the product has received and the star ratings are collected as nominal data. Nominal data is collected from counting the components and placing them into the category which is defined (Denscombe, 2010, p.243). Quantitative data indicates the measurements where numbers are used in a direct way to show the characteristics of something (Hair, Money, Page, Samouel, & Celsi, 2016). Finally, the text-typed comments are collected as data and are analysed with the quantitative content analysis. Content analysis belongs to a part of the procedure for systematic and replicable analysis of a text (Rose, Canhoto, & Spinks, 2015, p.1). As mentioned before, this research is related to quantitative analysis, thus it is analysed based on inductive research methodology. The intended purpose of the inductive approach is to identify the patterns in the data sample to come to a conclusion. (Hair et al., 2016)
This research paper investigates the subject behavioural analysis of online customer reviews, which is a subject that has not been studied with every aspect before. Therefore, exploratory type of research design is used to strengthen the understanding of a problem that already exists. According to Hair, Money, Page, Samouel, and Celsi (2016), to improve the understanding of an opportunity or problem, exploratory research design is used (Hair, Money, Page, Samouel, & Celsi, 2016). The behaviour to review and rate a product in e-commerce varies according to gender-based product categories. With exploratory research, this existing phenomenon is analysed and new insight is narrowed to the problem which relates issues to the subject that are found out. The outputs of this research provide answers to “How questions” which are in this paper: 1) How does the total number of reviews and star ratings of a gender-based product vary according to different product categories? 2) How does the word frequency and wording of text-typed comments change for different gender-based product categories? Hair et al. (2016) states that, exploratory research is accomplished when the researcher has limited information of a special problem or opportunity and wants to explore new relationships, patterns, themes, and ideas (Hair, et al., 2016).
3.2 Description of Methods for Data Collection
11
paper, this data is primary data in which it is mentioned as getting the original data by the researcher himself/herself (Singh & Mangat, 1996, p. 2).
There are two product categories in focus, and these are shoes and jeans. The brands for these product categories are Adidas and Levi’s, respectively. The reason why these brands are picked is that the size and design of shoes and jeans vary for different gender as it is mentioned before. Therefore, these brands are easier to be investigated in terms of gender-based products. The sample data is collected as follows. 4 search terms are used as “Adidas Shoes Men”, “Adidas Shoes Women”, “Levi’s Jeans Men'' and “Levi’s Jeans Women'' on Amazon’s homepage. After this step, the first 200 products which have a review became the focus of this paper. If there are not 200 products, the maximum number of products that have a review are collected. The number of the products is selected as maximum 200 since the number of rates thus the star rates and the number of comments decline at a very sharp slope and show a high variation which disarranges the whole data set. Products that have earned 0 reviews and rates were defined as outliers and were eliminated in the data collection part. Every products’ individual web page is visited to collect the total number of reviews, the value of the star rating of that product and text-typed comments inside this product’s web page. Sponsored products after searching the term are omitted since these products belong to other brands. In addition to this, the products that are not shoes or jeans after the search are excluded in the data gathering process as well. In terms of word distribution analyses, the first page of top reviews from the United States and all international reviews are used. International reviews sometimes consist of non-English-languages texts since people from other American countries such as Mexico comment on a product. In addition to this, these international comments also contain comments from other domains of Amazon such as German domain Amazon.de, Italian domain Amazon.it and others. Therefore, In case any language other than English occurs, these are translated by using the “Translate all reviews to English'' button in Amazon under “Top international reviews'' section to keep their originality as much as possible. All of the international reviews are translated according to Amazon's automated translator without any changes. Customer questions and answer section and product description are ignored whilst analysing the word distribution.
3.3 Description of Method of Analysis
12
is used. According to Denscombe (2010), the t-test gives the researcher the opportunity to obtain a statistical test of significance where it uses the means of two populations and tells the researcher the likelihood of the difference between the two data sets (p.257).
