Returns, the many-headed monster; An
experimental study regarding the effect of
online product display methods on
product returns.
Sanne Dominique van Rijs (11420677)
Master thesis – Versie 1.0
Master Business Administration –
Digital Business Track
Faculty of Economics and Business
University of Amsterdam
Supervisor: R. v. Kübler
22 June 2018
Statement of Originality
This document is written by Sanne Dominique van Rijs, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of content
Abstract ...6
Introduction ...7
1. Literature review ... 10
1.1. Product return ... 10
1.2. Online product display methods ... 12
1.3. Customer characteristics ... 14 1.3.1. Size... 14 1.3.2 Fit ... 15 2. Methodology ... 18 2.1 Research Design ... 18 2.1.1 Display selection ... 19 2.2 Survey ... 20 2.3 Variables ... 20 2.3.2. Moderated variables ... 21 2.3.3 Dependent variable ... 21 2.3.4 Control variables ... 22
2.4 Population and distribution ... 22
3. Results ... 24 3.1 Sample size ... 24 3.2 Data preparation ... 26 3.3 Hypotheses testing ... 28 4. Discussion ... 37 4.1 Major findings ... 38
4.2 Contributions ... 39 4.2.1 Theoretical implications ... 39 4.2.2 Practical implications ... 42 4.3 Limitations ... 43 4.4 Future research ... 44 5. Conclusion... 46 References ... 47 Appendix I Survey ... 55
List of tables
Table 1 Literature Overview ... 16
Table 2 Research Design ... 19
Table 3 Demographic information of respondents (n=415) ... 25
Table 4 Means, standard deviations, correlations and reliability values ... 27
Table 5 Descriptive Statistics on Product Return ... 31
Table 6 Factorial ANOVA – Tests of Between-Subjects Effects ... 32
Table 7 Sample Descriptives Using t-test for Equality of Means ... 36
Table 8 Sample Descriptives Using t-test for Equality of Means, Fashion Model ... 36
Table 9 Sample Descriptives Using t-test for Equality of Means, No Fashion Model ... 36
Table 10 Summary of Findings ... 38
List of figures
Figure 1 Conceptual Model ... 12Figure 2 Interaction effect, with a Fashion Model... 37
Abstract
The rise of the digital shopping stores creates a convenient shopping experience for customers, but also causes some inseparable challenges, like product return. The high impact of product returns in combination with a digital relevant future creates the need for businesses to adequately manage these influencing factors. Therefore, this academic paper will address the gap in the literature by exploring the relationship of online product display methods on (the amount of) product return. Furthermore, it will show how the moderating customer specific factors could affect this relationship. A deductive and explanatory experimental quantitative research design has been followed to test the hypotheses, by a 2 (product display methods: fashion model vs. no fashion model) x 2 (size: small sizes vs. big sizes) x 2 (physical fit vs. aesthetical fit) full factorial between-subject design and conducted through a survey (N = 415). The findings indicate that there is no significant decrease of product return when websites are personalized based on customer size for physical fit. However, product return is significantly less when websites are personalized based on customer size for aesthetical fit. Furthermore, the results suggest there is no significant difference on product return between customers with a small clothing size and customers with a big clothing size for products displayed by a fashion model, as well as products displayed without a fashion model specified per fit or not.
Key words: online shopping, apparel, product return, product display methods, customer
Introduction
The rise of the digital shopping stores creates a convenient shopping experience for customers, with fast product search, limitless options, quick payment, and specific timed delivery to their home (Fu et al., 2016). Nevertheless, the rising digital era causes some inseparable challenges as well. One of them is product return (Letizia, 2013). This topic is likely to be a familiar phenomenon for online shoppers. For instance, have you ordered products online, which had to be returned? If you have, what was your feeling when your just purchased item turns out not to fit you? Not to mention all the extra effort you have to undertake to return the product. It is probably not your favourite part of the online shopping experience.
Zalando, an international player, has a return rate of 50%. The company, founded in 2008, suffered from big losses until 2013. One of the explanations for their losses is their free return policy, which Zalando retains for a minimal threshold (Rijlaarsdam, 2016). Amazon experienced similar results (Jansen, 2014). Yet, Zalando and Amazon are not the only ones facing this challenge, they also apply to many other online shopping stores. A free return policy is not obligated by Dutch law. Web shops could charge return fees if they informed the customer upfront (Consumentenbond, 2018). This fact, increases the problem even further for web shops with free return policies. Why bother, if you as a customer can buy multiple sizes and therefore reach the free shipping amount faster and send unfitted items back for free as well. This represents a situation of a double win for the customer and a double loss for the web shop. How can this part of the customer journey be more convenient, and less expensive, in the future of online shopping, both for firms as well as online customers?
Ever since the substantial growth of e-commerce, research has been done on factors influencing product return, including web technologies (De, Hu, & Rahman, 2013), customer reviews (Minnema, Bijmolt, Gensler, & Wiesel, 2016; Zhu & Zhang, 2010) and online display
methods (Minnema, 2017; Yoo & Kim, 2014). One possible way to reduce product return is by using different types of display methods customized on an individual level. Why are online display methods essential? Effective online display methods not only attracts customers to a website, but also eases the decision making process for the customers due to the absence of direct product experiences. Product presentation impacts the shopping experience as well as the outcome (Kim & Lennon, 2008; Yoo & Kim, 2012). Basically, effective product display methods will make it easier for customers to visualize the product in reality. It is also important to note that the most effective online display method will not be the same for every type of customer. It is not optimal to simply generalize the most effective display method across different types and segments of customers, as their needs and wants can differ greatly. The moderating customer characteristics could truly be a game changer. The quantifiable effect of customer characteristics will be explored during this research for the first time ever.
The impact of product returns in combination with a digital relevant future creates the need for businesses to adequately manage these influencing factors. Therefore, this academic paper will address the gap in the literature by exploring the relationship of online product display methods on (the amount of) product return. As well as, how the moderating customer specific factors could affect this relationship.
What effect do online product display methods have on business-to-customer product returns within the fashion industry and how is this effect moderated by customer
characteristics?
research could provide insights on a customer specific level. It enables companies, especially with the opportunity of online marketing to personalized webpages, to target their customers on a more personalized level and reduce the possibility of product return. If every single customer would shop on a website, which is optimized for their specific needs related to online product display methods, the total amount of product returns is managed more optimal than before. Although, this paper focuses merely on the marketing perspective, the outcome of the improved product return management has vast impact beyond that of just marketing. For example, it could also lead to lower costs, increased customer engagement and/or increased competitive advantage.
Furthermore, this paper will not only review the findings of the relationship between online product display methods and product return which has previously been done in a limited framework, but it will also provide new insights. I will be the first one to investigate whether or not and how moderating factors of customer characteristics will impact the relationship between online display methods and product return.
