2013
Anoek Dina Vinke
S1779664
University of Groningen,
Faculty of Economics and Business
Msc. Marketing Intelligence
28-6-2013
Ordering online, fulfilling the order
or not?
by
Anoek Dina Vinke
University of Groningen
The conversion rates in online shops are much lower than in the traditional "brick and mortar" shops. Consumers abandon their order in different stages of the ordering process. The last stage of this process is the fulfillment policy. The conditions of this policy differ between web shops and these differences can lead to abandoning the order. Until now, it has never been researched which conditions in the fulfillment policy are valued the most by consumers. This study investigates the utility of the conditions of the fulfillment policy of online stores. The goal of this research is to determine which conditions give the consumers the highest utility.
The report starts with a literature review based on academic sources discussing the fulfillment policy conditions. The attributes that create the fulfillment policy are delivery costs, delivery time, convenience of delivery, convenience of payment and the price of the product.
In order to capture the relative importance of the conditions in the fulfillment policy, a choice based conjoint questionnaire was set up and answered by 203 respondents. Furthermore, this questionnaire was used to measure the price consumers are willing to pay for better conditions in the fulfillment policy. Respondents were also asked to answer questions regarding their shopping behavior and demographics. These variables were used to conduct a segmentation analysis. Moreover, an ordinary least square regression analysis was conducted to see what the effects of these variables on the importance of the attributes are and if they are in line with the choice based conjoint analysis results.
For the aggregated model, it became clear that consumers placed the highest relative importance on delivery cost. Second was delivery time and the third relative important attribute was price. The other two attributes, convenience of delivery and convenience of payment, do not play a big role in valuing the fulfillment policy. In the segmentation analysis, three segments were observed; “The mixed optimizers”, “Cheap and quick delivery” and “The price sensitive online shoppers”.
time, convenience of delivery, convenience of payment, price
Preface
This thesis is written to complete the master Marketing Intelligence at the University of Groningen. After graduating at the University of Groningen in Business Administration, I decided to follow my heart and study Marketing. After half a semester I came to the conclusion that Marketing Intelligence even more satisfied my interests. Therefore I decided to graduate in Marketing Intelligence. I am very glad I took this opportunity.
The reason that I chose this subject, is because I was ordering a book online and I saw that it took more than a week to be delivered. I went to another website and found that they did not only deliver it within three work days, they also did not include any shipping costs. This made me think about the different fulfillment policies web shops handle and how they influence my own buying behavior. Therefore, I decided to investigate the different fulfillment policies.
By handing in this thesis, my life as a student finally comes to an end after studying for five years in Groningen. I could not have graduated without some people that were there to support me the whole time. I would like to take this chance to thank a few people in particular.
First of all, I would especially like to thank Mariëlle Non for her clear advice, help in the right direction and feedback during the process of writing my thesis. Her door was always open and even though she had a very busy schedule, she always took the time to answer questions and solve difficult problems. I would also like to thank my second supervisor, Hans Risselada, for his suggestions. Moreover, I would like all the professors at the Marketing Department of the University of Groningen for their teaching and help during the process of writing my thesis.
Secondly, I would like to thank all my friends, family and boyfriend for helping me with my thesis, their ideas and moral support. I could not have done it without you. In addition, I would like to thank my parents for giving me the opportunity to study at the university, their support, patience and faith in me during these five years. By handing in my thesis, receiving my last credits and graduating with a Master’s Degree, I hope to repay them for their continuous support, patience and trust.
Management Summary ... 2 Preface ... 3 1. Introduction ... 1 2. Theoretical framework ... 6 2.1 Delivery ... 6 2.2 Convenience of Payment ... 12 2.3 Price ... 17 2.4 Moderators ... 18 3. Conceptual model ... 20 3.1 Hypotheses ... 21 4. Research Design ... 22 4.1 Data description ... 22 4.2 Conjoint analysis ... 24 5. Results ... 27 5.1 Descriptive variables ... 27
5.2 Choice-Based Conjoint Analysis ... 31
5.2.1 Aggregate model ... 31
5.2.2 Testing the hypotheses on the aggregate model ... 33
5.2.3 Choosing the number of segments ... 34
5.2.4 Predictive Validity ... 37
5.2.5 Describing the segments ... 37
5.2.6 Testing the hypotheses on the segmentation analysis ... 43
5.3 Ordinary least squares regression estimation ... 46
5.3.1 Testing the assumptions ... 47
5.3.2 Outcomes OLS and testing the moderating effects ... 48
6. Conclusion and Discussion ... 52
6.1 Which conditions in the order fulfillment policy of an online store have the highest utility for the customer? ... 53
6.2 What is the (negative) utility of the shipping costs of the order with respect to the fulfillment policy? ... 53
6.3 What is the utility of the delivery time of the order with respect to the fulfillment policy? ... 54
fulfillment policy? ... 56
6.6 What is the (negative) utility of price of the order with respect to the fulfillment policy? ... 57
6.7. Can there be different segments of customers distinguished with respect to the order fulfillment policy? ... 57
7. Implications and recommendations. ... 58
8. Limitations ... 59
9. Directions for further research ... 60
10. References ... 62
11. Appendix ... 68
11.1 Qualitative research ... 68
11.2 Efficiency scheme ... 70
11.3 OLS results with all independent variables included ... 70
Anoek Dina Vinke S1779664
1
1. Introduction
Anoek Dina Vinke S1779664
2
critical look at how they should shape and fill in this final stage.
Sismeiro and Bucklin (2004) did study a part of this final stage, namely the probability of task completion. Here they looked at the completion of product configuration, input of personal information and order confirmation. Within this order confirmation, several factors can play a role regarding the choice of buying or not. To show this from a business perspective, the following practical example gives some clarification. The check-out register in an online shop can take several forms. Whereas www.wehkamp.nl asks €5,95 for home delivery and €4,95 for delivery at a service point, www.Zalando.nl does not charge any costs for delivery. With the former web shop you will get your product within a day, when ordered before ten p.m. The latter takes about two to four working days and can only deliver at home. Another point on which the two companies differ is the paying method options. Zalando offers its customers the additional payment methods Paypal and Creditcard, which is not possible at Wehkamp. At Wehkamp it is only possible to pay by “Ideal” and “acceptgiro”. These differences can lead to preferring the one company over the other one by customers. Especially when acknowledging that both sites have a high brand awareness, are direct competitors and sell about the same products for about the same price. The competition between these two stores narrows down to the services, an important part of this is the fulfillment policy (Bhatnagar et al., 2000).
