The impact of gender, age, price and cognitive dissonance on
the e-commerce apparel product returns
Master Thesis, MSc Supply Chain Management
University of Groningen, Faculty of Economics and Business
June 22, 2015
Lingxi Huang Student number: s2659883 E-mail: L.huang.9@student.rug.nl
Supervisor / University Drs. J.C. (Jerry) van Leeuwen
ABSTRACT
To effectively manage product returns, it is important to understand the reason and heterogeneity of product returns behaviors. The thesis investigates the heterogeneity of product return behaviors in relation to three most observable variables for the online retailers, including age, gender and price, and furthermore seeks to explain the heterogeneity with cognitive dissonance, a cause of product return that is gaining increasing attention. Statistics tests were applied to analyze the data obtained from an e-commerce apparel company and a survey. Results showed that gender and price were positively related to product returns, while age was not; There was a significant positive relationship between cognitive dissonance and product returns, but only among female customers. Female customers were more likely to suffer from product dissonance. Neither age nor price was found to be related to cognitive dissonance. The findings provide some insights for understanding of return behaviors of e-commerce apparel customers.
Table of Content
ABSTRACT ... 2
1. INTRODUCTION ... 4
2. THEORETICAL BACKGROUND ... 6
2.1 E-Commerce Reverse Logistics ... 7
2.2 Age, Gender, Price and Product Returns ... 8
2.3 Cognitive Dissonance ... 9
2.3.1 Cognitive dissonance ... 10
2.3.2 Gender and cognitive dissonance ... 12
2.3.3 Age and cognitive dissonance ... 13
2.3.4 Price and cognitive dissonance ... 14
3. METHODOLOGY... 15
3.1 Research Design... 15
3.2 Data Collection ... 16
3.3 Measures ... 17
3.3.1 Instrument ... 17
3.3.2 Measure of cognitive dissonance ... 17
3.3.3 Measures of product returns ... 18
3.4 Data Analysis ... 19
4. RESULT ... 20
4.1 Sample Characteristics ... 20
4.2 Scale Reliability and Validity ... 21
4.3 Hypotheses Testing ... 22
4.3.1 Hypothesis 1: gender and product returns ... 22
4.3.2 Hypothesis 2: age and product returns ... 23
4.3.3 Hypothesis 3: price and product returns ... 23
4.3.4 Hypothesis 4: cognitive dissonance and product returns... 24
4.3.5 Hypothesis 5: gender, cognitive dissonance and product returns ... 25
4.3.6 Hypothesis 6: age, cognitive dissonance and product returns ... 26
4.3.7 Hypothesis 7: price, cognitive dissonance and product returns ... 27
5. DISCUSSION ... 27
6. CONCLUSION ... 32
REFERENCE ... 36
1. INTRODUCTION
For e-commerce supply chain management, one important dimension is the reverse logistics driven by consumer product returns. It is reported that some European online retailers of fashion industry have product return rates of 40% or higher (Accenture, 2012). Meanwhile it is very costly for online retailers to handle product return, for example, between $6 and $18 per returned item (The Economist, 2013). However, the main reason for product return is usually not product defects (Accenture, 2008). To develop proper strategies to control product return, it would be a relevant starting point for the online retailers to understand the product return behavior of their consumers.
Existing literature about reverse logistics optimization mainly focus on managing commercial return between retailers and distributors or end-of use product returns (e.g. Bernon, Rossi, & Cullen, 2011; Srivastava, 2007). The role of consumers was seldom addressed in such literature. However, in the e-commerce context, managing product return from the consumers perspective could be important. With the phenomenon of high product return rate and many non-defective products being returned, it gives rise to the question that whether there is a chance to reduce product return from consumers. Some attempts of consumer return management in the e-commerce were altering return policy or charging restocking fee (Shulman, Coughlan, & Savaskan, 2011). However it was found that more lenient return policy increase not only product return rates but also purchase rates in remote purchase environment (Wood, 2001). Moreover, different reasons and motivations for product returns were found to be related to customers’ return frequency (Foscht, Ernstreiter, Maloles, Sinha, & Swoboda, 2013). Thus it is proposed here to first understand the heterogeneity of product return behaviors, and therefore become more effective to influence the occurrence of product returns.
dissatisfaction of product performance, which was considered to be a main reason for product returns (Engel, Blackwell, & Miniard, 1995), there are many other reasons that could contribute to product returns, such as retail borrowing (Hess & Mayhew, 1997), buyer’s remorse (Fried, 2008). In an empirical study of Lee (2015), verbal return reasons were collected from customers, and many reasons reflects cognitive dissonance (Lee, 2015), which is the inconsistent relationship between one’s cognitions (Festinger, 1957). Other recent studies also found cognitive dissonance related to product returns (Fried, 2008; Powers & Jack, 2013).
The thesis was conducted with the help of a Dutch apparel e-commerce company, which is suffering from a product return rate of 27% approximately, and the handling cost for product returns is around 11 Euro per product. The high product return rate and high costs are not uncommon in the industry, thus necessitate the research to investigate the reasons and to seek for effective strategies to influence consumers’ product return behaviors. As stated above, the thesis will start from investigating the relationship between cognitive dissonance and product returns. The perspective is chosen because many studies have shown a great variance in return behavior across customers (Hess & Mayhew, 1997; Petersen & Kumar, 2009), while there are still limited literatures explaining such heterogeneity. Moreover, in the early phase of the thesis, it was found that product return behaviors show significant differences among different age, gender and price levels. Therefore whether such differences are related to cognitive dissonance will also be investigated. Hence, the research question is:
“What is the relationship between gender, age, price, cognitive dissonance and product returns in the e-commerce apparel context?“
provides a starting point to look into consumers’ product return behavior and to differentiate the strategies for product returners with different characteristics. Besides, it adds to the limited literature about consumer return management, which is a relevant field in e-commerce reverse logistics. It also has managerial implication since cognitive dissonance has been increasingly traced among the reasons for product returns, while online retailers are trying to reduce the costly and inefficient product returns in their reverse logistics. By understanding the impact of cognitive dissonance on consumer’s regret and product return behavior, online retailers could apply responsive strategies to influence the product returns. Moreover, the thesis also considered the variables of age, gender and price, which are very common data that online retailers would collect from their consumers, and therefore the results are more widely applicable and valuable for the managerial practice.
