WHY WE BUY
THE WAY WE BUY
Influencing the online
consumer decision making
process
MASTER THESIS
ABSTRACT
People do not behave the way they should behave based on rational thought, we behave irrational in all parts of our lives. Whether personally, socially or professionally; we are influenced by the way choices are presented to us. Which means, the person (or organisation) who has control over the choice architecture of our choices, has control over the outcome of our choice.
A.J. (Tommie) Dijstelbloem, 5855063
April 2015 Universiteit van Amsterdam Amsterdam Business School Executive Programme in Management Studies Strategic Marketing Management Prof. Dr. Ed Peelen
Statement of Originality
This document is written by Student Antonius Johannes Dijstelbloem who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
“We are all far less rational in our decision making than standard economic theory assumes. Our irrational behaviors are neither random nor senseless—they are systematic and predictable.”
Table of Contents
Abstract … … … p. 6
Chapter 1: Introduction … … … p. 8
1.1. Research objectives
Chapter 2: Literature review … … … p. 12
2.1 An important shift in the retailing landscape: Brick-and-click 2.2 Irrationality in the consumer decision making process
2.3 Different ways to apply framing in the choice architecture 2.4 To ship or not to ship, that’s the question
2.5 Theoretical framework and hypotheses
2.5.1 Role of default framing in shipping options 2.5.2 Hypotheses
Chapter 3: Data and methods … … … p. 24
3.1 Research design 3.2 Data collection
3.3 Methods of analysis and justification of the chosen methodology
3.3.1 Chi-square test 3.3.2 Regression analysis
Chapter 4: Results and discussion … … … p. 34
4.1 Chi-square test 4.2 Regression analysis 4.3 Hypothesis 1 4.4 Hypothesis 2 4.5 Hypothesis 3 4.6 Hypothesis 4 4.7 Results summary
4.8 Discussion
4.8.1 The influence of the use of default options
4.8.2 The influence of charging costs for delivery (and returns) 4.8.3 General discussion based on survey feedback and experiences
Chapter 5 Conclusions … … … p. 52
5.1 Conclusive findings
5.2. Evaluation of the research question and practical implications 5.3. Research limitations and suggestions for future research
Reflecting on 8 years of college … … … p. 58
Bibliography … … … … … p. 60
Appendix … … … … … … p. 64
SPSS data-output Online survey
Abstract
People do not behave the way they should behave based on rational thought, we behave irrational in all parts of our lives. Whether personally, socially or professionally; we are influenced by the way choices are presented to us. Which means, the person (or organisation) who has control over the choice architecture of our choices, has control over the outcome of our choice.
This premise is further investigated in this research to see to what extent the preference of consumers in an online retail environment is influenced by the use of default options in the choice architecture and by charging additional costs for shipping and returns.
For brick-and-click retailers this can be interesting seeing as, in the Dutch market, roughly 20% of all costs involved in online shopping consist of the cost for shipping and returns, varying between € 5,- and € 10,- for shipping and between € 10,- and € 20,- for returns. Amounting to a total cost of € 250 million for the sector.
This research, conducted among 271 respondents with similar backgrounds, has shown that the preferences of online consumers toward ‘delivery’ or ‘in store pick up’ are significantly influenced by the use of default options and the use of charging additional costs for shipping and returns.
I set out to discover a more consumer friendly way for brick-and-click retailers to reduce the costs involved with the delivery of online purchases. Seeing as some tactics nowadays are simply to discourage online consumers to return their products when they do not live up to their expectations.
This research has shown that consumers can be influenced in their choices concerned with the delivery of an online purchase. Now it is up to the retailers to use this knowledge, within the ethical limits, in order to reduce their costs for shipping and returns.
Chapter 1: Introduction
The introduction of the WorldWideWeb by Tim Berners-Lee in 1990 was the beginning of a fundamental change in the way we purchase products (InstantShift, 2010). E-commerce started taking consumer purchases from the physical world into the digital world. In 1995 Amazon.com started out and reached maturity in 2003, when it published its first yearly profit.
The total turnover of online shopping in the Netherlands in 2014 is estimated by the Thuiswinkel Markt Monitor (GFK, Post nl, Thuiswinkel.org) to be € 12 billion, which represents a 16% share of the overall consumer purchase market. With almost 10 million online consumers, the average amount spent shopping online is € 1.200,- per person per year. The total amount of transactions for 2014 is expected to be 120 million, averaging € 100,- spent on every online purchase. In ten years’ time the market has increased almost fivefold, total turnover 2005: € 2,8 billion; 6 million online consumers; average amount spend: € 441,- per person. But not all that glimmers in the online retail market is gold.
The growing numbers of returns is an important issue for online retailers (Financieele Dagblad, april 29th 2014). One of the biggest players in the Dutch online marketplace is
Zalando, originally a German webshop for shoes, clothing and fashion accessories. In 2013 they shipped over 12 million orders, with multiple purchases per order, only to have 50% of their products returned. Other online retailers, such as H&M and Zara, do not communicate their return percentages, but it is believed that the clothing industry has the highest average of all product categories, about 35%. In comparison, electronics and books keep returns under 10%.
According to Van Ossel (Omnichannel in Retail, 2014) online retailers spent on average € 5,- to € 10,- on delivery and € 10,- to € 20,- on returns, amounting to 20% of all costs for retailers spent on sending and returning consumer purchases. The Thuiswinkel Markt Monitor estimates the share of shipping costs for online retailers in 2014 at almost 2%. Adjusted for the market share of physical products (over services) based on spending (55%), shipping cost are responsible for 2.8% of the total online shopping turnover in the Netherlands in 2014, amounting to a total of € 250 million. Both statistics considered, it is no wonder that more and more online retailers look for ways to reduce the number of returns.
1.1. Research objectives
This research is aimed at contributing to literature and practice by experimenting with different methods of framing in the online consumer decision making process. In order to see if online consumers can be directed into picking up their online purchases in store instead of having their purchases delivered (at home). Which, as will be set out more precise in the upcoming paragraphs, can save online retailers a lot of costs otherwise concerned with shipping (and returns). In doing so opening the way for additional research to enhance the theoretical understanding and also its potential in practice.
The experiment will focus on the online purchase of products, seeing as the market share of products bought in the online retail market in the Netherlands is estimated by the Thuiswinkel Markt Monitor to be 80% (services making up the remaining 20%). And, more importantly, products –by nature- require shipping before enabling consumers to enjoy the advantages of their newest addition.