To follow further the example stated above, these online stores have the limitation of not being able to let the consumer physically “feel” the product, to have direct personal contact with a check-out clerk or to walk out with the product. Therefore it requires the presence of certain facilitating conditions and resources to create a perception of control over the purchasing process (Song and Zahedi, 2005). And even though price sensitivity is higher for online stores than offline stores (Nielsen et al., 2001), it becomes clear to retailers that web-presence and having a low price in the web shop are not the drivers of success on the online market. Furthermore, internet retailing has changed a lot since the 1990’s. Delivery of products can now be done within a day(s), consumer transaction costs are low and there are many locations available where orders can be placed as well as delivered (Alba et al., 1997).
Anoek Dina Vinke S1779664
3
conducted again and came to the same conclusion (Parasuraman et al., 2005). Gefen (2002) also found that in e-commerce, service quality has an impact on customers’ loyalty attitudes. Service quality includes meeting delivery deadlines, offering the right payment options and offering the right shipping options. Finally, Zeithalm et al. (2002) say that the most prevalent type of research on consequences of web site service quality is business research that identifies the reasons for abandoning shopping carts online. The reasons for abandoning the order are comparing online shops (61%), the total cost of items were too high (43%), the check-out process takes too long (41%) and the check-out requires too much personal information. This leads to the conclusion that the fulfillment options are critical in online shopping.
So it becomes clear that the fulfillment policy has a critical effect on the buying behavior of consumers. Not only on the decision whether to buy or not at the moment they get to the fulfillment policy (Montgomery et al, 2004; Sismeiro and Bucklin, 2004), but also whether they will buy again at a certain online shop and on the satisfaction with this web shop (Parasuraman et al., 2005; Zeithalm et al., 2002). Therefore it is very important for managers to look upon their fulfillment policy critically. And it is especially interesting to research which attributes within the fulfillment policy are important for consumers so that managers do not invest in the wrong attributes. For example, if it turns out that consumers put a much higher value upon delivery speed instead of delivery costs companies like Zalando.nl will lose it against companies such as Wehkamp.nl. For these reasons it would be interesting to see which attributes in the fulfillment policy are important to customers and what kind of check-out register has the highest utility among customers. But, as in every market, several segments can be distinguished (Kotler et al., 2008). The market is heterogeneous, which means that consumers differ in their preferences towards the fulfillment policy. It is therefore important for companies to target the segments they want to target in the right way and they need to know what their customers value in the fulfillment policy (Kotler et al, 2008).
Slogans like; “Ordered today, delivered tomorrow” are widely used among online shops, like bol.com and wehkamp.nl to lure people to buy at their shop. Others use “no delivery cost” as a convincing argument. Or “if you’re not at home, no problem! We’ll deliver it at a pick up point nearby”. These statements show that online shops are indeed using the service side of online shopping as a differentiation strategy. This “excellent” service is aimed at the fulfillment policy. So offering customers quick and cheap delivery at a convenient place with the opportunity to pay afterwards might help companies in the online market to create a competitive advantage.
Anoek Dina Vinke S1779664
4
web shop. Web elements can be all kinds of attributes related to a web site, but transaction efficiency is an attribute found to be far more critical than site attractiveness (Zeithalm et al, 2002). Furthermore, Song and Zahedi (2005) state in their research that these web elements influence a shoppers’ purchase intention by influencing their salient beliefs related to e-commerce, which in turn changes their attitudes, external subjective norms, and perceived behavioral controls, leading to changes in their purchase intentions. This makes important for managers to design the last step in the “online funnel” in a way that it can function to its full potential. In other words, converting as many visitors as possible into customers.
The lack of physical access and the time lag between purchase and delivery of the product makes web consumers more sensitive to the service of an online vendor. This also includes the payment method and delivery policies (Song and Zahedi., 2005). Furthermore the online stores have a much higher price sensitivity than offline stores. This makes communicating the value online difficult and even though price is more important in the online- than offline environment, non-price related elements become more important for e-commerce sites (Nielsen et al, 2001). As mentioned before, the payment process in an online store does not involve a check-out clerk and walking out with the product (Song and Zahedi., 2005). So online shopping requires the presence of certain facilitating conditions and resources to create a perception of control over the purchasing process. This purchase facilitation includes payment options and shipping options. Therefore, offering rich resources for purchasing facilitation enhances Web customers’ sense of control over the purchasing process. Zeithalm et al (2002) agree on this. They even say that 70% of a company’s resources should be devoted to creating a great customer experience and 30% should be spent on “shouting” about it. This is because service quality delivery through web sites is an essential strategy to success, possibly more important than low price and web presence. Moreover, service quality has a significant effect on customer satisfaction, in particular the fulfillment ratings are the strongest predictor of customer satisfaction and quality. It is also the second strongest predictor of intentions to repurchase at a site(Zeithalm et al., 2002). Online this means, on-time and accurate delivery, accurate product representation, and other fulfillment issues (Zeithalm et al., 2002, Parasuraman et al., 2005).
Anoek Dina Vinke S1779664
5
different attributes that create the fulfillment policy or clarified a distinction in importance. Therefore this study foresees a contribution to the existing literature.
In conclusion, this study will investigate the final stage of an online store. Ultimately, the overall aim of this study is to present the fulfillment policy where customers can derive the highest utility from. In order to investigate this problem, the following research question needs to be answered.
Which conditions in the order fulfillment policy of an online store have the highest utility for the customer?
This question can be answered by looking at the different attributes that create the fulfillment policy. When looking at several online shops like Wehkamp.nl, Zalando, Bol.com and Managementboek.nl, it becomes clear that they differ in delivery costs, delivery time, methods of delivery and the paying method. All these four attributes are in relation with the price of the product. This leads to the following sub-questions:
What is the (negative) utility of shipping costs to receive the order with respect to the fulfillment policy?
What is the utility of the delivery time of the order with respect to the fulfillment policy? What is the utility of the convenience of delivery of the order with respect to the fulfillment
policy?
What is the utility of the convenience of payment of the order with respect to the fulfillment policy?
What is the (negative) utility of price of the order with respect to the fulfillment policy? As mentioned before, segmenting the marketing is important. It makes this study complete. Therefore, this study will investigate whether segments can be distinguished based on different needs in the fulfillment policy. Segmentation will be done on a basis of demographics, behavior and psychographics, which is based on the principles of marketing (Kotler et al., 2008). This leads to the last sub-question:
Can there be different segments of customers distinguished with respect to the order fulfillment policy?
Anoek Dina Vinke S1779664
6
derive from the check-out service. This conceptual model is a graphical representation of the hypotheses, which will be tested by conducting a choice based conjoint analysis and an OLS analysis for the moderating effects. It will end with the conclusion and a discussion, the implications that result from these conclusion and finally the limitations of this study and direction for further research.