The remaining parts of the thesis will be organized as follow: The second part will be the theoretical background of product return and its relation with cognitive dissonance, age, gender and price. The third part will explain the methodology in this study, followed by the results, discussion and conclusions in the fourth, fifth and sixth sections respectively.
2. THEORETICAL BACKGROUND
FIGURE 1 - Conceptual model
2.1 E-Commerce Reverse Logistics
Product return in the reverse logistics can be divided into different types, including marketing returns, environmental returns, and consumer returns (Rogers, Rogers, & Lembke, 2010). Marketing returns (also called commercial return) are “product sent back from a position forward in the supply chain”, and environmental returns are “the disposal of hazardous materials or compliance to environmental regulation” (Rogers et al., 2010). Currently there are many literatures about product return optimization focus on managing the two types of product returns above (e.g. Bernon et al., 2011; Srivastava, 2007), while less attention has been paid to the consumer returns, which are the product returns from end consumers to the retailers. Meanwhile the optimization methods, such as quantitative models, contract management, network design, are mainly optimizing how to handle product return rather than how to reduce return rates. However, due to the high product return rate and expensive return handling costs, the perspective of reducing consumers’ product returns could be very important for the e-commerce reverse logistics.
Till now limited literature is found to investigate how to reduce consumer returns. Some attempts of return management in the e-commerce are applying a more strict return policy or charging restocking fee (Lee & Lund, 2003; Shulman et al., 2011).
However, Venkatesan and Kumar (2004) found that the influence of returns on customer purchase behavior is nonlinear, and a more lenient return policy might contribute more sales (Petersen & Kumar, 2009). Griffis, Rao, Goldsby and Niranjan (2012) also found in their research that return-experienced customers had higher average number of items per order and higher average item value in their post-return orders than in their pre-return orders. Therefore simply applying a more strict product return policy is not a desirable way to reduce consumer product returns. To become more effective in reducing product returns, it is useful for retailers to first understand the heterogeneity and causes of product return behaviors of their customers (Foscht et al., 2013).
2.2 Age, Gender, Price and Product Returns
including age, gender, education, and family income (Lightner, 2003; Zhou, Dai, & Zhang, 2007). Thus, it is possible that age and gender are also significant predictors of e-commerce apparel product returns.
Hypothesis 1. Gender is a significant predictor of e-commerce apparel product returns.
Hypothesis 2. Age is a significant predictor of e-commerce apparel product returns.
In addition, it is not surprising that people think expensive products are more likely to be returned. It was argued by Hart (1988) that “the consequences of failure for more expensive items are greater than the consequence for less expensive ones” (Hart, 1988; Griffis et al., 2012). De, Hu and Rahman (2013) investigated the product return rates in various product categories, and found that products with a higher list price are more likely to be returned (De et al., 2013). The possible reason given by De et al. (2013) is that customers who are not satisfied a low-priced item might not return it while customers buying an expensive item might return it as soon as the post-purchase dissatisfaction reaches a certain level. However, the argument was not further supported with empirical examination.
Hypothesis 3. Price is a significant predictor of e-commerce apparel product returns.
2.3 Cognitive Dissonance
were collected from customers, and many of the reasons can be summarized as “remorse”, “careless/hurried purchase”, “acquisition of additional information after purchase,” and “change of mind after brief use of the product” (Lee, 2015). It was argued that those consumers might experience buyer’s remorse, which is “an intense negative emotion consumers experience when they regret their purchase” (Lee, 2015). It was proposed that such remorse might be reflective of cognitive dissonance (Lee, 2015; Fried, 2008).
2.3.1 Cognitive dissonance Cognitive dissonance refers to the relations that
exist between pairs of (an individual’s) cognition elements which are inconsistent with each other (Festinger, 1957: 9). The cognition elements indicated here could be one’s knowledge, beliefs, attitudes, behaviors (Festinger, 1957). Festinger (1957) argued that as long as two cognition elements are relevant to each other, the relationship between them is either consonant or dissonant. Hence, there are many situations where cognitive dissonance occurs, such as when one’s behavior is inconsistent with his or her opinions of this behavior, or when one is exposed to new information which is inconsistent with his or her existing knowledge. It was addressed that the existence of cognitive dissonance gives pressures to reduce the dissonance, by “behavior changes, changes of cognition and circumspect exposure to new information and new opinions” (Festinger, 1957: 31). It is also worth noticing that cognitive dissonance has an emotional dimension, defined as “a person’s psychological discomfort subsequent to the purchase decision” (Soutar & Sweeney, 2003; Festinger, 1957)
could even happen during the decision making process, before receiving the products (Koller & Salzberger, 2007; Lee, 2015). As Festinger (1957) discussed specifically in his theory, cognitive dissonance could happen as a consequence of decisions between alternatives, as the cognition elements supporting the unchosen alternatives would be dissonant with the action once the decision is made. Since product purchase can be regarded as a decision of choosing multiple alternatives or a decision of choosing to buy or not, it gives rise to cognitive dissonance. As a result of pressures to reduce the dissonance, customers suffering from cognitive dissonance might return the products to “undo the effects of a regretted choice” (Powers & Jack, 2013)
Some researches provided empirical evidences that cognitive dissonance is in relation with product return (Lee, 2015; Maity, 2015; Milliman & Decker, 1990; Powers & Jack, 2013). Lee (2015) found that customers increasingly use product return as a strategy to reduce their post-purchase dissonance; Maity (2015) found that cognitive dissonance after purchase positively influence the return intention, especially in lenient return policy situations. Fried (2008) and Hibbs (2000) also found cognitive dissonance linked to back-out behavior, which is “the post-purchase action taken by a consumer that results in the return of merchandise or cancellation of the contract” and is reflective of cognitive dissonance (Donnelly & Ivancevich, 1970, Fried, 2008).