In order to reduce returns, at this moment, online retailers deploy several measures ranging from improving the product information made available to the consumer online to less consumer friendly ones such as making the return process more difficult by having consumers print out their own return forms (Financieele Dagblad, april 29th 2014). This
research tries to discover different, more consumer friendly, measures to influence the preferences of consumers in an online retail environment. In such a way that both the consumer and the online retailer are able to profit, by investigating the influence of the use of default options and by charging additional costs for shipping and returns.
Research question
To what extent is the preference of consumers in an online retail environment influenced by the use of default options in the choice architecture and by charging additional costs for shipping and returns?
Chapter 2 is dedicated to the provision and synthesis of existing (research)literature and also to the elaboration and justification of the hypotheses posed. In chapter 3 the research design, data collection, methods of analysis and a justification of the chosen methodology is presented. In chapter 4 the empirical results of the survey is provided. Leading to the
conclusive findings in chapter 5, providing retailers with practical suggestions to reduce returns of online purchases as well as suggestions for additional research.
Chapter 2: Literature review
2.1 An important shift in the retailing landscape: Brick-and-click
Every 50 years or so, retailing undergoes a disruption (Rigby, 2011). A century and a half ago, it was the growth of big cities and the rise of railroad networks that made possible the modern department store (brick and mortar). With the introduction of mass-produced automobiles 50 years later shopping malls soon were dotting the newly forming suburbs, effectively challenging these city-based department stores. In the 1960s and 1970s it were the discount stores undermining the old-style mall. Rigby (2011) goes on stating that each of these waves of change does not eliminate what came before it, but reshapes the landscape and redefines consumers expectations. And retailers relying on earlier formats either adapt or die out.
Like most of these disruptions, the digital technology got off to a shaky start. When internet-based retailers sprouted in the 1990s, they could pretty much run wild until the combination of ill-conceived strategies, speculative gambles and a slowing economy burst the dot-com bubble. In April 2000, over a four day period, the Nasdaq plummeted 600 points (from 4.188 to 3.321) a staggering 21% (CNN Money, 2000). The ensuing collapse wiped out half of all e-commerce retailers and provoked an abrupt shift from irrational exuberance to economic reality (Rigby, 2011). But nowadays it is widely accepted today that electronic channels (click) have the potential to become an important alternative avenue for firms to reach their customers and generate sales (Moe, 2003).
Online shopping has also been on the rise in the Netherlands. The industry is worth billions (€ 10,6 billion) with a market share of 10,5%. According to the Centraal Bureau voor de Statistiek (CBS, 2013), a Dutch governmental institution which gathers statistical information about the Netherlands, revenues in the online retail market has seen yearly double-digit growth from 2002 to 2011, even when the traditional (e.g. brick and mortar) retail market showed negative growth in 2005, 2009 and 2010, respectfully: -0,4; -4,1; -0,2%. Only slowing down a little to 8,5%, in 2013. With almost 10 million consumers shopping online in 2014, spending an average of € 1.200,- at one of the 45.000 online retail stores accumulating in a € 12 billion turnover, this disruption is here to stay.
However, physical stores remain the foundation of retailing. As evidenced by research of AT Kearney (2014) in the US retail market: 90% of all retail sales are transacted in stores and 95% of all retail sales are captured by retailers with a brick-and-mortar presence. Meaning
who shop online use a physical store before or after the transaction. Concluding that retailers with a brick-and-mortar presence collectively sell more on their websites than pure online retailers.
Ofek et al. (2011) state that many established retailers with a strong traditional brick and mortar presence have ventured into the online world since the stock market crash of 2000 wiped out many internet retailers. Allowing them to showcase their products online and appeal to consumers who value the convenience of online shopping, combine their brick and mortar presence with an online one resulting in a brick-and-click concept. But now they face the hidden costs of product returns, particularly evidenced in product categories where consumers need to ‘touch and feel’ the product to determine if it fits their needs and taste. These hidden costs encompass among other things, the collection, refurbishing and restocking of the unwanted product. Recent data showed that overall consumer returns are estimated at about 8.7% of retail sales, online retailers are faced with significantly higher number ranging from 18% up to 35%, depending on the category.
Research institute Ipsos shows that one in every two online purchase in the Netherlands is either clothing (30%) or consumer electronics (20%). Ofek et al. (2011) conclude that it is very difficult to judge the aesthetic appeal of the design, texture and colour of clothing without physically seeing it as opposed to viewing a digital image of it. Many relevant attributes for consumer decision making are ‘non-digital’ as they call it and therefore very difficult to communicate via a computer screen. On the other hand, the attributes of consumer electronics lend themselves better to be communicated digitally and as a result the returns in the first category are higher than in the second one.
Brick-and-click
• The stock crash of 2000 wiped out many online-only retailers, clearing the way for
brick-and-mortar retailers to venture in the online world. Resulting in the brick-and-click concept;
• But the hidden costs of product returns by online consumers are high, it is important for retailers to reduce these numbers, which can be as high as 35%;
• The two most important product categories for online purchases in the Netherlands are clothing (30%) and consumers electronics (20%);
• Attributes for clothing are less well communicated digitally than the attributes for consumers electronics, leading to a higher percentage of returns for the first product category.
2.2 Irrationality in the consumer decision making process
As outlined by Bettman et al. (1991) consumers constantly make decisions regarding the choice, purchase and use of products and services. The way in which consumers make these decisions is not only important to themselves but also to marketers. They determine a number of elements concerned in the consumer decision making process: the elements that compose a choice; alternatives; attributes of value; uncertainties; availability of information, both in terms of content (what is available) and structure (how is it organized); and other factors that may influence a consumer’ response to a choice.
Bettman et al. (1991) differentiate between two approaches of consumer decision making. One argues that consumers are rational beings, assuming that a consumer obtains complete information of all the alternatives and selects the most optimal alternative to suit their personal circumstances. This Homo Economicus, or the economic man, was coined by the godfather of economists John Stuart Mill. Who described in his essay ‘On the definition of political economy’ (1848) described a hypothetical subject, with narrow and well-defined motives which proved useful in economic analysis. The essence of his usefulness for economic models did not lay in the outcome of his choices but in the rational method in which his choices were made (Persky, 1995).
Therefore they are not able to be perfectly rational but decision makers try to make the best of the limitations to which they are subjected. With their Prospect Theory (1979) Kahneman and Tversky go even further pointing out that the (consumer) decision making process is not only influenced by structure but also by the human perception. Showing that by implementing small changes in the wording of the decision choice, for instance by pointing out a loss or a gain, different outcomes can be achieved.