All customers are different and therefore the product return issue is facing many different ‘heads’. Decreasing product return thus becomes a many-headed monster which will be tackled in this paper.
The paper will discuss the available literature on these topics in chapter 1, including the conceptual model. Followed by the methodology in chapter 2. The results of the experiment will be discussed in chapter 3, after which the discussion, limitations and future research are outlined in chapter 4.
1. Literature review
The literature review will clarify the constructs and variables presented in this study by reviewing the existing academic literature concerning product return, online product display methods and the customer characteristics: ‘customer size’ and ‘fit’. Followed by the corresponding hypothesis and an overview of the conceptual model.
1.1. Product return
Product return management is a well-mentioned topic in the e-commerce related academic literature. First of all, Bower and Maxham (2012) mentioned that “product returns are a widespread and expensive problem”. Gartner (2014) describes product returns as “the ticking time bomb of multichannel retailing”. Other researchers, like Wilson (2016) stated that high return as well as unwanted e-commerce orders are expensive, which possibly result in negative profitability. Guide et al. (2006) also noticed that product returns could impact profits substantially. Returned products may be late, incomplete or in a bad condition and therefore not re-sellable at full price. In addition, by the time the products are ready for a new buyer, products are often out of season and therefore sold at a discount, thus Financial Times reporter Ram (2016). Iain Prince (2017), supply chain director at KPMG cited the expensive costs related to product return as well: “It can cost double the amount for a product to be returned into the supply chain as it does to deliver it.”
Return policies allow customers to return initially purchased products. These policies divide the purchase decision process into two-stages. During the initial stage, the customer decides to buy the online product and in the next stage the customer decides to keep or to return the product (Anderson, Hansen, & Simester, 2009). Since e-commerce does not allow
product after receiving the product (Fu et al., 2016). In other words, the online purchase decision is based on imperfect information (Shulman, Cunha, & Saint Clair, 2015). After the purchase, full information will be available which allow customers to reconsider their previous decision (Wood, 2001). If the actual product does not meet the expectations formed by the information available during the purchase, the customer will be disappointed and it is more likely that the product will be returned (Bechwati & Siegal, 2005). Overall, the lack of physical contact is the key obstacle of online shopping (Lee & Park, 2014).
Barclaycard’s research (2016) discovered that 30% of the online shoppers intentionally over-purchase and afterwards return unwanted items. Another finding is that 19% of the online shoppers order multiple versions of the same item on purpose. Mr. Prince (2017) mentioned a similar finding. Almost a fifth of the online fashion shoppers over-order on purpose if shoppers know that the returns are free or cheap. Over-purchasing allow shoppers to try, touch or feel the product at home and decide whether the product meets their expectations. Although this behaviour increases the costs of return for the retailers, implementing a paid product return process is not likely. The same research of Barclaycard (2016) found out that approximately half of the online shoppers (47%) wouldn’t order an item if they had to pay the costs for the returns. Nevertheless, Kurt Salmon (2012) pointed out the benefits of free return shipping. It can lead to increased customer loyalty and therefore increase future sales. The same report estimates that at least half of the customers purchase one or more items during the return visit. The fashion industry has the highest percentage (25%) of total returns (KPMG, 2017). Poerink, Abraham, Van Welle & Matton (2016) found similar results, with a percentage of 28%. E-commerce is already a significant purchase channel of fashion. Yet, online fashion sales will explode even further in the future. The Boston Consulting Group predicts a growth of online fashion shoppers from 55-60 million today to 130-135 million by 2020. One-third of the predicted 130-135 million fashion shoppers are likely to buy online (Singhi, Jain, Ramesh,
Sanghi, & Bajaj, 2017). Finally, Forrester (2017) estimates that U.S. shoppers will spend nearly $460 billion online in 2017. The report also predicts that e-commerce will account for 17.0% of retail sales by 2022. As the internet channel continues to grow, so does the many-headed monster: product return.
The conceptual model is presented below in a visualized representation of the three hypotheses, which will be addressed in upcoming paragraphs. The control variables are stated in the upper right corner.
1.2. Online product display methods
Product information influences the online shopping experience and affects the decision to keep or return a purchased item. As a result, retailers offer multiple sources of information to inform customers as accurately as possible in order to avoid a possible mismatch (Minnema, 2017).
Product display methods are one of these information sources. They play a decisive role Product
return Online product
display methods +
The clothing size of the customer
Control variables: (1) age,
(2) education, (3) gender and (4) online experience.
Fit
H1 H2a,
H2b H3a,
H3b
minimize the gap between the physical and non-physical shopping experience. For instance, complementary apparel items (e.g. t-shirt and blazer together on a model) positively influence customer’s online shopping outcomes (Yoo & Kim, 2012). Another research of Yoo and Kim (2014) pointed out the importance of backgrounds. The majority of online retailers use fixed backgrounds (e.g. black or white – one colour) to present the products. Contrary to the behaviour of the majority, Yoo and Kim (2014) found evidence that concrete background pictures relevant to the product help customers to visualize the future consumption of the product. So, the findings of this study imply that online retailers could improve virtual product experience by increasing the reality of the context. Jeong et al (2009) mentioned similar findings. Rich and more complex pictures (e.g. when the product is shown by a model in a relevant context) lead to a higher shopping experience than standard images (e.g. plain views of the product), because complex pictures stimulate various emotional and cognitive experiences. As a result, both visual and verbal information have significant effects on customers’ affective and cognitive attitudes towards apparel products (Kim & Lennon, 2008).
In contrast with abovementioned literature, Minnema (2017) stated that fashion models are far from realistic and therefore not optimal for minimizing the gap between the offline and online shopping experience, including product return. They are good looking, have ideal sizes and possibly get too large clothes pinned for the perfect fit. These idealistic pictures are far from the reality and therefore possibly result in a mismatch between expectations and reality. Letizia (2013) mentioned similar findings, product returns are a result of the abovementioned mismatch. Therefore, if online product display methods (in this case product images) could be adjusted to customer characteristics (e.g. size) on a personalized level, it could minimize the gap between expectations and reality, which in turn should result in a decrease of returns. Therefore, the following is hypothesized.
H1: Personalized product display methods decrease the number of returned products in comparison to impersonalized product display methods.
1.3. Customer characteristics
Every single customer is different and could be characterized by an infinite number of features. The infinite number of features together shape the customer’s behaviour, which can be explained as “the behaviour that customers display in searching for, purchasing, using, evaluating, and disposing of products and services that they expect will satisfy their needs” (Arens & Bovee, 1996). This research explored two possible influencers: ‘size’ and ‘fit’.