Particularly for apparel online buying, it is very likely that the perceived uncertainty or risk is relative high because of the difficulty in communicating accurate product information and the experience attribute of apparel products. Since cognitive dissonance and perceived risks share many same characteristics (e.g. the importance of the decision, a lack of product information, a lack of familiarity with the product) (Soutar & Sweeney, 2003), the influence of cognitive dissonance in the context of apparel e-commerce might be even more significant. Therefore, a hypothesis is formulated below:
product returns.
2.3.2 Gender and cognitive dissonance Some studies found that there was no
significant relationship between gender and buyer’s remorse (Fried, 2008), and no significant relationship between gender and cognitive dissonance (Keng & Liao, 2009; Soutar & Sweeney, 2003; Graff, Sophonthummapharn, & Parida, 2012). However, these studies adopted different cognitive dissonance scales, since there has been no widely accepted scale for cognitive dissonance yet. Meanwhile, it is possible that for apparel products, the role of gender is more important, since the degree of involvement for apparel products between female and male is significantly different (O’cass, 2000), and the importance of decision is one of the antecedents of cognitive dissonance (Festinger, 1957).
of liberal return policies and both types of dissonance (Powers & Jack, 2013). These findings above suggested a closer relationship between cognitive dissonance and product returns among males than females. Therefore, it is also possible that compared to female customers, male customers are more likely to return apparel products because of cognitive dissonance. Accordingly, two hypotheses are formulated below:
Hypothesis 5a. Compared to female customers, male customers have a significantly higher level of cognitive dissonance.
Hypothesis 5b. Compared to female customers, male customers have a significantly higher relationship between cognitive dissonance and e-commerce apparel product returns..
2.3.3 Age and cognitive dissonance Soutar and Sweeney (2003) conducted an
search viewed as expense will decrease the satisfaction with the product and increase the intention to return product (Maity & Arnold, 2013). A study of young and mature customers in the fashion clothing industry argued that cognitively young elders were more confident in their decision making (Nam et al., 2007). Since cognitive dissonance after purchase can refer to an uncomfortable psychological state about conflicted cognitions regarding the purchase decision (Festinger, 1957), customers with more confidence in their buying decisions might have fewer conflicted cognitions and therefore are likely to have lower level of cognitive dissonance. Thus, a hypothesis is formulated below:
Hypothesis 6. Age has a positive indirect relationship with e-commerce apparel product returns, through cognitive dissonance.
2.3.4 Price and cognitive dissonance Though empirical studies found that price
tolerance was positively related to market competition, in other words, customers would be less willing to pay a high price when the availability of attractive alternatives increases (Anderson, 1996). The increased availability of alternatives is likely to lead to more cognitive dissonance, therefore it is possible that cognitive dissonance is one of the reasons that customers are less tolerable of the price of the purchase and choose to return the products. Hence, a hypothesis is formulated below:
Hypothesis 7. Price has a positive indirect relationship with e-commerce apparel product returns, through cognitive dissonance
3. METHODOLOGY
3.1 Research Design
Given the research question, statistical evidences are needed for testing the relationship between age, gender, price, cognitive dissonance and product returns. In relation with cognitive dissonance, existing literature mainly use surveys to measure the level of cognitive dissonance, and there are some existing scales (e.g. Sweeney & Hausknecht, 2000; Lee, 2015). Therefore survey was adopted as one of the research methods. The other required data can be obtained from both customer survey and the apparel e-commerce retailers. However, survey will rely on customer’s self reported information, and might be limited in the number of responses, while the data from e-commerce retailers are subject to availability, and there is limited possibility to conduct survey among the customers with the help of the retailers. Therefore to overcome the shortages, a combination of data sources was adopted in the research.
from e-commerce apparel retailers; In the second phase, the main purpose was to investigate the relationship found in the first phase, more specifically, how cognitive dissonance can be used to explain the relationships (H4, H5, H6, H7). A survey in the Netherlands was conducted to measure the cognitive dissonance level of respondents. In addition, the survey also collected the age, gender, price and product returns information, so that the hypotheses H1, H2, H3 were tested again in the phase 2.
3.2 Data Collection
In the first phase, customer transaction data were obtained from an apparel e-commerce company in the Netherlands. First an interview was conducted with a manager of the company, to collect the general information of the company and the aim and content of the research. The requirement of data needed for the research was then sent to the company. The data exported from the company information system contained two parts: one part was the customer-level information, including gender, age, the number of orders purchased, and orders (partially) returned per customer; the other part was the order-level information, including price, the state of orders (returned or not). The amount and types of data were subject to availability.
In the second phase, a survey was conducted. Since the original scale for cognitive dissonance was in English while the research was conducted in the Netherlands, the questionnaire was first designed in English and then translated into Dutch by a Dutch native speaker. The Dutch questionnaire was then translated into English again by another Dutch native speaker, and was compared with the original English questionnaire. Two Dutch words were altered for better comprehension and several spelling mistakes were corrected. Finally both English and Dutch questionnaires were sent to five native speakers to see if the questionnaire can be understood well.
the shopping mall and the university campus). The respondents were restricted to people currently living in the Netherlands, since product price was included as one of the questions in the survey and the impact of price potentially caused by different economy situations in different regions should be eliminated. The topic of the questionnaire, namely the apparel shopping experience, was communicated in the introduction part of the questionnaire. The respondents were not offered with incentives. A pre-test was conducted with 20 respondents on the street and no questions were deleted. The questionnaires were then sent out on a larger scale.