2. Theoretical framework
This chapter contains the theoretical framework regarding the different attributes that create the fulfillment policy. It starts with a review of the literature about the delivery. This section contains the delivery costs, delivery time and convenience of delivery. During this process, the different hypotheses will be formed. After these three attributes, the literature about the convenience of payment and price of the product(s) will be discussed and the corresponding hypotheses will be formed. Finally, the chapter concludes with a conceptual model of the theoretic framework.
2.1 Delivery
2.1.1 Delivery costs
The importance of delivery costs for companies in relation to their customers has been researched several times in the past (Koukova et al., 2012; Lewis, 2006; Trocchia and Janda, 2003). The growth in E-commerce has accentuated this fact even more (Koukova et al., 2012). Furthermore, shipping costs were the basis for the main complaints by 50 percent of the shoppers (Jupiter, 2000; Hua et al., 2012) and 60 percent of shoppers have abandoned an order when shipping fees were added (Ernst and Young, 1999). This fact makes it of vital importance for E-tailers to take a critical look upon this matter. The most important explanation of why people complain about shipping fees or abandon their purchase can be found in the fact that these fees are often added at the end of a transaction (Lewis, 2006). These fees are an addition to the product price and make the online order more expensive.
Nonetheless, shipping fees, are like promotions an element of price when shopping online. Therefore it can reduce order volume, because higher delivery surcharges increase the sacrifice asked of the consumer, without changing the utility of the products received (Lewis, 2006). This can also lead to consumers thinking that the retailer is using shipping fees to make additional profits rather than the cost of doing business (Koukova et al., 2012). Instead of seeing the delivery costs as inevitable, consumers think companies charge for them to gain more money.
Anoek Dina Vinke S1779664
7
certain threshold (Koukova et al., 2012). This threshold can be at a certain order value level, which can influence consumers’ buying behavior. Whereas order values below the threshold are coded as a loss for consumers, order values above the threshold are coded a bonus relative to the natural referent of free shipping. This causes buyers to become motivated to increase their order value to pass the threshold for free shipping, but it may also result into negative attributions and affect future purchase decisions negatively (Koukova et al., 2012). In general, there are three types of shipping policies used by retailers. These are unconditional free shipping (UFS), contingent free shipping (CFS) and shipping fees that increase with order size (Leng and Becererril-Arreola, 2010). UFS means that for every order a customer makes the shipping is for free, no matter what the price of the order is. CFS is as mentioned above, free delivery for the customer above a certain threshold. And finally, shipping fees that increase with order size speaks for itself (Leng and Becerril-Arreola, 2010). From an E-tailer’s perspective, the CFS strategy is the most profitable, especially when wanting to stimulate consumers to acquire more products. But it must be taken into account that this is only effective when the cutoff level is not too high and should only be used when an E-tailer wants to increase order sizes (Lewis, 2006; Leng and Becerril-Arreola, 2010). These low cutoff levels are more effective in relatively homogeneous markets than when implemented in relatively heterogeneous ones (Leng and Becerril-Arreola, 2010). Of course for a customer free delivery is the best (Frischmann et al., 2012), but from an E-tailer’s perspective this means that they have to pay for the shipping costs, which leads to a reduction of their profits. The need to make many individual shipments to thousands or even millions of customers rather than a single big delivery to one store is the downside of E-tailing (Boyer, 2001). On the other side, if consumers have to pay for their delivery after they have put the product(s) in their basket, they might abandon the process completely (Hua et al., 2012). According to Bertin and Wathieu (2008), the UFC helps to reduce this problem. When consumers go to the “check-out” of the web shop they will not leave the shopping process because unforeseen costs are already added.
Anoek Dina Vinke S1779664
8
increase order incidence rates, but lead to smaller order amounts. This has implications for managers, because they may be able to acquire more customers to their web shop, sell more, but at lower prices. So in the end the profit of these “free-shipping” promotions is not even that high. Koukova et al. (2012) also show that order value influences consumers’ response towards purchase intentions. They suggest that providing important alternative referents such as price promotion or reduction draws away attention from the referents used in evaluating shipping fee structures. This strategy is likely to be more effective for online retailers with unknown or low reputations , because consumers pay more attention to surcharges from those online retailers. Drawing attention away from shipping fees may encourage consumers to scale up their order values and complete the purchase process. But, although consumers may be motivated to increase order value to pass the threshold for free shipping, it may also result in negative attributions and affect future purchase decisions negatively (Koukova et al., 2012).
This may also depend on the kind of customer. Whereas people who shop more often at a certain website are probably more price-sensitive towards shipping fees, people who do not shop that often are expected to be less price sensitive (Dolan, 1987). Also, free shipping can effectively encourage a customer to order more goods, to the extent of ordering four times of the optimal order size without free shipping (Hua et al., 2012). Furthermore, the “zero-risk bias” suggests that people find the difference between one cent and zero cent much larger than the difference between two cents and one cent. This “zero-risk bias” also includes that people experience a more positive affect when they are exposed to a free offer and that this leads to a significant and major increase in demand (Finucane, 2000).
Consumers regard free shipments not only as a decrease in costs, but also as an extra benefit. This is caused by the fact that people are risk averse, which is even higher in the online environment (Frischmann et al., 2012). This is also an interesting fact for E-tailers, since it can create extra benefits when implementing free delivery costs. If this high free shipping fees preference is supported in this study it can create enormous benefits for E-tailers, since they can include the costs of delivery in their selling price instead of charging it separately. This however, should be seen in relation with the importance of the selling price and willingness to pay.
Anoek Dina Vinke S1779664
9
gross margins for retailers, while products with zero or high shipping fees, lead to substantially higher gross product prices. This result can be explained by the earlier mentioned “zero-risk bias”. The conclusion can be that delivery costs are negatively related to the utility consumers derive from it. Since people are more sensitive to free offerings and experience the difference between zero and one cent bigger than the difference between one cent and two cents, it is expected that this negative relation becomes weaker when shipment prices go up (Finucane, 2000). Also, consumers who shop more often tend to be more price-sensitive and are therefore expected to have a more negative utility for shipping costs than consumers who do not shop that often (Dolan, 1987). This leads to the following hypotheses:
H1a: Delivery costs have a negative effect on the utility derived from the fulfillment policy, H1b:The effect of delivery costs becomes less strong once the costs go up.
H1c: The effect of delivery costs will be stronger for customers who shop more often
2.1.2 Delivery time
Delivery time is the second attribute of the order fulfillment policy. To start with, delivery time is coherent with time risk. Time risk can be referred to as the extent to which buying, using, or disposing of the offering is perceived to have the potential to lead to loss of time (Hoyer et al., 2013). When consumers order online, they face the risk that the order is not delivered on time (Luo et al, 2012). This risk gets higher if consumers need their order within a specific time frame. If the product is not delivered in this promised time frame, it can cause large dissatisfaction and time loss. It can even happen consumers do not have use for the ordered product if they receive the order later than planned.
