0
What Characteristics of Fashion Leaders Relate to The Apparel Product Return : A Study of Product Return in
E-commerce
MSc Supply Chain Management University of Groningen Economics and Business Faculty
Tingting Wang Student number: s2490005
Westerbinnensingel 17a 9718BR Groningen Tel: +31 62620 7215 E-mail: t.wang.2@student.rug.nl First Supervisor: drs. J.C. (Jerry) van Leeuwen Second Assessor: prof. dr. ir. J.C. (Hans) Wortmann
6/22/2015
ACKNOWLEDGEMENT
The process of earning a master degree and writing a thesis is arduous, and it undoubtedly cannot be done single-handedly.
First and foremost, I would like to acknowledge with deep appreciation and gratitude the invaluable and extraordinary support of Dr. J.C. (Jerry) van Leeuwen, for giving me guidance and encouragement in this thesis process.
I would also like to thank prof. Dr. ir. J.C. (Hans) Wortmann for his expertise and advice.
Thanks my fellows, Lingxi and Alwin for their help and cooperation during the whole writing process.
And I sincerely thank my parents for encouraging me and giving me all the loves unconditionally. Also, I heartily thank all my friends helping me out when I encountered difficulties.
Finally, I especially appreciate Sytze for his encouragement and supports all the time.
ABSTRACT
This paper seeks to investigate the relationships between apparel return and the factors of fashion leaders, including gender, income level, age, and fashion magazine readership. Impulse buying is also discussed in this paper. A real company data analysis and a survey study together measured apparel returns, gender, income level, age, fashion magazine readership, and buying impulsiveness. Welch t-test and non-parametric correlation test were used to analyze the data. It was found that apparel return behavior was positively related to female, fashion magazine reader and impulse buying behavior. Gender and fashion magazine readership were two significant predictors of frequency of apparel returns. However, neither gender nor fashion magazine readership were moderator of the coefficient between impulse buying behavior and apparel product return frequency. Age and income level were not significantly related to both impulse buying behavior and frequency of apparel returns.
The findings are useful for e-commerce companies to have better understanding of which characteristics of fashion leaders have impact on impulse buying behavior and apparel product returns.
Key words: online shopping; impulse buying; gender; age; income; fashion magazine readership; fashion leader; apparel product return, reverse supply chain
Contents
ACKNOWLEDGEMENT ... 1
ABSTRACT ... 2
INTRODUCTION ... 4
THEORETICAL BACKGROUND AND CONCEPTUAL MODEL ... 6
The Causes of Product Return for Apparel Products in E-commerce ... 6
Product Returns in Reverse Logistics ... 6
Fashion Leader and Apparel Product Return ... 7
Gender and product return ... 8
Age and product return ... 9
Income and product return ... 10
Fashion magazine readership and product return ... 10
Conceptual Model ... 12
METHODOLOGY ... 12
Data Collection ... 12
Measurement ... 13
Return frequency of online shopping ... 13
Income level ... 13
Fashion magazine readership ... 14
Impulse buying... 14
DATA ANALYSIS ... 14
Gender, Age and Apparel Product Return ... 14
Income and Apparel Product Return ... 17
Fashion Magazine Readership ... 17
Impulse Buying ... 17
DISCUSSION ... 20
CONCLUSION AND FURTHER STUDY ... 22
APPENDIX A ... 24
APPENDIX B ... 24
REFERENCE ... 25
INTRODUCTION
The amount of online shoppers has increased in recent decades. According to CBS (2014), in 2013, 83% of the online shoppers (around 10.3 million people) were between the ages of 12 and 74. Frequent buyers mostly ordered clothes (CBS, 2014).
Due to the lack of the opportunity for customers to examine the products physically, which is \ different from offline shopping, online shopping causes lots of product return (Dholakia, Zhao, & Dholakia, 2005). An increase in product returns leads to bad results such as shrinking profit margins, increasing maintaining inventory costs and opportunity costs, etc. (Jeszka, 2014). To prevent the negative results of product returns, reverse supply chain management for online e-retailers is essential. Besides the traditional forward supply chain, reverse supply chain is another element in closed loop supply chain (Morana & Seuring, 2007). To manage the reverse supply chain, forecasting the backflow of a closed-loop supply chain is needed ( Krapp, Nebel, &
Sahamie, 2013). Because of the uncertainties within product returns, for instance the uncertainty of time and quantity, several numerical studies have built the forecast systems to predict future returns (De Brito & Van der Laan, 2009; Krapp et al., 2013).
Since return management is rooted in both marketing and logistic disciplines, Mollenkopf, Russo, and Frankel (2007) have provided a link between logistics and marketing, claiming that both functional integration and supply chain orientation are related to a firm’s management of the returns process. As one of the external environment elements, customers can influence a firm’s return management, such as physical flows, information flows and financial flows in the supply chain (Mollenkopf et al., 2007). In addition, customers act as an external force to drive the reverse logistics system, which depends on who the customers are, what the customers need, and how the customers support it (Dowlatshahi, 2000). However, an in-depth study of the relation between customers and product returns is still lacking. For e-retailers, it’s important to understand the underlying reasons for frequent returns from customers’
aspects when planning for the return management process, but not only from logistical and economical aspects (Lee, 2015).
