The impact of dynamic pricing when it becomes an
industry norm
by NIENKE DOORNBOS Master Thesis University of Groningen Faculty of Economics and BusinessMSc Marketing Intelligence Supervisor: Dr. A.E. Vomberg
2
ABSTRACT
Dynamic pricing has become a very powerful source for retailers. Dynamic pricing is a pricing strategy in which prices are differentiated based on demand or supply, previous customer behaviour or on other circumstances. A proper implementation can lead to an increase of profits by 8% till 25% which makes it a very attractive pricing strategy. However, adopting dynamic pricing can also lead to negative reactions from customers. This study gives insights in the customers’ perceptions regarding price fairness, search intention, and purchase intention and two types of dynamic pricing, namely time-based and behavioural-based pricing. Also, we examine the effect whether customer respond differently towards dynamic pricing when it becomes an industry norm. Moreover, when dynamic pricing becomes an industry norm, prices differentiate every hour or per customer which leads to more uncertain prices. Customers react differently towards uncertain situations. Therefore, the risk characteristics of customers are considered when looking at the price fairness, search intention and purchase intention. An online experiment with 122 participants has been used and we found the following main points: (1) behavioural-based pricing as an industry norm has a negative effect on the purchase intention, (2) time-based pricing as an industry norm leads to a higher purchase intention, (3) risk taken customers reconsider prices as more unfair than risk averse customers but have a higher purchase intention when dynamic pricing is an industry norm.
Keywords: Industry Norm, Behavioural-based Pricing, Time-based Pricing, Price Fairness, Search Intentions, Purchase Intentions, Risk Characteristics, ANOVA, Kruskall-Wallis, Factor Analysis, Linear Regression Analysis.
3
ACKNOWLEDGEMENTS
Three months ago, my life looked very different from now. It was a normal procedure to attend lectures, talk face-to-face and meet for group assignments with students from all around the world. Due to the Corona Virus, this is might be not so normal anymore. This period helps us appreciate our freedom even more.
The University of Groningen quickly responded to this new development by giving online lectures, online group meetings, and online Q&A. We were really relying on the digital resources when writing our thesis. Therefore, I would like to thank my supervisor Arnd Vomberg. He helped me by giving online feedback and support when writing my thesis at home.
Moreover, I would like to thank all the respondents who participated in the experiment. Also, my friends and family who helped me with getting 122 respondents by spreading the online experiment with their network.
4
TABLE OF CONTENTS
1 INTRODUCTION ... 6
2 LITERATURE REVIEW ... 7
2.1 Dynamic Pricing ... 8
2.1.1 Different Types of Dynamic Pricing. ... 8
2.2 Industry Norms ... 9
2.3 Dynamic Pricing as an Industry Norm ... 10
2.4 Search Intentions ... 11
2.5 Purchase Intentions ... 12
2.6 Risk-taking behaviour as a Moderating Effect ... 13
2.7 The Type of Dynamic Pricing ... 14
3 METHODOLOGY ... 16 3.1 Study Design ... 16 3.2 Procedure ... 17 3.3 Pre-Test ... 18 3.4 Measurements ... 18 3.4.1 Price Fairness. ... 19 3.4.2 Search Intentions. ... 19 3.4.3 Purchase Intentions. ... 19 3.4.4 Risk aversion. ... 19 3.5 Analysis ... 23 4 RESULTS ... 23 4.1 Manipulation Checks ... 23 4.1.1 Industry Norm. ... 24 4.1.2 Dynamic Pricing. ... 24
4.1.3 Smaller Sample Size. ... 24
4.2 Assumptions ANOVA ... 24
4.3 High Perceptions of Price Fairness ... 25
4.4 Lower Search Intentions ... 25
4.5 Higher Purchase Intentions ... 26
4.6 Risk Characteristics as Moderator ... 26
4.7 Risk negatively influence Price Fairness ... 27
4.7.1 Interaction effects. ... 28
5
4.7.3 Smaller Sample Size. ... 29
4.8 Risk as Moderator of Search Intentions ... 31
4.8.1 Interaction Effects. ... 32
4.8.2 Main Effects.... 32
4.8.3 Smaller Sample Size. ... 32
4.9 Risk Positively moderates Purchase Intention ... 32
4.9.1 Interaction Effects and Main Effects. ... 33
4.9.2 Smaller Sample Size. ... 33
4.10 Behavioural-based and Time-based Pricing ... 33
4.10.1 Price Fairness. ... 34
4.10.2 Search Intention. ... 34
4.10.3 Purchase Intention. ... 34
4.10.4 Behavioural-based and Time-based Pricing as an Industry Norm. ... 35
5 CONCLUSION AND DISCUSSION ... 37
5.1 Findings ... 37
5.2 Future Research and Limitations ... 40
APPENDIX A: QUESTIONNAIRE ... 49
6
1
INTRODUCTION
The scenery of a typical shopping street has changed over the last years. Five years ago, more than 90% of the customers only used offline stores in order to search and purchase products (Kim, Libaque-Saenz, & Park, 2019). Nowadays customers use online and offline channels to search for more price information and to compare prices across retailers (Verhoef, Neslin, & Vroomen, 2007; Kim, Libaque-Saenz, & Park, 2019). The internet gives retailers the opportunity to generate customer data to target customers and to get insights in the customers’ shopping behaviour. Based on this new development, dynamic pricing is becoming a powerful resource for retailers to set prices for customers based on their shopping behaviour or based on the demand of a product (Grewal, et al., 2011).
7 norms, especially because it is cost-beneficial for retailers and can add value for customers (Ellickson, 2001).
The purpose of this study is to give insights in the customers’ behaviour responding to time-based and behavioural-time-based pricing when it becomes an industry norm. Our study makes three important contributions to current literature. First, our study goes beyond the fact how customers react towards dynamic pricing, it also includes the effect when it becomes an industry norm. Moreover, this study is one of the first studies that examined the effect of risk characteristics on the relationship between dynamic pricing and customers responds, such as perceived price fairness, search intention, and purchase intention. At last, this study differentiates from others by addressing two types of dynamic pricing and give insights in how customers react towards these pricing strategies regarding price fairness, search intention, and purchase intention. The current gap in literature leads to the following research question: “How do customers respond to dynamic pricing when it becomes an industry norm?”.
This study helps manager to get an understanding of the customer purchasing behaviour and explores whether it is interesting for a manager to differentiate itself from the industry norm on the aspect of dynamic pricing. Moreover, this study highlights whether it is beneficial for a manager to invest in dynamic pricing and how customers’ perceptions and behaviour may change to when dynamic pricing becomes an industry norm.
The structure of this study is as follows: the important related studies are discussed in the literature review (chapter 2), in which we discuss the definition of dynamic pricing and industry norm and the important factors of purchase behaviour. Furthermore, the methodology of our study is explained (chapter 3), which is the procedure of how our study is conducted. Afterwards we will discuss the results of our study in detail (chapter 4). Finally, the conclusion and discussion are provided, in which we reflect on the research question and the limitations of this study (chapter 5).