To analyse the outputs of text-typed comments which customers made on the gender-based product categories, quantitative content analysis is used as a method. The quantitative content analysis contributes a structured form of analysing the provided data which is normally open-ended and comparatively unstructured (Rose, Canhoto, & Spinks, 2015, p.2). With the help of quantitative content analysis, the research can be enriched by identifying the frequency patterns and afterwards exploring the relation with inferential statistics (Boettger & Palmer, 2010, p.346). As described in the description of methods for the data collection part, the comments are collected for the analysis phase. Afterwards, the data is categorized according to their product descriptions like: Adidas Shoes Men, Adidas Shoes Women, Levi’s Jeans Men and Levi’s Jeans Women. The words that have not any subjective values are taken out of the sample size before being analysed. These words are the ones that include search terms as Adidas, shoes, shoe, men, women, Levi’s, levis and jeans. In addition to these words, pronouns such as I, you, we, they; and combination of the pronouns with variations of be or tense abbreviations such as I’m, it’s, I’ve are also removed from the data sample. In the further step, with the support of the data mining tool, word clouds and word frequencies are created. Word clouds are created by the tool according to the importance and repetition of the word. If the word repeats often it is assumed as a significant word and has a large size in the cloud. The frequency list is also created according to the relative frequencies in the provided text and a comparison is made between Adidas Shoes Men and Adidas Shoes Women and also between Levi’s Jeans Men and Levi’s Jeans Women.
As mentioned before, the software Minitab is used to accomplish the statistical analysis that is needed to procure an answer for the first research question. This software guides to spot trends, provide a solution for problems and explore valuable insights in data by delivering statistical analysis and process improvement tools. Minitab makes it simple to get deep insights to data ("About Us", n.d.). In order to procure an answer to the second research question, Voyant Tools is used for quantitative content analysis. It is a web-based text reader which can be accessed via voyant-tools.org ("About", n.d.). It provides a visual distribution of the words and also mentions the frequencies of the words used in the text sample.
3.4 Evaluation and Justification of Methodological Choices
13
research is already gathered by other parties (p.2). For our research, the data set that is investigated is not collected from any other parties since they are retrieved by the authors of this paper manually and directly. This suits with the definition of primary data mentioned by Singh and Mangat (1996); if the data is collected from the original source, it is called primary data (p.2).
There are three types of research designs which are exploratory, descriptive and hypothesis-testing. Exploratory research aims to investigate a problem for more accurate and detailed knowledge. Descriptive research is to describe the characteristics of a phenomenon of interest. Hypothesis-testing studies are experimental studies where the investigator tests the causal connection between variables (Nallaperumal, 2014, pp.68-71). The focus of this paper is on how customer behaviour varies according to gender-based products. Since this is not a piece of new knowledge but has not been well understood, an exploratory research design is used. This paper also investigates the word frequency and wording in the text-typed comment section of the selected products. Since there are words to be investigated, this can be thought of as a qualitative content analysis methodology. However, the qualitative content analysis focuses on the essentials of the meaning of the texts (Wildemuth, 2017, p.319). Since the word frequency is created, the numerical value of the words are used. This makes this approach quantitative content analysis as it is described by Wildemuth (2017); Quantitative content analysis is used to count textual elements (p.319).
3.5 Limitations
The first limitation is the defined sample size. The amount of listed products on Amazon.com varies between the defined product categories which limit the data that can be collected. To obtain a similar size for the data set, the collection of a maximum of 200 products for each category was defined. Secondly, while collecting the text-typed comments which are not English, Amazon's automated translator was used. This translation button is under the comment and changes the existing comment to English. The translation was assumed as correct and the data was collected accordingly. Another limitation is that amazon.com does not list all the reviews that are made for the product. Therefore, the research just contains the reviews that are demonstrated by amazon.com. Additionally, this research does not use the selling amount of the products as an additional variable.