Researches on commercial returns have shown that the reasons for product returns can be: defects, product incompatibility with user needs, and deficiencies in product performance, which is relative to customer expectations (Ferguson, Guide, Souza, 2006; Guide, Souza, Van Wassenhove, & Blackburn, 2006; Rao, Rabinovich, & Raju, 2014). According to Lee (2015), the main reason for product returns during every early stage of consumption is not “dissatisfaction”. In addition, in the study of Lawton (2008), the main reason of returns is narrowed down to: “products are not meeting consumer’s needs”. As a consequence of the characteristics of online shopping, consumers cannot touch or feel the products, especially the apparel products, causing customers to perceive more risks when making the decision to place the order (Foscht, Ernstreiter, Maloles III, Sinha, & Swoboda, 2013).
Different from common customers who are trying to avoid risks when purchasing apparel products online, fashion innovators are more likely to do impulse buying of apparel products, and they don’t mind spending more money on fashion products (Phau & Lo, 2004). In previous studies, scholars have found that when compared with non-fashion innovators, fashion innovators are younger, having more tendency to be fashion opinion leaders, and spending more money on new fashion (Goldsmith &
Stith, 2011). Likewise, Shim & Mahoney (1992) have found that the heavy catalog users of fashion products are more likely to have higher fashion-consciousness, higher income and higher frequency in social events.
Some scholars have already revealed that demographics and lifestyle have influence on consumers’ online shopping attitudes (Hashim, Ghani, & Said, 2009). Kang &
Johnson (2009) examined the relation between fashion innovators and apparel product return, and they claimed that fashion innovativeness was not significantly related to consumers’ return behavior, but the impulsive purchase was. However, they neglected the interrelation between fashion innovators and impulsive buying behaviors and they also ignored that fashion innovativeness is not the only factors of fashion innovators.
Some scholars have already summarized the general demographic profiles and lifestyles of fashion leaders, and they implied that younger females, with higher income and who strongly associate with fashion magazine readership tend to be fashion leaders (Summers, 1970). The gender, income level, and exposure to fashion magazines readership can be considered as characteristics of fashion leaders, and the relation between these characteristics and impulse buying and apparel product returns need to be investigated. Therefore this paper implements the study of Kang and Johnson (2009), and further investigates the following question:
What Characteristics of Fashion Leaders Relate to the Apparel Product Return in E-commerce?
It is an interesting theme for e-retailers of apparel web shops, because by answering this question, this paper can provide implications to them about how to predict the future returns, thus they can redesign the product return process for different types of consumers, or make logistics strategies to handle the product returns. For the theoretical contribution, this paper is trying to fill the gap between consumer characteristics and the impacts of logistics, and also adds the consumer perspective into the product return forecasting system in reverse supply chain.
The rest of this paper is organized as follows. Section two presents the framework of the resource dependency, which provides a theoretical basis for the research objectives of this article. In the third section, details of the research methodology are explained, including data collection procedures, variables measurement. Hypotheses testing and results are presented in section four. In the fifth section, discussion is offered. Finally, in the concluding section, the suggestions for further investigation and managerial implications are highlighted.
THEORETICAL BACKGROUND AND CONCEPTUAL MODEL
The Causes of Product Return for Apparel Products in E-commerce
Research on commercial returns have shown that the reasons for product returns could be defects, product incompatibility with user needs, and deficiencies in product performance relative to customer expectations (Ferguson, Guide, Souza, 2006; Guide, Souza, Van Wassenhove, & Blackburn, 2006; Rao, Rabinovich, & Raju, 2014).
However, the main reason of returns suggested by Lawton (2008) is that products are not meeting consumer’s needs. Because of the characteristic of online shopping, customers cannot feel or touch the products; the product information becomes important for customers to familiarize themselves with the product characteristics.
Lee (2015) also indicates that one of the reasons consumers return products is
“purchase with incomplete product knowledge”. He explains that when consumers encounter problems because they lack information about how to use the complex product after purchased, they will become anxious or uncertain with the purchased product and then tend to return it (Lee, 2015). However, not every customer has a perfect idea of his own preference, especially when impulse buying occurs. Kang and Johnson (2009) claim that the impulse buyers tend to purchase products without in-depth evaluation or careful consideration. Regret and return are always consistent with impulsive purchases (Rook, 1987; Park & O’Neal, 2000). As a result, return of impulse purchases occurs.
Even though the description of products is thorough and the product is expected to perform well, customers may not be able to assess whether a product’s attribute match their preference, this is named product fit uncertainty (Hong & Pavlou, 2014). Product fit relates to experiential product attributes, such as the fit of clothes (Hong & Pavlou, 2014). Consumers who purchase apparel product online can only know whether the product fits him or her after receiving it, and if the product-preference is mismatched, it will lead to dissatisfaction and product return (Hong & Pavlou, 2014). In addition, Lee (2015) has found that one of the reasons consumers returned products is
“acquisition of additional informgation after purchase”. What’s more, if consumers get some negative information from the third party about the product they purchased, they will probably return it.
Product Returns in Reverse Logistics
The implementation of reverse logistic is complex because “there are many unknown parameters such as product return quantity, time and quality in the process” (Temur, Balcilar, & Bolat, 2014). Lots of scholars regard the cost of product returns, even the cost of non-defective returns, as very costly expense since the return process includes the collection, transportation, testing, sorting, cleaning, packaging, and redistribution,
etc. (Ruiz-Benítez, Ketzenberg, & Van der Laan, 2014). Cost-efficient reverse supply chain is attempting to reduce the cost, however, it does not focus on time efficiency, resulting the returns go through a long time delay before they are reintroduced to the market (Blackburn, Guide Souza, & Van Wassenhove, 2004). The longer the return takes, the more likelihood of the significant obsolescence cost occurs (Rogers &
Tibben-Lembke, 2001; Blackburn et al., 2004). Fashion-related merchandise, especially the clothing, experiences such value decline over time (Rogers & Tibben‐
Lembke, 2001). Since apparel product is time-sensitive, the later it is reused, the less value it has. Therefore, it is important to deal with the returns quickly and efficiently.