1.3.1. Size
Product return is a result of a mismatch between expectations and reality (Letizia, 2013). Minnema (2017) stated that clothes shown by models lead to higher product return, due to the same reason as Letizia (2013) mentioned. However, Minnema did not investigate customer characteristics, like size, as possible moderator on the outcome.
If the product display methods visualize the product closer to the reality of the customer, it could possibly minimize the gap and therefore reduce the likelihood of product return. For instance, if the clothes are shown without a fashion model or with a fashion model with similar characteristics (e.g. clothing size) as the customer it could improve the reality and with that increase the possibility of future consumption. In other words, if a customer looks like the model (e.g. small clothing size) showing the clothes by a fashion model could also decrease product return, whereas customer who do not looks like the model (e.g. big clothing size) showing the clothes without a fashion model could also decrease product return. Therefore, the following is hypothesized:
H2a: The product return rate of clothes displayed by a fashion model are significantly lower for customers with a small clothing size in comparison to customers with a big clothing size.
H2b: The product return rate of clothes displayed without a fashion model are significantly lower for customers with a big clothing size in comparison to customers with a small clothing size.
1.3.2 Fit
Fit refers ‘to be the right size or shape for someone or something’ or ‘to be suitable for someone or something’ (Cambridge Dictionary, 2018). ‘Fit’ has several directions and therefore, this research specifies the term ‘fit’ into two more specific terms: ‘physical fit’ and ‘aesthetical fit’. Physical fit attributes to the fit between the product and the physical conditions (the build of body) of the customer, e.g. fat or thin, tall or short, etc. (Das & Alagirusamy, 2010). Whereas, aesthetical fit attributes to the fit between the product and the social factors of the customer, e.g. the residence, cultural background, gender, occupation, occasion, social status etc. (Das & Alagirusamy, 2010). As well as, the match between the product and the personality of the customer (Aziyatum, Holland, Harrison, & Kazi, 2010).
If the clothing match with the aesthetic aspect of the customer depends on much more factors than the clothing size. It depends on the color, the pattern, the structure, the shape, the fabric etc. As mentioned before, Kim and Lennon (2008) explored that rich and more complex pictures lead to a higher shopping experience than standard images. In other words, the rich and more complex photo with the fashion model will provide more information of the product, which is needed in order to determine the aesthetic fit. Therefore, the following is hypothesized:
H3a: The effect of clothes displayed by a fashion model on product return, as hypothesized in H2a, is stronger for aesthetic fit, than for physical fit.
H3b: The effect of clothes displayed without a fashion model on product return, as hypothesized in H2b, is stronger for physical fit, than for aesthetic fit.
In the past, researchers have been able to identify multiple effects or reasons of product return. A list of the most important or relevant literature, related to this research, is stated below in Table 1 Literature Overview.
Table 1 Literature Overview
Author(s) Effect(s) researched
Anderson, Hansen, & Simester, (2009)
The two stages of the purchase decision process and the effect of on product return.
Bechwati & Siegal (2015)
The effect of a mismatch between expectations and reality, which leads to product return.
Das & Alagirusamy (2012)
The difference between physical fit and aesthetical fit, including the conditions for a match.
Fu et al. (2016) The uncertainty of the inability to touch, feel or try the product before the purchase that leads to product return.
Guide et al. (2006) The effect of product return on finance of companies. Joeng et al. (2009) The effect of rich and complex pictures in comparison to
standard images on shopping experiences.
Kim & Lennon (2008) The effect of visual and verbal information on customers’ affective and cognitive attitudes towards apparel products. Lee & Park (2014) The effect of the lack of physical contact on online shopping.
Letizia (2013) The effect of a mismatch between expectations and reality, which leads to product return.
Minnema (2017) The effect of fashion models, as product display method, on product return.
Shulman, Cunha, & Saint Clair (2015)
The effect of (in)complete information on product return.
Wood (2001) The effect of (in)complete information on the previous purchase decision and thereby product return.
Yoo & Kim (2012) The effect of product display methods, as an information source, on online shopping outcomes.
Yoo & Kim (2014) The effect of fixed and concrete backgrounds on shopping outcomes.
Little research has been done regarding customer characteristics and its effect on product return or online product display methods. Nor has a potential effect on the relationship of these two factors been investigated. So far, no academic research found any significant effects on return likelihood for customer size or fit. This paper will investigate which of the above-mentioned factors are involved regarding product return and will thereby address this gap in existing literature.
2. Methodology
The methodology will outline the research design. Followed by an explanation of the survey and the corresponding distribution. Then, the content of the variables will be illustrated and finally the population of this study will be defined.
2.1 Research Design
A deductive and explanatory experimental quantitative research design has been followed to test the conceptual model, including the corresponding hypotheses (Saunders, Lewis, & Thornhill, 2012). The experiment will be a 2 (product display methods: fashion model vs. no fashion model) x 2 (size: small sizes vs. big sizes) x 2 (physical fit vs. aesthetical fit) full factorial between-subject design and conducted through a survey.
In total, there are eight survey versions, one for each level of the moderator (with/without a fashion model) in combination with one for each level of fit (physical fit/aesthetical fit) specified per gender (male/female). This results in four versions per gender and eight versions in total. Each respondent was randomly exposed within their gender group to one of the four versions (with/without a fashion model in combination with physical fit/aesthetical fit). The versions are mutually exclusive: whereas version one displayed three different types of outerwear showed by a fashion model and version two showed the same types of outerwear, but without a fashion model. These versions are similar across the different types of fit. A visual representation of the research design is displayed in table 1. Furthermore, the three different types of outerwear are shown on individual pages, including the corresponding statements and/or questions. The questions were the same in all the versions of the survey.
Table 2 Research Design
Physical fit Aesthetical fit
Small Sizes (m/f) Big Sizes (m/f) Small Size (m/f) Big Sizes (m/f) Fashion model (V1) Fashion
model and Physical fit Fashion model and Physical fit Fashion model and Aesthetical fit Fashion model and Aesthetical fit
No fashion model (V2) No fashion model and Physical fit No fashion model and Physical fit No fashion model and Aesthetical fit No fashion model and Aesthetical fit 2.1.1 Display selection
When selecting the product display of apparel products, the objective was to minimize the noise of other influencers. Therefore, the online web shops related to the apparel products were left out. Additionally, all the extra information related to the product (e.g. brand, reviews or prize) were deleted as well, because brand could influence the perception of the respondent (Laroche, Kim, & Zhou, 1996) and price or reviews could influence the level of risk (Bhatnagar, Misra, & Rao, 2000). By deleting all those external factors, the products will be at comparable level of risk across the experiments. So, the respondents were only exposed to the images of the products and did not know the sources of the products nor the extra information related to the product.
Thereby, all respondents were exposed to three different but comparable product images per experiment in order to correct the possible model preference and to ensure the construct validity. This is because the statements were similar per product image.