2
LITERATURE REVIEW
8 instance search intentions and purchase intention in section 2.4. In addition, there previous studies show that customers have a negative attitude towards dynamic pricing. Research suggest that the underlying nature of the negative response on dynamic pricing depends on the risk-taking behaviour of customers. Therefore, we explore risk-risk-taking behaviour of customers as a moderator of the customers perceptions and behaviour towards dynamic pricing in section 2.5. Furthermore, little research has been done to determine if there is a different perception of behavioural pricing strategy or time-based pricing strategy when dynamic pricing becomes an industry norm or not. Therefore, in the last section the perception of different types of dynamic pricing is described.
2.1 Dynamic Pricing
Which of the various pricing strategies a retailer should implement and invest in, depends on various aspects, for example the objectives of a retailer, the industry conditions, and the conditions of the retailer (Elgar, 2009). One of the upcoming pricing strategies is dynamic pricing, which is considered as a powerful pricing strategy for a retailer in order to increase their profits, target the right customers, and generate a competitive advantage (Grewal, et al., 2011). According to Haws and Bearden (2006) dynamic pricing can be defined as “a pricing strategy in which prices varies based on a time period, consumer or circumstances”. Within this study, generally two types of dynamic pricing are distinguished, namely behavioural-based and time-based pricing. Dynamic pricing gains popularity due to the increasing development of big data on the internet (Vaidyanathan & Baker, 2003; Kannan & Kopalle, 2001). Several industries already implemented dynamic pricing, for example the hospitality and travelling industry, while other industries and retailers are doubting whether dynamic pricing is worth the investment (Weisstein, Monroe, & Kukar-Kinney, 2013). Sahay (2007) even state that implementing dynamic pricing in a proper manner leads to a grow in the customer base and an increase of profits by 8% to 25%. However, dynamic pricing can also lead to perceptions of price unfairness, a lower purchase intention and less satisfied customers (Weisstein, Monroe, & Kukar-Kinney, 2013). These factors are especially the case when customers perceive that retailers implement dynamic pricing for their own financial benefit (Maxwell, 2002).
9 Kopalle, 2001). The second type is behavioural-based pricing. This pricing strategy is based on customers’ personal data, for instance internet browsing behaviour or by the tracking of cookies (King & Jessen, 2010). In addition, the prices differ according to time but also across customers (Kannan & Kopalle, 2001). The third type is geographical pricing. During a geographic-based pricing strategy, prices are set based on the location of a customer (Peeters & Thisse, 1996). Due to the increasing use of internet, more customers react stronger to prices set based on individual browsing behaviour or based on economic conditions (Koschate-Fischer & Wüllner, 2017) because time-based and behavioural-based pricing strategies are widely used on the internet (Koschate-Fischer & Wüllner, 2017; Vaidyanathan & Baker, 2003). Therefore, this study mainly focuses on behavioural-based and time-based pricing.
2.2 Industry Norms
Within a specific industry, retailers target the same potential customer and is therefore in a competition with other retailers to persuade potential customers. A common reaction of retailers is to start with charging different prices for products or developing new products in order to persuade potential customers. Over time, a certain strategy pattern occurs, which becomes institutionalized into a set of industry norms (Thomas & Soldow, 1988).
10
2.3 Dynamic Pricing as an Industry Norm
When an industry norm is violated, customers can response negatively towards the retailer with feelings of price unfairness. According to the economy theory, customers evaluate prices based on the economic acceptability and their own benefit (Maxwell, 2002). Feelings of price unfairness occurs when there is a difference among prices and sellers while the price of a similar retailer is justified (Xia, Monroe, & Cox, 2004). A justified price is based on the reference price which is the price that customers are expected to pay (Kahneman, Knetsch, & Thaler, 1986). When the actual price is higher than the preference price, customers get feelings of price unfairness. Only a lower price than the reference price is perceived as fair. However, the perceived price fairness differs across customers due to the subjective judgement and the purchase characteristics (Weisstein, Monroe, & Kukar-Kinney, 2013).
According to Choi & Mattila (2003) industry norms can change over time. A new efficient development or technology can trigger a change in norms, especially when it is cost-beneficial for retailers (Ellickson, 2001). Several studies argue that a proper implementation of dynamic pricing leads to growing revenue and more satisfied customers (Vaidyanathan & Baker, 2003; Kannan & Kopalle, 2001; Sahay, 2007). Therefore, dynamic pricing may change industry norms over time. Furthermore, also the involvement of other competitors or industries may trigger changing industry norms (Ellickson, 2001). An example is the implementation of dynamic pricing in the travelling industry. When dynamic pricing was adopted by some companies in the travelling industry, customers considered it as unfair. However, when dynamic pricing was used by more companies and evolve into an industry norm, it became more acceptable by consumer (Kahneman, Knetsch, & Thaler, 1986; Kimes, 1994; Kimes, 2002).
11 customers are more likely to begin to expect dynamic pricing in some industries which will be more accepted and leads to less negative reactions. Therefore, it can be concluded that familiarity with a certain price strategy in an industry increases the perceived fairness of a pricing strategy. Moreover, when retailers adopt the same pricing strategy, customers get familiar with it and get higher perceptions of price fairness. Therefore, it is expected that dynamic pricing will be accepted as fair in some industries when it becomes an industry norm. This results in the following hypothesis:
H1: “When dynamic pricing is an industry norm, there are high perceptions of price fairness.”
2.4 Search Intentions
An important activity during the pre-purchase stage is the consumer’s price searching. Price search can be divided into two aspects. The first aspect is temporal dimension which means that consumers compare prices across one store within different time periods. The second aspect is spatial dimension which means that customers compare product prices across multiple retailers (Elgar, 2009). In this study the focus is on the spatial dimension. The amount of time a customer wants to spend on searching for a reliable price depends on the searching costs. The addition gain must be higher than the expected costs which means that the amount of money saved during the search should be higher than the searching costs itself. When searching costs are high, customers are not likely to collect more information on alternatives or search at all (Elgar, 2009). Also, the amount of effort for price searching depends on the type of product and customer’s characteristics. Customers are more likely to search for price of durable products because the addition gain and price distinctions are larger. Also, customers who are price sensitive are more engaged in price searching (Kukar-Kinney, Walters, & MacKenzie, 2007). Previous study shows that customers who buys a high involvement products (e.g. washing machines, cameras, cars) are more likely to perceive stronger feelings of price unfairness and therefore, will have a higher search intention (Jin & You, 2019). For that reason, this study focuses on durable products.
12 (Haws & Bearden, 2006). In addition, when prices change continuously, customers experience a time pressure which reduce the search intention. Customers who are time pressured rely more on their product knowledge and previous experience when making choices than search for new information (Sundaram & Taylor, 1998).
Therefore, we expect that the searching intentions will be lower when dynamic pricing becomes an industry norm, because little to no price distinctions arise between retailers. In addition, customers rely more on their knowledge when there is time pressure which is the case when time-based pricing becomes an industry norm. Based on the little price distinctions and time pressure, customers are less likely to search for (cheaper) alternatives.
H2: “When dynamic pricing becomes an industry norm, customers have lower intentions to search.”