In his book ‘Predictably Irrational, the hidden forces that shape our decisions’ (2008) Ariely shows us what happened when he came across an offer to subscribe to The Economist. The offer contained three choices (i) an online subscription, for $59,00, (ii) a print subscription, for $125,00 or (iii) a combined online and print subscription, for $125,00. To him this second choice seemed odd, because, as the economic man theory dictates, consumer always choose the option which is most economically attractive. Thus, having two offers costing the same amount of money while one obliviously offers more is illogical. To test his presumption he asked two (identical) groups of 100 MIT-students to give their preference. The first group choose as follows (i) 16, (ii) 0 and (iii) 84. Which, of course, delighted Ariely.
The option which was deemed illogical to begin with, and was turned down by everybody in the first group of students, was removed. So for the second group of students there remained a choice between two subscriptions. Surprisingly, in this situation, where nothing had changed, except the removal of un unused and economically unfavourable option, these students chose for (i) 68 and (iii) 32. Leading Ariely to state that our mind is wired a specific way. Namely, we are always looking at the things around us in relation to others. Where people tend to not only compare things with one another but in addition also tend to focus on comparing things that are easily comparable. Like subscriptions (ii) and (iii) are comparable on price and thus leading for the students to choose the most value for their money. With the removal of the unused option (ii) the reference point of students changed. The students were no longer able to compare the subscription based on price caused their reference point to change to look at the usability of the product. This change of reference point caused a drastic change in the outcome of the second experiment. Because now the majority of the students favoured accessibility of information and thus preferred the unlimited digital access over the limited access of receiving a hard copy at their home address. This finding has become known as the decoy-effect.
This choice-experiment is a representation how preferences can be altered based on the way the choices are formulated. Where value should be treated as a function in two arguments: the asset position that serves as a reference point and the magnitude of the change (either positive or negative) from that reference point. Confirming Kahneman and Tversky (1979) that a change of reference point alters the preference order for prospects.
Consumer decision making process
• The consumer decision making process contains several elements, such as the
organization of information;
• Consumers have bounded rationality and therefore endure limitations to process
information;
• The decision making process can be influenced by human perception: applying small
changes in the decision choice can lead to different outcomes.
2.3 Different ways to apply framing in the choice architecture
Decisions are made in an environment where many features (can) form an influence. This environment is created by a choice architect, this is someone who indirectly influences the choices other people make by organizing the context in which people make decisions. The choice architecture can be used to ‘nudge’ people a certain way, without forcing certain outcomes upon anyone. In their book Nudge (2008) Thaler and Sunstein suggest as a rule of thumb to assume that everything matters. They sketch six principles of what they consider to be good choice architecture. Being: iNcentives, Understand mappings, Defaults, Gove feedback, Expect error, structure complex choices, i.e. NUDGES. According to the authors, with keeping these principles in mind, choice architects can nudge people a certain way.
Hansen and Jespersen (2013) suggest that nudging is not necessarily about manipulation, nor necessarily about influencing choice. As they present a framework identifying four types of nudges. In order to do so, they define a nudge as: any attempt at influencing behavior in a predictable way without forbidding any previously available courses of actions or making alternatives appreciably more costly in terms of time, trouble, social sanctions, and so forth. In other words, nudging is not necessarily about the manipulation of choice. Furthermore Hansen and Jespersen suggest a distinction between two types of
thinking. But while type 2 nudges are aimed at influencing the attention and premises of, and hence the behavior anchored in, reflective thinking (i.e. choices), via influencing the automatic system, type 1 nudges are aimed at influencing the behaviour maintained by automatic thinking, or consequences thereof without involving reflective thinking. And claim that manipulation, whether of choice or behavior, does not necessarily work. They produce a matrix delineating four different types of nudges, given the two distinctions developed (type 1 and type 2 nudges) and epistemic transparency and non-transparency, as shown in figure 1: Transparent type 2 nudges
This type of nudge intervention engages the reflective system in a way that makes it easy for the citizen to reconstruct the intentions and means by which behaviour change is pursued. An example is the fly-in-the-urinal intervention.
Transparent type 1 nudges
For this type of nudges reflective thinking is not engaged in what causes the behavior change in question. Rather, reflective thinking occurs as a by-product, but in a way that easily allows for the reconstruction of ends and means. A paradigm case of this type of nudges is the playing of relaxing music while passengers board a plane in order to calm them.
Non-transparent type 2 nudges
For this type of nudges to be successful, the reflective system has to be engaged, but
it does not happen in a way that by itself gives people epistemic access to the intentions and means by which influence is pursued. A paradigm example is the clever framing of risks aimed to influence one’s decision-making, e.g. when choosing between medical treatments.
Non-transparent type 1 nudges
This type of nudges cause behaviour change without engaging the reflective system and in a way that does not make it likely to be recognized and transparent. An example is that by reducing the size of plates in a cafeteria leads people to serve and eat 22 % less calories.
Fig 1. Different types of nudges as constituted by Thaler and Sunstein
One of these principles, and the one which is unavoidable in a choice architecture system because in that system there has to be decided what happens when no choice is made, is the use of a default option or ‘padding the path of least resistance’. For any given choice, there is a default option. Meaning, the option that will obtain if the person choosing does nothing. For a variety of reasons -from laziness to fear and distraction- it is to be expected that a large number of people end up with the default option. And this behavioural tendency toward doing nothing can be reinforced if the default option comes with the suggestion (implicit or explicit) that it represents the normal or recommended course of action.
To illustrate this Johnson and Goldstein (2004) researched the effect of a default option on consent rates for organ donations in a number of European countries. In their example different countries have chosen different default options for the decision to become an organ donor. Some countries required explicit consent (opt-in), others presumed consent (opt-out) to become donors. It revealed a dramatic affect with donation rates being almost twice as high when the default choice was to opt-out instead of opt-in. So default options make a large
maintaining an opt-in default. They accompanied the creation of a national register with an extensive educational campaign and mass mailing, in which they send out over 12 million letters (on a total of 15,8 million citizens), asking people to register. But failed to change the effective consent rate. Leaving Johnson and Goldstein to conclude that the idea that preferences are constructed provide an important alternative to views that incentives are required to increase the rate of donation. Showing that defaults not only make a difference in what is chosen, they can also make decisions easier.
Johnson et al. (2012) elaborates on the use of defaults defining it as one of the most powerful (and popular) tools available to the choice architect. A default determines the way consumers encounter a decision choice, for example whether or not to pick up an online purchase at the store or have it delivered (at home). On electronic forms a default can take the shape of a pre-checked box.
Choice architecture
• Decisions are made in an (online) environment where many features can form an
influence. This environment is created by a choice architect, someone who (indirectly) influences the choices people make by organizing the context in which the decisions are made (i.e. the choice architecture);
• This choice architecture can be used to nudge people a certain way. There are six
different types of nudges, such as the use of defaults;
• A default determines the way (online) consumers encounter a decision choice, for
example whether or not to pick up an online purchase at the store or have it delivered (at home);
• The default is the most powerful tool available to the choice architect. On electronic forms a default can take the shape of a pre-checked box.