The choice of product images rather than visiting the web shop on their own internet browser was for the same reason, to increase internal validity by ensuring consistent instrumentation. Even though an interactive approach was more realistic it has the risk of allowing other influencers to play a part in the experiment. Not to mention the chance that the website could change the lay-out or content, which may influence the respondents as well.
2.2 Survey
After completing all the versions of the survey, the entire survey was translated from Dutch to English by a peer student of the university and then translated back from English to Dutch by another peer student of the university. Small language differences were discussed by the two peer students and a third independent peer student until all three peer students were pleased. Qualtrics, an online survey platform, was used to make the survey available for respondents. The survey was published in two languages: English and Dutch.
All the questions had to be answered in order to fulfil the response, no exceptions allowed. Furthermore, to avoid participant bias, the true goal of the research ‘the effect of online display methods on product return, possible influenced by customer characteristics’ was not defined in the survey. It was defined as ‘causes of product return’. All the versions of the survey are compressed to one complete overview, which can be found in Appendix I.
2.3 Variables
The independent variable (online display methods), moderator variables (clothing size and fit), dependent variable (product return) and control variables (gender, age, education and online experience) will be explained and operationalized in the following section.
Previous studies did not use or publicly publish adaptable scales, therefore all scales are developed with care per individual variable. Further tests will be conducted to ensure construct validity.
2.3.1 Independent variables
The independent variable ‘online display methods’ is operationalized as a nominal variable with two levels: (1) with a fashion model or (2) without a fashion model. The variable will be tested through A/B construct. So, a respondent will be exposed to either product displays with a fashion model or without a fashion model.
2.3.2. Moderated variables
There are two moderated variables within the experiment: ‘clothing size’ and ‘fit’. The clothing size refers to the clothing size of the customer and is operationalized as an ordinal variable with eight levels: (1) XXS, (2) XS, (3) S, (4) M, (5) L, (6) XL, (7) XXL, (8) XXXL and after the experiment categorized as small (XXS, XS, S and M) and large (L, XL, XXL and XXXL).
The fit refers to match between the product and the customer. The variable fit is operationalized as a nominal variable with two levels: (1) physical fit and (2) aesthetical fit. Physical fit refers to the match between the product and the physical body of the customer, regardless of the taste of the customer whereas aesthetical fit refers to the match between the product and the personality and lifestyle of the customer (e.g. color, design, pattern).
2.3.3 Dependent variable
The dependent variable ‘product return’ is operationalized as an ordinal variable with seven levels: (1) strongly disagree, (2) disagree, (3) somewhat disagree, (4) neither disagree or agree (5) somewhat agree (6) agree and (7) strongly agree.
2.3.4 Control variables
There are four control variables within the experiment: ‘age’, ‘education’, ‘gender’ and ‘online experience’. Age refers to the age of the customer and is operationalized as an ordinal variable with nine levels: (1) younger than 18, (2) 18-24, (3) 25-34, (4) 35-44, (5) 45-54, (6) 55-64, (7) 65-74, (8) 75-84 and (9) 85, or older than 85.
Education refers to the highest level of education of the customer and is operationalized as an ordinal variable (1) primary education, (2) secondary education (4 years: lbo/vmbo), (3) high school (mavo), (4) secondary vocational education (4 years: mbo), (5) secondary education (5 years: havo), (6) secondary education (6 years: vwo), (7) university of applied sciences (bachelor or master) (8) university (bachelor, master or PhD)
Gender refers to the gender of the customer and is operationalized as nominal variable with two levels: (1) female and (2) male.
Online experience refers to the level experience of online shopping of the customer and is operationalized as an ordinal variable with three levels: (1) no experience with online shopping, (2) neutral experience with online shopping or (3) experience with online shopping.
2.4 Population and distribution
The population of this study is every person who could potentially buy clothes on an online platform. The survey of the experiment was distributed in The Netherlands, among mostly Dutch citizens. Initially, the survey was accessible for all ages. However, Dutch citizens below 18 years are legally not obligated to purchase goods or services without permission of their parents or guardians (art. 1:233-234 BW). Afterwards, all respondents younger than 18 years were deleted from the data set.
were approached via online networks (Facebook, email and WhatsApp) and social networks (face-to-face). Several contacts shared the survey as well among their own network. The minimum number of responses needed for the results to be analyzable was calculated using the estimation of Wilson Vanvoorhis and Morgan (2007). This study has eight (8) different cells, with a minimum of 30 respondents per cell, a sample of 240 respondents is necessary.
3. Results
The results will outline findings. At first, an overview of the sample size will be given. Followed by an explanation of the data preparation. Then, the actual hypothesis will be tested, including a description of the corresponding analysis.
3.1 Sample size
In total, 425 respondents completed the survey. 180 respondents only answered part of the survey but did not answer all the questions. These responses were not submitted. A possible explanation could be that some respondents opened the link on one device, for instance their mobile phone, and preferred to answer the survey on another device, like a laptop or tablet. Another possibility is that some respondents opened the survey just to quickly explore the topic and the length of the survey. Last, it is also possible that respondents felt like the picture did not change per page, especially for the experiment with the white t-shirt without a fashion model. Upfront, the respondents did not know that the experiment consist of three different images. These explanations are in line with the completion rate of the unfinished surveys, which is either between 1-10% completion or around 70% completion of the survey. This first completion rate corresponds with the beginning of the survey and the second completion rate corresponds with the end of the second picture or the beginning of the third picture. Overall, there is a completion rate of 70.2%.
Furthermore, 10 completed surveys were eliminated before the data analysis due to the age of the respondents (<18). These cases were excluded from the population of this study. So, this results in 415 valid respondents (n=415). The demographic information of the respondents is presented in Table 3.
Table 3 Demographic information of respondents (n=415)
Variable Item Freq. Percent
Gender Male Female 142 273 43.2 65.8 Age 18-24 25-34 35-44 45-54 55-64 65-74 75-84 85 or older 88 121 64 95 38 8 1 0 21.2 29.2 15.4 22.9 9.2 1.9 0.2 0 Education Primary education
Secondary education (4 years; vmbo) Secondary education (5 years; havo) Secondary education (6 years; vwo) Secondary vocational education (mbo) University of Applied Sciences (hbo) University/PhD (wo) 1 34 46 15 89 141 89 0.2 8.2 11.1 3.6 21.4 34.0 21.4 Size XXS XS S M L XL 1 17 75 140 102 58 0.2 4.1 18.1 33.7 24.6 14.0
XXL XXXL 15 7 3.6 1.7 Online experience
No experience with online shopping Neutral experience with online shopping Great experience with online shopping
46 209 160 11.1 50.4 38.6 3.2 Data preparation
In total, the experiment existed of 425 respondents. However, 10 completed surveys were eliminated before the data analysis due to the age of the respondents (<18). So, this resulted in 415 valid respondents (n=415).