Studies recognized that the low productivity and efficiency for return processing were often due to erratic volume flows and increased handling (Mollenkopf, Frankel, &
Russo, 2011). In order to predict and eliminate the fluctuation of apparel product return volume and reduce the amount of apparel products entering into the return flow, to understand the return behavior of consumers is essential.
In the study of Temur et al. (2014), they mention that customer segment (De Brito, 2004) and income (WEEE, 2003) are two socioeconomic factors that impact return quantity. By analyzing the customers, suppliers can tailor the suitable return policies that help customers achieve their own commercial goals (Mollenkopf et al., 2011), therefore reducing the product return.
Fashion Leader and Apparel Product Return
Fashion leader is a broad term that includes fashion innovators and fashion opinion leaders. Bailey and Seock (2010) conclude that consumers who have higher fashion innovativeness and opinion leadership tend to be fashion leaders (Bailey & Seock, 2010). Fashion opinion leaders are distinguished from non-leaders by demographic, sociological, attitudinal, communication, and fashion involvement measures (Summers, 1970). Fashion opinion leaders are identified to be predominately young women, usually they have higher incomes, higher occupational status, and associate more with fashion magazine readership than non-fashion leaders (Summers, 1970).
Fashion leaders are more likely to experience fashion oriented impulse buying since they will decide to buy a new style product which refers to the fashion ability (Stern, 1962; Phau & Lo, 2004). During the impulse buying, consumers do their purchasing without an in-depth evaluation of the product information or even neglect the possibility of undesirable post-purchase experiences (Kang & Johnson, 2009). A study analyzed that consumers who do impulsive buying frequently are more likely to shop for hedonic needs, and once the product was purchased to satisfy hedonic needs, it might quickly lose its value (Kang & Johnson, 2009). Impulsiveness is one of the traits of innovators in psychology (Phau & Lo, 2004). Even though Kang and Johnson (2009) suggest that frequent apparel return behavior is not related to fashion innovativeness, however, they don’t consider the implication of web shops, since internet plays a role as impulsive channel (Phau & Lo, 2004). Impulse buying is an
unplanned behavior, and such purchase may leave the financial or other consequences out of consideration (Park, Kim, Funches, & Foxx, 2012), therefore, it easily leads to regret after purchasing the product.
To better analyze which characteristics of fashion leaders have an impact on apparel product return, this paper will investigate the relationship between demographic profiles (gender, income, age) and lifestyle (fashion magazine readership) of fashion leaders and apparel product return.
Gender and product return
Hashim, Ghani and Said (2009) have suggested that demographic characteristics are regarded as external influences upon consumer behavior, while beliefs, attitudes, personalities, perceptions are internal factors. Since demographic characteristics are easy to assess when compared to internal factors, they have often been used to segment the consumer population for better marketing (Phang, Kankanhalli, Ramakrishnan, & Raman, 2010). Haque and Khatibi (2006) and Hashim et al. (2009) have examined the demographic factors that have significant influence on consumers’
online shopping attitudes. Gender, income, age, are commonly used as demographic characteristics which people use to label themselves and others (Hashim, et al., 2009;
Naseri, & Elliott, 2011).
Females are more willing to spend on fashion, and they are found to have more fashion innovativeness and fashion leadership than males (Goldsmith, Stith, & White, 1987; Kwon & Workman, 1996; Studak & Workman, 2004; Ha & Stoel, 2004).
Leisure and enjoyment are two important shopping goals of females, who prefer hedonic browsing over search or deliberation (Phang, 2010). Since women’s online shopping tends to be hobby or interest driven, female consumers always have higher playfulness, which results in a more positive mood, greater shopping satisfaction and impulsive buying behavior (Wolfinbarger & Gilly, 2010). Female consumers have always regarded impulse buying as a way to release stress and manage their moods, therefore they are more likely to be engaged in impulse buying than male consumers (Coley & Burgess, 2003). Impulse buying has been proven to be positively related to post-purchase regret (Saleh, 2012). The regretful consumers probably tend to return the product after they thought they had made a bad decision (Chebab & Gharbi, 2010).
Women are found to have more feelings of regret or mixed feelings of pleasure and guilt than men after the impulse buying (Coley & Burgess, 2003; Saleh, 2012), thus we assume that women are more related to impulse buying and product return than men for the online shopping. In addition, different from women, men always regard shopping as a task with clear objective to fulfill, thus males shoppers can be regarded as a goal-focused shopper (Phang, 2010). Goal-focused shoppers can directly and quickly purchase what they need without distraction, since the majority of them have planned online purchase in advanced (Wolfinbarger & Gilly, 2010). Compared to women consumers, men spend more time on gathering useful information online before they do purchasing on web shop and have a lower probability of impulsive
buying. Male consumers desire to accomplish efficient online purchases successfully so they tend to get more useful information of products. Sufficient knowledge of the products and better preparation of purchasing can lower the probability of neglecting the product information, and consequently lead to lower chance of return. Thus, we hypothesized that:
H1: Women have higher intention in impulse buying than men.