3.3 Measures
3.3.1 Instrument The survey questionnaire consisted of two sections. The first
section contained demographic questions of age and gender, and questions about frequency of online apparel purchase and return in the past 6 months. The second section required customers to answer the questions based on their most recent apparel online shopping experience. The questions included the price of the product, the occurrence of the purchase, whether the product was returned, and 16 items of cognitive dissonance scale. Finally respondents were required to score their online apparel return frequency on a 7 point Likert scale. The questionnaire can be found in the appendix.
3.3.2 Measure of cognitive dissonance The sixteen-item scale of cognitive
made the right choice”). The measure has exhibited Cronbach’s alpha reliability coefficients of 0.98 and 0.88 for emotional dissonance and product dissonance respectively (Powers & Jack, 2013). The order of items were randomized.
3.3.3 Measures of product returns In the existing literature about product returns,
there were various measures for the product return frequency, including a nominal scale of self-evaluated return frequency (e.g. Foscht et al., 2013), the absolute number or the percentages of return shopping trips (e.g. Kang & Johnson, 2009), return frequency self-evaluated by the respondent on a 7-point Likert scale (Powers & Jack, 2013), etc. However there are disadvantages of these measures: The nominal scale cannot quantify the return frequency and might suffer from subjectivity; The percentage of return shopping trips would define a customer who only bought one product and returned it as a frequent returner with 100% return rate, while that is usually not the case (Kang & Johnson, 2009); The number of return trips does not involve the frequency of purchase, and therefore cannot reflect the probability of return. Moreover, since the scale of cognitive dissonance was designed to measure one shopping experience (example item: “After I bought this product, I feel sick”), it might be more appropriate to also measure whether the product involved in the specific shopping experience was returned or not.
To avoid the bias of product return measures caused by the limitations above, different measures were applied in this research. Since the data provided by the e-commerce apparel retailer include the number of orders purchased and returned by customers during a given period, the percentage of orders returned can be calculated. The company data also include whether product item was returned or not. Therefore to be comparable with the company data, the above measures of product returns were also adopted in the survey questionnaire. Additionally, survey respondents were asked to self-evaluate their return frequency. As a result, there were four ways to measure the product return with this questionnaire (see appendix), including:
(2) the percentage of (e-commerce apparel) orders returned in the past 6 months; (3) the self-evaluated (e-commerce apparel) return frequency reported on a 7 point Likert scale;
(4) whether the most recently purchased apparel product online was returned or not.
3.4 Data Analysis
Due to different data characteristics, the hypotheses were tested with different statistics tests. In the first phase, the sample size was large (n=31622 for customer sample; n=227876 for order sample), therefore normality of data can be assumed. The relationship between age and the number of orders returned was analyzed with regression analysis. The relationship between price and whether the product was returned was analyzed with binary logistics regression analysis. Gender difference was tested with Welch’s t-test because of the unequal sample size between female (n=23674) and male (n=7948).
each hypothesis.
4. RESULT
4.1 Sample Characteristics
In the first phase, the original sample exported from the e-commerce apparel company consisted 44174 customers. As many customers wrongly reported their age, only customers with reported age ranging from 18 to 80 were considered. As a result, the final sample of customers was composed of 31623 customers. The original sample also consisted 227876 order line information purchased between Sep 1, 2014 and April 1, 2015. Among the given data, 79.12% of the order lines were not returned, 20.88% of the order lines were (partially) returned.
In the second phase, the final sample consisted of 123 respondents. 41 respondents were eliminated as they had missing values in their questionnaires or did not buy any apparel items in the past 6 months. 27.1% respondents bought an apparel product online within 1 week before the survey, 17.9% bought within 1-2 weeks, 11.1% bought within 2 weeks to 1 month, and 43.9% bought over 1 month. In this most recent apparel online shopping experience, 18.7% respondents returned the product, and 81.3% did not. The variables demographics in the phase 1 and phase 2 is presented in Table 4.1.
Variables Phase 1: Company data Phase 2: Survey data
Age Mean 37.23 Mean 25.33
Gender 74.9% Female, 25.1%
Male
64.2% Female, 35.8% Male
Price Mean 20.67 Mean 57.6
number of orders purchased Mean 1.43 Mean 4.59
number of orders returned Mean 0.44 Mean 1.17
4.2 Scale Reliability and Validity
In order to confirm the division of dimensions in the 16-item scale of cognitive dissonance, factor analysis was conducted. An examination of the Kaiser-Meyer Olkin measure of sampling adequacy suggested that the sample was factorable (KMO=0.91). Bartlett's test of sphericity was significant (p<0.001). Two components were extracted in the factor analysis (Table 4.2). The first factor included all 16 items in the scale. The Cronbach’s alpha for this factor was 0.95. The second factor included 4 items concerned “product dissonance” (Powers & Jack, 2013). The Cronbach’s alpha for this factor is 0.84. The Cronbach’s alpha for the first factor can be improved to 0.97 by deleting the 4 items in the second factor. The improved factor 1 consisted 12 items concerned “emotional dissonance” (Powers & Jack, 2013).
Factor loadings
Item Factor 1 Factor 2
Product Dissonance a 3 0.39 0.72 7 0.58 0.64 11 0.54 0.60 14 0.54 0.58 Emotional Dissonance b 1 0.76 2 0.79 4 0.87 5 0.88 6 0.87 8 0.84 9 0.87 10 0.85 12 0.87 13 0.87 15 0.89 16 0.83 a,b
: Division of constructs in the original scale by Powers and Jack (2013)
4.3 Hypotheses Testing
4.3.1 Hypothesis 1: gender and product returns The first hypothesis predicted a
significant relationship between gender and apparel product returns. An independent sample t-test was performed to examine the difference between the gender groups. Female returned an average of 0.47 product and a percentage of 29.52% of the products purchased during the given period, while male returned an average of 0.34 product and a percentage of 22.15% of the products purchased in the same period. In the results, female significantly returned more products than male (t=12.86, p<0.001), and female significant retuned a higher percentage of products they bought online than male did (t=13.71, p<0.001).