Firstly, there has been checked if the data contained any outliers, errors or missing values. None of these where found. Secondly, twelve counter indicative items were recoded into different variables. Thirdly, to test the reliability of the construct, the internal consistency method is used. The product return scale has high reliability, with Cronbach’s Alpha = .881. The corrected item-total correlations indicate that all the items have a good correlation with the total score of the scale (all above .30). Also, none of the items would substantially affect reliability if they were deleted. Fourthly, the scale means have been computed for a new variable product return. The variable product return existed of twenty-one (21) individual items. Fifthly, Shapiro-Wilk test was used to evaluate the assumptions of normality. All independent variables, (1) product display method, (2) customer clothing size and (3) fit were normally distributed. Finally, a correlation matrix is developed to provide an overview of all the variables used in this study, the overview contains: means, standard deviations, correlations and reliability values. The overview is presented in Table 4.
Table 4 Means, standard deviations, correlations and reliability values Variable M SD 1 2 3 4 5 6 7 8 1. Gender .66 .48 - 2. Online experience 2.27 .65 .14** - 3. Age 3.8 1.38 .09 -.20** - 4. Education 5.26 1.55 .00 .16** -.30** -
5. Product display method 0.51 .50 -.00 .03 -.04 -.01 -
6. Customer clothing size .44 .50 -.19** -.08 .34** -1.8** -.03 -
7. Fit .51 .50 .02 .11* .08 -.01 .01 -.07 -
8. Product return 4.12 0.90 -.19* .08 .04 -.04 .00 -.02 -.03 (.88)
** Correlation is significant at the 0.01 level (2-tailed) * Correlation is significant at the 0.05 level (2-tailed)
3.3 Hypotheses testing
The personalized product display method (N = 215) refers to respondents with small clothing sizes who were exposed to product displays with fashion models, whereas respondents with big clothing sizes were exposed to product displays without fashion models. Impersonalized product display method (N = 200) refers to respondents with small clothing sizes who were exposed to product displays without fashion models, whereas respondents with big clothing sizes were exposed to product displays with fashion models. To test the hypothesis if personalized product display methods, based on size, decrease the amount of returned products in comparison to impersonalized product display methods an independent-samples t-test was conducted.
The results suggest that there was no significant difference in scores for the personalized product display method (M = 4.19, SD = 0.92) and the impersonalized product display method (M = 4.05, SD = .88) conditions; t(413) = -1.58, p = .116, two-tailed. Additionally, the assumptions of homogeneity of variances were tested and satisfied via Levene’s F test, F(413) = .28, p = .864. Thus, product display methods, based on size, personalized or not does not affect the product return in general.
Another independent-samples t-test was conducted to compare product return in personalized product display method and impersonalized product display method conditions specified per fit. The results specified for the physical fit indicate that there was no significant difference in scores for the personalized product display method (M = 4.07, SD = .91, N = 109) and impersonalized product display methods (M = 4.11, SD = .83, N = 95) conditions; t(202) = .35, p= .726, two-tailed. Again, the assumptions of homogeneity of variances were tested and
However, the results specified for the aesthetical fit suggest that there is a significant difference in score for the personalized product display method (M = 4.30, SD = 0.92, N = 106) and impersonalized product display method (M = 3.98, SD = 0.93, N = 105) conditions; t(209) = -2.50, p = 0.013, two-tailed. Last, the assumptions of homogeneity of variances were tested and satisfied via Levene’s F test, F(209) = .08, p = .780.
These results suggest that personalized or impersonalized product display methods do not have an effect on product return in general or specified for the physical fit. However, there is a significant positive effect of personalized product display methods specified for aesthetical fit on product return. Therefore, hypothesis 1 is partly accepted.
H1: Personalized product display methods decrease the number of returned products in comparison to impersonalized product display methods.
A factorial between group analysis of variance (ANOVA) was used to compare the average product return of eight (8) groups of respondents: (a) respondents with small clothing sizes and products displayed by a fashion model, based on physical fit, (b) respondents with small clothing sizes and products displayed by a fashion model, based on aesthetical fit, (c) respondents with small clothing size and products displayed without a fashion model, based on physical fit, (d) respondents with small clothing size and products displayed without a fashion model, based on aesthetical fit, (e) respondents with big clothing sizes and products displayed by a fashion model, based on physical fit, (f) respondents with big clothing sizes and products displayed by a fashion model, based on aesthetical fit, (g) respondents with big clothing size and products displayed without a fashion model, based on physical fit and (h) respondents with big clothing size and products displayed without a fashion model, based on aesthetical fit.
Shapiro-Wilk and Levene’s tests were used to evaluate the assumptions of normality and homogeneity of variance respectively. Neither was violated.
The main effect of product display methods on product return was statistically significant, F(1, 403) = 19.31, p < .01, η² = 0.05. The Pairwise Comparisons tests revealed that the level of product return showed to be significantly lower for product display methods with a fashion model (M = 4.32, SD = 0.81, N = 213) compared to product display methods without a fashion model (M = 3.91, SD = 0.95, N = 202) (p < .01). Furthermore, there was no significant effect found for fit, F(1,403) = 0.03, p = 0.86, η² = 0.00 or size, F(1,403) = 1.5, p = 0.22, η² = 0.00. Last, there was a marginally significant interaction effect between product display methods, fit and size on product return F(1, 403) = 3.21, p = .07, η² = .01. No significant effect was found between independent variables.
The control variables (gender, age, education and online experience) were tested as well. The control variables gender, F(1, 403) = 8.58, p = .01, η² = .02 and online experience F(1, 403) = 5.61, p = .02, η² = .01 were significant. The control variable age, F(1, 403) = 3.19, p = .08, η² = .01 was marginally significant and the control variable education (p = .48) was not significant.
The output of the factorial ANOVA is presented in Table 5 (Descriptive Statistics on Product Return), Table 6 (Factorial ANOVA – Tests of Between-Subjects Effects), Figure 2 (Interaction effect, without a Fashion Model) and Figure 3 (Interaction, with a Fashion Model).