H2: Women have higher intention in apparel product return of online shopping than men.
H3: Impulse buying is positively related to apparel product return of online shopping.
Age and product return
According to CBS (2014), over 60% of individuals between 12 and 64 years old are frequent online shoppers for sport articles and clothes, while only 32% older than 65 are frequent online shoppers. On one side, some studies have proven that the majority of online shoppers are more likely to be younger and tend to have higher computer skills (Lokken & Cross, 2003; Phang et al., 2010). However, there is few literature supports that age has correlation with consumers’ online shopping behavior (Hernández, Jiménez, & Martín, 2011; Román, 2010).
On the other side, age relates to impulse buying (Ren & Zhang, 2010). Younger consumers are more likely to buy impulsively because they have fewer shopping experience and poorer judgment on intuition than elder consumers ( D'Astous, Maltais,
& Roberge, 1990; Ren & Zhang, 2010). In addition, fashion leaders are significantly younger than non-fashion leaders, and they are found to be more interested in adopting new apparel fashion trends, occupying the early majority section of the innovation cycle (Quigley & Notarantonio, 2009; Rahman, Saleem, Akhtar, Ali, &
Khan, 2014). In the study of Pentecost and Andrews (2010), they have found the group of consumers who from 18 to 28 years old have higher fashion fanship than any other age groups. In the meanwhile, they also have found that the group from 18 to 28 years old has significant more impulse buying than the other groups except the group from 29 to 40 years old (Pentecost & Andrews, 2010). Bashar, Ahmad, and Wasiq (2013) collected the data from consumers who are mainly between 25 to 39 years old and analyzed that there is a significant strong correlation between age and impulsive buying behavior. What’s more, young consumers have less leisure time thus have a lower tendency for information seeking when they are compared to the elder generations (Phang, et al., 2010). The lack of information searching can reduce the fair understanding of the product so that lead to lower satisfaction when consumers receive the product. It is likely that age has a strong relationship with impulsive buying and product return. Thus, our hypotheses are:
H4: Age is negatively related to impulse buying.
H5: Age is negatively related to product return of online shopping.
Income and product return
Usually, fashion leaders tend to have high income. They are heavy users of fashionable clothing (Goldsmith, 2000; Goldsmith & Stith, 2011). In the research of Rahman, Saleem, Akhtar, Ali, and Khan (2014), they proved that social values (including opinion leadership, and social status) have a positive relationship with the consumers’ intention to purchase new fashion clothes. New fashion clothes are always costly and unique, conveying the message of wealth and high status (Rahman et al., 2014). Since fashion leaders usually have higher status and greater wealth, they need to purchase appropriate fashionable clothing for their status and lifestyle, and need more fashionable products for different social activities (Summers, 1970; Bailey &
Seock, 2010).
Research suggested that high income consumers tend to shop online more than low income consumers since they believe that online shopping can save time (Naseri &
Elliott, 2011; Punj, 2011). High-income consumers, who lack leisure time, value their time more, because the time is related to opportunity cost (Phang et al., 2010). The study of Ciunova-Shuleska (2012) supports that lacking time for shopping and having plenty money on hand can positively influence impulsive buying behaviors. Moreover, Bashar, Ahmad, and Wasiq (2013) have proven that consumers who have good earnings are more likely to display impulsive-buying behaviors because they have more disposable income to spend on shopping. Once the consumer makes a purchasing decision under time pressure and insufficient information, they tend to experience regret after the purchase (Lee, 2015). In the study of Foscht, Ernstreiter, Maloles III, Sinha and Swoboda (2013), they have found that “heavy returners”
perceive cost and time saving more important than other groups (“medium returners”
and “occasional returners”) in mail order purchase, while “occasional returners”
perceive risk avoidance most. “Heavy returners” are more likely to be indifferent about ordering more products online than they need in reality since they have little mind to spend extra time on other activities rather than shopping (Foscht et al., 2013).
Therefore they are likely to return a product that they bought in such situation (Foscht et al., 2013). Thus, the hypotheses are:
H6: Income is positively related to impulse buying.
H7: Income is positively related to apparel product return of online shopping.
Fashion magazine readership and product return
Fashion leaders have a strong brand and fashion consciousness, which is positively
related to information searches and online purchase of apparel products. This leads them to spend more money and effort on new fashion, searching for information before purchase products online (such as visit apparel websites to check the latest styles and new fashion trends), or expanding their scope of fashion knowledge through fashion magazines (Phau & Lo, 2004; Seock & Bailey, 2008). According to Lee (1995), a person’s hobby, consumption habits and lifestyle can be told from his or her choice of magazines. Venkatesh (1980) suggests that people in the same segment have similar characteristics and different segments have different magazine reading habits. Fashion leaders are most likely to be exposed to fashion magazines, and use the sources of fashion magazines (Summers, 1972; Bailey & Seock, 2010).
Even though most existing studies suggest that more product information leads to fewer returns because of less product uncertainty and more realistic product expectations, De, Hu, and Rahman (2013) have proposed a different conclusion. They claim that too much product information (e.g. use the visualization system) will raise the level of impression-based information, this may indeed lead to more returns (De et al., 2013). Once the perceived quality turns out to be lower than the expected quality, customer will feel dissatisfied, therefore product returns will occur (De et al., 2013).