In the survey data, the Levene’s test showed that there were deviations of variances between the two gender groups (p<0.01). Therefore Welch’s t-test was performed to examine the difference between gender. There were significant gender differences in the number of orders returned in 6 months (p<0.01), percentage of orders returned in 6 months (p=0.02), self-evaluated return frequency (p<0.01), but there was no significant gender difference in whether the most recently bought apparel product was returned or not (p=0.10). The mean values of the indicators of product returns can be found in Table 4.3. Thus given both the retailer and the survey data, H1 was supported.
Measures Female Male
Number of orders returned in 6 monthsa 0.47 0.34
Number of orders returned in 6 monthsb 1.54 0.50
Percentage of orders returned in 6 months 30.00% 15.91%
Self-evaluated return frequencyc 2.77 1.93
Whether the product was returned or notd 0.23 0.11
a: Company data; b: Survey data; c: On a 7 point Likert scale; d: Returned=1, Not returned=0.
4.3.2 Hypothesis 2: age and product returns The second hypothesis stated that there was a significant relationship between age and product returns. In phase 1, a linear regression revealed that there was no significant relationship between age and the number of orders returned (p=0.54). There was also no significant relationship between age and the percentage of orders returned (p=0.26).
In phase 2, Spearman's rank correlation coefficient test revealed that there was no significant relationship between age and all four measurements of product returns: the number of orders returned in 6 months (p=0.25); the percentage of orders returned in 6 months (p=0.76); the self-evaluated return frequency (p=0.20); whether the most recently bought apparel product was returned (p=0.31). Thus given both the retailer and the survey data, H2 was rejected.
4.3.3 Hypothesis 3: price and product returns The third hypothesis predicted a
significant relationship between price and product returns. The independent sample t-test showed that the difference between the prices of products returned (Mean=25.59) and the prices of products not returned (Mean=19.37) was significant (p<0.001). Furthermore, a binary logistics regression was performed to ascertain the effect of price on whether the product was returned. The results indicated that increasing price was associated with an increasing likelihood of the product returns (B=0.02, p<0.001). Therefore with the company data, H3 was supported.
The contradictory results in the company data and survey data will be discussed in section 5.
4.3.4 Hypothesis 4: cognitive dissonance and product returns Hypothesis 4
predicted a positive correlation between cognitive dissonance and product returns. Spearman's rank correlation coefficient test was run to determine the relationship between cognitive dissonance and the four measures of product returns. There was no significant correlation between cognitive dissonance and the three measures of product return frequency (p=0.15 for the number of items returned in 6 months; p=0.18 for the percentage of items returned in 6 months; p=0.21 for the self-evaluated return frequency). There was a significant positive correlation with whether the most recently bought apparel product was returned or not (rs(123)=0.25, p<0.01).
Meanwhile, the two factors extracted from the factor analysis, namely emotional dissonance and product dissonance, were tested separately to investigate their relationship with product returns. The Spearman's rank correlation coefficient test revealed that product dissonance was positively correlated with whether the product was returned (rs(123)=0.31, p<0.001), the number of orders returned in 6 months
(rs(123)=0.20, p=0.03), and the percentage of orders returned in 6 months
(rs(123)=0.22, p=0.02). However, there was no significant correlation between
product dissonance and self-evaluated return frequency (rs(123)=0.12, p=0.19). There
was no significant correlation between emotional dissonance and all four measurements of product returns.
4.3.5 Hypothesis 5: gender, cognitive dissonance and product returns
Hypothesis 5a predicted that males had a significantly higher level of cognitive dissonance. In order to compare the cognitive dissonance level between female and male, Welch t-test was performed. As shown in Table 4.4, The gender difference was significant in both emotional dissonance (p=0.01) and product dissonance (p=0.10). Female scored significantly higher than male in both cognitive dissonance dimensions (see Table 4.4 below). However this was contradictory to the hypothesis that males had a higher level of cognitive dissonance than females. Thus, H5a was rejected.
Variable Female Male Significance of Difference between
Gender
Cognitive dissonance 2.3987 1.9730 p=0.04
Emotional dissonance 2.0000 1.6515 p=0.01
Product dissonance 3.4715 2.7955 p=0.01
TABLE 4.4 - Gender and cognitive dissonance
relationship with the number of orders returned in 6 months (r=0.26, p=0.02), the percentage of orders returned in 6 months (r=0.22, p=0.05), and whether the most recently bought apparel product was returned or not (r=0.34, p<0.01), and there was also a significant positive relationship between cognitive dissonance and whether the most recently bought apparel product was returned or not (r=0.29, p<0.01). However, this was contradictory to the hypothesis 5 that males would have a significant higher relationship between cognitive dissonance and product returns than females. Thus, H5b was rejected. Correlation Emotional dissonance Product dissonance Cognitive dissonance Number of orders returned in 6 months F a: p=0.25 M b: p=0.76 F: p=0.02** M: p=0.58 F: p=0.08 M: p=0.62 Percentage of orders returned in 6 months F: p=0.74 M: p=0.99 F: p=0.05** M: p=0.77 F: p=0.26 M: p=0.91 Self-evaluated return frequency F: p=0.45 M: p=0.16 F: p=0.14 M: p=0.56 F: p=0.28 M: p=0.80 Return or not F: p=0.08 M: p=0.83 F: p=0.00*** M: p=0.20 F: p=0.01*** M: p=0.65 a,b F=Female, M=Male; ** Significant at 0.05 level; *** Significant at 0.01 level.