3.6 Ethical Consideration
14
4. Findings and Analysis
4.1 Analysis of the Number of Total Customer Reviews
4.1.1 Adidas Shoes
First of all, the summary report is created in the software Minitab for the products which include the keywords “Adidas Shoes Men” and “Adidas Shoes Women” in their product description. The output of this report indicates headings like: The distribution of the data set with a Histogram, the mean, standard deviation, variance, skewness, kurtosis, sample size (N) and the 95% confidence interval (CI) for the mean, median and standard deviation. This report is created as step 1 to have an overview of the data set before applying the 2-sample t-test used in step 2.
Graph 1: Summary Report of Total Number of Reviews for “Adidas Shoes Men” Category
The data of the gender-based product category Adidas Shoes Men in Graph 1 has a mean of 555,69 total number of reviews and a standard deviation of 1.166,53. The standard deviation for this data set shows that a lot of numbers are spread out from the average. The 95% CI for the mean shows that the range of values between 394,03 and 719,35 contains 95% the true mean of the sample. Additionally, the 95% CI for the standard deviation gives an output of range between 1.062,32 and 1.293,60.
15
Graph 2: Summary Report of Total Number of Reviews for “Adidas Shoes Women” Category
The output for Graph 2 shows that this category has a mean of 345,03 total number of reviews and the CI for the means shows that the range between 227,75 and 462,31 contains 95% the true mean of the sample. The 95% CI for the median is between 68,04 and 133,63 for this data set. Furthermore, this sample has a standard deviation of 841,12 and the 95% CI for standard deviation is between 765,97 and 932,74. In this case, the standard deviation is lower than the product category Adidas Shoes Men but it still shows that a lot of numbers are spread out from the average.
To compare the total number of reviews of the product categories “Adidas Shoes Men” and “Adidas Shoes Women” 2-sample t-test is applied. With the application of the 2-sample t-test, it is tested whether the means of the two samples are statistically equal. Therefore the null hypothesis is H₀: μ₁ - µ₂ = 0 and the alternative hypothesis is H₁: μ₁ - µ₂ ≠ 0. μ₁ indicates the mean of the total number of reviews for Adidas Shoes Men and µ₂ stands for the total number of reviews for Adidas Shoes Women. As shown below, to strengthen the outputs of the test, equal variances were assumed for this case. Below are the outputs of the 2-sample t-test:
Two-Sample T-Test and CI: Adidas Shoes Men Total Number of Reviews; Adidas Shoes Women Total Number of Reviews
Method
μ₁: mean of Adidas Shoes Men Total Number of Reviews µ₂: mean of Adidas Shoes Women Total Number of Reviews Difference: μ₁ - µ₂
Equal variances are assumed for this analysis.
16
Sample N Mean StDev SE Mean
Adidas Shoes Men Total Number of Reviews
200 557 1.167 82 Adidas Shoes Women Total Number of
Reviews
200 345 841 59
Estimation for Difference
Difference Pooled StDev 95% CI for Difference 212 1.017 (12; 412) Test Null hypothesis H₀: μ₁ - µ₂ = 0 Alternative hypothesis H₁: μ₁ - µ₂ ≠ 0 T-Value DF P-Value 2,08 398 0,038
17
Graph 3: Boxplot of Total Number of Reviews for “Adidas Shoes Women” and "Adidas Shoes Men" Category
As shown above, in graph 3, the boxplot displays the distribution of the data for the compared product categories. The points which are mentioned with stars are the so-called outliers in the data set. Products which have earned 0 total reviews are defined as outliers but were determined in step 1 and excluded directly in the data collection part. The line in the box represents the median and is the measure of the center of the data set. The interquartile range box represents the middle 50% of the data. The interpretation of graph 3 is that both of the samples have a lot of outliers of the median and the median of Adidas Shoes Men total number of reviews is higher than Adidas Shoes Women total number of reviews.