Table 5 Descriptive Statistics on Product Return
Product display Size Fit N M SD
No fashion model Small size Physical fit 49 3.87 0.80 Aesthetical fit 61 3.88 1.00
Total 110 3.86 0.91
Big size Physical fit 51 3.83 1.00
Aesthetical fit 41 4.10 1.00
Total 92 3.95 1.00
Total Physical fit 100 3.85 0.91
Aesthetical fit 102 3.97 1.00
Total 202 3.91 0.95
Fashion model Small size Physical fit 58 4.28 0.76 Aesthetical fit 65 4.43 0.85
Total 123 4.36 0.81
Big size Physical fit 46 4.38 0.78
Aesthetical fit 44 4.12 0.83
Total 90 4.25 0.81
Total Physical fit 104 4.32 0.77
Aesthetical fit 109 4.31 0.85
Total 213 4.32 0.81
Total Small size Physical fit 107 4.09 0.80
Aesthetical fit 126 4.17 0.96
Total 233 4.13 0.89
Big size Physical fit 97 4.09 0.94
Aesthetical fit 85 4.11 0.90
Total 182 4.10 0.92
Total Physical fit 204 4.09 0.87
Aesthetical fit 211 4.14 0.94
Total 415 4.12 0.90
Note. N = Number of cases. M = Mean. SD = Standard Deviation. Product Return ranges from 1 (strongly disagree) to 7 (strongly agree), whereas 1 refers to high likelihood of product return and 7 refers to low likelihood of product return.
Table 6 Factorial ANOVA – Tests of Between-Subjects Effects
Dependent variable: Product Return
df SS MS F η² Sig. Gender 1 6.52 6.52 8.58 .02 .01 Online experience 1 4.27 4.27 5.61 .02 .02 Age 1 2.42 2.42 3.19 .01 .08 Education 1 0.39 .39 .51 .01 .48 Product display method 1 14.67 14.67 19.31 .05 .01 Customer size 1 1.14 1.14 1.50 .01 .22 Fit 1 0.03 .03 .03 .01 .86 Product display method * Customer size 1 0.73 .73 .96 .01 .33 Product display method * Fit 1 1.03 1.03 1.36 .01 .25
Customer size * Fit 1 0.12 .12 .15 .01 .70
Product display method * Customer Size * Fit 1 2.44 2,44 3.21 .01 .08 Error 403 306.21 .76 Total 415 338.34
a. R Squared = ,095 (Adjusted R Squared = ,070) b. Computed using alpha = ,05
The effect of products displayed by a fashion model on product return could be moderated by the customer clothing size. To test the hypothesis if the product return rate of clothes displayed by a fashion model are significantly lower for customer with a small clothing size in comparison to customer with a big clothing size an independent-samples t-test was conducted.
conditions for products displayed by a fashion model; t(211) = .95, p = .343, two-tailed. Additionally, the assumptions of homogeneity of variances were tested and satisfied via Levene’s F test, F(211) = .11, p = .743. Thus, there was no significant difference between the product return of a customer with a small clothing size and customer with a big clothing size for products displayed by a fashion model. Therefore, hypothesis 2a is rejected.
H2a: The product return rate of clothes displayed by a fashion model are significantly lower for customers with a small clothing size in comparison to customers with a big clothing size.
Another independent-samples t-test was conducted to compare product return between customers with a small clothing size and customers with a big clothing size for product displayed without a fashion model. The results indicate that there was no significant difference in scores for small clothing sizes (M = 3.86, SD = 0.91, N = 110) and big clothing sizes (M = 3.95, SD = 1.00, N = 92) conditions for products displayed without a fashion model; t(200) = -.56, p = .575, two-tailed. Again, the assumptions of homogeneity of variances were tested and satisfied via Levene’s F test, F(200) = .92, p = .338. Thus, there was no significant difference between the product return of a customer with a small clothing size and customer with a big clothing size for products displayed without fashion model. Therefore, hypothesis 2b is rejected.
H2b: The product return rate of clothes displayed without a fashion model are significantly lower for customers with a big clothing size in comparison to customers with a small clothing size.
Both findings of hypothesis 2a and hypothesis 2b are in line with the results of the factorial ANOVA. The effect of products displayed by a fashion model or without a fashion
model on product return is not significantly moderated by the customer clothing size. The output of the independent-samples t-test is presented in Table 7.
To test the hypothesis if the effect of aesthetical fit on product return is stronger than the effect of physical fit on product return for product display methods with a fashion model (specified per clothing size group) an independent-samples t-test was conducted. The results suggest that there was no significant difference in scores for the physical fit (M = 4.28, SD = 0.76, N = 58) and the aesthetical fit (M = 4.43, SD = .88, N = 65) conditions for the small clothing size group; t(121) = -1.03, p = .307, two-tailed. Additionally, the assumptions of homogeneity of variances were tested and satisfied via Levene’s F test, F(121) = 1.03, p = .312.
Thereby, the results of the difference in score for the physical fit (M = 4.38, SD = 0.78, N = 46) and aesthetical fit (M = 4.12, SD = 0.83, N = 44) conditions for the big clothing size group were not significant either; t(88) = 1.50, p = .137, two-tailed. Again, the assumptions of homogeneity of variances were tested and satisfied via Levene’s F test, F(88) = 0.01, p = .981. The output of the independent-samples t-test is presented in Table 8. Thus, there is no significant difference between physical fit and aesthetical fit on product return for small clothing sizes and big clothing sizes for products displayed by a fashion model. Therefore, hypothesis 3a is rejected.
H3a: The effect of clothes displayed by a fashion model on product return, as hypothesized in H2a, is stronger for aesthetic fit, than for physical fit.
revealed that there was no significant difference in scores for the physical fit (M = 3.87, SD = 0.80, N = 49) and the aesthetical fit (M = 3.88, SD = 1.00, N = 61) conditions for the small clothing size group; t(108) = -0.97, p = .923, two-tailed. Additionally, the assumptions of homogeneity of variances were tested and satisfied via Levene’s F test, F(108) = 1.72, p = .193.
Furthermore, the results of the difference in score for the physical fit (M = 3.83, SD = 1.01, N = 51) and aesthetical fit (M = 4.10, SD = 1.00, N = 41) conditions for the big clothing size group were not significant either; t(90) = -1.27, p = .207, two-tailed. Again, the assumptions of homogeneity of variances were tested and satisfied via Levene’s F test, F(90) = 0.02, p = .884. The output of the independent-samples t-test is presented in Table 9. Thus, there is no significant difference between physical fit and aesthetical fit on product return for small clothing sizes and big clothing sizes for products displayed without fashion model. Therefore, hypothesis 3b is rejected.
H3b: The effect of clothes displayed without a fashion model on product return, as hypothesized in H2b, is stronger for physical fit, than for aesthetical fit.
Hypothesis 3a and hypothesis 3b were both not significant. The factorial ANOVA suggested a marginal significant interaction effect between product display method (fashion model or no fashion model), customer clothing size (small clothing size or big clothing size) and fit (physical fit and aesthetical fit). However, both independent samples t-tests were not significant. So, there is no significant difference, per clothing size group, on product return between physical fit and aesthetical fit. The output of the independent-samples t-test are presented in Table 8 and 9 and the interaction graphs of the factorial ANOVA are presented in figure 2 and 3.