Similar to the theory of De at al. (2013), Maity and Arnold (2013) have proven that information search is an expense rather than experience. They point out that more effort in research processes leading to high expectation about product performance (Maity & Arnold, 2013). Once the received products cannot meet consumers’
expectation, the satisfaction level will decrease, leading to higher product return intention (Bailey & Seock, 2010; Maity & Arnold, 2013). In addition, the mass advertising in fashion magazines can be a cause for impulse buying even though advertisements usually help consumers to do planned shopping (Stern, 1962). The mass advertising in fashion magazine also has reminder benefits that influence consumers buy the items which are unplanned but already gained the knowledge from magazines before (Stern, 1962). To sum up, the next hypotheses are as follows:
H8: Fashion magazine readership is positively related to impulse buying.
H9: Fashion magazine readership is positively related to apparel product return of online shopping.
Conceptual Model
Figure 1
Conceptual model
METHODOLOGY Data Collection
This research is a quantitative research, which is more suitable for rigorous tests of hypotheses (Pekrun, Goetz, Titz, & Perry, 2002). There were two stages for collecting the data. In the first stage, we cooperated with a Dutch company in the garment industry, which has a web shop, to get direct transaction data (gender, age, purchase orders, and return frequency) from its IT department. Since they couldn't provide the data of income and fashion magazine readership of each consumer, a designed questionnaire was used to conduct an in-depth analysis.
In the second stage, a questionnaire was developed by using multi-item scales drawn from the literature. Gender, age and return frequency were also addressed in the questionnaire in order to reinforce or reject the results from the first stage. Income level, fashion magazine readership and impulse buying were added in the questionnaire for the extended in-depth analysis.
Since the research was conducted in the Netherlands, the questionnaire was translated into the Dutch version from the English version. There were 3 steps to translate the questionnaire (Del Greco, Eastridge, & Walop, 1987; Harkness & Schoua-Glusberg, 1998). First, the English questionnaire was translated into the Dutch version by a bilingualist. After that another translator helped to translate the Dutch version back to the English, then we compared the difference between those two versions. Some spelling mistakes were corrected and two Dutch words were changed for better comprehension. Third, both the English and Dutch versions were sent to 5 native
H1（+）
H 2 (+) （+
H 3 (+)
H 4 (-) H 5 (-) H6 (+)
H7 (+)
H 8 (+)
H 9 (+)
speakers to see whether they can understand the questionnaire well. After translating the questionnaire, a pretest was applied. 10 respondents were asked to finish the questionnaire; we found that half of them were not willing to answer the question of the annual income. Therefore, we changed the open question of income into the scale of income catalogue. In addition, we asked respondents to select at most three types of favorite magazines instead of asking them to choose exactly three types of favorite magazines. In this way, it was not compulsory for respondents to choose 3 types of magazines, but only to choose their favorite ones.
After the new questionnaire was designed, it was sent to random respondents through internet, and conducted through random interviews on the shopping streets of Groningen. 164 respondents participated in this study. Two master students collected all the questionnaires and summarize the data in excel. After that, a SPSS analysis was conducted.
Measurement
Return frequency of online shopping
Using the frequency of product returns as the indicator of return was more accurate than other measurements. For example, many e-retailers encourage customers to buy more items to try them on, then return the rest back. Thus a consumer may return a lot of items at once. However, it doesn’t mean that this consumer is a frequent returner (Kang & Johnson, 2009). If we use the total number of return items, the measurement would be inaccurate. Kang and Johnson (2009) also suggest that individual return rates are not a proper indicator either. They give an example that if a person only places an order twice a year but return once, the return rate will be as high as 50%.
Thus researchers should not focus on this kind of returners, but those who order frequently and return frequently (Kang & Johnson, 2009). Therefore, in this paper we use the frequency of apparel product returns within last six months to measure the product return. What’s more, in the survey, we also asked the purchase frequency of respondents within last six months. If they didn’t purchase any apparel products online within the last six months, it was not necessary to count their return frequency.
Only respondents who purchased more than once would be taken in the analysis process.
Income level
Respondents were not willing to share the detailed income information even the questionnaire was anonymous. Thus, it was easier to asking income catalogues in the questionnaire instead of asking the open question of income. Four income catalogues were designed: less than €10,000 is scored as 1, between € 10,001 to 35,000 is scored as 2; between € 35,001 to 70,000 is scored as 3, over € 70,000 is scored as 4.
Impulse buying
I buy apparel items at a whim on the internet
During online shopping I buy apparel products without a lot of thinking I tend to think about it after purchasing
I tend to buy things I have no desire to buy during online shopping When I find something I like on the Internet I purchase it immediately
Fashion magazine readership
To distinguish fashion magazine readers and non-fashion magazine readers, respondents were asked to list at most 3 types of favorite magazines (among fashion, movie, music, autos, sports, or others). Respondents who chose fashion magazine as one of their favorite magazines were grouped into group 1 as fashion magazine readers, and others were grouped into group 2 as non-fashion magazine readers.
Impulse buying
Five items of impulse buying were modified from existing scales (Rook & Fisher, 1995; Park, Kim, & Forney, 2006; Park et al., 2012). A 7-point scale, with anchors from 1 as “very unlikely” to 7 as “very likely”, measured each item. After collecting the data, a reliability test of Cronbach’s alpha for these five items is conducted with minimum value of 0.6 to continue further analysis (Bland & Altman, 1997).
