The paradox of choice in online retail:
When less is more
A study on the use of consumer preferences for manipulating online retail environment attributes
to avoid information overload and the influence on consumer purchase probability.
E. E. Neijzen
2
Thesis of Master’s programme
Communication Science: Corporate Communication
University of Amsterdam
Graduate School of Communication
Supervisor: Dr. J. Slevin
By
Eveline Neijzen
Student number: 11903007
evelineneijzen@icloud.com
Amsterdam, June 2019
E. E. Neijzen
3
“We are drowning in information, while starving for wisdom”
E. E. Neijzen
4
Abstract
Because consumers differ in their preferences (Rosen & Purinton, 2004), online retailers have to offer a wide variety of products and product information on their web shop to be able to match all of their consumers’ needs in order to beat the competition (Hayes & Allison, 1998; Schwarz, 2015).
The abundance of offered information can cause an overload of information for consumers, which can lead to misunderstanding and deter further consumer engagement (Rosen & Purinton, 2004). To avoid this, retailers have to carefully design their web shops to optimize the interface of the process to boosts information processing fluency for consumers (Boatwright & Nunes, 2001). By matching their web shop to consumer preferences, retailers can positively influence consumer information processing fluency and thus, can increase consumer browsing behaviour and purchase probability.
By searching for an answer on the research question “How can online retailers best match
consumer preferences to avoid information overload using their size of product assortment and product description in order to influence consumer browsing behaviour and purchase probability by looking at consumer cognitive style?” this research seeks to create more understanding on how to
improve the structuring of web shops with different sizes of assortments and product descriptions by examining consumers’ preferences through their cognitive style and the effect on their browsing behaviour and thus, the acquisition in online retail environments.
An experimental web shop was built in order to gather click-stream data to examine the effect of assortment size and product description size on consumer browsing behaviour and purchase intentions. Furthermore, a survey was conducted to test the influence of consumer cognitive style on consumer preferences for assortment size and product description size. Data obtained by the
experiment and survey was examined using multiple regression analyses to test for mutual relationship between the variables.
Results proved an opposite effect of claims made by previous research to be true: an increase in assortment size raises the purchase probability. When testing the influence of product description size, no significant results were proved. Yet, when testing the influence of consumer cognitive style on the relationship between product description size and purchase probability, it proved that enlarging the product description causes a positively influences purchase probability. Also, one of the cognitive styles proved to be a significant moderator: having a reflective cognitive style negatively influences the positive effect that a large product description has on consumer purchase probability. This result opposes conclusions drawn in previous research. Furthermore, it was found that consumer cognitive style influences the time consumers spent on browsing the web shop.
E. E. Neijzen
5
Table of contents
Introduction
7
The evolution of online retail 7
Product information in retail: is it too much? 7
Previous research on retail environment influencers 8
Determinants of purchase probability
9
An overload of information 9
Consumer preferences 9
Product assortment size 10
Product description size 10
Consumer cognitive style 11
Overview and visualization of the hypotheses 15
Operationalization
16
Research design 16 Sample 16 Research procedure 16 Experimental design 17 Construction of variables 17Pre-testing the manipulated variables 17
Consumer browsing behaviour & Purchase probability 18
Product assortment size & Product description size 18
Consumer cognitive style 19
Control variables 19
Construction of dataset 20
Analyses and Results
21
Testing the influence of information quantity on purchase probability 21
The effect of product assortment size 21
The effect of product description size 22
The effects of the interaction between the sizes of product assortment & product description 22
Testing the influence of consumer cognitive style as a moderator 23
The moderating effect of consumer cognitive style 23
The effect of consumer cognitive style on browsing behaviour 24
Overview and visualization of the results 26
Conclusion and Discussion
28
Concluding the analyses on the influence of consumer preferences 28
Extending product assortment size increases purchase probability 29
No direct effect of product description size 29
The moderating effect of the analytical visualizer cognitive style on the preferred size of product
descriptions 30
The effect of the reflective and intuitive cognitive styles on browsing behaviour 31