Table 7 Sample Descriptives Using t-test for Equality of Means
Small clothing sizes Big clothing sizes
M SD M SD t-test
Fashion Model 4.36 0.81 4.25 0.81 ns
No Fashion Model 3.86 0.91 3.95 1.00 ns
Note. M = Mean. SD = Standard Deviation.
Table 8 Sample Descriptives Using t-test for Equality of Means, Fashion Model
Physical fit Aesthetical fit
M SD M SD t-test
Small clothing size 4.28 0.76 4.43 0.88 ns
Big clothing size 4.38 0.78 4.12 0.83 ns
Note. M = Mean. SD = Standard Deviation.
Table 9 Sample Descriptives Using t-test for Equality of Means, No Fashion Model
Physical fit Aesthetical fit
M SD M SD t-test
Small clothing size 3.87 0.80 3.88 1.00 ns
Big clothing size 3.83 1.01 4.10 1.00 ns
Figure 3 Interaction effect, without a Fashion Model Figure 2 Interaction effect, with a Fashion Model
4. Discussion
The discussion elaborates on the results. At first, an overview of the major findings will be given. Followed by the contribution of the findings of this research. Then, the limitations of the study will be discussed, continued by suggestions for further research.
4.1 Major findings
The purpose of this research was to explore the gap in the literature between the relationship of online product display methods and (the amount of) product return. As well as, how the moderating customer specific factors could affect this relationship. This study contributes to related literature through one major finding.
The results suggest personalized product display methods only reduce product return significantly for aesthetic relevant apparel. There was no significant effect for physical relevant apparel. Furthermore, the findings indicate that the customer characteristic size and fit do not significantly affect the relationship between product display methods and product return. An overview of the corresponding hypothesis, including the outcome, is stated in Table 10.
Table 10 Summary of Findings
Hypothesis Description Significant effect (p < .05)
H1 Personalized product display methods decrease the number of returned products in comparison to impersonalized product display methods.
Partially, only for aesthetical fit
H2a The product return rate of clothes displayed by a fashion model are significantly lower for customers with a small clothing size in comparison to customers with a big clothing size.
with a big clothing size in comparison to customers with a small clothing size.
H3a H3a: The effect of clothes displayed by a fashion model on product return, as hypothesized in H2a, is stronger for aesthetic fit, than for physical fit.
No
H3b The effect of clothes displayed without a fashion model on product return, as hypothesized in H2b, is stronger for physical fit, than for aesthetical fit.
No
4.2 Contributions
This study addresses a relatively unknown academic topic of the relationship between online product display methods and (the amount of) product return and how this relationship is influenced by customer characteristics (size and fit). So the results of this study, significant or not, offer valuable insights for both theoretical implications and practical implications.
4.2.1 Theoretical implications
The findings from this study confirm, challenge and add to the existing literature on product display methods and product return in online shopping environments.
Firstly, the results showed that personalized or impersonalized product display methods do not have an effect on product return in general or specified for the physical fit. However, there was a significant positive effect of personalized product display methods specified for aesthetical fit on product return. Jeong et al (2009) mentioned that rich and more complex pictures lead to a higher shopping experience than standard images. In this experiment, a personalized product display match of customers with a bigger clothing size refers to product display methods without a fashion model, which is far from rich and complex pictures. Furthermore, Ghafffari (2011) found out that experienced customers could estimate the physical fit of the product better when the clothes are presented on multiple models of various
body types, whereas less experienced customers could indicate their physical fit better when clothes are presented without a fashion model. So, these findings could explain the first outcome.
In line with the additional outcome, Minnema (2017) stated that fashion models are not realistic at all and are therefore not ideal for minimizing the gap between the offline and online shopping experience. Letizia (2013) mentioned similar findings, product returns are a result of a mismatch between expectations and reality. Therefore, if online product display pictures could be adjusted to customer characteristics (e.g. size) on individual level it could minimize the gap between expectations and reality, which in turn should result in a decrease of returns.
Other studies have defined two dimensions of apparel fit as well: (1) physical fit and (2) aesthetical fit. Whereas aesthetical fit refers to the appearance of the garment in relation to the taste and social factors of the customer, which is comparable to our condition. However, functional fit or physical fit refers to the physical specifications of the customer’s body, but also the comfort and performance of the garment due to the fit (Eckman et al., 1990; Outling, 2007). This online experiment does not address the offline part of comfort and performance, which possibly explains the non-significant outcome of personalized display methods for the physical fit. A product display picture simply does not give any (or minimal) information regarding the comfort or performance.
Secondly, the findings suggest that the effect of products displayed by a fashion model or without a fashion model on product return is not significantly moderated by the customer’s clothing size. Furthermore, the product return rate was significantly lower for apparel displayed with a fashion model than apparel displayed without a fashion model.
Rosa et al. (2006) mentioned that apparel is a high body involving product and therefore the consumption is related to fit. Whereas, Fallon (1990) stated that the body image is a person’s mental picture of their bodies. Other researchers found a relation between the customer’s body image and the consumption of apparel (Cash and Cash, 1982; Solomon and Douglas, 1985). It is possible that the product display method is not related to the actual customer fit, but to the customer’s mental picture of their body. The respondents are exposed to several conditions based on their actual size, instead of their own body image. This insight could explain the non-significant results.
In contrast with the above-mentioned finding, Minnema (2017) stated that apparel items displayed by fashion models lead to significant higher product returns. An entire team of make-up artists, hairstylists and photographers are working hard to make sure everything looks perfect. As a result, these fashion models are good looking, have ideal sizes and possibly get too large clothes pinned for the perfect fit. In other words these photos do not reflect the reality and do certainly not reflect the average people. Kim and Damhorst (2014) recommend, based on their findings, that it may help customers to imagine the fit of the garment when retailers have models with different body sizes with the same outfit. In addition, extra information related to the model about the length or size they are wearing could also improve the imagination. They suggest advanced personalized model (e.g.) avatars, which may also help to minimize the gap between online and offline. So, maybe a single static picture is not sufficient enough for customers to decide whether the product will fit them or not.
Finally, the data indicates that there is no significant difference, per clothing size group, on product return between physical fit and aesthetical fit.
The physical fit is has two aspects. On the one side, the match depends on the physical built of the customer’s body (e.g. fat or thin, tall or short), but can also be more specific like firm calves or big breasts and therefore require special specific physical requirements for clothes because bodies vary widely from person to person. On the other side, the physical fit requirements depend also on the given environmental conditions (e.g. the weather: hot or cold and the activity: the gym or work) (Das & Alagirusamy, 2010). Unfortunately, this experiment measured the effect on a compound variable ‘physical fit’, which include all those lower variables, and did not take the environmental conditions into account. These could both have an impact on the non-significant outcome.