TABLE 1
Items of impulse buying
DATA ANALYSIS Gender, Age and Apparel Product Return
In the first phase, the real transaction data of the garment company were collected.
There were a total of 42,703 consumers included in this data. Since the sample size of women (31,897) and men (10,806) were different, Welch robust t-test should be used (Ramsey, 1980). The result in Table 2 shows that women (mean = 1.43) were more significantly (p-value =0.00 < 0.05) related to apparel product returns than men (mean
= 1.36) of online shopping. Thus, Hypothesis 2: “Women have higher intention in apparel product return of online shopping than men” was supported in the first phase.
TABLE 2
Independent Variable Sample size Mean Std. Deviation Sig. (Welch)
Female 31897 1.433 .958 .000
Male 10806 1.355 .008
Total 42703 1.413 .931
Gender and apparel product return (data from the garment company)
From the data of the garment company, the age of purchasers was mainly from 23 to 53 (63.9% of total purchasers), therefore in the analysis we used the data of purchasers who were between 23 and 53. In K-S Test (Table 3), p-value = 0.00 < 0.01, which means the null hypothesis, that the data is following a normal distribution, was rejected. Neither the data of age or return frequency were following the normal distribution, thus a non-parametric test should be used. Hypothesis 5: “Age is negatively related to apparel product return of online shopping” was rejected since there was no significant correlation between the age and apparel product return (Table 4, p-value = 0.059 > 0.05) in the first analysis phase.
TABLE 3
Age 23-53 Apparel Product Return
N 2728 2728
Mean 37,707 1,551
Std. Deviation 6,227 1,800
Sig. (2-tailed) ,000 ,000
Kolmogorov-Smirnov Test for the data of age and return
TABLE 4
Age 23 -53 Apparel Product Return
N 2728 2728
Correlation Coefficient ,036
Sig. （2-tailed） ,059
Correlation of age and return (data from the garment company)
The results from the second stage of analysis supported the results from the first stage.
There were totally 164 respondents that participated in the survey. 22 surveys were deleted from the data because they didn’t do any online shopping for apparel products within the last 6 months. Among the total of valid respondents, 93 were female and 49 were male. Therefore, robust tests should be used when testing the unequal size groups. As the results in Table 5 shows, gender had a significantly different impact (p-value = 0.00) on apparel product return of online shopping. Female consumers had significant higher average return frequency (mean = 1.4) then male consumers (mean
= 0.49)
TABLE 5
Independent Variable N Mean Std. Deviation Sig. (Welch)
Female 93 1.4 2.622 .003
Male 49 .49 .845
Total 142 1.08 2.217
Gender and apparel product return (data from the survey)
The age of respondents were ranged from 14 to 61years old, however, 81% of the respondents were between 20 and 30 years old. None of the data were following the normal distribution (Table 6), therefore non-paramatric correlation test should be used.
According to the result in Table 7, there was no significant linear relationship found in age (20 to 30) and apparel product return frequency of online shopping (p-value = 0.56> 0.05). In addition, in order to compare the result with the analysis in phase one, respondents who were between 23 to 53 years old were also tested. As Table 7 shows, the result supported the finding in the stage one. There was still no significant
correlation between age and apparel product return of online shopping (p-value = 0.22 > 0.05). The result of the additional test (Table 8) reinforced that there was no significant difference between age group 1 (from 23 to 30), group 2 (from 31 to 40), and group 3 (from 41 to 53) since p-value = 0.21 > 0.05. Thus, Hypothese 5: “Age is negatively related to apparel product return of online shopping” was still rejected.
TABLE 6
Age Return Age 23-53 Return23-53 Age 23-30 Return23-30
N 142 142 110 110 97 97
Mean 25,4577 1,08 25,7455 1,0909 24,1443 1,1443
Std. Deviation 7,44114 2,217 5,35087 2,26884 1,88740 2,35403
Sig. (2-tailed) ,000 ,000 ,000 ,000 ,000 ,000
Kolmogorov-Smirnov Test
TABLE 7
Age 23 -30 Return23-30 Age 23-53 Return23-53
N 97 97 110 110
Correlation Coefficient -,061 -,118
Sig. （2-tailed） ,556 ,219
Correlation between age and product return
TABLE 8
Age N Mean Std. Deviation Sig. (Welch)
23 – 30 373 1.576 .092 ,206
31 – 40 1434 1.493 .043
41 - 53 921 1.631 .067
Total 2728 1.551 .034
Difference between different age group
Income and Apparel Product Return
As an in-depth study after the first phase of analysis, the second stage included income. Since income catalogues were regarded as ordinal data, and returns were not following the normal distribution, a non-parametric correlation test was used. Table 9 shows that no correlation was found between income and apparel product return of online shopping (p-value = 0.93 > 0.05). Therefore, Hypothesis 7: “Income is positively related to apparel product return of online shopping” was rejected.
TABLE 9
Income Apparel Product Return
N 137 137
Correlation Coefficient ,007
Sig. （2-tailed） ,932
Correlation between income and return
Fashion Magazine Readership
125 of 142 respondents gave information in the survey about their favorite magazine.
69 of respondents prefer to read fashion magazine and 56 of them prefer to read other kinds of magazine better than fashion magazine. Again, Welch t-test was used since the sample sizes of each group were different. The P-value of Welch t-test was 0.04 (<0.05). The result in Table 10 indicates that respondents who preferred reading fashion magazines had a significantly higher apparel product return average (mean = 1.52) than consumers who didn’t like reading fashion magazines (mean = 0.73). Thus, Hypothesis 9: “Fashion magazine readership is positively related to apparel product return of online shopping” was supported.