Secondly, delivery time and time risk is associated with performance risk. Performance risk can be referred to as the possibility that the offering will perform less than expected (Hoyer et al., 2013). So a combination of time and performance risk occurs when an order is not performing as expected. For example, a dress that does not fit and has to be send back.
Anoek Dina Vinke S1779664
10
have the product as soon as possible, rather than waiting for it for a long time. Rapid delivery can lead to a competitive advantage for companies and even enable them to enjoy a higher price premium and larger market share (Li and Lee, 1994). This leads to the assumption that delivery speed is very important for customers and it even makes them willing to pay more for their online orders. Therefore it is expected that the faster an order is delivered, the more benefits a consumer derives from the delivery. This leads to the following hypothesis:
H2: Delivery time has a negative effect on the utility derived from the fulfillment policy
2.1.3 Convenience of Delivery
Convenience of delivery is another aspect of the order fulfillment policy. It means that customers have a choice in where they want to have their orders to be delivered. This can be done at home, but also at a service point, which is often located in a local store. In the Netherlands many service points are located in a store that has extended opening times, like a supermarket. These extended opening times allow customers to get their online delivery after the “regular” stores are closed. Furthermore, when a service point is located at a local supermarket, customers can receive their delivery in the same place they do their grocery shopping.
Consumers have different motivations for ordering online. While some people buy online, because they do not have time to go to a brick and mortar store, others shop online, because they just do not want to drive to the store and they want to enjoy the convenience of getting their products delivered at home. The difficulty of delivering the products and the profitability of it has been acknowledged in the retail literature. A lot of research has been conducted in this field on how much to charge for home deliveries and the differences between home and pick up deliveries. (Kämäräinen et al., 2001; Punakivi et al., 2001).
Anoek Dina Vinke S1779664
11
(Kämäräinen et al., 2001).
The difference between home delivery and delivery at a service point is stated in table 1 below, which is derived from the research of Kämäräinen et al. (2001) on reception boxes in the e-grocery market.
Home delivery Service point
Personal service Yes by courier Yes by store personnel
Final Delivery point Customer Store
Customer’s dependence Has to be home Does not need to be home
E-tailer’s value offering Order fulfillment (quick delivery) “Inventory management”(enough products)
Available delivery hours Fixed timetable At any time/time window
Delivery frequency Variable, depending on
customer needs
Usually fixed
Delivery window Fixed delivery hours Operating hours
Delivery time per household
Long Short
Table 1. Differences between Home Delivery and Delivery at a Service-Point (Kämäräinen et al., 2001)
As can be seen in table 1, home delivery and delivery at a service point differ on the level of personal service, the final delivery point, the E-tailer’s value offering, available delivery hours, delivery frequency, delivery window and delivery time.
Anoek Dina Vinke S1779664
12
different delivery addresses every day and a different number of addresses to deliver to. This makes the exact time of delivery less certain and therefore broadens the time window (Lin and Lee, 2009). Since consumers differ in their preference towards the shipping method it can be concluded that for consumers having the choice between home delivery and a service point will have a higher preference if there is only the option of home delivery. Which leads to the following hypothesis: H3a: The convenience of delivery has a positive effect on the utility derived from the fulfillment policy Especially for consumers that are not at home that often, which can be caused by a busy workweek, it is expected that they will put a higher value on delivery at a service point. This is caused by the fact that they are not dependent on a delivery or bound on a fixed timetable. This leads to the following hypothesis:
H3b: The effect of the convenience of delivery is higher for people who work more often
2.2 Convenience of Payment
This section explores the expected utility that is derived from having the convenience of payment. The convenience of payment refers to having the option to choose between paying before ordering the product and the option to pay after receiving the order.
2.2.1 Trust and Risk
Anoek Dina Vinke S1779664
13
(Song and Zahedi, 2005).
So ordering online brings several sorts of risk. For this section (payment method) financial risk is applicable. This is referred to as the extent to which buying, using or disposing an offering is perceived to have the potential to create financial harm (Hoyer et al., 2013). Financial risk is created by the fact that consumers have to pay on beforehand, but endure the risk of never receiving their order and thereby losing their money.
Since people are born as being risk-averse (Frischmann, 2012), it is expected that most people will prefer having the option to pay afterwards. Risk-aversion in this study refers to the risk of ordering from an E-tailer that is not reliable. This reliability means that the E-tailer does not deliver the order as promised (Zeithalm et al., 2002). Hence, if the web shop promises that an order is delivered the next day, it has to be delivered the next day. If consumers have the option to pay afterwards they do not have to worry about the reliability of the E-tailer, because they will not have the risk of losing money when the E-tailer turns out not to be trustworthy. On the other hand, if consumers have bought before at a certain web shop and the E-tailer did what he promised to do, it might be less important for consumers to have the option to pay later on. At least, from a trust perspective. Consumers can also have other reasons for wanting to have an afterwards payment option.
In general, the seller of a product, in this case a web shop, would like to receive the payment sooner than later. While on the other hand, the consumer frequently seeks to delay making these kinds of decisions in order to avoid risk. Therefore some consumers like to delay their payment, even though the purchase decision has already been made. This can be explained by the fact that time influences how consumers rate the goodness or badness of outcomes (Mowen and Mowen, 1991). It is not surprising that some selling techniques are focused on delaying payments. Holding off the obligation to pay right away, gives consumers the gratification of receiving the offer, while delaying the financing. This is explained by the time and outcome valuation model of Mowen and Mowen (1991). In the context of shopping online, on the one hand, a time lag between ordering and receiving enhances the risk and therefore decreases the chance of ordering. While on the other hand, the option of delaying the payment until after you received the product, reduces the risk and therefore increases the chance of ordering (Mowen, 1992). So in order to reduce the risk and thereby increasing the chance of placing the order, E-tailers should offer the option for consumers to pay afterwards.
Anoek Dina Vinke S1779664
14
effect of when a payment must be made for a product on purchase intentions. Present-time oriented consumers have a preference towards paying sooner than later, while future-time oriented consumers want to delay their payment as long as possible. For online shopping this would mean that these future-time oriented consumers will probably put a significant value on whether they have the opportunity to pay afterwards. They might not even order the product if they do not have this opportunity. Therefore, it is expected that consumers that are future-time oriented, derive a higher utility from the convenience of payment.