References
33
Appendices
39
Appendix A – Overview Literature & Hypotheses 39
Appendix B – Research procedure 41
Appendix C – Survey Design 42
Procedure of the survey 42
Appendix D – Customer journey & Screenshots of the web shop 47
E. E. Neijzen
6
Screenshots of the web shop 48
Appendix E – Product categories 54
Appendix F – Pre-test 55
Procedure of the pre-test 55
Screenshots of the pre-test survey 56
Results of the pre-test 58
Appendix G – Experimental manipulations 59
Overview of the manipulations 59
Screenshots of the manipulations in the web shop 61
Appendix H – Construction of the consumer cognitive style 67 Appendix I – Construction of the control variables 69
Appendix J – Construction of the dataset 70
Data gathering 70
Structure of the dataset 71
Data cleaning 73
Descriptive statistics 73
Reliability and validity 75
E. E. Neijzen
7
Introduction
The evolution of online retail
Applying commerce on the Internet has created an efficient and effective environment for shopping. Electronic commerce (e-commerce)enables consumers to shop anytime and anywhere (Yoon & Occeña, 2015). Great amounts of people spread all over the world have made online purchases and continuously assembling products in great magnitude. In 2010-2014, online retail kept expanding with a 16% annual growth rate (Mintel, 2015). With the rapid movement of prioritizing online shopping and purchasing on electronics and appliances, nothing could prevent shoppers plunging into online retail in the contemporary online era (Shultz & Block, 2015). Today, online retail accounts for 11.6% of all global retail business, and is envisioned to be 15.5% in 2021 (Statista, 2018).
The enormous growth in online retail and the ease of opening an online store resulted in the existence of a wide range of online shopping stores, and thus, a fierce competition between stores. Because consumers differ in their preferences, online retailers have to offer a wide variety of products and information on their web shop to be able to match all of their consumers’ needs in order to beat the competition (Schwarz, 2015).
Product information in retail: is it too much?
While opening their online stores a majority of retailers do not bear in mind the burden that offering a large amount of products and information on their web shop has on their consumers’ minds. Retailers and consumers alike assume that more alternatives relates to greater selections and more gratification (Schwartz, 2004). But providing consumers with too much information can cause a problem:
consumers can feel overwhelmed because of the overload of information on a cognitively demanding online environment, which can negatively effect sales (Rosen & Purinton, 2004). Retailers can try to remedy this by manipulating their web shop to match consumer preferences on the amount of information offered. An efficient web shop that matches consumer preferences can positively contribute to the way consumers process information (Rosen & Purinton, 2004), which can increase the consumer browsing behaviour and purchase probability. To match consumer preferences, it is imperative for retailers to know which variables varying in quantity of information on their web shop prove to impact consumer browsing behaviour and purchase probability. Thus, the overall research question of this study is as follows:
RQ: “How can online retailers best match consumer preferences to avoid information overload using their size of product assortment and product description in order to influence consumer browsing behaviour and purchase probability by looking at consumer cognitive style?”
E. E. Neijzen
8
Previous research on retail environment influencers
Earlier studies provided multiple variables affecting consumer browsing behaviour and purchase probability. Different studies show that assortment size and the quantity of product content in shops have an impact on consumers’ browsing behaviour and purchase intentions (Malhotra, 1982; Iyengar & Lepper, 2000; Chernev, 2006; Shah & Wolford, 2007; Wells, Valacich & Hess, 2011; Dvir & Gafni, 2018). Furthermore, Hayes & Allinson (1994) deem that a person’s cognitive style impacts their information processing and decision-making. Although previous studies provide insight for retailers into the influence of various web shop attributes and consumer preferences on consumer browsing behaviour and motivation for purchasing, not much understanding of the behaviour of e-consumers (online e-consumers) could have been obtained on actual acquisitions (Payne, 1976; Ornstein, 1977; Broniarczyk, Hoyer & McAlister, 1998; Bown, Read & Summers, 2003; Senecal & Nantel, 2004; Zhou, Dai, & Zhang, 2007; Braun, Lee, Urban, & Hauser, 2009; Karimi, Papamichail, & Holland, 2015). Furthermore, existing research on the influence of product content and assortment sizes on consumer decision-making and purchase intentions predominantly concentrates on offline shopping (Payne, 1976; Ornstein, 1977; Malhotra, 1982; Broniarczyk, Hoyer & McAlister, 1998; Iyengar & Lepper, 2000; Bown, Read & Summers, 2003; Chernev, 2006; Shah & Wolford, 2007).