TABLE 10
Independent variable N Mean Std. Deviation Sig. (Welch)
Fashion magazine reader 69 1.52 .340 .042
Non-fashion magazine reader 56 .73 .156
Total 125 1.17 .208
Fashion magazine readership and apparel product return
Impulse Buying
There were five questions to measure impulse buying, thus a reliability test should be performed to ensure that the scale of impulse buying was reliable. After conducting the reliability test in SPSS, we got the result that impulse buying scale had high reliability coefficients (Cronbach’s alpha = 0.74). The scores of each item could be summed up to the total impulse buying scores of each respondent. The higher the scores, the more impulse buying tendency the respondent had. Welch t-test was applied to test whether impulse buying was different between females and males.
Hypothesis 1: “Women have higher intention in impulse buying than men” was supported by the result. As Table 11 displays, the p-value of Welch t-test was 0.02, which means females (mean = 3.14) had significant higher impulse buying tendency than males (mean = 2.64).
TABLE 11
Independent Variable N Mean Std. Deviation Sig. (Welch)
Female 93 3.144 1.477 .020
Male 49 2.643 1.080
Total 142 2.971 1.331
Gender and impulse buying
Since age and scores of impulse buying didn't follow the normal distribution (Table 12, p-value = 0.00 < 0.05), Spearman’s rho test was conducted to test the correlation between them. Table 13 and 14 shows that no significant correlation was proved between the age (between 20 and 30 years old, p-value = 0.54 > 0.05) or age (between 23 and 53 years old, p-value = 0.73 > 0.05) of consumers and their apparel product return frequency. Therefore, Hypothesis 4: “Age is negatively related to impulse buying” was rejected.
TABLE 12
Impulse buying
N 142
Mean 2,9708
Std. Deviation 1,33090 Sig. (2-tailed) ,001
Kolmogorov-Smirnov Test for data of impulse buying
TABLE 13
Age23-30 Impulse buying
N 97 97
Correlation Coefficient -,063
Sig. （2-tailed） ,539
Correlation between age (23 – 30) and impulse buying
TABLE 14
Age23-53 Impulse buying
137 110 110
Correlation Coefficient ,033
Sig. （2-tailed） ,733
Correlation between age (23 – 53) and product return
Neither income had significant correlation with impulse buying. As the result Table 15 shows, the p-value of the non-parametric correlation test was 0.54 (> 0.05), which didn't support the Hypothesis 6: “Income is positively related to impulse buying”.
TABLE 15
Income Inpulse buying
N 137 137
Correlation Coefficient -,053
Sig. （2-tailed） ,541
Correlation between income level and product return
The result in Table 16 supports the Hypothesis 8: “Fashion magazine readership is positively related to impulse buying”, fashion magazine readership had a positive relation to apparel product return of online shopping (p-value = 0.05). Consumers who were fashion magazine readers had a significantly higher return frequency (mean
= 3.20) than consumers who were non-fashion magazine readers (mean = 2.74).
TABLE 16
Independent variable N Mean Std. Deviation Sig. (Welch)
Fashion magazine reader 69 3.200 1.339 .048
Non-fashion magazine reader 56 2.737 1.245
Total 125 2.992 1.313
Fashion magazine readership and impulse buying
Hypothesis 3: “Impulse buying is positively related to apparel product return of online shopping” was also strongly supported by the non-parametric correlation test on impulse buying and apparel product return of online shopping. Table 17 shows the result that there was a significant positive correlation between impulse buying and apparel product return (p-value = 0.00 < 0.05, r = 0.25). The higher scores of impulse buying the consumer had, the higher return frequencies he or she had.
TABLE 17
Income Inpulse buying
N 142 142
Correlation Coefficient ,252
Sig. （2-tailed） ,003
Correlation between impulse buying and product return
The relations between gender, age, income, fashion magazine readership and impulse buying were consisted with the relations between them and apparel product return frequency of online shopping. However, the differences of z-scores between the correlations with gender was 1,026 (p-value = 0.05>0.05), while the differences of z-scores between the correlations with fashion magazine readership was 0.023 (p-value = 0.98 > 0.05). These results (in Table 18) indicated that the correlation
between impulse buying and product return is not significantly different in men and women or in fashion magazine readers and non-fashion magazine readers.
TABLE 18
Gender Fashion magazine readership
R N R N
Female ,227 93 Fashion magazine reader ,242 69
Male ,045 49 Non fashion magazine reader ,238 56
Z difference 1,026 Z difference ,023
Sig. (2-tail) ,305 Sig. (2-tail) ,98
The moderating test of fashion magazine readership
TABLE 19
Hypothesis Support/Rejected
H1: Women have higher intention in impulse buying than men. Supported H2: Women have higher intention in apparel product return of online shopping
than men.
Supported
H3: Impulse buying is positively related to apparel product return of online shopping.
Supported
H4: Age is negatively related to impulse buying. Rejected
H5: Age is negatively related to product return of online shopping. Rejected H6: Income is positively related to impulse buying. Rejected H7: Income is positively related to apparel product return of online shopping. Rejected H8: Fashion magazine readership is positively related to impulse buying Supported H9: Fashion magazine readership is positively related to apparel product
return of online shopping.