2.2.3 The pain of payment
Other research contradicts the preference for paying afterwards. The idea of “the pain of payment” explains that consumers experience pain in spending their hard earned money (Hahn et al., 2012; Soman, 2003; Soman, 2001; Prelec and Loewenstein, 1998) and it can be referred to as the emotion that consumers experience in parting with their money. Thinking about the costs of a purchase undermines the pleasure derived from it, while thinking about the benefits of a purchase can blunt the pain of having to pay for it (Soman, 2003). Prelec and Loewenstein (1998) were the founders of this concept and came up with a model that predicts a preference of consumers for flat rate pricing. Flat rate pricing means that a fixed price is charged no matter how much is used (Essegaier et al., 2002). For example, a fixed monthly price for unlimited internet usage on your smartphone or in this study, a fixed delivery price no matter how much you order.
Anoek Dina Vinke S1779664
15
payment can also occur and can influence the degree of pain.
Applying the withdrawing money from an ATM example to consumers who pay on beforehand in an online store, there can be concluded that this leads to a higher chance of decoupling the thoughts of paying from the pleasure derived from the consumption of the product(s). Furthermore, consumers have a strong debt aversion (Prelec and Loewenstein, 1998) and find debt unpleasant. When consumers get rid of debt it feels good, especially when the purchase turns out to be an unpleasant one (Prelec and Loewenstein, 1998). On the other hand, consumers could also want to pay after they receive their online order can be that they cannot afford it at the moment. They could either start saving or pay later (Hahn et al., 2012). When they choose to pay afterwards it allows them to have the desired product immediately, while extending the payment. So consumers have to outweigh the need and desire to want the product with the aversion of being in debt with the online shop.
When looking at consumers who would rather pay with their credit card, but see this as a monthly payment which leads to decoupling and a smaller “pain of payment”, there can be concluded that having the option to pay afterwards is still interesting. There are consumers who can decrease the “pain of payment” when they are able to pay afterwards. Furthermore, consumers who paid by credit card rather than check seemed to experience less of a pain and hence were more willing to incur a given expense (Soman, 2001). Credit card users are even willing to spend more compared to when they have to pay by check (Prelec and Simester, 2001). This leads to the assumption that it is very interesting for E-tailers to offer an option to pay afterwards, since this could lead to consumers spending more on their web shop.
Consumers are also constantly outweighing the utility derived between the pain of payment and the pleasure deriving from the consumption of the product. In other words, when ordering a product online they ask themselves; "How much is this pleasure costing me?" (Prelec and Loewenstein, 1998). From an hedonic perspective, consumers want to minimize the thoughts of paying and therefore rather pay afterwards. They do however, need to know how much they have to pay. When they have the option to pay afterwards, they do not unduly think about the costs and will probably be more likely to purchase. Or they just do not have the patience to wait until they have the money to order the desired product online and therefore want to pay afterwards. This impatience can be explained by the fact that people prefer a positive outcome sooner rather than later.
Anoek Dina Vinke S1779664
16
consumption of the product that is ordered online (Shafir and Thaler, 2006). So having the obligation to pay before receiving an order during online shopping could be influenced by whether consumers would like to savor their pleasure or if they are just impatient to receive it even though they do not have the money. Furthermore, the results from the study of Hahn et al. (2012) show that product related emotions are more positive when buyers pay before they consume a product. Paying afterwards sometimes even leads to more negative related product experienced emotions.
Overall, it can be concluded that the payment method is expected to be strongly influenced by consumer characteristics. Firstly, lack of trust in E-tailers make consumers want to delay their payment. They do not want to pay before they actually have seen and felt the product they ordered (Büttner and Göritz, 2008; Nicalaou et al., 2013; Song and Zahedi, 2005) . But once consumers are experienced with online shopping, they will probably be less risk averse for paying on beforehand. On the other hand the “pain of payment” shows that decoupling thoughts of paying for the product from consuming the product is in general easier for consumers who pay on beforehand (Prelec and Loewenstein, 1998). Furthermore, consumers tend to be debt-averse and are therefore expected to be more willing to pay on beforehand. Or they want to savor their pleasure, by paying on beforehand and enjoy consuming the product even more. However, it could also be the case that they just not have the money at the moment and do not want to wait before they saved enough. This leads to the choice to pay afterwards so that they can enjoy the pleasure of consuming the product as soon as possible(Hahn et al., 2012).
Another factor that may influence the preference in payment method is the frequency of buying online. It is positively related to online shopping tendency and negatively related to the likelihood to abort an online transaction (Zhou et al., 2007). So the more often consumers shop online, the less likely it becomes that they do not pursue their order. Shopping online more often also decreases the financial risk of buying online, since the shoppers are familiar with the concept (Bhatnagar et al., 2000). Therefore it is expected that frequency of buying online has a moderating effect on the preference for the paying method.
All this taken together leads to the following hypothesis:
H4a: The convenience of payment has a positive effect on the utility derived from the fulfillment policy
Anoek Dina Vinke S1779664
17
the web shop, will probably trust the E-tailer more and perceive less risk. Therefore the following hypotheses is formed:
H4b: The effect of the convenience of payment will be lower for people who shop online more often Consumers can be categorized in future-time oriented and present-time oriented consumers. As explained before, future-time oriented consumers are expected to have a high preference for paying their order after they received it. This leads to the following hypothesis:
H4c: The effect of convenience of payment is higher for people who are more future-time oriented
Finally, it can be concluded that decoupling the thoughts of payment and pleasure is easier for consumers if they have already paid for it before they can use it (Pralac and Loewenstein, 1998; Hahn et al., 2012). Therefore it can be said that in the light of the "pain of paying" paying on beforehand is more preferred by consumers and makes the option of having the choice to pay on beforehand or afterwards less important than the other attributes of the fulfillment policy. However, the payment method characteristics affect payment use more than the demographics attributes of the consumers who conduct the transactions (Stuh and Stavins, 2012). The characteristics of the payment method refer to the costs that are associated with paying online. So if paying afterwards is associated with extra costs, this influences the preference in paying on beforehand or afterwards. This is however not researched in this study, because it outside the scope of this paper.
2.3 Price
Anoek Dina Vinke S1779664
18
any information as long as the marginal gains from the search are higher than the marginal costs (Sen et al., 2006). Even though search cost are low on the internet, consumers will still only compare web shop prices as long as the search costs are lower than the benefits they derive from having a lower price.
Of course the price sensitivity also depends on the reference price of consumers. This can be internal as well as external. The external reference price is provided by the marketer, while the internal reference price is the price consumers associate with a product or service (Hardesty and Suter, 2005). Also, the purchase environment has been long established to have an influence on a consumers’ price decision. Thaler (1985) already suggested that the price perceptions of consumers go up when they were told to buy cold beverages near a hot beach. This can also count for the online environment. From a logical point of view and extracting this from the research of Thaler (1985) it would be expected that consumers will expect a higher price when the product ordered online is delivered within a day. Or the other way around, expecting a lower price when it takes a long time before the product is delivered. Furthermore, consumers already have a lower price perception when comparing online stores to traditional brick-and-mortar stores. In other words, they expect to pay less when visiting an online store than when visiting an offline store (Hardesty and Suter, 2005). Overall, consumers are expected to rather pay less than more for their product. This leads to the following hypothesis.