2.5 Purchase Intentions
Before making a purchase decision, customers reconsider searching costs and prices (Kolter, 2002). Knowing that customers pay different prices for similar products can influence consumers’ purchase intention (Gelbrich, 2011; Haws & Bearden, 2006). Customers can create a negative attitude towards the retailer when the industry norm is violated (Garbarino & Maxwell, 2010). A punishment response towards the retailer consists out of a lower willingness to purchase, developing a different search intention, and complain publicly (Maxwell, 2002). However, little research has been done to investigate if the same principle occurs when dynamic pricing becomes an industry norm.
13 To conclude, customers judge the willingness to purchase from a certain retailer based on the fairness of a pricing strategy by looking at circumstances and pricing strategies of other competitors. Therefore, when retailers have the power and implement dynamic pricing, customers have no reason to switch from retailers which cause a higher purchase intention. Looking at the hospitality and traveling industry, where dynamic pricing is already an industry norm, customers do have a higher purchase intention and therefore, the following hypothesis is conducted:
H3: “When dynamic pricing becomes an industry norm, the purchase intention will increase.”
2.6 Risk-taking behaviour as a Moderating Effect
According to literature, risk is a broad term which is used in different manners (Byrnes, Miller, & Schafer, 1999). Bran & Vaidis (2019) defined risk-taking behaviour as: “behaviour that refer to actions or inactions involving potential risks”. According to Byrnes, Miller, & Schafer (1999) three types of risk are widely used in the literature: reported risk behaviour, actual risk behaviour, and projected risk behaviour. First, reported risk behaviour focuses on the participants’ risk behaviour in the current state but also evaluate the risk behaviour in the past. Second, actual risk behaviour focuses on determining participants actual risk-taking behaviour by using risk-taking tasks. Finally, projected risk behaviour aims to discover risk-taking intentions or decisions of participants regarding a hypothetical choice. Projected risk is commonly used in experiments in order to determine the risk perceptions regarding a certain shopping scenario. Therefore, projected risk behaviour is also used in this study.
14 as risk-neutral, risk-averse or risk-seeking. An understanding of these behaviour is important for retailers when implementing dynamic pricing.
2.5.1 Risk-taking behaviour as moderator. Risk averse customers use different channels to search for information regarding prices in order to make a right choice (Moorthy, Ratchford, & Talukdar, 1997). Urbany (1986) states that customers who are certain about a product price are less likely to search for more information and alternatives. In addition, the need for searching more information is depending on the state of uncertainty. Risk-averse customers tend to avoid uncertainty (Bao, Zhou, & Su, 2003). Implementing dynamic pricing increases uncertainty due to privacy concerns and continuously fluctuating prices. Furthermore, fluctuating prices leads to less price knowledge which makes prices in the market more uncertain (Gelbrich, 2011). In order to reduce uncertainty, customers invest more time in searching for the optimal price, in particular risk averse customers (Bao, Zhou, & Su, 2003). So, customers who are risk averse will have stronger negative reactions when dynamic pricing will be implemented by retailers. Therefore, we expect that the risk-taking behaviour of customers influence the perceptions of price fairness, search intentions, and purchase intention when dynamic pricing becomes an industry norm. This gives the following hypotheses: H4a:“Risk taken characteristics positively moderates the relationship between dynamic pricing and price fairness such that the relationship is stronger when dynamic pricing is an industry norm.”
H4b:“Risk taken characteristics positively moderates the relationship between dynamic pricing and search intentions such that the relationship is stronger when dynamic pricing is an industry norm.”
H4c:“Risk taken characteristics positively moderates the relationship between dynamic pricing and purchase intentions such that the relationship is stronger when dynamic pricing is an industry norm.”
2.7 The Type of Dynamic Pricing
15 based on the degree of independent self-construal or interdependent self-construal. Whereas, independent construal customers focus more on themselves while interdependent self-construal customers focus more on the connection of other customers. Therefore, interdependent self-construal customers find individual price discrepancy more unfair than independent self-construal customers (Chen, 2009). When the price setting of a retailer is based on economic conditions, such as competition or costs, customers consider prices discrepancy as fairer than individual price discrepancy (Xia, Monroe, & Cox, 2004). The reason for adopting a price strategy of a retailer influence price fairness and the purchase intention of customers. The time-based pricing strategy in the travelling industry is widely accepted (Eisen, 2006). However, studies about behavioural pricing indicate a more negative attitude towards price changes based on individual price discrepancy (Garbarino & Maxwell, 2010; Grewal, Hardesty, & Iyer, 2004; Haws & Bearden, 2006; Weisstein, Monroe, & Kukar-Kinney, 2013). Therefore, we expect that customers respond differently towards time-based pricing and behavioural-based pricing when it is an industry norm. The following hypotheses are formed:
H5a: “When Behaviour-based pricing becomes an industry norm, it influences the perceptions of price fairness differently than when time-based pricing is an industry norm.”
H5a: “When Behaviour-based pricing becomes an industry norm, it influences the perceptions of search intentions differently than when time-based pricing is an industry norm.”
H5a: “When Behaviour-based pricing becomes an industry norm, it influences the perceptions of purchase intentions differently than when time-based pricing is an industry norm.”
16 violate the social norms and therefore might be considered as more unfair than time-based pricing. In figure 1 an overview of our study is pictured as a conceptual framework.
FIGURE 1: CONCEPTUAL FRAMEWORK
3
METHODOLOGY
In this chapter, the methodology is described. The aim of this section is to show how this study is constructed. In section 3.1 we described the design of the study. Followed in section 3.2 which explains the procedure of our study. In section 3.3, the pre-test is discussed. The measurements of the experiment are explained in section 3.4. Finally, in section 3.5, the analyses are further discussed.
3.1 Study Design
During an online scenario-based experiment, participants were shown two different pricing strategies (time-based pricing vs. behavioural-based pricing) in a way that it is an industry norm or not an industry norm (i.e., one retailer charge dynamic prices compared to all retailers charge dynamic prices). Moreover, a 2 (time-based pricing vs. behavioural-based pricing) x 2 (industry norm: yes vs. no) between-subject design was used to test the perceptions of customers regarding price fairness, purchase intentions and search intentions. The main goal of this experiment was to explore the effect of dynamic pricing on the customers perceptions of price fairness, search intentions and purchase intentions. The participants were exposed to a short scenario which described a purchase situation wherein they would purchase a camera from an online retailer. The same camera was shown to all respondents but with different pricing strategies and industry norms, see table 1 for the four conditions.