4.1.2 Levi’s Jeans
18
Graph 4: Summary Report of Total Number of Reviews for “Levi’s Jeans Men” Category
Data for 148 products for Levi’s Jeans Men were collected as shown in graph 4. The sample size is 148 instead of 200 since this product category has just 148 products which have earned a review. This sample has a mean of 1.082,1 total number of reviews and a standard deviation of 2.787,2. The 95% CI for the mean shows that the range of values between 629,3 and 1.543,9 contains 95% the true mean of the sample. The 95% CI for the standard deviation is between 2.501,8 and 3.146,8 for the product category Levi’s Jeans Men.
Before applying the 2-sample t-test the summary report for the second sample Levi’s Jeans Women was created which is shown in graph 4. The sample size (N) of this category is 195. The difference of the sample size between Levi’s Jeans Men and Levi’s Jeans Women is by reason of limited provided data on Amazon.com for those categories.
19
Levi’s Jeans Women has a mean of 402 total number of reviews and a standard deviation which is 1.430. The minimum of this data set is 1 and the maximum is 13.451 customer reviews. 95% CI for the mean is between 200 and 603,9. The 95% CI for the standard deviation is between 1.300,7 and 1.587,9. The median of this data set is 26 customer reviews and the 95% CI for the median is between 10 and 47.
In the 2-sample t-test for the categories Levi’s Jeans Men and Levi’s Jeans Women, the null hypothesis is H₀: μ₁ - µ₂ = 0 and the alternative hypothesis is H₁: μ₁ - µ₂ ≠ 0. The test is again accomplished with the assumption of equal variances.
Two-Sample T-Test and CI: Levi’s Jeans Men Total Number of Reviews; Levi’s Jeans Women Total Number of Reviews
Method
μ₁: mean of Levi’s Jeans Men Total Number of Reviews µ₂: mean of Levi’s Jeans Women Total Number of Reviews Difference: μ₁ - µ₂
Equal variances are assumed for this analysis.
Descriptive Statistics
Sample N Mean StDev SE Mean
Levi’s Jeans Men Total Number of Reviews 148 1.082 2.787 229 Levi’s Jeans Women Total Number of
Reviews
195 402 1.430 102
Estimation for Difference
Difference Pooled StDev 95% CI for Difference 680 2.124 (225; 1.136) Test Null hypothesis H₀: μ₁ - µ₂ = 0 Alternative hypothesis H₁: μ₁ - µ₂ ≠ 0 T-Value DF P-Value 2,94 341 0,004
20
that the difference between the means is statistically significant. Therefore H₀ is rejected. As a conclusion, it can be stated that the product category Levi’s Jeans Men has a higher mean for total number of reviews than the product category Levi’s Jeans Women.
Graph 6: Boxplot of Total Number of Reviews for “Levi’s Jeans Men” and “Levi’s Jeans Women” Category
The graph 6 shows that Levi’s Jeans Men has products that have earned more than 20.000 reviews and that the maximum for Levi’s Jeans Women is below 15.000 reviews. Most of the number of reviews for Levi’s Jeans Women are accumulated between 0 and 5.000 reviews. From the summary report and boxplot for these product categories it can be stated that Levi’s Jeans Men has a higher median and mean than Levi’s Jeans Women in terms of total number of reviews.
4.2 Analysis of Star Ratings
4.2.1 Adidas Shoes
21
Graph 7: Summary Report of Star Ratings for “Adidas Shoes Men” Category
As shown in graph 7, the mean of star ratings for Adidas Shoes Men is 4,45 and the standard deviation is 0,3875. The minimum is 1 and the maximum is 5 which shows the range of star ratings that is possible to earn for a product. Adidas Shoes Men category has a 95% CI for the mean between 4,3938 and 4,5052 and a 95% CI range between 0,3522 and 0,4307 for the standard deviation.
Graph 8 shows the summary report for the category Adidas Shoes Women which has a sample size of 193 products. The sample size consists of 193 products since 193 products out of 200 have earned a star rating in this category.