H5:Price has a negative effect on the utility derived from the fulfillment policy
2.4 Moderators
As mentioned in the introduction, segmenting the market is always important. Segmenting the market enables a company to gain deep knowledge about what the consumers in that market want and desire. This helps companies to decide which segment(s) they can serve best, which segment(s) are most profitable for the company and how they should target them (Leeflang, 2003). Some people value a quick delivery, which leads to paying the premium price, while others can wait a reasonable amount of days if this means they can buy the item(s) for a bargain (Boyer and Frohlich, 2006). In general these “good deal” seekers tend to have a more modest income in proportion to the number of family members in their household. So large households are expected to value the cheapest way of ordering online, especially when this is combined with a modest income.
Anoek Dina Vinke S1779664
19
Therefore, early research was not able to find any significant difference among online shoppers. Nowadays, this age gap has become smaller, but the effect of age on online shopping is still unclear (Zhou et al., 2007). Even though recent research is not clear about the effect, it would still be interesting to see whether age has an effect on the formation of the segments based on the utility people derive from the order fulfillment process.
Another interesting moderator that can be distinguished is gender. According to Bhatnagar et al. (2000), men have greater confidence when it comes to online buying. This is especially true for products like hardware and software. Furthermore Zhou et al. (2007) show that men make more online purchases than women and spend more when shopping online. Women also tend to be more skeptical towards e-business than men. All this leads to the expectation that gender will have a significant effect on the preference of the attributes that make the fulfillment policy. Especially the preference for payment method is expected to differ between men and women. Men will probably have greater confidence in the web shop and are therefore more willing to pay before receiving their product. As mentioned above, men buy hard- and software more often and have greater confidence in this product group, therefore it is also interesting to see in what kind of product group consumers shop online and in what product category most often.
Anoek Dina Vinke S1779664
20
3. Conceptual model
From the theoretical framework, several hypotheses were formed. The model below displays these expected effects of the different attributes on the fulfillment policy and their moderating effects.
Anoek Dina Vinke S1779664
21
3.1 Hypotheses
As a clarification, the hypotheses that will be tested are displayed here again.
Delivery cost
H1a: Delivery costs have a negative effect on the utility derived from the fulfillment policy, H1b:The effect of delivery costs becomes less strong once the costs go up.
H1c: The effect of delivery costs will be stronger for customers who shop online more often Delivery time
H2: Delivery time has a negative effect on the utility derived from the fulfillment policy Convenience of delivery
H3a: The convenience of delivery has a positive effect on the utility derived from the fulfillment policy H3b: The effect of the convenience of delivery is higher for people who work more often
Convenience of payment
H4a: The convenience of payment has a positive effect on the utility derived from the fulfillment policy
H4b: The effect of the convenience of payment will be lower for people who shop online more often H4c: The effect of convenience of payment is higher for people who are more future-time oriented Price
Anoek Dina Vinke S1779664
22
4. Research Design
In this chapter, the research design of the study will be discussed. A research design can be described as a framework or blueprint for conducting the research (Malhotra, 2010). This chapter will describe the procedures necessary for obtaining the information that is needed for the research discussed. First the data collection and description will be discussed, followed by the study design and ending with the plan of analysis.
4.1 Data description
This empirical study focuses on the fulfillment policy of online web shops. In order to obtain the data that can be used to do the analysis, a survey is used. The questionnaire can be found in appendix 11.4. It starts by asking for some general demographics, like age, gender and income. After this, several questions regarding the moderators and shopping behavior are asked. The last part of the questionnaire contains the choice-based conjoint questions.
Sample
Choice-based conjoint analysis can only be done when there are at least 200 respondents (Hair et al., 2010), therefore snowball sampling is used. Doing it in another way is not possible due to lack of resources. Snowball sampling is a nonprobability sampling technique in which an initial group of respondents is selected, and afterwards is asked to find others who belong to the target population of interest (Malhotra, 2010). In this study the initial group of respondents consists of family, friends and colleagues of the researcher. These people are reached by email and social media like Facebook and Twitter. Since anyone who speaks Dutch and shops online is allowed to fill in the questionnaire, this initial group of respondents has been asked to forward it to anyone they know. For the analysis, 231 people filled in the survey. However, after deleting the missing values, only 203 respondents were left for analysis. These missing values were found in the choice based conjoint questions as well as the other questions. Therefore, they needed to be deleted. Fortunately, the minimum requirement of 200 respondents to conduct a choice-based conjoint analysis is met.
Measures
Anoek Dina Vinke S1779664
23
Demographics; age, gender, income, household size, household composition and education.
Plumer (1974) distinguishes several characteristics of demographics: age, gender, education, income, occupation, family size, geography and city size. Therefore, these demographics are included in this study. The demographics age and household size were asked in a numeric way. In other words, respondents could fill in the number themselves. The other demographic questions were asked on ordinal scales that are commonly used in Dutch surveys. As an example, income is asked on a scale from below, equal or higher than the average income in the Netherlands.
Moderators; Work hours per week, frequency of online shopping and time orientation.
The moderator of how often a consumer works a week is set on an ordinal scale of five options. These are based on the standard hours that could be worked in a week in the Netherlands. The other moderator, frequency of online shopping, is set on an ordinal four point scale. These vary between weekly and shopping online less than once a month. By conducting a pre-test among ten people with different ages, genders, incomes and household compositions the scale was tested on validity. The last moderator was carried out on a five point likert scale. Respondents were asked to fill in whether they focused more on the future or on the present in their daily life. This was also pre-tested in the same way as the frequency of online shopping.
Other variables; Shopping preference (Offline vs. Online), likability of offline and online shopping, online shopping category, internet usage in hours and days, shopping hours in general, online and offline, spending behavior in general, online and offline, online shopping allocation (for whom), last time bought online.
These questions are set on an ordinal scale, which were also tested in a pre-test and extracted from the research of Hong and Kim (2012). The variables were chosen in order to get more insights in the respondents' shopping behavior. This enables the researcher to see if the respondents are actually shopping online and whether there is a difference in the fulfillment policy between consumers who shop more often online or offline. Furthermore, the analyses will show if these variables have an influence on the different segments with regard to the fulfillment policy.