2.4 To ship or not to ship, that’s the question
One of the big issues for online retailers, some even call it the Holy Grail (Lewis et al., 2006), is understanding the connection between online consumer decision making process and shipping costs. Because, in contrast to traditional retailing, when transactions take place with a distance between customers and products the firm incurs the costs of order and assembly, delivery and returns.
The growing numbers of returns is an important issue for online retailers (Financieele Dagblad, april 29th 2014). One of the biggest players in the online marketplace is Zalando,
originally a German webshop for shoes, clothing and fashion accessories. In 2013 they shipped over 12 million orders, with multiple purchases per order, only to have 50% of their products returned. Other online retailers, such as H&M and Zara, do not communicate their return percentages, but it is believed that the clothing industry has the highest average of all product categories, about 35%. In comparison, electronics and books keep returns under 10%.
According to Van Ossel (Omnichannel in Retail, 2014) online retailers spent on average € 5,- to € 10,- on delivery and € 10,- to € 20,- on returns, amounting to 20% of all costs for retailers spent on sending and returning consumer purchases. This does not cover the costs, seeing as at this moment online retailers are already accustomed to use the cost of shipping as a marketing tool. According to Frankwatching (2012) the prices range from free to € 1,95 (industry standard) up to € 4,95 (which is considered high).
The Thuiswinkel Markt Monitor estimates the share of shipping costs for online retailers in 2014 at almost 2%. Adjusted for the market share of physical products (over services) based on spending (55%), shipping cost are responsible for 2.8% of the total online shopping turnover in the Netherlands in 2014, amounting to a total of € 250 million. Therefore, a key marketing decision for online retailers is how to save money on delivery costs and how to reduce returns to a minimum.
As Chatterjee (2011) pointed out, there has been a considerable amount of research regarding the framing effect of shipping costs. The way the shipping costs are framed is particularly relevant in a remote channel such as an online shop, a channel in which the consumer pays a base price for the product they have set their eye on and a surcharge for the delivery of the product (at home). This research focusses on the framing effect, in the online consumer decision making process, of a default option in the choice architecture of the ordering process (in this case, for the pick up or delivery option). By varying the default option between ‘product pick up at the store’, ‘product delivered (at home)’ and by having no default for different research groups it is possible to see if the (different) default option is significantly preferred over the other option as compared to the base reading of the research group where no default option will be selected.
Shipping options
• Understanding the connection between the online consumer decision making process
and shipping costs (deliveries and returns) is considered by some as the Holy Grail;
• Online retailers spent on average € 5,- to € 10,- on delivery and € 10,- to € 20,- on returns,
amounting to 20% of all costs;
• One of the biggest online clothing retailer sees 50% of the purchases of online
consumers returning. The returns of consumer electronics are kept under 10%;
• Previous research has shown that consumers display different preferences based on the framing effect of shipping costs.
• This research will show if the use of a default option as a framing effect in the choice architecture can influence the outcome of the online consumer decision making process.
2.5 Theoretical framework and hypotheses
The dependent variables in this research are the default option, as a part of the choice architecture, and the use of additional costs for shipping and returns, in the online ordering process. These are expected to moderate the relation between the consumer decision process and the outcome (i.e. the consumer choice), as shown in figure 2. A major difference between this research and previous ones, is that it examines the effect of default options in the pick up or delivery stage of the online ordering process.
Fig. 2. The relationship between the consumer decision making process and the consumer choice is moderated by the choice architecture
2.5.1 Role of default framing in shipping options
This research will focus on the two main product categories in the online retail market: clothing and consumer electronics. These two product categories combined represent 50% of the purchases in the online retail market. Also, these product categories differ substantially
from one another: clothing is a hedonistic or ‘touch and feel’-product, whereas, electronics is a utilitarian product. Which can be bought based solely on the technical specifications.
Based on the average online purchase, € 100,- according to the Thuiswinkel Markt Monitor 2014, three different price levels will be used which are the same for both categories: € 20,- (low), € 100,- (regular), € 180,- (high). The same goes for the shipping costs, which averages about € 1,95 according to Frankwatching (2012). The low option will be € 0,- and the high option will be € 4,95.
In total there will be three different research groups. For all of whom the above settings will be equal. The only differences between these groups is that group 1 will have ‘pick up at the store’ as the default option. Group 2 will have ‘delivery (at home)’ as the default option. Group 3 will be the control group, which will have no default option selected.
The selection of the total number of items in the shopping basket will be the same for both categories within the same price level. So chances are limited that other factors than the default option will play a role, such as averaging the shipping costs over the number of items in the shopping basket.
2.5.2 Hypotheses
Decisions, made in an online environment, can be influenced in many ways. Literature has shown the most powerful one is the use of default options.
H1a: The use of the default option ‘in store pick up’ will result in a higher consumer preference
for ‘in store pick up’ of their online purchase compared to the preferences of the control group
H1b: The use of the default option ‘delivery (at home)’ will result in a higher consumer
preference for ‘delivery (at home)’ of their online purchase compared to the preferences of the control group
A substantial distinction between clothing and electronics is that the first is a hedonistic product and the second a utilitarian product. Clothing needs to be touched and felt, before purchasing. Whereas electronics can be bought solely based on the technical specifications.
H2a: Consumers prefer to pick up their online clothing purchases in store
H2b: Consumers prefer to have their online consumer electronic purchases delivered (at
Online retailers spent up to 20% of all costs on shipping (and returns). Previous research has shown preferences of consumers tend to differ based on the framing effect of shipping costs.
H3a: Charging additional cost for shipping and returns for ‘delivery (at home)’ for online
clothing purchases will result in a higher consumer preference for ‘in store pick up’
H3b: Charging additional cost for shipping and returns for ‘delivery (at home)’ for online
consumer electronics purchases will result in a higher consumer preference for ‘in store pick up’
As the total amount of the online purchase(s) goes up, the (relative) height of the additional costs for shipping (and returns) will go down.
H4a: Online consumers are more inclined to pay shipping costs for high price level purchases
than for low price level purchases
H4b: Online consumers are more inclined to have their online purchases delivered (at home)
Chapter 3: Data and methods
3.1 Research design
The research design is fixed and quantitative. Fixed, because the variables are based on theory which has been discussed in chapter 2. And quantitative, because numerical data from observable behaviour of three research groups is collected and statistically analysed in order to generalise –unbiased- the specific behavioural results for a larger population. This research is confirmatory of nature.