22
The products which include the keywords Adidas Shoes Women in their description have earned star ratings with a mean of 4,3228 and a standard deviation of 0,5226. The 95% CI for the mean of this category is between 4,2486 and 4,3970.
As in the comparison for the total number of reviews between the categories “Adidas Shoes Men” and “Adidas Shoes Women”, the 2-sample t-test is applied for the comparison of the means for the star ratings. Equal variances are assumed and the null hypothesis is H₀: μ₁ - µ₂ = 0 and the alternative hypothesis is H₁: μ₁ - µ₂ ≠ 0.
Two-Sample T-Test and CI: Adidas Shoes Men Star Rating; Adidas Shoes Women Star Rating Method
μ₁: mean of Adidas Shoes Men Star Rating µ₂: mean of Adidas Shoes Women Star Rating Difference: μ₁ - µ₂
Equal variances are assumed for this analysis.
Descriptive Statistics
Sample N Mean StDev SE Mean
Adidas Shoes Men Star Rating 192 4,450 0,388 0,028 Adidas Shoes Women Star Rating 193 4,323 0,523 0,038 Estimation for Difference
Difference Pooled StDev 95% CI for Difference 0,1272 0,4602 (0,0350; 0,2194) Test Null hypothesis H₀: μ₁ - µ₂ = 0 Alternative hypothesis H₁: μ₁ - µ₂ ≠ 0 T-Value DF P-Value 2,71 383 0,007
23
Graph 9: Boxplot of Star Ratings for “Adidas Shoes Men” and “Adidas Shoes Women” Category
Graph 9 above shows the boxplot for the product categories Adidas Shoes Men and Adidas Shoes Women. For both of the categories, a lot of outliers exist in the range of 5 stars. The accumulated data points below 3-star ratings are more in the product category Adidas Shoes Women than in Adidas Shoes Men. From the summary report and boxplot for these product categories, it can be stated that Adidas Shoes Men has a higher median and mean than Adidas Shoes Women in terms of star ratings.
4.2.2 Levi’s Jeans
24
Graph 10: Summary Report of Star Ratings for “Levi’s Jeans Men” Category
Customers uploaded star ratings with a mean of 4,1757 and a standard deviation of 0,6599 for the product category Levi’s Jeans Men. The 95% CI for the mean is between 4,0685 and 4,2829. Additionally, the 95% CI for the standard deviation is between 0,5923 and 0,7450. Graph 11 shows the outputs for the product category Levi’s Jeans Women and has a sample size of 195 products which include the keywords Levi’s, Jeans and Women in their description.
Graph 11: Summary Report of Star Ratings for “Levi’s Jeans Women” Category
25
Below is the 2-sample t-test for “Levi’s Jeans Men” and “Levi’s Jeans Women”. The 2-sample t-test is created with the assumption of equal variances of the two samples and the null hypothesis is H₀: μ₁ - µ₂ = 0 and the alternative hypothesis is H₁: μ₁ - µ₂ ≠ 0.
Two-Sample T-Test and CI: Levi’s Jeans Men Star Rating; Levi’s Jeans Women Star Rating Method
μ₁: mean of Levis Men Star Rating µ₂: mean of Levis Women Star Rating Difference: μ₁ - µ₂
Equal variances are assumed for this analysis.
Descriptive Statistics
Sample N Mean StDev SE Mean
Levis Men Star Rating 148 4,176 0,660 0,054 Levis Women Star Rating 195 4,106 0,748 0,054 Estimation for Difference
Difference Pooled StDev 95% CI for Difference 0,0695 0,7113 (-0,0830; 0,2220) Test Null hypothesis H₀: μ₁ - µ₂ = 0 Alternative hypothesis H₁: μ₁ - µ₂ ≠ 0 T-Value DF P-Value 0,90 341 0,371
26
Graph 12: Boxplot of Star Ratings for “Levi’s Jeans Men” and “Levi’s Jeans Women” Category
The boxplot for the product categories Levi’s Jeans Men and Levi’s Jeans Women in graph 12 indicates that the outliers are below the median and the mean. They are mainly accumulated below 3 star ratings in this case. The mean of Levi’s Jeans Men star ratings is 4,1757 and the median is equal to 4,3. Levi’s Jeans Women has a mean of 4,1062 and a median of 4,2. These two samples have almost the same median and mean. To be on the same point, the 2-sample t-test has indicated that there is not a statistical significant difference between these two samples.