TABLE 4.5 - Spearman’s coefficients between dissonance and product returns sorted by gender
4.3.6 Hypothesis 6: age, cognitive dissonance and product returns Hypothesis
relationship (Frazier, Tix, & Barron, 2004; Cohen & Cohen, 1983: 366), therefore there was no significant indirect relationship between age and product returns through cognitive dissonance, emotional dissonance or product dissonance. Thus, H6 was rejected.
4.3.7 Hypothesis 7: price, cognitive dissonance and product returns Hypothesis
7 predicted an indirect relationship between price and product returns through cognitive dissonance. A Spearman's rank correlation coefficient test was performed to test the relationship between price and cognitive dissonance. Price did not show significant relationship with emotional dissonance (p=0.17), product dissonance (p=0.73), or cognitive dissonance (p=0.27). Therefore there was no significant relationship between the predictor variable and mediator variable. Thus, H7 was rejected.
5. DISCUSSION
respondents reported their cognitive dissonance level based on the shopping experience within 1 week before the survey, while the rest 73.4% respondents were recalling an even earlier emotional feeling. The long time gap between purchase and filling the survey for emotional dissonance might also contribute to the insignificant relationship between emotional dissonance and product returns.
the importance of decision. Therefore, it is possible that female could be more likely to return apparel products because of higher involvement for the products which lead to higher level of product dissonance. Meanwhile the contradictory findings might imply that the impact of cognitive dissonance on product return could vary among different product categories.
Customer age was not found to be significantly related to product return frequency, neither in the survey nor in the customer data provided by the retailer. This finding is not very surprising, since some previous researches already found that age did not correlate with back-out behaviors (Fried, 2008; Hibbs, 2000). Age also show no correlation with the cognitive dissonance, which contradicts with the finding of Soutar and Sweeney (2003). However, in the statistical result of Soutar and Sweeney (2003), the significance of relationship between age and dissonance was not very strong (p<0.05 for the car stereo customer group; p<0.10 for the furniture customer group). Moreover, many studies found no relationship between age and cognitive dissonance (Graff et al., 2012; Alhammad, Gulliver, Wiafe, & Nakata, 2013). In a study about the relationship between demographics, cognitive dissonance and product returns, Hibbs (2000) found significant differences between the means of age groups, income levels, and back-out behaviors, but no correlation between cognitive style and back-out behaviors.
data and the survey data, the reported price in the survey data (Mean=57.6, Median=40) was much higher than the reported price in the company data (Mean=20.68, Median=17.95). A possible reason for this difference is that the retailer tend to target at the price-conscious customers according to the interview with the company representative. It is possible that the those customers of the retailer are more price conscious than apparel customers in general, and their decisions of purchasing or returning are more likely to be influenced by price. However, as price consciousness was not examined in this research, this argument is not supported with empirical evidences. And since the retailers were providing a more objective and massive data of item prices and returned status than the survey, we can still conclude that price is very likely to have impact on product returns.
importance of decision (Korgaonkar & Moschis, 1982; Oliver, 1997; Powers & Jack, 2013), but the degree of influence might differ among individuals due to different level of price sensitivity. Therefore it might be interested to also investigate the relationship between price, price consciousness and cognitive dissonance. Meanwhile, other reasons that could lead to the positive relationship between price and product return need to be investigated. For example, the price premiums was proposed to relate to the product fit uncertainty, a factor that contribute to product returns especially for experience goods like apparel products (Hong & Pavlou, 2010); It was also proposed that an individual customer’s zone of tolerance might vary with the product price (Santos & Boote, 2003). These possible explanations for the impact of price on post-purchase behaviors can be investigated in the future research.
6. CONCLUSION
customers will spontaneously try to reduce their post-purchase dissonance by rationalizing their purchasing decisions in their mind, seeking additional information etc. (Lee, 2015), therefore it might be useful for retailers to provide positive information of the product choice, such as positive product reviews from other customers, placing advertisements showing satisfied brand owners, sending letters of congratulation to owners with a free gift, or sending favorable press reports (Mitchell & Boustani, 1994). Post-purchase communications with customers have been shown effective to reduce the rate of product returns and order cancellations (Mitchell & Boustani, 1994; Donnelly & Ivancevich, 1970). However, existing literature about what retailers can do to help customers to reduce their post-purchase dissonance is still very limited. Since a considerable proportion of customers are returning products with reasons related to post-purchase dissonance (Lee, 2015), it would have important managerial implications to investigate what effective dissonance reduction strategies that e-commerce retailers can apply to their customers.
Besides, this research also found empirical evidences that price was positively related to e-commerce apparel product returns. However the survey didn’t find cognitive dissonance as a mediator of the positive relationship. As the reason of why price can influence product returns still lacks empirical examinations, it can also be an interesting direction for future researches, since it would be effective if the e-commerce retailers understand the cause of returns of customers and orders with different characteristics, and take actions accordingly. It would also probably contribute to forecasting consumer return, which leads to a better decision making in strategic, tactical and operational areas of the organization (Potdar & Rogers, 2012).
dissonance items were only significantly related to whether the product was returned but not related to the other three measures of return frequency, and the customer gender only showed differences with the number and percentage of orders purchased, but not with whether the single product was returned or not. This finding is reasonable since the cognitive dissonance items adopted in this research are describing the situation after purchasing one single product, and respondents were asked to answer the cognitive dissonance questions and whether the product was returned based on the most recent shopping experience, while whether the product was returned in one shopping experience might not be a good predictor of the overall return frequency among different customer groups. In this way, the measures of product return match with the different needs for data analysis. Another advantage of this research is that, in the first phase data was sourcing from the massive transaction data of a real company while most researches on product return only rely on the survey as source of data. The company transaction data provided numerous and reliable data for analysis especially concerning price.