This study will focus on the usability of the web shop by looking at the various influences during online shopping and will seek to create more understanding for retailers on how to improve the structuring of web shops by examining consumer preferences on different product assortments and varying sizes of product descriptions. Studying the usability offers great insight into a user’s experience using a web shop because the facilitator observes users in action and thus, examines real browsing behaviour (Snider & Martin, 2012). This research segments consumers on their preferences by looking at consumer cognitive styles and their preferences on quantity of offered information and will focus on changes in consumer browsing behaviour because of consumer preferences, which can lead to better consumer information processing fluency and thus, an increase in acquisition in online retail.
E. E. Neijzen
9
Determinants of purchase probability
An overload of information
With the evolution of online retail and the rising number of consumers shopping online, the way consumers shop changed. Consumers can compare a wide variety of products and retailers for the best match to fulfil their needs and preferences, which increases the competition between retailers. Brand names and consumer loyalty are important in the online retail, where it all comes back to the
availability of information (Degeratu, Rangaswamy, & Wu, 2000; Danaher, Wilson, & Davis, 2003). Rosen and Purinton (2004) disclose that inadequate product information or availability can hinder acquisitions. However, retailers yielding to the demand of consumers to offer a wide variety of products and information on web shops for consumers to choose from can have a drawback on sales. Consumers can feel overwhelmed because of an overload of information on a cognitively demanding online environment and thus, this overload of information can lead to misunderstanding and deter further consumer engagement (Rosen & Purinton, 2004). When retailers offer consumers a web shop with a great amount of content to browse through, optimizing the processing fluency for consumers is important. For attracting and maintaining consumers to their store, retailers are faced with a challenge: consumers want a wide variety of products to chose from, but at the same time the interface of these products must be easy to process (Boatwright & Nunes, 2001).
Consumer preferences
Online retailers can reduce the consumers’ feeling of being overwhelmed by making use of an efficient way on how information on a web shop is presented (Rosen & Purinton, 2004). A way of doing this is by manipulating their site to match consumer preferences and, in this way, personalizing the content of their web shop based on the individual preferences of the consumer. For this technique is identifying which online aspect greatly influences consumer preferences and the information processing fluency, and thus the determinants of consumer decision-making and motivation in the online shopping process, imperative knowledge for marketers in retail.
Previous studies show that in an online retail environment the consumer processing fluency can be influenced by the sizes of product assortment and product descriptions in the web shop
(Malhotra, 1982; Iyengar & Lepper, 2000; Chernev, 2006; Shah & Wolford, 2007; Wells, Valacich & Hess, 2011; Dvir & Gafni, 2018). Matching these influencers of information quantity in the retail environment to consumer preferences is crucial to retailer success; online retail can use web shop characteristics for consumer customizing in order to improve the purchase probability.
E. E. Neijzen
10
Product assortment size
The studies of Iyengar and Lepper (2000) and Chernev (2006) show the influence of the size of an online product assortment on consumer browsing behaviour. Consumers found the increasing amount of decisions to be made difficult when choosing a product from a larger product assortment versus a smaller product assortment, and thus preferred a smaller assortment. The study of Shah and Wolford (2007) shows that the amount of purchases rises when consumers have to choose from a product assortment consisting from 2 to 10 options, than from a larger choice set of 14 to 20 products. Taking these results into account, the following hypothesis can be phrased:
» Hypothesis 1: Increasing the product assortment size, negatively influences purchase probability.