4.3 Text-Typed Comments and Word Frequencies
27
4.3.1 Adidas Shoes Men
As mentioned earlier, there are 110.183 words collected for the products which are accessed by using the search term “Adidas Shoes Men”. After data is analysed, the result shown in Figure 1 is found.
Figure 1: Word distribution and frequencies for "Adidas Shoes Men"
Figure 1 shows the word distribution and word frequencies for Adidas Shoes Men. Whilst the left figure shows the word distribution, the right one displays the word frequencies. Ten most used words for this criteria are; size, comfortable, like, great, fit, good, wear, pair, just and look respectively.
4.3.2 Adidas Shoes Women
28
Figure 2: Word distribution and frequencies for "Adidas Shoes Women"
Figure 2 displays the word cloud and the word frequencies for the search term Adidas Shoes Women. Top ten words for this product category are; size, comfortable, fit, wear, like, love, feet, great, pair and just.
4.3.3 Adidas Shoes Men and Women Comparison
29
Figure 3: Word Frequency comparison of "Adidas Shoes Men" and "Adidas Shoes Women"
The figure 3 shows the top twenty words how much they used in numbers and also gives the proportional values in total number of the sample size, frequencies, as percentages for the product categories Adidas Shoes Men and Adidas Shoes Women.
Seventeen words in the top twenty words are in both product categories. However, their frequencies differ from each other. Top two words that are used most for both product categories are size and comfortable with frequencies of 0,679% and 0,661% for Adidas Shoes Men and 1,156% and 0,789% for Adidas Shoes Women. Whilst the frequency difference for the word comfortable is 0,128%, the frequency difference for the word size is 0,477%. When other words are investigated, it can be seen that most of the words’ frequency difference between Men and Women for Adidas Shoes is less than 0,1%. The words that have more than 0,1% higher frequency in Adidas Shoes Women in comparison to Adidas Shoes Men are stated with the percentage differences; wear by 0,161%, love by 0,290%, feet by 0,164%, ordered by 0,187%, big by 0,227% and perfect by 0,165%. The words that are used in Adidas Shoes Men with more than 0,1% difference in comparison to Adidas Shoes Women are; great by 0,134 and good by 0,183.
4.3.4 Levi’s Jeans Men
30
Figure 4: Word distribution and frequencies for "Levi’s Jeans Men"
Figure 4 shows that the first ten words that are most used for Levi’s Jeans Men are; fit, pair, size, like, good, pants, just, quality, great and bought.
4.3.5 Levi’s Jeans Women
Last investigated category is Levi’s Jeans Women. 85.119 words are analyzed for this category and the results that show the word cloud and word frequencies can be seen in Figure 5.
Figure 5: Word distribution and frequencies for “Levi’s Jeans Women”
31
4.3.6 Levi’s Jeans Men and Women Comparison
The comparison of the word frequencies between Levi’s Jeans Women and Levi’s Jeans Men is made with the same approach that used in Adidas Shoes product categories. The words are shown in Figure 6 as in order by their percentages of typed in their data sample for that specific product category.
Figure 6: Word Frequency comparison of "Levi’s Jeans Men" and "Levi's Jeans Women"
Figure 6 displays of the top twenty words that are typed in their product categories. Where the count column states how many times that word is used in that data sample, the frequency column shows the percentages of the count value for that category.