The research is designed to see to what extent the preferences of online consumers toward ‘delivery’ or ‘in store pick up’ vary (i) between significantly different product groups (clothing and consumer electronics) and/ or (ii) when delivery cost are being charged for ‘delivery’ and/ or (iii) when different amounts of delivery costs are being charged (iv) when the option for ‘delivery’ or ‘in store pick up’ is preselected (default). When influenced by the use of default options and the use of charging additional costs for shipping and returns.
The impact of the first three variables can be measured within one research group. In order to do so six sets of questions are posed to the respondent (the online consumer) about the preference to have their online purchase to be delivered at home or to be picked up in store. The six sets are divided along two variables: product group (with varying delivery costs) and price level. Two distinct product groups have been selected: clothing and consumer electronics. Together they form 50% of the total of products which are bought online (Ofek et al., 2011). Also, the characteristics of these two product groups differ significantly. With one (clothing) being a touch and feel product which, according to theory at least, online consumers prefer to try out (check colours in different lighting and fit to determine the right size) before proceeding toward purchasing. Whilst the other (consumer electronics) are chosen based on their technical capabilities, which can be determined from the specific technical information provided online, as presented in figure 3.
Fig. 3 The two different product groups: product group 1 Clothing (Hedonistic), product group 2 Electronics (Utilitarian)
For the two product groups a different amount of delivery costs is being used: in both ‘free or no shipping costs are being charged in the first stage of an online purchase but in the second stage € 1,95 (industry standard) is being charged for the shipping costs of clothing and € 4,95 (high) for the shipping costs of consumer electronics. In the third stage the respondent is being made aware of the return policy of the store, which charges of € 1,95 to return clothing purchases by mail and € 4,95 to return a consumer electronics purchase by mail, as presented in figure 4.
Fig. 4 The two different levels of delivery (and returns) costs: product group 1 Clothing uses delivery costs of € 1,95, product group 2 Electronics uses delivery costs of € 4,95
The other variable within these two different product groups is the price level. Three different price levels have been introduced: low price level (€ 20,-), average price level (€ 100,-), high price level (€ 180,-). The average price level has been derived from the Thuis Winkel Monitor (2014), which shows that the average online purchase amounts to € 100,-, as presented in figure 5.
Fig. 5 The three different price levels within one product group: price level ‘Low’: € 20,-, price level ‘Average‘: € 100,-, price level ‘High’: € 100,-
The impact of the fourth variable (default) can only be measured when comparing the answers to the same questions by similar research groups with different default options. For this purpose a total of three research groups have been set. The questions of the first group ‘GD’
(‘Geen Default’ in Dutch or ‘No Default’ in English) did not show a preselected option. The questions of the second group ‘DB’ (‘Default Bezorgen’, ‘Default Delivery’) showed a preselected option to online consumers to have their purchases delivered (at home). The questions of the third group ‘DA’ (‘Default Afhalen, ‘Default In store pick up’) showed a preselected option to online consumers to pick up their purchase in store. In this setup the first group functions as the control group, in this respondents are not being ‘nudged’ either toward the option which favours ‘delivery’ nor the option which favours ‘in store pick up’. The results of this group is the normal behaviour. The results of the second and third group will show a deviation of these results, either towards a preference of respondents to have their online purchases delivered (group 2) or a preference to pick up their purchases at the store itself. The deviation from the first group of the second and/ or third group will be the impact of the ‘nudge’, which is the pre-selected option, as presented in figure 6.
Fig. 6 The three different research groups, based on the presence of a default option: research group 1 ‘No Default’, research group 2: ‘Default Delivery’, research group 3: ‘Default pick Up’
3.2 Data collection
The quantitative data needed to test the hypothesis of this research is collected through online surveys. For every research group 80 respondents participated, with three different
research groups, the total amount of respondents participating is 240. In order to reach this amount a combination of the accidental sampling method and the snowball method is used, both of which are a form of non-probability sampling. The justification for this approach lies in the fact that the target audience needed to be collected, with minimal resources, in a short period of time. According to Powell (1997), the advantage of the accidental sampling method is that the sample needed can be drawn from a population which can be accessed directly and easily. Respondents are asked to participate in one of the three different surveys via e-mail and selection is based on a first come-first served basis, no preferential selection was made between respondents. According to Goodman (1961), the advantage of the snowball sampling method is that each individual respondent in the sample is asked to ask others (including but not limited to his/ hers best friends, family members and colleagues) to also participate. This procedure is repeated until all three research groups had sufficient respondents. In total, three e-mails have been sent to a target audience, consisting of friends, family members and colleagues, of 80 (accidental sampling method). In these e-mails the target audience was asked to do the same (snowball sampling method). The combination of these two sampling methods was successful, meaning that within two weeks of launching the online surveys the target of total of respondents needed was met and even exceeded. The final total amount of respondents was 271, with 89 respondents in research group 1 (No Default), 89 respondents in research group 2 (Default Delivery) and 93 respondents in research group 3 (Default Pick Up). None of the respondents were aware of the detailed research questions before entering the online survey. Nor were the respondents aware of the specific testing conditions most specifically of which the differences in default options between the three surveys. All of the respondents were made aware that only one of the three surveys needed to be conducted. Some questions were asked about the underlying reasons for this, these were not disclosed until after finishing the online survey. By (partly) disclosing the underlying reasons respondents were discouraged to enter multiple online surveys. To ensure a sufficient amount of respondents within a limited time period, the respondents were made aware of the maximum amount of time which would be spent on filling out the online survey, by mentioning that the maximum time span would be 7 minutes. This maximum time span was set based on several test runs with the surveys. In addition to myself, also four persons not involved with the setup of the research were, consecutively, asked to participate in a test run.
amongst the target audience, these test runs were used to obtain insight in a number of ambiguities in the research in general and some questions in particular and have led to a reformulation of these questions in order to make it immediately clear to each respondent what was expected from them.
In order to conduct the online survey an online research tool was used: Qualtrics. Qualtrics is a US-based software company, founded in 2002. Their software enables researchers to collect data online and is used in both professional and academic journals (Albaum, 2006).
3.3 Methods of analysis and justification of the chosen methodology
The goal of this thesis is to research the link between a number of variables concerning consumer preferences towards online shopping. A connection is suggested between the independent (or predictor) variables, such as price level, cost of shipping (and returns) and product groups, on the one hand at the dependent (or outcome) variable, online purchase being delivered (at home) or picked up in the store. The dependent variable depends on the independent variable, in such a way that changes in (one of) the independent variables will lead to a (significant) change in the dependent variable. This research is aimed at uncovering the details surrounding the (suspected) influence of the independent variable(s) on the dependent variable. The research question is divided into different, testable hypotheses. Every hypothesis is presented in such a way that an effect is present. This type of hypotheses are known as alternative hypothesis (Field, 2009).