17 TABLE 1: FOUR CONDITIONS OF THE EXPERIMENT
Industry norm
Yes No Camera is priced based on time which is an industry norm Camera is priced based on time which is not an industry norm Time basedD
y
na
m
ic pr
ici
n
g
Camera is priced based on behaviour which is an industry norm Camera is priced based on behaviour which is not an industry norm Behavioural-based 3.2 Procedure18 TABLE 2: SAMPLE DESCRIPTION PER CONDITION
Conditions 1 2 3 4
Gender
Female 17.5 33.3 23.8 25.4
Male 28.8 23.7 27.1 20.3
Age (%) 19 years or younger 50.0 50.0
20 – 29 years 26.0 26.0 20.5 27.4 30 – 39 years 30.0 10.0 40.0 20.0 40 – 49 years 33.3 50.0 16.7 50 – 59 years 16.0 44.0 24.0 16.0 60 years or older 33.3 16.7 33.3 16.7 Degree (%) Highschool 3.6 14.3 3.2 3.6 MBO 10.7 22.9 29.0 14.3 HBO 53.6 42.9 38.7 39.3 University 32.1 20.0 29.0 32.1 Annual Income (%) < €5.000 17.9 21.2 9.7 17.9 €5.000 - €9.999 7.1 6.1 25.8 10.7 €10.000 - €19.999 25.0 3.0 9.7 179 €20.000 - €29.999 10.7 18.2 9.7 10.7 €30.000 - €39.999 10.7 15.2 9.7 17.9 €40.000 - €49.999 3.6 3.0 6.5 10.7 > €50.000 10.7 18.2 25.8 3.6 Do not want to answer 14.3 15.2 3.2 10.7 3.3 Pre-Test
Before conducting the survey, a small pre-test has been done. The aim of the pre-test was to identify unclear questions, questions which participants did not understood well, and possible incompleteness of the survey (Vomberg & Klarmann, 2020). A small sample size (N = 6) have been asked to fill in the online experiment. At the end of the survey, they could give tips to improve the survey. A little adjustment was made relating to the questions. Every question started by referring to the written scenario, so the participants keep the scenario in their mind when answering the question.
3.4 Measurements
19 perception of price fairness, search intentions and purchase intentions differ per customer. Therefore, also the risk-taking behaviour of customers were measured to explore how risk influence customers’ price fairness, search intentions and purchase intentions. To control for random measurement errors and increase validity, multiple items are asked per construct (Vomberg & Klarmann, 2020). In addition, some items are dropped to increase validity and reduce a certain response pattern.
3.4.1 Price Fairness. To test the price fairness, participants were asked to evaluate perceived fairness by three statements on a scale from 1 (strongly disagree) to 7 (strongly agree) and one statement on a scale from 1 (very unfair) to 7 (very fair) deployed by Garbarino & Maxwell (2010). The Cronbach Alpha is assessed and indicated that the four statements can be used as one construct, see table 3.
3.4.2 Search Intentions.In order to measure search intentions, three statements from the study of Grewal et al. (1998) were asked on a 7-point Likert scale from 1 (strongly disagree) to 7 (strongly agree) and adapted to the context of purchasing a camera, see table 3.
3.4.3 Purchase Intentions.Six items are used to determine the purchase intentions of the participants to shop at the retailer described in the scenario. Three statements regarding the intention to purchase were asked on a 7-point Likert scale from 1 (strongly disagree) to 7 (strongly agree) as employed by Garbarino & Maxwell (2010). The other three statements are used from Grewal, et al. (1998) and based on ratings from 1 (very low possibility) till 7 (very high possibility). Based on the Cronbach’s Alpha, all items of purchase intention are internally consistent and used as one construct for further analysis, see table 3.
3.4.4 Risk aversion. Two components of risk were measured on a 7-point scale. First, the risk component regarding purchasing new products were used to measure the risk-taken behaviour of participants. Moreover, other studies also measured risk based on risk-taking behaviour regarding prices. Therefore, the second component of risk measured the risk-taking behaviour of customers regarding prices. Three items created by Slovic (1972), Cable & Judge, (1994), and Judge, et al. (1999) are adapted to the context of this study.
20 consistency (α=0.495) (M=3.866; SD=0.419). Deleting items does not increase the Cronbach’s Alpha.
TABLE 3: SURVEY CONSTRUCT MEASUREMENTS
Constructs CA SD M Price Fairness deployed by Garbarino & Maxwell [2010]. 0.804 0.157 4.76 Seven-point Likert scale (1 = “strongly disagree”, and 7 = “strongly agree”)
How fair do you consider this way of pricing in this situation? I feel this sort of pricing is unfair.
This pricing practice is fair to retailer (dropped). This pricing practice is unfair to the buyer.
Search Intention adapted from Grewal et al [1998]. 0.697 0.028 5.883 Seven-point Likert scale (1 = “strongly disagree”, and 7 = “strongly agree”)
Before purchasing this camera, I would need to search for more information about prices of alternative cameras.
Before purchasing this camera, I would search for other retailers that sells cameras to check their prices.
Before purchasing this camera, I would visit other retailers for a lower price.
Purchase Intention employed by Garbarino & Maxwell [2010]. 0.930 0.036 3.919 Seven-point Likert scale (1 = “strongly disagree”, and 7 = “strongly agree”)
Purchase something from this retailer?
Purchase this item from this retailer if you were in the market for it? Shop at this retailer in the future?
Purchase probabilities adapted from Grewal, et al [1998].
Seven-point Likert scale (1 = “strongly disagree”, and 7 = “strongly agree”) The likelihood of my purchasing this camera is…
My willingness to buy this camera is …
The probability that I would consider buying this camera is…
Risk-taken Behaviour developed by Raju [1980] and used by Bao, Zhou, & Su [2003].
Seven-point Likert scale (1 = “strongly disagree”, and 7 = “strongly agree”)
Risk- taken behaviour regarding purchasing new products/brands 0.244 0.449 3.842 I am cautious in trying new and different products.
I would rather stick with a brand I normally purchase than try something I am not very sure of (dropped).
21 Risk-taken behaviour regarding prices created by Slovic (1972), Cable &
Judge, (1994), and Judge, et al. (1999)
0.495 0.419 3.866
To hold out for the best price on something, even if it means waiting a long time.
Choose a product with fluctuating price over a product with a steady price. Play it safe, even if it means occasionally losing out a good opportunity (dropped)
Manipulation Checks
Seven-point Likert scale (1 = “strongly disagree”, 7 = “strongly agree”, 8 = “no answer possible”)
Based on the scenario, the retailer’s price strategy was typically for the industry.
1.845 4.56
Based on the scenario, the varying pricing was based on time. 1.611 5.03 Based on the scenario above, the varying pricing was based on browsing
history and cookies.
1.569 5.57
Due to the low Cronbach’s Alpha, a factor analysis is conducted to reduce the number of items and determine which items belongs to which risk construct. First, the Pearson correlation is checked to indicate which variables are correlated to each other. Item: “to hold out for the best price on something, even if it means waiting a long time” and “choose a product with fluctuating price over a product with a steady price” are moderate correlated (r = 0.33, p <.01). Also, item: “play it safe even if it means occasionally losing out a good opportunity” and “to hold out for the best price on something, even if it means waiting a long time” are moderate correlated (r = 0.21, p <.01).
The factor analysis is appropriate because the Kaiser Meyer Olkin requirements are met (KMO = 0.589) and the sample size (N = 122) is large enough (Field, 2009). Also, the Bartlett’s test of Sphericity is significant (p <0.01) which rejects the null hypothesis that the correlation matrix is an identity matrix. According to Field (2009) when the Bartlett’s test of Sphericity is significant, the correlated items below 0.2 should be deleted from the analysis. Therefore, the item “play it safe even if it means occasionally losing out a good opportunity” is deleted.
22 two factors can be extracted from the analysis which means that choosing more factors leads to a smaller explained variance and thus, more information loss, see figure 2.