32
5. Conclusion
The review of past research shows that the fashion retail industry was not analyzed intensely according to specified product categories in terms of customer behaviour analysis. This paper investigated the analysis of customer behaviour to review and rate gender-based products. Two research questions are established to obtain a deep understanding of this aspect. The research questions are: 1) How does the total number of reviews and star ratings of a gender-based product vary according to different product categories? and 2) How does the word frequency and wording of text-typed comments change for different gender-based product categories?. To answer the mentioned research questions, quantitative research methods were used. Descriptive statistics and two-sample t-test were implemented for the first question and quantitative content analysis for the second one.
After the implementation of the analysis for the first research question, the following interpretations are made: The mean between the product categories Adidas Shoes Men total number of reviews and Adidas Shoes Women total number of reviews is statistically different and the product category Adidas Shoes Men has a higher average of Total Number of Reviews than the product category Adidas Shoes Women. This shows that the product category Adidas Shoes receives more customer reviews for the products that include the keyword Men in their description. The means of the samples for Levi’s Jeans Men total number of reviews and Levi’s Jeans Women total number of reviews is statistically different according to the p-value in the 2-sample t-test. Levi’s Jeans Men has a higher average of Total Number of Reviews than Levi’s Jeans Women. The p-value of the 2-sample t-test for the samples Adidas Shoes Men star rating and Adidas Shoes Women star rating indicate that the mean of these two samples is statistically different. Adidas Shoes Men has a higher average of Star Ratings then Adidas Shoes Women. Finally, the 2-sample t-test was applied for the product categories Levi’s Jeans Men star rating and Levi’s Jeans Women star rating. The p-value of this test shows that the difference between the means of these two samples is not statistically different. Therefore, a conclusion is not made for the comparison of star ratings of these categories. Previous literature states that men tend more to post messages to give information as discussed in the Literature Review chapter. With considering the limitation that the sale amount of the chosen products that may affect the number of reviews directly proportional is not included as an additional variable, the following can be stated: Levi’s Jeans Men has a higher total number of reviews than Levi’s Jeans Women and Adidas Shoes Men has a higher total number of reviews than Adidas Shoes Women. Additionally, in the previous literature, it was expressed that men are more emotionally satisfied with online shopping and have a more positive attitude towards internet shopping. The analysis of this paper shows a link to the literature where it is possible to observe that the products in the category Adidas Shoes Men have a higher average of star ratings than in the product category Adidas Shoes Women. These findings correspond to the literature in terms of the behavioral changes of customers when the keywords of the product description change. These changes consist of the type and gender of the product. As stated above, men are keener to give information and show a more positive attitude while they shop online which can be related to the analyzed gender-based product categories in this research paper. The importance of these findings is that the companies and/or sellers can pay more attention to increase the satisfaction rate of the products which consists of the keyword women in their product description. That is how they may increase the rating star of a product and may increase the sale amount.