3.3.1 Chi-square test
The independent and dependent variables used in this research are a categorical. To be able to see if a relationship between these exist the Pearson’s chi-square test (Pearson, 1900) is used. As shown by Field (2009) this statistic is based on the (simple) idea of comparing frequencies in certain categories (independent variables, such as price level, cost of shipping and product groups) to the frequencies in other categories (preferences of online consumers towards their purchases being delivered of pick up at the store). Because categorical variables are not continuous this type of data is not distributed normally. The Pearson’s chi-square test has two assumptions: (i) the different data entries are independent, each entry represents one cell in the cross tab. And (ii) the expected frequencies is greater than 5, if the amount of
frequencies is below this number it results in a loss of statistical power which leads to possible failure to detect a genuine effect between the variables.
With the help of the IBM SPSS Statistics Software edition 20 the Pearson’s chi-square test examines if there is an (significant) relationship between two categorical variables. For this to be the case the significance value must be less than ,05.
In the next chapter the results of the Pearson’s chi-square test will be reported as follows: the value of this test statistic, its degrees of freedom, the significance value and the cross tab. The test statistic is denoted as χ2.
3.3.2 Regression analysis
When it is established that a (significant) relationship between the variables exists, the next step in the analysis is to find out if one variable can be predicted by looking at the other. According to Field (2009), this is the essence of regression analysis: to use a model to predict the values of the dependent variable from one or more independent variables in the collected data. Regression analysis is used to predict the value of the dependent variable from one (of the) independent variable(s). In this research there are several predictor variables, such as price level, cost of shipping (and returns) and product groups, thusly multiple regression analysis will be used. With this method I’ll be able to test the test the assumptions regarding the expected influence of the use of default options and charging additional costs in the choice architecture of an online retail environment. These assumptions are based on the literature review and set out in the hypotheses.
In order to measure the adequacy of this analysis we look at the relative improvement of the prediction following the usage of the regression analysis. This value is R2 and will be
used to show the variance in the outcome which is explained by the regression analysis. The curve in the regression analysis is also called a coefficient and will be represented as b: the change in outcome resulting from a unit change in the predictor (Field, 2009). If a predictor variable (significantly) predicts the outcome of the outcome variable it has a b-value different from zero. This difference can be both positive and negative, seeing as it represents the change in the outcome variable resulting from a one unit change in the predictor variable. A one unit change in the predictor variable can both lead to an increase of the outcome variable as well as a decrease of the outcome variable. This is tested using a test. With the
t-two variables. If the significance is less than ,05 then we can assume the b is (significantly) different from 0 i.e. the predictor variable is significant in predicting the outcome variable. To assess the contribution of predictor variables in logistic regression the Wald statistic is used, which shows whether the b-value is significantly different from 0. In this research multiple predictor variables are used. For this reason a method of cross validation is used known as adjusted R2, which not only accountants for the number of respondents but also the number
of predictor variables.
The variables, both independent and dependent, in this research are categorical. This implies that an extension of the regression analysis needs to be used, also known as logistic regression: a multiple regression with an outcome variable which is categorical and predictor variables which are either continuous or categorical (Field, 2009). This particular variant of regression analysis shows which of two possible categories will be chosen, in this research this means that the online consumer will prefer to have their purchase either delivered (at home) or to be picked up at the store. Seeing as there are only these two categorical outcomes the specific logistic regression analysis used is known as binary logistic regression. In binary logistic regression a variant of b is used, the odds ratio or Exp(B): this is an indicator of the change in odds which results from a one unit change in the predictor variable and is used for the interpretation of the type of regression analysis.
In the next chapter the results of the binary logistic regression analysis test will be reported as follows: the Wald statistic, adjusted R2 and Exp(B).
Chapter 4: Results and discussion
The data collection has led to a total of 271 respondents in total, spread over the three different research groups with (i) no default option, (ii) delivery as the default option and (iii) in store pick up as the default option. Overall 50 of these were incomplete, bringing the total useful data set to 221, the results of the analyses are presented and discussed in this chapter.
4.1 Chi-square test
As established in the previous chapter, Pearson’s chi-square test is used to see if an association between categorical variables exist (Field, 2009). For this research it is important to establish if the outcome variable Delivery is dependent on the predictor variables Price Level (there are three price levels used in this research: low € 20,-; average € 100,-; high € 180,-) and/ or Cost (there are three different amounts of total costs involved when respondents made the choice to have their online purchase(s) delivered (at home) or not: € 0,-; € 1,95; € 3,90 for the product category Clothing and € 0,-; € 4,95; € 9,90 for the product category Electronics). If the significance of the test-statistic is small enough, then we can assume the predictor variables and the outcome variable are in, in some way, related. The test statistic is denoted as χ2. Also
its degrees of freedom and its significance are reported below, for both of the product categories.
Clothing Pearson’s chi-square value df Significance
Price Level 1,904 2 ,387
Cost 119,596 2 ,000
Table 1. In the product category Clothing a relationship between consumer choice and Cost exists
Within the product category Clothing the predictor variable Price Level has a χ2 of 1,904 and
a significance of ,387. The significance has to be less than ,05, as shown in table 1. This means that Price Level as a predictor variable and the outcome variable Delivery are independent of one another. The predictor variable Cost has a χ2 of 119,596 and a significance of ,000. This
means that Cost as a predictor variable and the outcome variable Delivery are related. Which means that for the product category Clothing it can be assumed that the decision making process, of respondents in this research, is influenced by the height of the additional costs
charged for shipping (and returns). But the preference of online consumers is not influenced by the height of the total amount of the online clothing purchases.
Electronics Pearson’s chi-square value df Significance
Price Level 225,425 2 ,000
Cost 123,148 2 ,000
Table 2. In the product category Electronics a relationship between consumer choice and Price Level and Cost exists
Within the product category Electronics the predictor variable Price Level has a χ2 of 225,425
and a significance of ,000. This means that Price Level as a predictor variable and the outcome variable Delivery are related. The predictor variable Cost has a χ2 of 123,148 and a significance
of ,000, as shown in table 2. This means that Cost as a predictor variable and the outcome variable Delivery are also dependent. Which means that for the product category Electronics it can be assumed that the decision making process, of respondents in this research, is both influenced by the total amount of the online electronics purchase as well as the height of the additional costs charged for shipping (and returns).