Das & Alagirusamy (2010) mentioned that the aesthetical fit is related to the social factors of the customer. However, they mentioned a lot of different social factors, like the residence, cultural background, gender, occupation, occasion and social status. Other researchers mentioned that an aesthetical match is related to the match between the product and the personality of the customer (Aziyatum, Holland, Harrison, & Kazi, 2010). All of those individual variables could influence the effect on a lower level. Again, this experiment measured the effect on a compound variable ‘aesthetical fit’, which includes all those lower variables. This could affect the non-significant outcome.
4.2.2 Practical implications
As mentioned before, the rise of the digital shopping stores creates a convenient shopping experience for customers, but also causes some inseparable challenges for companies, like product return. The results of this research enable companies to rethink about their current online shopping environment, with a particular emphasis on their website design.
is significantly less when websites are personalized based on customer size for aesthetical fit. So, companies with a majority of aesthetical relevant items could benefit from this insight by adjusting their product displays based on their customer size. They could also implement an option where customers could change the product display (with or without a fashion model) themselves.
Secondly, the findings suggest that there is no significant difference on product return between customers with a small clothing size and customers with a big clothing size for products displayed by a fashion model, as well as products displayed without a fashion model. So, product displays methods are not significantly influenced by the size of the customer. However, the product return for products displayed without a fashion model is significantly higher than products displayed with a fashion model. In other words, regardless of the customer’s size this research suggests that companies should display their products with a fashion model if they want to minimize their product return.
Finally, the results suggest that there is no significant difference on product return between customers with a small clothing size and customers with a big clothing size for products displayed by a fashion model, as well as products displayed without a fashion model specified per fit. So, product display methods are not significantly influenced by different kinds of fits. In addition to above-mentioned implication, regardless of the customer size and the different types of fit this research suggests that companies should display their products by a fashion model if they want to minimize their product return.
4.3 Limitations
Although this study has explored some untouched research areas, like most other empirical studies, it has some downsides: the limitations.
Firstly, the research model does not consider all possible factors influencing product return. The focus of the research is specified on the customer characteristic size and the two dimensions of fit on product return. Thus, the scope of this research is more restricted compared to the actual online shopping world. Even though it is not necessarily a bad thing, it should be taken into account considering the results.
Secondly, the data was collected by an online experiment with a corresponding online survey. Questions like ‘Based on the product display picture ... I am able to judge whether the product will suit me’ were asked hypothetically. So, whether the product suits the customer in real-life, as they expected or not, is not clear. The offline part was out of scope of the experiment, due to limited time and budget. However, this could possibly impact the overall outcome: product return.
Finally, the environment of the survey was fixed. Respondents did not browse to a certain website or select a certain item. This implies the possibility that respondents may have to answer statements about apparel they do not like or there may have been a lack of the ‘uncertainty’ feeling since they know their choices are purely hypothetical. There are no extra costs or effort related to possible ‘wrong’ decisions. It was a total set-up and therefore far from reality, which also may or may not have influenced implications based on the results.
4.4 Future research
Every research paper aims to find answers for related questions. Whether the study can find those answers or not, new questions will arise for sure. Based on this study future research could explore three new questions.
bigger clothing sizes as product display methods without a fashion model. What if the personalized display methods for customers with a bigger clothing size are matched with fashion models only with a bigger size? Future research could explore whether the effect on product return will significantly increase or decrease due to this change.
Secondly, this research addresses whether the customer characteristic size has affected the relationship of product display methods on product return. Other customer characteristics, like gender or age, could affect that relationship as well. Perhaps they also influence the optimal product display method. For instance, a man with a bigger clothing size could be less sensitive, in terms of product return, for products displayed by a thin model than a woman with a bigger clothing size. Future research could explore these other customer characteristics. It is also possible that customer size could influence other factors instead of product display methods. Perhaps tall customers value product information, like the model length and corresponding size more than regular length customers or maybe customers with a bigger clothing size attaches more value to reviews than customers with regular sizes. So, future research could explore other website factors that could be influenced by the customer characteristic size.
Finally, the entire experiment was fixed and therefore far from the real shopping experience. Thereby, the experiment was hypothetical as well. Future research could repeat the complete experiment, but only in a real-life situation, including actual products, payments and returns in order to check whether the outcomes are similar.
5. Conclusion
Due to the rise of internet the world of retailing is changing and e-commerce is booming. However, the successful e-commerce has also several unwished side effects, like high product return. Product return is a many-headed monster due to the enormous amount of possible influencing variables, which are not all explored by the existing academic literature yet. This research paper addresses an academically unknown area by answering the following research question:
What effect do online product display methods have on business-to-customer product returns within the fashion industry and how is this effect moderated by customer
characteristics?
Overall, product display methods do have a significant effect on business-to-customer product return within the fashion industry. Apparel displayed by a fashion model have a significantly lower product return likelihood than apparel displayed without a fashion model.
However, the data suggest that there was no significant moderated effect of customer characteristic. So, neither the customer size nor the product fit does affect the relationship of product display methods on business-to-customer product return within the industry.
References
Anderson, E. T., Hansen, K., & Simester, D. (2009). The Option Value of Returns: Theory and Empirical Evidence. Marketing Science, 28(3), 405–423.
https://doi.org/10.1287/mksc.1080.0430
Arens, W. F., & Bovee, C. L. (1996). Contemporary advertising (Second). Boston: Irwin Professional Publishing. https://doi.org/10.1017/CBO9781107415324.004
Aziyatum, B., Holland, R., Harrison, D., & Kazi, T. (2010). The future design directions of smart clothing development. Journal of Textile Institute, 96(4), 199–210.
https://doi.org/10.1533/joti.2004.0071
Barclaycard. (2016). Emergence of “serial returners.” Retrieved January 25, 2018, from https://www.home.barclaycard/media-centre/press-releases/emergence-of-serial-returners-hinders-growth-of-UK-businesses.html
Bechwati, N. N., & Siegal, W. S. (2005). The Impact of the Prechoice Process on Product Returns. Journal of Marketing Research, 42(3), 358–367.
https://doi.org/10.1509/jmkr.2005.42.3.358
Bhatnagar, A., Misra, S., & Rao, H. (2000). On Risk, Convenience, and Internet Shopping Behavior. Communications of the ACM, 43(11), 98–105. Retrieved from
http://delivery.acm.org.proxy.uba.uva.nl:2048/10.1145/360000/353371/p98-bhatnagar.pdf?ip=146.50.98.29&id=353371&acc=ACTIVE
SERVICE&key=0C390721DC3021FF.86041C471C98F6DA.4D4702B0C3E38B35.4D4 702B0C3E38B35&__acm__=1525177588_4efd1c94b3e8191c5fe1787c0