The influence of the quantity of information on consumer purchase probability, in addition to being caused on product assortment size level, can also be caused by the variety of information on a product description level.
Product description size
An effective and easy way for consumers to process the presentation of all product information on a web shop improves sales and consequently, the successfulness of an online retailer (Wells, Valacich & Hess, 2011). Various studies show that the amount of product information given by different types of product information sources on a web shop effect consumer evaluation and purchase probability. Research conducted by Huizingh (2000) shows that offering plentiful product description is key for the success of a web shop. However, the studies of Wells, Valachic and Hess (2011) and Dvir and Gafni (2018) suggest that having less product content on a web shop is a determinant of acquisitions. When consumers are given too much information, information overload occurs, which leads to poorer decision-making and less purchases (Malhotra, 1982). It can be concluded that when the amount of product description increases, a cognitive overload on consumers occurs, which forces consumers to selectively process the information to deal with the high load of information present. This brings us the next hypothesis:
» Hypothesis 2: Increasing the product description size, negatively influences purchase probability.
Research on the effect of quantity of product information on consumer browsing behaviour and the success of e-commerce web shops thus suggest for retailers with large online assortments to only
E. E. Neijzen
11
include the most important information and attributes. Combining the findings on information quantity and product attributes, the following hypothesis can be made:
» Hypothesis 3: The interaction between the size of product assortment and product description significantly influences purchase probability.
Consumer cognitive style
Braun, Lee, Urban, and Hauser (2009) show that the success of utilizing web shop characteristics, such as the amount of information offered through the size of product assortments and product descriptions, relies on the cognitive style of the consumer. The online behaviour of a consumer is being influenced by that persons’ cognitive style; a person’s favourite method on how to process information (Rosen & Purinton, 2004). Hayes & Allinson (1994) deem that a person’s cognitive style impacts their information processing and decision-making. Furthermore, cognitive style influences how individuals assemble, process and evaluate information (Hayes & Allinson, 1998). For a web shop where every consumer has his or her own cognitive style and thus their own preferences for their ideal product, it is important for the online retailer to offer a variety of products for consumers to chose from (Hayes & Allison, 1998; Schwarz, 2015). This research examines the effect of consumer preferences for the size of product assortments and product descriptions on purchase probability by looking at consumer cognitive styles. Various cognitive style labels exist because of the numerous methods that were adopted in previous studies investigating consumer cognitive style. Studying this categorization of different styles is important, because the usability of a web shop in terms of acquisitions depends on consumer preference, which in turn can be affected by the cognitive style of the consumer (Braun et al., 2009). Studies show that matching information exposure to consumer cognitive styles increases consumer comprehension, click-trough-rates and purchase conversions (Hauser, Urban, Liberali, & Braun, 2009; Urban, Liberali, MacDonald, Bordley, & Hauser, 2013). However, not much is known on the ideal quantity of information in a web shop to show consumers and if an abundance of information effects consumer processing fluency differently for different cognitive styles. Rosen and Purinton (2004) state that an overload of product information can create selectively processing of information and discourage further engagement, but a shortage of
information or product availability can thwart acquisitions.
In this study, the influence of consumer cognitive style is examined on the expected moderating effect it has on the relationship between the two influencers of information quantity (size of product assortment and product description) on consumer browsing behaviour and purchase probability. This brings us to the following hypotheses:
E. E. Neijzen
12
Sub hypotheses aiding in examining the fourth hypothesis are:
» Hypothesis 4A: Consumer cognitive style moderates the relationship between product assortment size and purchase probability.
» Hypothesis 4B: Consumer cognitive style moderates the relationship between product description size and purchase probability.