FIGURE 2: SCREE PLOT FACTOR ANALYSIS
The factor analysis is conducted and shows that items “to hold out for the best price on something, even if it means waiting a long time” and “choose a product with fluctuating price over a product with a steady price” belong to the first component; price risk related factors. The loadings of the two items are above 0.8. The items “I would rather stick with a brand I normally purchase than try something I am not very sure of” and “I never purchase something I do not know about at the risk of making a mistake” belong to the second component; product risk related factors. The loading of the first component is above 0.8 and the loading of the second component is above 0.6, see table 4.
TABLE 4: ROTATED COMPONENTS MATRIX
Components
1 2
I would rather stick with a brand I normally purchase than try something I am not very sure of.
.043 .821
I never purchase something I do not know about at the risk of making a mistake.
-.187 .675 To hold out for the best price on something, even if it means waiting a
long time.
.813 -.072
Choose a product with fluctuating price over a product with a steady price.
.822 -.079
Note: Rotation converged in 3 iterations.
When looking at the explained total variances, the two factors explain 62.8% of the variances which is above the required amount of 60% (Malhora, 2010). However, the two items regarding
23 price cannot be combined as one component according to the requirements of the Cronbach’s Alpha (α =.495) (M= 4.127; SD =.429). Also, the two items about product risk cannot be combined as a component due to the low internal consistency (α =.244) (M= 3.842; SD =.449). Therefore, the four items are used separately for further analyses.
3.5 Analysis
Three ANOVA tests will be performed to explore the differences of the means of the two groups relating to dynamic pricing as an industry norm or not on price fairness, search intentions, and purchase intentions. By making use of a regression, the effect of risk on the relationship between dynamic pricing as an industry norm on the price fairness, search intention and purchase intention has been tested. To determine the different perceptions of price fairness, search intention and purchase intention when dynamic pricing is time-based or behavioural-based, also an ANOVA test is used.
4
RESULTS
In this chapter, the results of the analysis of our online experiment are written. In section 4.1 the manipulation check can be found. Before conducting an ANOVA test, the assumptions are checked and described in section 4.2. Moreover, in section 4.3 the first hypothesis is tested with an ANOVA analysis. To continue, the results of the second hypothesis can be found in section 4.4. In section 4.5, the third hypothesis is answered. A regression analysis is used for the fourth hypothesis which can be found in section 4.6, 4.7, 4.8 and 4.9. Finally, the differences between behavioural-based and time-based pricing are analysed and discussed in section 4.10.
4.1 Manipulation Checks
24 conduct an ANOVA test. Therefore, also the Homogeneity test is used to determine if the variances are equal. This test also shows a significant value (p <.05) which means that the variances are not equal. Due to the violation of both assumptions of an ANOVA test, the Kruskal-Wallis H Test is used for further analysis. This test does not require a normal distribution and equal variances.
4.1.1 Industry Norm. First, a dummy variable is created based on the industry norm conditions whereas dynamic pricing is not an industry norm serves as a reference group. The Kruskal-Wallis H test shows that there is not a significant difference between the mean of the dynamic pricing as industry norm and dynamic pricing not being an industry norm, χ²(2) = 1.328, p =.249, with a mean of 4.56. Therefore, the manipulation regarding industry norm did not work out.
4.1.2 Dynamic Pricing. Similar as for the industry norm manipulation check, dummy variables were created for the dynamic pricing conditions. Two variables are made based on the behavioural-based pricing scenario or time-based pricing scenario. First the manipulation check about time-based pricing was asked. The outcome of the Kruskal-Wallis H test shows a significant difference between the means of the two groups regarding the time-based scenario, χ²(2) = 5.947, p =.015, with a mean of 5.03. Also, the manipulation check regarding the behavioural-based pricing scenario shows a significant difference, χ²(2) = 4.909, p =.048, with a mean of 5.57. Therefore, the manipulation check of dynamic pricing has been successful.
4.1.3 Smaller Sample Size. The manipulation check of the industry norm was not significant which means that the manipulation about the industry norm did not work out. Therefore, further analyses about the industry norm are done with the whole sample and a smaller sample to determine if the outcomes are different compared to participants who passed the manipulation check and participants who did not passed the manipulation check. The smaller sample consists of the data of the participants who did pass the manipulation check.
4.2 Assumptions ANOVA
25 participants took part in the experiment which is an adequate sample size for conducting an ANOVA test. Price fairness, search intention, and purchase intention show no univariate outliers. To test if the variables are normally distributed, the Shapiro-Wilk test is used. The result shows that price fairness, search intention, and purchase intention are not normally distributed (p <.05). However, the size in each condition (N = 30) is large enough to conduct an ANOVA test. Furthermore, the assumption of homogeneity of variance is met with the Levene’s test. The results show that price fairness F (3,118) = 1.329, p =.268, search intentions F (3,118) = 1.381, p =.252, and purchase intention F (3,118) = 0.639, p =.591 are insignificant. Therefore, an ANOVA test can be used for further analyses. See APPENDIX B4 for further details.
4.3 High Perceptions of Price Fairness
To test the first hypothesis: “When dynamic pricing is an industry norm, there are high perceptions of price fairness”, an ANOVA test has been used. First, the ANOVA test is done with the whole sample (N = 122) and shows an insignificant effect F (1,120) = 0.441, p =.508 which means that there is no effect of whether dynamic pricing is an industry norm or not on the perceived price fairness.
Not all the participants passed the manipulation check regarding the industry norm. Therefore, a smaller sample (N = 57) is used with all the participants who passed the manipulation check. Two groups are formed based on the condition of the participant; a group who experienced the prices as an industry norm (N = 31) compared to the group who did not experienced the prices as an industry norm (N = 26). The ANOVA test also shows an insignificant value F (1,55) = 0.700, p =.406. The results of both groups indicate that whether dynamic pricing is an industry norm or not, it does not have an effect on the perceived price fairness. Therefore, there is not enough evidence to argue that dynamic pricing as an industry norm influences the price fairness. See APPENDIX B5 for the supporting details of the ANOVA.
4.4 Lower Search Intentions
26 Also, a smaller sample (N =57) is used as earlier described. That consists of the participants who passed the manipulation check about the industry norm manipulation. An ANOVA test cannot be conducted due to the significance of the homogeneity of variance test, F (1,55) = 5.832, p =.019, and the small sample size. Therefore, a Kruskal-Wallis Test is used. This test shows an insignificant value, χ²(2) = 1.464, p =.226, with a mean of 17.6. The results of both samples assume that industry norm has no effect on the search intentions. Therefore, the second hypothesis cannot be accepted. APPENDIX B5 provide more details.
4.5 Higher Purchase Intentions
The third hypothesis: “When dynamic pricing becomes an industry norm, the purchase intention will increase” is also tested with an ANOVA. To test the third hypothesis an ANOVA test is used. The ANOVA test indicate an insignificant effect F (1,120) = 0.428, p =.514 which shows that the means of the two groups do not differ.