Anoek Dina Vinke S1779664
24
be used to conduct an ordinary least square analysis. This is a control mechanism for the outcomes of the choice-based conjoint analysis as well as a method to test the moderators.
4.2 Conjoint analysis
The conjoint analysis part of the survey is discussed in this sub-section. Conjoint analysis is a multivariate technique developed specifically to understand how respondents develop preferences for any type of object. Objects can include products, services, but also ideas. It is based on the simple premises that consumers evaluate the value of an object by combining the separate amounts of value provided by each attribute (Hair et al., 2010). This value can be described as the utility. It is a subjective judgment or preference unique to each individual, which is the most fundamental concept in conjoint analysis and the conceptual basis for measuring value (Hair et al., 2010). There are several types of conjoint analyses. This study uses choice-based conjoint analysis. Traditional conjoint analysis assumes that the judgment task, based on ranking or ratings, captures the choice of the respondent. However, this is not the most realistic way of depicting a respondent’s actual decision process. Furthermore, there is a lack of formal theory linking these measured judgments to choice (Hair et al., 2010). Choice-based conjoint analysis has the inherent face validity of asking the respondents to choose a full profile from a set of alternative profiles known as a choice set. It is much more representative of the actual process of selecting a product from a set of competing products. Normally, this type of conjoint analysis includes a none-option, which is also an advantage of the choice-based conjoint analysis. However, this study does not include a none-option, because this research investigates a choice in which the choice of buying the product has already been made and the respondents only have to choose between the different fulfillment options. Even though this leads to not being able to determine the willingness to pay, it still enables the calculation of the equalization price. Which in other words, is the price consumers are willing to pay for better conditions in the fulfillment policy.
Anoek Dina Vinke S1779664
25
workdays, with €2.50 delivery cost, choice between home delivery and delivery at a service point, payment only possible on beforehand and a product price of €49.95. Each scenario that is shown at the respondents represent a different fulfillment policy. The last step is deciding on how many profiles should be evaluated. The number of profiles that must be evaluated by each respondent can be calculated by the total number of levels across all factors – numbers of factors + 1 = minimum number of profiles according to Hair et al. (2010).
This study uses twelve different profiles and one hold out profile. The design of the profiles are generated by Sawtooth SSI web 6.6.6. According to Hair et al. (2010) the minimum number of profiles that should be used are 9 (13-5+1=9). However, if nine profiles would be used, it would lead to too many profiles that would have one obvious winner and respondents would most likely always choose this option. This would not lead to very insightful findings. Moreover, if the respondents were to answer nine profiles, the questionnaire would become too long and complex. The choice for using twelve profile is created by Sawtooth SSI web 6.6.6. By entering the attributes and levels, the minimum number of profiles is generated based on the efficiency scheme (Appendix 11.2). If only one version of the questionnaire would be used, respondents would have to answer at least ten choice sets. This is too complex for the respondents and therefore two versions were used in order to improve the efficiency. The two versions are randomly assigned to the 203 respondents. This makes six choice sets sufficient to create efficiencies higher than 95%.
Anoek Dina Vinke S1779664
26
Delivery costs € 0.00 € 2.50 € 5.00
Delivery time 1 day 2-3 work days 1 work
week Convenience of
Delivery
Home delivery
Choice between home delivery and delivery at a Service point
Convenience of payment
On
beforehand
Choice between on Beforehand or Afterwards
Price € 49.95 € 51.95 € 53.95
Table 2. Attributes and its levels
Anoek Dina Vinke S1779664
27
5. Results
This chapter will discuss the results of the analyses. It will start with exploring the data by explaining the descriptive variables. Then the results of the choice based conjoint analysis will be explained. This starts with an aggregate model and will end with three chosen segments which ultimately will
determine if the hypotheses can be accepted or need to be rejected. Lastly, an ordinary least square analysis will be shown, which will give additional insight in the moderators.
5.1 Descriptive variables
Before analyzing the data thoroughly, it is important to get an understanding about the data. This can be done by looking at the descriptive variables. From the 203 respondents in this study, 96 are male (46.3%) and 107 (53.7%) are female. So gender is almost equally distributed. Age ranges from 10 to 72 years old with an average of 31 years (table 3). The households where the respondents live in, range from 1 to 15 people, but 89.7% live in a household from 1 to 4 people. From all the respondents, 83.3% live in the north of the Netherlands and 103 people (50.7%) live in the city and 98 people (49.3%) live in a village. When looking at the income of the respondents, there can be seen that more than half (58.6%) has an income that is less than the average in the Netherlands, 11.8% has an income that is about average and 21.2% has an income that is more than €33.000,-. Even though most of the respondents have an income less than average, 35% works more than 32 hours a week, but is this is followed by 33% that work less than 8 hours a week (table 4).
Table 3. Age distribution
Working hours a week Freq. Percentage
Less than 8 hours 67 33%
Between 8 and 12 hours 16 7.9%
Between 12 and 20 hours 21 10.3% Between 20 and 32 hours 28 13.8% More than 32 hours (Fulltime) 71 35%
Total 203 100%
Anoek Dina Vinke S1779664
28
When looking at the shopping behavior of the respondents in offline and online shopping, there can be looked upon where they shop most often, how much they enjoy shopping offline and online, in what product category they shop most often online, how much time and money they spend on shopping and finally for whom they shop most often when shopping online.
First of all, more respondents shop more often in offline shops than online shops. 51.7% shops more often in offline shops, while only 23.2% shops more often online. The other 24.6% says they spend about the same time shopping online as offline (table 5).
I shop Freq. Percentage
Mostly in Offline shops 37 18.2%
More often in Offline shops 68 33.5%
As often In Offline as well as in Online shops 50 24.6%
More often in Online shops 32 15.8%
Mostly in Online shops 15 7.4%
Table 5. Offline vs. Online shopping
There is not much difference in the enjoyment the respondents feel for offline or online shopping. Both types are not enjoyed by about one fifth of the respondents. 55.7% really enjoys shopping offline and 50.7% really enjoys shopping online (table 6).
Enjoy Offline shopping freq. Percentage Online shopping Freq. Percentage
Strongly disagree 9 4.5% 10 4.9%
Disagree 30 14.9% 33 16.3%
Neither agree or disagree 50 24.9% 57 28.1%
Agree 80 39.8% 88 43.3%
Stronglyagree 32 15.9% 15 7.4%
Total 201 100% 203 100%
Table 6. Enjoyment offline and online shopping
There is however, a difference between the time the respondents spend shopping offline and online. Whereas almost 40.4% spends less than 5 hours a month on shopping offline, 71.4% spends less than 5 hours on online shopping (table 7). So in conclusion, the respondents spend more time on offline shopping than online shopping. This is not that surprising, since online shopping has lower search costs, lower waiting queues and is just a few clicks away. Whereas offline shopping requires consumers to go to the physical store, search in the store and wait in line to pay.