It is established that a significant relationship between the variables exists. In the following paragraphs, among other things, the results of the regression analysis are presented. In order to see if the hypothesis are supported, partially supported or not supported.
4.2 Regression analysis
As established in the previous chapter, the regression analysis is used to predict the value of the dependent variable from one (of the) independent variable(s). In this research: price level, cost of shipping (and returns) and product groups. Below the Wald, Significance and Exp(B) are reported, for each product category in all three different research groups.
Clothing Wald Significance Exp(B)
1. Cost 20,515 ,000 ,610
2. Cost 35,745 ,000 ,505
3. Cost 59,844 ,000 ,429
Comb. Price Level 3,263 ,071 ,998
Comb. Cost 65,260 ,000 ,380
Table 3. In the product category Clothing the consumer choice is influenced heavily by Cost
The regression analysis shows that the predictor variable Price Level is not significant in the product category Clothing, as shown in table 3. This means that the different total costs of the online clothing purchases do not influence the preferences of the respondents in this research. The additional costs of shipping and returns do influence the preferences. Also, this effect is greater when the default option to pick up their online purchases in store is pre-selected.
Electronics Wald Significance Exp(B)
1. Price Level 69,053 ,000 1,012 1. Cost 18,238 ,000 ,621 2. Price Level 67,131 ,000 2,530 2. Cost 47,445 ,000 ,441 3. Price Level 64,131 ,000 1,011 3. Cost 34,250 ,000 ,525
Comb. Price Level 6,187 ,013 1,003
Comb. Cost 132,078 ,000 ,249
Table 4. In the product category Electronics the consumer choice is influenced heavily by Cost and somewhat by Price Level
For the product category Electronics both the use of default options and the use of charging additional costs for shipping and returns influence the preferences, of the respondents in this research. In the first (no default) and third (default in store pick up) research group the influence of the variable Price Level is modest. But in the second (default delivery (at home))
influence of the use of charging additional costs for shipping and returns is significant, as shown in table 4.
Overall, it can be stated that of the two predictor variables, Price Level and Cost (of shipping and return), the additional costs charged for shipping and returns is the most influential. The first only has a modest effect and only in the product category Electronics. For every additional unit (€) in the total cost of the online purchase, a total of 0,3% more respondents, in this research, prefer to have their purchase delivered. The second predictor variable however has a significant influence on the preference of the respondents, in this research. For every additional unit (€) charged, for cost of shipping (and returns), 62% of the respondents prefer to pick up their online clothing purchases in store and even 75,1% of the respondents regarding their online electronics purchase.
4.3 Hypothesis 1
In the online survey, respondents were asked about their preferences about online shopping and specifically whether they preferred home delivery of their purchases or preferred to pick these up themselves in the store. In order to find out if there are any differences between the preferences when different default options are used, three different research groups were used. Research Group 1 did not have a default option preselected. Whereas Research Group 2 and 3 did have a default option preselected, respectively a default towards having their online purchases delivered (at home) in the second group and a default towards picking up their online purchases in store. The literature review showed a preference of online consumers towards sticking with the preselected default option. Which has led to the following hypotheses:
H1a: The use of the default option ‘in store pick up’ will result in a higher consumer preference for ‘in store pick up’ of their online purchase compared to the preferences of the control group
H1b: The use of the default option ‘delivery (at home)’ will result in a higher consumer preference for ‘delivery (at home)’ of their online purchase compared to the preferences of the control group
The data set showed a clear preference of consumers towards wanting to pick up their online purchase(s) in store when a default option had preselected the ‘in store pick up’ option. Within the product group Clothing, respondents in the control group (research group 1, which did not have any default option), tended to pick up their online purchases in store 32% of the time (34% at the low price level of € 20,- and 30% at the high price level of € 180,-). In research group 3, which had the default option where ‘in store pick up’ was preselected, the Exp(B) for predictor variable Cost is ,429 (predictor variable Price Level was not significant in either of the research groups in this product category), compared to research group 1 with Exp(B) ,610. Meaning more respondents chose to pick up their online clothing purchases in store in this research group than within the control group. Within the product group Electronics, respondents in the control group, tended to pick up their online purchase in store 35% of the time. What has to mentioned, the volatility of this preference between the low price level and the high price level was very high, receptively 65% and 5%. Seeing as the high price level purchase in the Electronics product category was a table model refrigerator, the preference towards having this purchase delivered (at home) was impacted highly. Nevertheless, also in this product category research group 3 scored better on the Exp(B) than the control group did, respectively ,525 and ,621 (the differences for the predictor variable Cost are negligible for either product group). Meaning that research group 3 preferred to pick up their online purchase(s) in store more in comparison with the control group. Therefore, hypothesis H1a is
supported.
Also the data set showed a preference of consumers wanting to have their online purchase(s) delivered (at home) when a default option had preselected the ‘Delivery (at home)’ option. Within the product group Clothing, respondents in the control group, tended to have their online purchases delivered (at home) 55% of the time (48% at the low price level and 60% at the high price level). In research group 2, which had the default option where ‘delivery (at home)’ was preselected, the Exp(B) for predictor variable Cost is ,505, compared to research group 1 with Exp(B) ,610. Which means less respondents choose to have their online clothing purchases delivered (at home) in this research group than within the control group. Nevertheless within the product group Electronics, respondents in the control group, tended to have delivered their online purchase (at home) 54,5% of the time (29% at the low price
‘delivery (at home)’ was preselected, the Exp(B) for predictor variable Price Level is 2,530, compared to research group 1 with Exp(B) 1,012. Which means much more respondents choose to have their online electronics purchase delivered (at home) in this research group than within the control group. Therefore, hypothesis H1a is partially supported.
4.4 Hypothesis 2
In the online survey, respondents were asked about their preferences about online shopping and specifically whether they preferred home delivery of their purchases or preferred to pick these up themselves in the store. In order to find out if there are any differences between the preferences for specific product categories, two distinct product categories were selected. The literature review showed a preference of online consumers towards picking up ‘touch and feel’-type (or hedonistic) products, such as clothing. Whilst for purely utilitarian products, such as electronics, a preference for delivery (at home) was implied. Which has led to the following hypotheses:
H2a: Consumers prefer to pick up their online clothing purchases in store
H2b: Consumers prefer to have their online electronic purchase delivered (at home)
In this research no clear preference is showed of consumers wanting to pick up their online clothing purchases in store. If anything, they prefer to have their online clothing purchases to be delivered (at home). A total of 31% prefers to pick up their online clothing purchases in the store, regardless if shipping costs and/ or costs for returns are being charged. A total of 56% prefers to have their online clothing purchases to be delivered (at home), even if shipping costs and costs for returns are being charged. The remaining 13% prefers to have their online clothing purchases delivered (at home), even when shipping costs are being charged. When, additionally, costs for returns are being charged they prefer to pick up their online clothing purchases in store. Therefore, hypothesis H2a is not supported.