33
To answer the second research question of this paper, the text-typed comments by the reviewers are analysed for different product categories. The results show that there are similarities in the words that are written for different search terms in the same product category since most of the words in the top twenty most used words are very similar to each other. This result can be interpreted that even though the words belong to different gender-based product categories, the text-typed comments of the reviews are similar. When taking the word frequencies into consideration, most of the words have close frequencies to each other. However, the word frequencies in Adidas Shoes Women and Levi’s Shoes Women are slightly higher than Adidas Shoes Men and Levi’s Jeans Men in general for most of the words in the top twenty most used words. There are some salient words that show more difference. These can be stated as size, love, perfect and comfortable. Both for Adidas Shoes and Levi’s Jeans, these words are used more in Women searched terms than the Men. The frequency difference for size is 0,477%, for love is 0,290%, for perfect is 0,165%, for comfortable 0,128% in Adidas Shoes, and for Levi’s Jeans product category, the difference for these words are 0,455%, 0,265%, 0,157% and 0,096% respectively. Another difference in the word frequency analysis is that the words good and quality are used more in Men search terms than the Women. Where the frequency difference for the word good is 0,183% in Adidas Shoes and 0,097% in Levi’s Jeans, the difference for the words quality is 0,093% and 0,200% respectively. Other words do not show similar characteristics when both product categories are taken into consideration. It is stated in the literature review part that men are task-oriented and women pay more attention to their social interaction and confidentially of their personal information. When looking into gender-based products, there is not any relation to this remark since there are not significant frequency differences that can be associated. Another subject that is mentioned in the literature review is that women are more inclined to raise their personal issues. In the analyzed gender-based product categories the words do not show that the product categories which include the keyword women raise their personal issues. However, the comments in these product categories show a higher likelihood to mention the personal opinion. In the product categories with the keyword women the words love, perfect and comfortable are more used than in the product categories with the keyword men which can be linked to a statement of a personal opinion. These findings can be interpreted as the words are written more densely for the product categories which consist of the keyword women than the product categories with the keyword men in their description, since their frequencies are higher. Even though the difference is low, it can still be beneficial. The significance of this finding is that the companies and/or individual sellers can focus more on the most used words in the product description to draw attention to those products. Using these words may make it easier for the possible buyer to get more precise and necessary information. Additionally, they can pay more attention to the words that express the reflection on the personal opinions for the products which consist of women in their search term which may also provide better information to the potential customer.
34
35
6. Future Research
36
7. References
#Envision2030 Goal 12: Responsible Consumption and Production Enable. (n.d.). Retrieved March 20, 2020, from https://www.un.org/development/desa/disabilities/envision2030-goal12.html
#Envision2030 Goal 13: Responsible Consumption and Production Enable. (n.d.). Retrieved March 20, 2020, from https://www.un.org/development/desa/disabilities/envision2030-goal13.html
Abdul-Muhmin, A. G. (2010). Repeat Purchase Intentions in Online Shopping: The Role of Satisfaction, Attitude, and Online Retailers Performance. Journal of International Consumer Marketing, 23(1), 5–20. doi: 10.1080/08961530.2011.524571
About. (n.d.). Retrieved May 5, 2020, from https://voyant-tools.org/docs/#/guide/about About Us. (n.d.). Retrieved May 2, 2020, from https://www.minitab.com/en-us/company/ Alexandrov, A., Lilly, B., & Babakus, E. (2013). The effects of social- and self-motives on the intentions to share positive and negative word of mouth. Journal of the Academy of
Marketing Science, 41(5), 531–546. doi: 10.1007/s11747-012-0323-4
Awad, N. F., & Ragowsky, A. (2008). Establishing Trust in Electronic Commerce Through Online Word of Mouth: An Examination Across Genders. Journal of Management
Information Systems, 24(4), 101–121. doi: 10.2753/mis0742-1222240404
Bae, S., & Lee, T. (2010). Gender differences in consumers’ perception of online consumer reviews. Electronic Commerce Research, 11(2), 201–214. doi: 10.1007/s10660-010-9072-y Boettger, R. K., & Palmer, L. A. (2010). IEEE TRANSACTIONS ON PROFESSIONAL COMMUNICATION. Quantitative Content Analysis: Its Use in Technical Communication, 53.
Bounie, D., Bourreau, M., Gensollen, M., & Waelbroeck, P. (2008). Do Online Customer Reviews Matter? Evidence from the Video Game Industry. SSRN Electronic Journal. doi: 10.2139/ssrn.1091449
Boyd, E. A., & Bilegan, I. C. (2003). Revenue Management and E-Commerce. Management Science, 49(10), 1363–1386. doi: 10.1287/mnsc.49.10.1363.17316
Bradley, G. L., Sparks, B. A., & Weber, K. (2016). Perceived prevalence and personal impact of negative online reviews. Journal of Service Management, 27(4), 507–533. doi:
10.1108/josm-07-2015-0202