REFERENCE
Accenture. 2008. Most returned products work fine, study says. Available at: http://www.pcworld.com/article/146576/article.html?null (Jun 2, 2008)
Accenture. 2012. European cross-border e-commerce: the challenge of achieving profitable growth. Available at:
http://www.accenture.com/SiteCollectionDocuments/PDF/Accenture-ERRT-Brochure .pdf.
Alhammad, M., Gulliver, S., Wiafe, I. and Nakata, K. 2013. Culture, cognition and e-commerce behavior. In: 14th International Conference on Informatics and Semiotics in Organisation (ICISO) (pp.113-119), Stockholm, Sweden.
Anderson, 1996. Customer Satisfaction and Price Tolerance. Marketing Letters, 7(3): 265-274.
Bernon, M., Rossi, S., & Cullen, J. 2011. Retail reverse logistics: a call and grounding framework for research. International Journal of Physical Distribution & Logistics Management, 41(5): 484-510.
Black, D. W. 2007. A review of compulsive buying disorder. World Psychiatry, 6(1): 14-18.
Cho, Y., Im, I., Hiltz, R., & Fjermestad, J. 2002. An analysis of online customer complaints: implications for web complaint management. In System Sciences, 2002. HICSS. Proceedings of the 35th Annual Hawaii International Conference (pp. 2308-2317). IEEE.
firm performance in an e-commerce market. International Journal of Physical Distribution & Logistics Management, 38(5): 336-359.
Cohen, J., & Cohen, P. 1983. Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum.
Davenport, K., Houston, J. E., & Griffiths, M. D. 2012. Excessive eating and compulsive buying behaviours in women: An empirical pilot study examining reward sensitivity, anxiety, impulsivity, self-esteem and social desirability. International Journal of Mental Health and Addiction, 10(4): 474-489.
De, P., Hu, Y., & Rahman, M. S. 2013. Product-oriented web technologies and product returns: An exploratory study. Information Systems Research, 24(4): 998-1010.
Dholakia, R., & Uusitalo, O. 2002. Switching to electronic stores: consumer characteristics and the perception of shopping benefits. International Journal of Retail & Distribution Management, 30(10): 459-469.
Donnelly, J. H., & Ivancevich, J. M. 1970. Post-purchase reinforcement and back-out behavior. Journal of Marketing Research, 7(3): 399-400.
Engel, J. F., Blackwell, R. D., & Miniard, P. W. 1995. Consumer behavior (8th ed.). New York: Dryden.
Festinger, L. 1957. A theory of dissonance cognitive. Evanston, IL: Row Peterson.
Foscht, T., Ernstreiter, K., Maloles III, C., Sinha, I., & Swoboda, B. 2013. Retaining
or returning? Some insights for a better understanding of return
113-134.
Frazier, P. A., Tix, A. P., & Barron, K. E. 2004. Testing moderator and mediator effects in counseling psychology research. Journal of Counseling Psychology, 51(1): 115-134.
Fried, T. A. 2008. The relationship between psychological types, demographics and post-purchase buyer's remorse. Florida: ProQuest.
Graff, J., Sophonthummapharn, K., & Parida, V. 2012. Post-purchase cognitive dissonance–evidence from the mobile phone market. International Journal of Technology Marketing, 7(1): 32-46.
Griffis, S. E., Rao, S., Goldsby, T. J., & Niranjan, T. T. 2012. The customer consequences of returns in online retailing: An empirical analysis. Journal of Operations Management, 30(4): 282-294.
Hart, C. W. 1988. The power of unconditional service guarantees. Harvard Business Review, 66 (4): 54-62.
Hess, J. D., & Mayhew, G. E. 1997. Modeling merchandise returns in direct marketing. Journal of Interactive Marketing, 11(2), 20-35.
Hibbs, R. D. 2000. The relationship between cognitive style, demographics, and back-out behavior in the post-purchase phase of the consumer decision making process. Doctoral dissertation, Nova Southeastern University.
Hong, Y., & Pavlou, P. A. 2010. Fit does matter! An empirical study on product fit uncertainty in online marketplaces. Available at:
Inman, J. J., Winer, R. S., & Ferraro, R. 2009. The interplay among category characteristics, customer characteristics, and customer activities on in-store decision making. Journal of Marketing, 73(5): 19-29.
Kang, M., & Johnson, K. 2009. Identifying characteristics of consumers who frequently return apparel. Journal of Fashion Marketing and Management: An International Journal, 13(1), 37-48.
Keng, C. J., & Liao, T. H. 2009. Consequences of postpurchase dissonance: The mediating role of an external information search. Social Behavior and Personality: An International Journal, 37(10): 1327-1339.
Kincade, D. H., Giddings, V. L., & Chen-Yu, H. J. 1998. Impact of product-specific variables on consumers' post-consumption behaviour for apparel products: USA. Journal of Consumer Studies & Home Economics, 22(2): 81-90.
Koller, M., & Salzberger, T. 2007. Cognitive dissonance as a relevant construct throughout the decision-making and consumption process-an empirical investigation related to a package tour. Journal of Customer Behaviour, 6(3): 217-227.
Korgaonkar, P. K., & Moschis, G. P. 1982. An experimental study of cognitive dissonance, product involvement, expectations, performance and consumer judgment of product performance. Journal of Advertising, 11(3): 32-44.
Kumar, V., George, M., & Pancras, J. 2008. Cross-buying in retailing: Drivers and consequences. Journal of Retailing, 84(1): 15-27.
Marketing, 32(1): 49-64.
Lee, J., & Lund, M. (2003, May). Strategies to optimize return on investment (ROI) through effective reverse supply chain programs. In Electronics and the Environment, 2003. IEEE International Symposium (pp. 335-340). IEEE.