Supported
Summary of Hypotheses
DISCUSSION
Our primary purpose was to examine which factors of fashion leaders can predict the apparel product return of online shopping. As the impulse buying behavior was proved to be a predictor of frequent returners (Kang & Johnson, 2009), we also investigated that whether those factors of fashion leaders had impact on the impulse buying behavior and how they influenced the correlation between impulse buying and product return.
This study has demonstrated that two main factors of fashion leaders are gender and fashion magazine readership, can influence the product return frequency of apparel products in e-commerce. Similar to the result of product return, gender and fashion magazine readership were proven to have impacts on impulse buying. Even though both gender and fashion had an impact on product return in impulse buying, and the relation between impulse buying and product return were strongly correlated, it is
somewhat surprised that neither gender nor fashion magazine readership had a significant impact on the correlation between impulse buying and product return.
The data from the transaction data of the garment company demonstrated that the majority of purchasers of apparel products were female (75% of the total sample).
This phenomenon consisted with the previous studies, that females predominated the majority of fashion leaders (Summers, 1970; Kwon & Workman, 1996). They purchased more apparel products, but also returned more. Besides, women did more hedonic browsing than men, whilst purchasing on the internet (Phang et al., 2010). In concordance with the previous study that hedonic browsing could lead to a higher possibility of impulse buying (Park et al., 2012), the results of this paper proved that women more frequently made impulse purchases of apparel products than men did (Coley & Burgess, 2003). Fashion leaders who were exposed to fashion magazine often had a high involvement of fashion since fashion magazines contain lots of fashion trends, apparel product information and celebrity news. Those fashion content were often used as inspiration sources of high fashion involvement shoppers (Kinley, Josiam, & Lockett, 2010). The result of our study affirmed that fashion leaders who had higher fashion involvement directly affected impulse buying behavior of apparel products (Park, 2006). In addition, fashion leaders were sensitive to the latest fashion trend, they were more willing to purchase the new apparel products which they saw in the fashion magazines. However, since the new collection couldn’t meet their expectations or looked different from the picture on fashion magazines, fashion leaders would tend to return the products. Interestingly, we found that female respondents in this study showed more interests in fashion magazines, while male respondents preferred other types of magazines (Appendix B). It further explained that why female consumers had more apparel product returns than male consumers.
Nonetheless, there was no evidence to support gender or fashion magazine readership had moderating effects on the coefficient between impulse buying and product return.
Even though female and fashion magazine readers had a positive relation with product return and impulse buying behavior, it didn't mean that male or non-fashion magazine readers never did impulse buying or returned the apparel product. So once people did impulse buying, they would have equal possibilities to return the apparel products.
In contrast to the theory in the previous sections, the other two demographic characteristics of fashion leaders, age and income level, had no significant influence on the product return. In the same way, age and income level scarcely had significance in the explanation of impulse buying either. The previous section claimed that fashion leaders tended to be younger, but had less time for information searching and lacked experience which would in turn lead to product return and impulse buying behavior. However, Hernández et al. (2011) argued that online purchasing behavior did not depend on age but experience. Since young people are a huge group of internet users, even if they had less offline shopping experience than elder consumers, it doesn't mean they have less online shopping experience than elder consumers.
McCloskey (2006) stated that age could influence the initial decision of purchasing on
the internet or not, but it could not influence the subsequent behavior of e-consumers.
We argued that fashion leaders often had high income and they preferred to shop online since they wanted to save time. However, we neglected that the web shops in recent years are offering more attractive alternatives for more price-conscious consumers, so low income consumers also prefer to do the online shopping because they could buy the same product online cheaply compared to offline (Hernández et al., 2011). As those consumers were price-conscious, they might return products they bought during the promotions but they did not really need. Since there were some limitations of this study, we could not assert that age and income were not the predictor of impulse buying and apparel product return of online shopping.
There was a limitation of the survey, which might weaken the result. The age of our respondents of the survey were mainly between 20 to 30 years old (81% of the total respondents), while the age of consumers of the garment company were between 23 to 53 years old. Even though the results of the relation between age and apparel return frequency in the two stages were consistent, they were unconvincing. Besides, in the survey study, the majority of respondents were young people so it was difficult to compare the difference between older and younger people. Similarly, because the respondents of our survey study were mainly students who only had a part-time job or no income at all, 66% of them had the annual income below 10,000 euro, 94% below 30,000 euro, and none of them above 70,000 euro. The comparison between each group was limited. Therefore, we couldn't make a conclusion that age and income level had no correlation with impulse buying or apparel product return.
To summarize, female gender predominance and fashion magazine readership were two characteristics of the fashion leaders. It helped to explain how fashion leaders were more likely to take the risk of impulse buying and to return the apparel products during the online shopping.
CONCLUSION AND FURTHER STUDY
Different from the previous studies, this paper has identified two predictors of apparel product return of online shopping from the features of customers rather than those from logistics. Two characteristics of fashion leaders are: gender and fashion magazine readership have been found to be critical predictors for the apparel product return of online shopping. By identifying the female consumers and the fashion magazine readers, there has been noticed that they tend to return more apparel products and engage in impulse buying during online shopping. E-retailers can come up with strategies to improve the reverse supply chain management for these frequent returners. A quick response to frequent returners can reduce the time of return processes so as to limit the reduction of the value of returned apparel products.
Besides the quick response, the improved return forecast systems can also help to reduce the amount of apparel products that enter in the return flow. For example, there