Consumer cognitive style & information processing fluency
The most adopted cognitive styles in previous research are the three categories used for the twenty-two cognitive styles presented by the Allinson-Hayes Cognitive Style Index (CSI) (Hayes & Allinson, 1996), ranked in a two-way classification. However, other researchers have opposed the quality and importance of cognitive style as the foundation for comprehending individual dissimilarities in behaviour, arguing that the differing aspects of cognitive style can be otherwise be classified by one encompassing category. Hodgkinson and Sadler-Smith (2003) present in their assertion a substituting scoring procedure of a two-categorical model with interacting determinants to provide a superior predictor of responses than the unifactorial solutions of the Allinson-Hayes CSI (Allinson & Hayes, 1996). However, Hayes, Allinson, Hudson and Keasey (2003) claim that the reasoning of Hodgkinson and Sadler-Smith (2003) is flawed, presenting misinterpretations that were formed on the essence and use of the CSI and demonstrate it by conducting exploratory factor analyses to refute the opposing claims.
One of the cognitive styles of the Allinson-Hayes CSI regards to the processing of information: the holistic versus analytic cognitive style (Entwisle, 1981; Miller, 1987). Holistic thinkers process the whole situation at once and blend all usable information, whereas analytical thinkers process information in a consecutive, linear way (Ornstein, 1977). Other researchers created different labels regarding to the processing of information: the dependent versus independent
cognitive style (Goodenough, 1986) and the analytic versus nonanalytic cognitive style (Messick & Kogan, 1963). A cognitive style regarding a consumers preferred visualization of information processing is the verbal versus visual cognitive style, which distinguishes between individuals processing information verbally or in mental images (Childers, Houston, & Heckler, 1985; Paivio, 2013; Richardson, 1977; Riding, 2001). A cognitive style derived from the holistic versus analytic style describes the intuitive and the analytical individual (Hayes, Allinson, Hudson & Keasey, 2003). Individuals who are intuitive in processing information tend to process in a holistic way and first observe cues of the provided information (Olson, 1985). Hodgkinson and Sandler-Smith (2003) state that the measurement of the intuition-analysis cognitive style is formed on a unitary perception of the
E. E. Neijzen
13
design. However, Hayes, Allinson, Hudson and Keasey (2003) conclude that that claim fails to offer a real challenge to either the theoretical or empirical arguments supporting the construct validity of the CSI. The study of Soane, Schubbert, Lunn and Pollard (2015) supports the use of CSI and explain that analytical thinkers process information structurally and like to base their decision on an abundance of information; however, intuitive individuals process only the information given at once and thrives on more text-based web shops.This research studies the possible different effects of various cognitive styles on consumer preferences for the ideal quantity of information in a web shop. Thus, the hypothesis will be:
» Hypothesis 4C: Purchase probability is higher for consumers with an analytical cognitive style than for consumers with an intuitive cognitive style, when product descriptions size increases.
Consumer cognitive style & browsing behaviour
Studying various cognitive styles regarding consumer browsing behaviour can aid in uncovering consumer preferences and thus, can be used to better personalize webs shop content of online retail for possible greater purchase conversions. Interesting outcome variables of browsing behaviour to study consumer preferences are: time spent in web shop, number of clicks, and average time spent per
webpage, where a webpage indicates a page in the web shop (e.g. the landing page or check-out page).
The average time spent in one web shop session is a quotient of both the amount of page views and the durations of the views (Bucklin & Sismeiro, 2009).
A cognitive style that can aid in uncovering consumer preferences is the leader versus
follower cognitive style (Braun, Lee, Urban, & Hauser, 2009). Because this leadership cognitive style
is pegged as a crucial indicator for the adaptation of new information sources and products (Rogers, 1962; Rogers & Stanfield, 1968), it is expected to influence consumer preferences for information quantity in web shops.