Average time spend in Offline shopping
Freq. Percentage Average time spend in Online shopping
Freq. Percentage
Less than 5 hours a month 82 40.4% 145 71.4%
5-10 hours a month 72 35.5% 37 18.2%
10-15 hours a month 28 13.8% 10 4.9%
15-20 hours a month 12 5.9% 4 2%
More than 20 hours a month 8 3.9% 5 2.5%
Total 202 99.5% 201 99%
Anoek Dina Vinke S1779664
29
The monetary value spent offline is not that high. About 70% does not spend more than €100,- a month (table 8). However, it is still higher than what is spend online. The average spending offline per month is about €85,-, while the average spending online is about €51,- per month (table 8 ).
Monetary value offline shopping per month Freq. Percentage Monetary value online shopping per month Freq. Percentage Less than €50,- 70 34.5% 129 63.5% Between €50,- and €100,- 71 35% 52 25.6% Between €100,- and €150,- 33 16.3% 16 7.9% Between €150,- and €200,- 15 7.4% 2 1% Between €200,- and €250,- 4 2% 2 1% More than €250,- 10 4.9% 2 1% Total 203 100% 203 100%
Table 8. Offline and online monetary value
Finally, from the figures below (fig 1,2 and 3), there can beseen that almost 90% of the respondents mostly buy online for themselves, followed by 7% who buy for friends and 4% that buy for their partner. The most frequently bought product categories are clearly books and clothes. Finally, for most of the consumers (63) €100 to €249,- is the highest amount of money they have ever spent online.
87% 1% 7% 1% 4% 0%
For whom is bought most
often
Myself Friends Family Partner Children Acquaintances 118 22 10 32 54 62 79 45 14 114 Less than €20,- €20,- to €99,- €100,- to €249,- €250,- to €500,- More than €500,- 8 49 63 39 44Most ever spent online
Number of respondents
Fig. 1 Most ever spent online
Anoek Dina Vinke S1779664
30
Respondents were also asked to declare how often they shop online and when the last time was they bought something online. As can be seen from table 9, 81.3% have bought within the last 3 months. The frequency of shopping however, is quite low. Almost half (49.8%) buys less than once a month (table 10). However, there is still a considerable amount of respondents that buys more often than once a month.
Time since last online purchase Freq. Percentage
This week 43 21.2%
This month 68 33.5% Last 3 months 54 26.6% 3-6 months ago 20 9.9% Longer than 6 months ago 17 8.4%
Total 202 99.5%
Besides from using full profile cards to determine the importance of the attributes, respondents were also asked to rate on a scale from 1 to 5 how important they find the attributes separately from each other. This is displayed in table 11. From the table, there can be seen that price of the product is considered most important (more than 80% is in the category above neutral), then delivery cost (more than 60%), then convenience of delivery (slightly more than 60%) and finally delivery time (about 60%). Surprisingly, more respondents consider convenience of payment less important than important. More than 50% answered that they found it not at all or slightly important. This is not in line with the results from the choice-based conjoint analysis, which is explained further on in this paper.
Level of importance Price Delivery cost Delivery time Convenience of delivery
Convenience of payment Freq. Perc. Freq. Perc. Freq. Perc. Freq. Perc. Freq. Perc.
Not at all important 7 3.4% 9 4.4% 6 3% 9 4.4% 48 23.6%
Slightly important 7 3.4% 14 6.9% 21 10.3% 23 11.3% 57 28.1% Neutral 20 9.9% 39 19.2% 47 23.2% 50 24.6% 36 17.7% Moderately important 85 41.9% 87 42.9% 75 36.9% 74 36.5% 36 17.7% Very important 83 40.9% 53 26.1% 53 26.1% 46 22.7% 25 12.3% Total 202 99.5% 202 99.5% 202 99.5% 202 99.5% 202 99.5%
Frequency of online shopping Freq. Percentage More than 4 times a month 14 6.9% 2-4 times a week 30 14.8% 1-2 times a week 58 28.6% Less than once a month 101 49.8%
Total 203 100%
Table 10. Frequency of online shopping Table 9. Recency online shopping
Anoek Dina Vinke S1779664
31
5.2 Choice-Based Conjoint Analysis
This part of the results explains the outcomes of the choice-based conjoint analysis. It starts by exploring the aggregate model and ends with the analysis of the segments.
5.2.1 Aggregate model
For the choice-based conjoint analysis, the attributes asked in the questionnaire will be used. These are price, delivery cost, delivery time, delivery method and payment method. From the aggregate model, the parameters belonging to these attributes determine the corresponding utilities. These utilities show if the attributes consisting of three levels should be treated as linear or part-worth. By putting the parameters in a graph, there can easily be seen if the attributes are linear. This is done by estimating a one class model without including any covariates.
Even though price and delivery cost show a little bend, they can still be treated as linear. When including both variables as linear in the model, the degrees of freedom increase and the fit of the model improves as can be seen from table 12. The BIC as well as the CAIC values go down and the degrees of freedom go up. Therefore, the two attributes will be treated as linear.
Table 12. Model fit aggregate model
-1 -0,5 0 0,5 1 €49,95 €51,95 €53,95
Price
Utility -2 -1 0 1 2 €0,- €2,50 €5,00Delivery cost
Utility1 class model BIC(LL) CAIC(LL) Degrees of freedom Part-worth Linear 1855.6149 1849.3941 1863.6149 1855.3941 195 197
Graph 1. Part-worth price
Anoek Dina Vinke S1779664
32
Delivery time shows a very linear line, but the levels belonging to this attribute do not have the same distance between them and therefore, the attribute should still be treated as nominal. It is also of importance to take a look upon the internal hit rate. When randomly predicting the choices of the respondents this would be 33.33%. For the aggregate model this is 68.8% ((270+234+334)/3). So it does a better job than the random prediction.
After setting the first two variables as numeric the following preferences for the attributes are given.
As can be seen from the graph above (graph 4), respondents consider the delivery cost relative to the other attributes as very important. If they were to be asked to place a value on every attribute adding up to 100, they would give delivery cost about 43 points. Delivery time has a value of 27.67, price 20.84 and convenience of payment 7.45. The convenience of delivery is considered to be not that important, because it has a relative importance of less than 1%.
-1 0 1
Next day 2-3 work days 1 work week
Delivery time
Utility 0 0,1 0,2 0,3 0,4 0,5 Price Convenience of payment Convenience of delivery Delivery Cost Delivery Time 0.2084 0.0745 0.0077 0.4327 0.2767Relative importance of the attributes
Graph 3. Part-worth delivery time