Also, the smaller sample (N = 57) with the participants who passed the manipulation check are used to determine if there are different means on the level of purchase intentions between the group in which dynamic pricing was an industry norm and in which it was not an industry norm. The homogeneity shows that the variances are equal F (1,55) = 4.545, p =.037. However, the conditions are not large enough to conduct an ANOVA test. Therefore, a Kruskal Wallis test is used. The outcome of the test shows an insignificant effect, χ²(2) = 2.156, p =.142, with a mean of 23.3. We can conclude that the means of dynamic pricing as an industry norm and no industry norm do not differ from each other and therefore, react not differently towards purchase intention. The third hypothesis cannot be accepted. See APPENDIX B5 for further details.
4.6 Risk Characteristics as Moderator
27 intentions with a value of 2. Finally, search intention shows no autocorrelation with a value of 1.6 (Field, 2009). The risk items are mean-centered due to the different scales between the independent variable and moderator. The independent variable industry norm is kept as a dummy variable. See APPENDIX B6 for more supporting details regarding the assumptions. For more details regarding the results of the regression, see APPENDIX B7. The following linear regression equation (1) is used:
(1) 𝑌 = 𝛼 + 𝛽1𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚 + 𝛽2𝑅𝑖𝑠𝑘1 + 𝛽3𝑅𝑖𝑠𝑘2 + 𝛽4𝑅𝑖𝑠𝑘3 + 𝛽5𝑅𝑖𝑠𝑘4 +
𝛽6𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚 ∗ 𝑅𝑖𝑠𝑘1 + 𝛽7𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚 ∗ 𝑅𝑖𝑠𝑘2 + 𝛽8𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚 ∗ 𝑅𝑖𝑠𝑘3 + 𝛽9𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚 ∗ 𝑅𝑖𝑠𝑘4 + ɛ
Y Dependent variable
𝛼 Intercept
𝛽1𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚 Dummy variable coefficient of industry norm where 1 is for industry norm and 0 for no industry norm (reference group)
𝛽2𝑅𝑖𝑠𝑘1 Coefficient of risk item 1: I would rather stick with a brand I normally
purchase than try something I am not very sure of
𝛽3𝑅𝑖𝑠𝑘2 Coefficient of risk item 2: I never purchase something I do not know
about at the risk of making a mistake
𝛽4𝑅𝑖𝑠𝑘3 Coefficient of risk item 3: to hold out for the best price on something,
even if it means waiting a long time
𝛽5𝑅𝑖𝑠𝑘4 Coefficient of risk item 4: choose a product with fluctuating price over
a product with a steady price
𝛽6𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚 ∗ 𝑅𝑖𝑠𝑘1
Interaction effect of industry norm (dummy variable 1 = industry norm, 0 = no industry norm) and risk item 1: I would rather stick with a brand
I normally purchase than try something I am not very sure of
𝛽7𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚 ∗ 𝑅𝑖𝑠𝑘2
Interaction effect of industry norm (dummy variable 1 = industry norm, 0 = no industry norm) and risk item 2: I never purchase something I do
not know about at the risk of making a mistake
𝛽8𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚
∗ 𝑅𝑖𝑠𝑘3
Interaction effect of industry norm (dummy variable 1 = industry norm, 0 = no industry norm) and risk item 3: to hold out for the best price on
something, even if it means waiting a long time
𝛽9𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑁𝑜𝑟𝑚
∗ 𝑅𝑖𝑠𝑘4
Interaction effect of industry norm (dummy variable 1 = industry norm, 0 = no industry norm) and risk item 4: choose a product with fluctuating
price over a product with a steady price
ɛ Disturbance term
4.7 Risk negatively influence Price Fairness
28 which means that the model is relative high in explaining the variances of perceived price fairness. Other circumstances may influence those choices of customers as well. This should be taken into account when intepreting the results.
4.7.1 Interaction effects. The result shows that the interaction effect of risk “I would rather stick with a brand I normally purchase than try something I am not very sure of” and “choose a product with fluctuating price over a product with a steady price” are insignficant (p =.433, p = .135). In addition, the interaction effect of risk items: “I never purchase something I do not know about at the risk of making a mistake” and “To hold out for the best price on something, even if it means waiting a long time” are marginally significant (β 1.269, β =-1.134, p <.10). When dynamic pricing is an industry norm, the moderator risk item: “to hold out for the best price on something, even if it means waiting a long time” and risk item: “I never purchase something I do not know about at the risk of making a mistake” have a negative influence on the relationship between industry norm and price fairness. This implies that item “to hold out for the best price on something, even if it means waiting a long time has a stronger negative effect on price fairness when dynamic pricing is an industry norm than when it is not an industry norm, namely decreases by 1.269. The full industry norm effect is when risk increases with one, price fairness decreases by .388 (= -1.269 + .881) when dynamic pricing is an industry norm. Moreover, item “I never purchase something I do not know about at the risk of making a mistake” has a strong negative effect of 1.134 on price fairness when dynamic pricing is an industry compared with not being an industry norm. The full effect is when risk increases with one, price fairness decreases by 0.253 (= - 1.134 +.881) when dynamic pricing is an industry norm.
29 TABLE 5: COEFFICIENTS OF REGRESSION
Estimates P values
Intercept -.236 .727
Interaction effects
I would rather stick with a brand I normally purchase than try something I am not very sure of * industry norm
-1.134 .086*
To hold out for the best price on something, even if it means waiting a long time * industry norm
-1.269 .083*
Choose a product with fluctuating price over a product with a steady price * industry norm
1.048 .135
I never purchase something I do not know about at the risk of making a mistake * industry norm
-.544 .433
Main effects
I would rather stick with a brand I normally purchase than try something I am not very sure of
.332 .495
To hold out for the best price on something, even if it means waiting a long time
1.113 .043**
Choose a product with fluctuating price over a product with a steady price
-1.361 .011**
I never purchase something I do not know about at the risk of making a mistake
-.321 .534
Industry norm .811 .388
Note: Dependent variable: Price Fairness. * p < 0.10, one-sided; ** p < 0.05, one sided; *** p < 0.01, one-sided 4.7.3 Smaller Sample Size. Due to the failure of the industry norm manipulation among participants, another linear regression is conducted. The smaller sample size (N =52) is used to determine the effect of risk on the relationship between industry norm and price fairness. There are no high significant correlations (r >.8) determined between the independent variables, moderator and dependent variable search intention, thus no multicollinearity. The residuals of the independent variable, moderator, and dependent variables are equally distributed. Altough there are some residuals which are a bit skewed, there are no drastic deviations. Variance of the error terms of the independent variables are equally distributed, which means that the assumption of homoscedasity is met. The linear regression can be conducted. The R-square is lower, namely .43. The Durbin Watson test shows a value of 1.3 which might not be a concern of autocorrelation yet. A value below 1 shows a generally concern of autocorrelation (Field, 2009). The overall model is signficiant (p <.01).
30 seems to be insignficant (β =1.355, p =.180). When dynamic pricing is an industry norm, the moderators “to hold out for the best price on something, even if it means waiting a long time”, “I never purchase something I do not know about at the risk of making a mistake”, and “I would rather stick with a brand I normally purchase than try something I am not very sure of” have a negative significant influence on the relationship between industry norm and price fairness. Item “to hold out for the best price on something, even if it means waiting a long time has a stronger negative effect on price fairness when dynamic pricing is an industry norm than when it is not an industry norm, namely decreases on average by 2.454. The full interaction effect is when dynamic pricing is an industry norm, an increase in risk leads to a decrease of 4.258 ( = -2.454-1.804) in price fairness.