Lehman, A., O’Rouke, N., Hatcher, L. & Stepanski, E.J. 2005. JMP for basic univariate and multivariate statistics: A step-by-step guide. North Carolina: SAS Institute.
Lightner, N. J. 2003. What users want in e-commerce design: effects of age, education and income. Ergonomics, 46(1-3): 153-168.
Maity, D. 2015. Forming product return intentions to reduce cognitive dissonance after purchase. In: Marketing Dynamism & Sustainability: Things Change, Things Stay the Same… (pp. 456-456). Springer International Publishing.
Maity, D., & Arnold, T. J. 2013. Search: An Expense or an Experience? Exploring the Influence of Search on Product Return Intentions. Psychology & Marketing, 30(7): 576-587.
McGoldrick, P. J., & Collins, N. 2007. Multichannel retailing: profiling the multichannel shopper. International Review of Retail, Distribution and Consumer Research, 17(2): 139-158.
Milliman, R. E., & Decker, P. J. 1990. The use of post-purchase communication to reduce dissonance and improve direct marketing effectiveness. The Journal of Business Communication, 27(2): 159-170.
post-purchase risk perception and reduction. European Journal of Marketing, 28(1): 56-71.
Mittal, V., & Kamakura, W. A. 2001. Satisfaction, repurchase intent, and repurchase behavior: Investigating the moderating effect of customer characteristics. Journal of Marketing Research, 38(1): 131-142.
Mowen, J.C. 1995, Consumer behaviour (4th ed.). Prentice-Hall, Englewood Cliffs, NJ.
Nam, J., Hamlin, R., Gam, H. J., Kang, J. H., Kim, J., Kumphai, P., Starr, C., & Richards, L. 2007. The fashion‐conscious behaviours of mature female consumers. International Journal of Consumer Studies, 31(1): 102-108.
O’Cass, A. 2000. An assessment of consumers product, purchase decision, advertising and consumption involvement in fashion clothing. Journal of Economic Psychology, 21(5): 545-576.
Petersen, J. A., & Kumar, V. 2009. Are product returns a necessary evil? Antecedents and consequences. Journal of Marketing, 73(3): 35-51.
Pinto, D. 2001. Driving anger, articulated cognitive distortions, cognitive deficiencies and aggression. Doctoral Dissertation Hofstra University, Hempstead, New York.
Potdar, A., & Rogers, J. 2012. Reason-code based model to forecast product returns. Foresight, 14(2): 105-120.
Rogers, D. S., Rogers, Z. S., & Lembke, R. 2010. Creating value through product stewardship and take-back. Sustainability Accounting, Management and Policy Journal, 1(2): 133-160.
Ruxton, G. D. 2006. The unequal variance t-test is an underused alternative to Student's t-test and the Mann-Whitney U test. Behavioral Ecology, 17(4): 688-690.
Santos, J., & Boote, J. 2003. A theoretical exploration and model of consumer expectations, post-purchase affective states and affective behaviour. Journal of Consumer Behaviour, 3(2): 142-156.
Schaninger, C. M., & Sciglimpaglia, D. 1981. The influence of cognitive personality traits and demographics on consumer information acquisition. Journal of Consumer Research, 8(2): 208-216.
Shulman, J. D., Coughlan, A. T., & Savaskan, R. C. 2011. Managing consumer returns in a competitive environment. Management Science, 57(2): 347-362.
Slama, M. E., & Tashchian, A. 1985. Selected socioeconomic and demographic characteristics associated with purchasing involvement. Journal of Marketing, 49(1): 72-82.
Soutar, G. N., & Sweeney, J. C. 2003. Are there cognitive dissonance segments?. Australian Journal of Management, 28(3): 227-249.
Srivastava, S. K. 2007. Green supply chain management: a state-of-the-art literature review. International Journal of Management Reviews, 9(1): 53-80.
multidimensional scale. Psychology & Marketing, 17(5): 369-385.
The Economist. 2013. Return to Santa - E-commerce firms have a hard core of costly, impossible-to-please customers. Available at:
http://www.economist.com/news/business/21591874-e-commerce-firms-have-hard-co re-costly-impossible-please-customers-return-santa (Dec 21, 2013)
Willert Jr., M.G. 1995. Coping strategies and emotional and physical status of family members of mental health consumers. Doctoral Dissertation, The University of North Dakota.
Wood, S. L. 2001. Remote purchase environments: the influence of return policy leniency on two-stage decision processes. Journal of Marketing Research, 38(2): 157-169.
APPENDIX: Survey Questionnaire
1. Your gender? Female/Male
2. Your age? ___ years old
3. How often did you buy apparel items online in the past 6 months? ___
4. How often did you return apparel items bought online in the past 6 months? ___
For below questions, please recall one of your most recent apparel online shopping experiences. (If you bought more than one products in the purchase, please randomly choose one of them)
5. What is the price of the product? (in Euro) ___
6. When did the purchase happen? Within 1 week / Within 2 weeks / Within 1 month / Over 1 month before
7. Did you finally return the product? Yes/No
8. Please describe how you agree with below items (1-strongly agree, .., 7-strongly disagree)
After I bought this product I felt scared
After I bought this product I wondered if I'd been fooled
After I bought this product I wonder if I really need this product
After I bought this product I felt hollow
After I bought this product I wondered if they had spun me a line
After I bought this product I felt uneasy
After I bought this product I was in pain After I bought this product I felt depressed
After I bought this product I wondered whether there was something wrong with the deal I got)
After I bought this product I wonder if I have made the right choice After I bought this product I felt I'd let myself down After I bought this product I felt furious with myself
After I bought this product I wonder if I have done the right thing in buying this product
After I bought this product I felt sick
After I bought this product I was in agony