For online electronics purchase a clear preference towards delivery (at home) is showed for the consumers in this research. A majority of 52% prefers to have their online electronics purchase delivered (at home), regardless if shipping costs and/ or costs for returns are being
charged. Nevertheless a substantial part (39%) still seems to prefer to pick up their online electronics purchase in store. Looking more closely at the data shows a big difference between the price levels of the online purchases. When a low price level (€ 20,-) online electronics purchase is done 65% of the online consumers in this research prefer to pick up their online electronics purchase in store. However, when a high price level (€ 180,-) online electronics purchase is done then only 5% of the online consumers in this research prefers in store pick up and 80% prefers to have their online electronics purchase to be delivered (at home). The other 15% at the high price level online electronics purchase prefers delivery (at home) up until the point that, in addition to shipping costs (€ 4,95), costs for returns (€ 4,95) are being charged. At the average price level (€ 100,-) no clear preference either towards delivery (at home) nor in store pick up can be seen, respectively 47% and 48%. Therefore, hypothesis H2b
is partially supported, when it comes to the preference of consumers to have their high price
level online electronics purchase to be delivered (at home).
4.5 Hypothesis 3
In the online survey, respondents were asked about their preferences about online shopping and specifically whether they preferred home delivery of their purchases or preferred to pick these up themselves in the store. In order to find out what the influence of shipping costs is on their preference this question, whether they prefer delivery or in store pick up, had to be answered three consecutive times. The first time the delivery of their online purchase(s) was free delivery, the second time shipping costs were charged and the third time also shipping costs were charged if the respondent wished to return (a part of) the purchase. The literature review showed a preference of online consumers towards delivery when no additional costs were being charged and were more prone to pick up their online purchases in store when additional costs were charged. Which has led to the following hypotheses:
H3a: Charging additional cost for shipping and returns for ‘delivery (at home)’ for online clothing purchases will result in a higher consumer preference for ‘in store pick up’
H3b: Charging additional cost for shipping and returns for ‘delivery (at home)’ for online consumer electronics purchases will result in a higher consumer preference for ‘in store pick up’
For both product categories the effect of charging costs for shipping (and returns) is significant. The Wald score for Clothing is 65,260 and the significance is ,000. For Electronics Wald is 132,078 and the significance is ,000. Both scores differ greatly from 1, which means that the predictor variable (Cost) has an significant effect on the outcome variable (delivery (at home)). The impact of the predictor variable differs between the two product categories. In the product category Clothing the Exp(B) is ,380. Which means that for every extra unit in the predictor variable, the outcome variable loses 62%. In practice this will mean that every extra Euro which is being charged for costs for shipping (and returns) 68% less consumers will choose to have their online purchase delivered. For the product category Electronics this difference is even greater: 75,1%, with Exp(B) at ,249. Therefore, both hypothesis H3a and
hypothesis 3b are supported. 4.6 Hypothesis 4
In the online survey, respondents were asked about their preferences about online shopping and specifically whether they preferred home delivery of their purchases or preferred to pick these up themselves in the store. In order to find out what the influence of the price level of their online purchases is on their preference this question, whether they prefer delivery or in store pick up, had to be answered for six different purchase situations. In two product categories (Clothing and Electronics), consumers had to give their preference for three different price levels. These three different price levels were the same for both product categories. The total purchase amount for the first price level (‘Low’) came to a total of € 20,. The second price level (‘Average’) came to a total of € 100,- and the third price level (‘High’) came to a total of € 180,-. In practice, online retailer often offer free shipping upward of a certain purchase amount, ranging from anywhere between € 20,- (Bol.com) and € 150,-
(Tommy Hilfiger). Also, between the two product categories different levels of costs for shipping (and returns) are used. For clothing the average amount of € 1,95 is charged, whilst the costs for shipping and returns charged for electronics is high (€ 4,95). The literature review has showed that with the service of free shipping and returns high costs are associated. While the literature also shows people whittle away the (extra) costs. One reason may be that the extra cost for shipping (and return) is relatively smaller compared to the purchase price when this is higher. Which has led to the following hypotheses:
H4a: Online consumers are more inclined to pay shipping costs for high price level purchases than for low price level purchases
H4b: Online consumers are more inclined to have their online purchases delivered (at home) when the cost of shipping (and returns) are low
Fig. 7 Differences in preference regarding delivery and pick up between online purchases of a low- and high price level and between the average cost of shipping (and return) for clothing and high costs for electronics
In this research a clear preference is shown of consumers being willing to pay for the (extra) shipping costs (and returns) when the price level of their online purchase is higher. This goes for both product categories. Overall, there is a difference of 32 percentage points observed for the preference for delivery (at home) between online purchases at a low price level (39%) and the online purchases at a high price level (71%). The volatility of the preference between the two product groups varies highly. Were the preference for delivery (at home) for online clothing purchases differs from 48% at the low price level and 62% at the high price level, a total volatility of 24 percentage points, as shown in figure 7.
Fig. 8 Respondents preferences regarding delivery and pick up for online purchases of clothing and electronics for different price levels (Low € 20,-; Average € 100,-; High € 180,-)
The preference for delivery (at home) for online electronics purchases differs from 29% at the low price level and 80% at the high price level, a total volatility of 51 percentage points. In Figure 8 the impact of the predictor variable (Price Level) on the outcome variable (delivery) is shown. For both product categories the effect of the height of the total purchase amount is significant. The Wald score for clothing is 3,263 and the significance is ,071 (so predictors are significant with a 90% certainty), for clothing Wald is 6,187 and the significance is ,013. Both scores differ from 1, which means that the predictor variable (Price Level) has an significant effect on the outcome variable (delivery (at home)). The impact of the predictor variable differs between the two product categories. In the product category Clothing the Exp(B) is ,998. Which means that for every extra unit in the predictor variable, the outcome variable loses 0,2%. In practice this will mean that every extra Euro which is being charged for costs for shipping (and returns) 0,2% less consumers will choose to have their online purchase delivered. For the product category Electronics this difference is: ,03%, with Exp(B) at 1,003. Which implies that for every extra Euro in the total purchase amount ,03% more people chose to have their purchases delivered (at home). Therefore, hypothesis H4a is partially supported. Consumers show no clear preference towards the average cost of shipping (and return) nor the high cost of shipping (and return), respectively charged for online clothing purchases (€ 1,95) and online electronics purchases (€ 4,95). When the additional cost of shipping are