Also, item “I never purchase something I do not know about at the risk of making a mistake” react negatively strong on the relationship between industry norm and price fairness. When dynamic pricing is an industry norm, the risk item reacts on average 2.937 times more negative on price fairness than when dynamic pricing is not an industry norm. The full moderating effect when dynamic pricing is an industry norm, is a decrease in risk leads to a 4.741 (= -2.937 - 1.804) decrease in price fairness. Figure 3 shows how the linear relationship will evolve during dynamic pricing when it is an industry norm and whether it is not an industry norm.
FIGURE 3: LINEAR RELATIONSHIP BETWEEN PRICE FAIRNESS AND RISK 2
In addition, item “I would rather stick with a brand I normally purchase than try something I am not very sure of” also shows a stronger negative reaction towards the relationship between industry norm and price fairness. When dynamic pricing is an industry norm the risk item react on average 1.511 more negatively on price fairness than when dynamic pricing is not an industry norm. The full industry norm effect is when risk increases by one the price fairness
16 18 20 22 24 26 28 30 32 0 1 2 3 4 5 6 7 P rice Fair n ess Risk
31 decreases by 3.315 (= -1.511 – 1.804), when dynamic pricing is an industry norm. See table 6 for an overview of all the coefficients of the linear regression.
TABLE 6: REGRESSION COEFFICIENTS SMALLER SAMPLE SIZE
Note: Dependent variable: Price Fairness, * p < 0.10, one-sided; ** p < 0.05, one sided; *** p < 0.01, one-sided
The linear regression formula (1.1) when dynamic pricing is an industry norm with only the significant values from table 6:
(1.1) 𝑃𝑟𝑖𝑐𝑒 𝐹𝑎𝑖𝑟𝑛𝑒𝑠𝑠|𝑤ℎ𝑒𝑛 𝑑𝑦𝑛𝑎𝑚𝑖𝑐 𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑖𝑠 𝑎𝑛 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑛𝑜𝑟𝑚 = 20.388 + 𝛽1(1) + 𝛽2𝑅𝑖𝑠𝑘1 + 1.578𝑅𝑖𝑠𝑘2 + 𝛽4𝑅𝑖𝑠𝑘3 − 1.526𝑅𝑖𝑠𝑘4 + 1 ∗ −1.511 + 1 ∗ −2.937 +
1 ∗ −2.454 + 1 ∗ 𝑅𝑖𝑠𝑘4 + 𝜀𝑖
Hypothesis 4a cannot be accepted.
4.8 Risk as Moderator of Search Intentions
To determine whether risk influence the relationship between industry norm and search intention and thus, to test this hypothesis (4b): “Risk taken characteristics positively moderates the relationship between dynamic pricing and search intentions such that the relationship is stronger when dynamic pricing is an industry norm”, a linear regression is conducted. The
R-Estimates Sig.
Intercept 20.388 .000***
Interaction effects
I would rather stick with a brand I normally purchase than try something I am not very sure of * industry norm
-1.511 .095*
To hold out for the best price on something, even if it means waiting a long time * industry norm
-2.454 .044**
Choose a product with fluctuating price over a product with a steady price * industry norm
1.355 .180
I never purchase something I do not know about at the risk of making a mistake * industry norm
-2.937 .002***
Main effects
I would rather stick with a brand I normally purchase than try something I am not very sure of
.232 .716
To hold out for the best price on something, even if it means waiting a long time
1.461 .100
Choose a product with fluctuating price over a product with a steady price
-1.526 .056*
I never purchase something I do not know about at the risk of making a mistake
1.578 .024**
32 square of our model is .96 which means that risk and industry norm are relative high in explaining the variances of search intention.
4.8.1 Interaction Effects. The output shows that the interaction effect between the risk items and industry norm on search intention was insignificant (p =.524, p =.351, p =.127, p =.795). Also, the main effects of industry norm (p =.164), item “I would rather stick with a brand I normally purchase than try something I am not very sure of” (p =.704), item “I never purchase something I do not know about at the risk of making a mistake” (p =.685), item “choose a product with fluctuating price over a product with a steady price” (p =.277) are all insignificant.
4.8.2 Main Effects. The main effect of item “to hold out for the best price on something, even if it means waiting a long time” is found to be signficiant (β =.793, p <.05) which means that an increase in risk leads to an increase in search intention by 0.793. The moderator analysis shows that the effects of risk do not influence the relationship between the industry norm and search intention.
4.8.3 Smaller Sample Size. According to the failure of the industry norm manipulation, a sample sample size is used (N =52). The smaller sample size is used to determine the effect of risk on the relationship between industry norm and search intention. First the assumptions of a linear regression are checked. The residuals are normally distributed. Also, the error terms do not differ across the values of the independent variables which met the assumption of the homoscedasticity. There are no high correlations between the variables. The correlation matrix tells us that there are no high correlations (r >.8, p <.05) between the independent variable, risk items and dependent variables. The Durbin Watson test is used to determine autocorrelation. The test shows a value of 2.2 which means that the autocorrelation assumption is hold. The regression can be conducted. The model fit is increased and has a R-square of .36 which is still not very satisfying. The model is overal significant (p <.05).
All interaction effects show insignificant values (p >.05). Also, most mean effects show a insignificant value except for industry norm. Dynamic pricing as an industry norm negatively influence search intention by 1.77 compared to whether dynamic pricing is no industry norm (β = -1.77, p <.01). Hypothesis 4b cannot be accepted.
4.9 Risk Positively moderates Purchase Intention
33 characteristics positively moderates the relationship between dynamic pricing and purchase intention such that the relationship is stronger when dynamic pricing is an industry norm”. The r-square of the linear model is .84 which means that the indicators of the model explain the variances of purchase intention very well. The overall model is significant.
4.9.1 Interaction Effects and Main Effects. The interaction effects of the risk items and industry norm seems to be insignificant. Also, all the main effects of risk and industry norm are all insignficant.
4.9.2 Smaller Sample Size. The smaller sample size is used for further analysis. The residuals are a bit skewed. However, the sample size is large enough to continue with a linear regression. The residuals are randomly distributed which assumes homoscedasticity. Therefore, the assumption of homoscedasticity is met. Also, there are no high correlations between the independent variable, risk items, and dependent variable. The linear regression can be conducted. The r-square is .37 which means that the model is not very good in explaining the variances of the purchase intention. In addition, the autocorrelation assumption is hold. The Durbin Watson test shows a value of 2. The overal model is significant (p <0.01).
Moreover, the interaction effect of item “I never purchase something I do not know about at the risk of making a mistake” and industry norm is signifciant (β =3.847, p <.05). Risk positively influence the relationship between dynamic pricing as an industry norm and purchase intention by 3.847 compared to when dynamic pricing is not an industry norm. The full industry norm effect is when risk increases with one during the period that dynamic pricing is an industry norm, the purchase intention increases by 7.54 (= 3.847 + 3.693).
Also, the main effect of “I never purchase something I do not know about at the risk of making a mistake” is marginal significant (β = -2.021, p <.10) which means that an increase in risk leads to a decrease in purchase intention. Moreover, item “I never purchase something I do not know about at the risk of making a mistake” negatively influenced purchase intention by 2.021. The more risk taken participants are the less likely they purchase. Hypothesis 4c cannot be accepted.
4.10 Behavioural-based and Time-based Pricing
34 condition and a second variable based on the time-based pricing condition. See APPENDIX B8 for the assumptions and output of all ANOVA tests.
4.10.1 Price Fairness. The price fairness regarding behavioural-based and time-based are tested with an ANOVA, which shows a significant value F (1,61) = 4.131, p =.044. The results show that the means of the two groups differ regarding price fairness. Moreover, the Levene’s test shows an insignificant value F (1,120) = 2.889, p =.92 which means that there are equal variances assumed. To determine the difference between groups, an Independent samples t-test is conducted. This test shows a significant value, t(120)=2.032 (p <.05) which states that the behavioural-based group experiences the prices on average 1,897 fairer than the average fairness of the time-based condition, see figure 4. Therefore the hypothesis 5a can be accepted. FIGURE 4: MEAN DISTRIBUTION BETWEEN BEHAVIOURAL-BASED AND
TIME-BASED CONDITIONS
4.10.2 Search Intention. Also, the search intention will be tested by the aid of an ANOVA test. The result of the test is insignificant F (1,120) = 0.282, p =.597. The means of the two groups do not differ based on their search intentions. Therefore, the hypothesis 5b cannot be accepted.
4.10.3 Purchase Intention. The means between groups regarding purchase intention are also tested by the aid of an ANOVA test. The results show a significant value F (1,120) = 6.113, p =.015. Furthermore, the Levene’s test has been conducted. The test shows an insignificant value F (1,120) = 0.501, p =.083 which means that we can assume that there are equal variances. To determine the effect of the two groups, an Independent samples t-test is assessed. This test
15 17 19 21 23 25 27 Behavioural-based Time-based Me an s Su m m ed
35 is significant, t(120)=-2.473 (p <.05) which explains that the behavioural condition has rated their purchase intention on average 3.691 lower than the average purchase intention of the participants in time-based conditions, see figure 4.
4.10.4 Behavioural-based and Time-based Pricing as an Industry Norm. To determine the different perceptions of customers regarding price fairness, search intentions, purchase intentions when behavioural-based or time-based pricing is an industry norm, an ANOVA test is conducted. First, the ANOVA test is used to determine if there are different means between the groups. The ANOVA test shows an insignificant value regarding price fairness F (3,118) = 1.751, p =.160 and search intentions F (3,118) = 0.870, p =.459. However, the means between the four conditions significantly differ regarding purchase intention F (3,118) = 5.614, p <.05. Moreover, the Levene’s test is assessed and shows an insignificant value F (1,118) = 0.639, p =.592 which assumes that the variances are equal. Based on the Levene’s test, the Turkey HSD test is conducted. The Turkey HSD test shows a significant different mean between condition 1 (time-based pricing as an industry norm) and condition 3 (behavioural-based as an industry norm). Furthermore, condition 3 (behavioural-based pricing as not being an industry norm) and condition 4 (behavioural-based pricing with not being an industry norm) have significantly different means, see figure 5.
FIGURE 5: MEAN DISTRIBUTION OF PURCHASE INTENTION ALL CONDITIONS
The mean difference between time-based industry norm condition is 8.361 higher than the average mean of the behavioural-based industry norm condition. Also, the mean difference between behavioural-based with no industry norm is on average 5.432 higher than the average
15 17 19 21 23 25 27 29 Behavioural-based Time-based Me an s Su m m ed
36 mean of the behavioural-based with industry norm. Therefore, the hypothesis 5c can be accepted. The results of all tests and relating hypotheses are shown in table 7.
TABLE 7: OVERVIEW RESULTS
Hypotheses Test Values
H1 (N=122) “When dynamic pricing is an industry norm, there are high perceptions of price fairness”
Not supported ANOVA F (1,120) =
0.441, p =.508
H1 (N=57)
Not supported ANOVA F (1,55) =
0.700, p =.406
H2 (N=122)
“When dynamic pricing becomes an industry norm, customers have lower intentions to search”
Not supported ANOVA F (1,120) =
0.096, p =.757
H2 (N=57)
Not supported Kruskal-Wallis Test
χ2(2) = 1.464, p =.226
H3 (N=122)
“When dynamic pricing
becomes an industry norm, the purchase intention will
increase”
Not supported ANOVA F (1,120) =
0.428, p =.514
H3 (N=57)
Not supported Kruskal-Wallis Test
χ2(2) = 2.156, p =.142
H4 “Risk taken characteristics positively moderates the relationship between dynamic pricing and price fairness(H4a), search intentions(H4b) and purchase
intention(H4c) such that the relationship is stronger when dynamic pricing is an industry norm.”
H4a (N=122)
Price fairness
Risk 2: negative effect² Risk 4: negative effect⁴
Not supported Regression β =-1.134, p <.10² β =-1.269, p <.10⁴ H4a (N= 52) Price fairness
Risk 1: negative effect¹ Risk 2: negative effect² Risk 3: negative effect³
Not supported Regression β = -2.937,
p<.10¹ β =-1.511, p<.10² β = -2.454, p<.10³ H4b (N=112) Search intention
No significant interaction effect
Not supported Regression p >.10 H4b (N=52) Search intention
No significant interaction effect
Not supported Regression p >.10 H4c
(N=122)
Purchase intention
No significant interaction
37
(N=52) Risk 2: positive effect² Partially supported (1 out of 4 risk items)
(β =3.847, p <.05)²
H5 “When behaviour-based pricing becomes an industry norm, it influences the
perceptions of price fairness(H5a), search intentions(H5b), and purchase intentions (H5c) differently than when time-based pricing is an industry norm”
H5a (N=122)
Price fairness
Behavioural-based condition experiences prices 1.896 fairer than behavioural-based condition
Supported ANOVA & Independent Samples T-Test t(120)=2.032 (p <.05) H5b (N=122)
Search intention Not supported ANOVA F (1,120) =
0.282, p =.597
H5c (N=122)
Purchase intention
Behavioural-based condition rated their purchase intention 3.691 less likely than time-based condition
Supported ANOVA & Independent Samples T-Test
t(120)=-2.473 (p
<.05)
Purchase intention among all conditions Different means: Condition 1 – Condition 3 Condition 3 – Condition 4 Supported ANOVA F (3,118) = 5.614, p <.05
Note: ¹ “I would rather stick with a brand I normally purchase than try something I am not very sure of” , ² “I
never purchase something I do not know about at the risk of making a mistake”, ³“to hold out for the best price on something, even if it means waiting a long time”, ⁴ “Choose a product with fluctuating price over a product with a steady price” .
5
CONCLUSION AND DISCUSSION
This study examines the effects of dynamic pricing when it becomes an industry norm on the perceived price fairness, search intention and purchase intention with a moderating role of risk characteristics. Within this study, the following research question is answered: “How do customers respond to dynamic pricing when it becomes an industry norm?”.
5.1 Findings