Another category of the consumer cognitive styles most discussed regarding browsing
behaviour, especially the responsiveness of individuals, is: reflective versus impulsive cognitive styles. A reflective response is a response that demands effort and the implementation of learned rules and an impulsive response is an instinctive reaction that arises without much thought (Frederick, 2002). Different outcomes about this cognitive style have been brought about: individuals with a reflective cognitive style are more flexible in their thinking and have a higher willingness to invest time than individuals with a more impulsive cognitive style, but a reflective process is however more time-consuming (Strack & Deutsch, 2006). The following hypothesis is proposed:
E. E. Neijzen
14
» Hypothesis 5: Consumer cognitive style significantly influences the browsing time in the web shop.
» Hypothesis 5A: Time spent in the web shop is higher for people with a reflective cognitive style than for people with an impulsive cognitive style.
Building on the previous discussed literature that individuals with reflective cognitive styles are willing to spend more time in investigating (Frederick, 2005), it can be suggested that consumers with a reflective cognitive style will spent more time per webpage and will make more clicks to properly investigate the web shop than consumers with an impulsive cognitive style. These assumptions bring us to the following hypotheses:
» Hypothesis 6: Consumer cognitive style significantly influences the average time spent per webpage.
» Hypothesis 6A: The average time spent per webpage is higher for people with a reflective cognitive style than for people with an impulsive cognitive style.
» Hypothesis 7: Consumer cognitive style significantly influences the number of clicks.
» Hypothesis 7A: The number of clicks is higher for people with a reflective cognitive style than for people with an impulsive cognitive style.
Based on previous discussed literature, the same assumption can be made for people with an analytical cognitive style: analytical thinkers like to base their decision on larger amounts of information
compared to intuitive thinkers (Soane et al, 2015). Consumers with an analytical cognitive style will make more clicks to properly investigate the web shop, as argued in the following hypothesis:
» Hypothesis 7B: The number of clicks is higher for people with an analytical cognitive style than for people with an intuitive cognitive style.
E. E. Neijzen
15
Overview and visualization of the hypotheses
See Appendix A for an overview of the discussed literature (Table 1) and the proposed hypotheses (Table 2). The results of the most applicable literature debated were used to define the hypotheses. Based on the discussed literature and the formulated hypotheses, a conceptual framework is drawn (Figure 1).
E. E. Neijzen
16
Operationalization
Research design
Concerning the methodology of this study, a conclusive design centring on cause-and-effect relationships between multiple variables is conducted. The causality of these relationships is tested through a controlled experimental study in a factorial 2 x 2 between-subjects design, which examines the proposed hypotheses concerning the possible effects of consumer cognitive style, different sizes of product assortments and product descriptions by studying changes in consumer browsing behaviour and purchase probability (Figure 2). A between-subjects design was chosen to minimize the learning and transfer across conditions (Abutabenjeh & Jaradat, 2018). To assure that there is no negative effect on the external and internal validity, the groups are randomly selected with comparable exposure to external factors. The experiment will be administered online and will consist of a survey and a browsing assignment.
Figure 2. Experimental conditions
Sample
The sample of participants consists of all ages with the exception of consumers below the age of 18, for many web shops make use of minimum ages where minors need permission from parents to shop online. To ensure the heterogeneity of the sample, no restrictions are set on the nationalities of the respondents. More information about the sample of this study will be discussed in the paragraph describing the construction of the dataset.
Research procedure
The research occurs through an online survey in Qualtrics where participants complete a browsing assignment and answer multiple open and closed answer questions. A survey for consumers was conducted to provide insight in participant’s attitudes that can influence their behaviour and purchases (Groves, Fowler, Couper, Lepkowski, Singer & Tourangeau, 2011). See Appendix B for a description of the research procedure. A summary of the procedure is shown in Figure 3 in Appendix B. Every
E. E. Neijzen
17
participant is presented with the same survey, and thus, answers the same questions. See Appendix C for a description of the survey design. The browsing assignment shows how consumers behave online and tests for the most optimal customer journey, focussing on consumer in-store activities on browsing behaviour and acquisitions. For conducting a browsing assignment, an experimental web shop is built. The experimental part of this study lies in the browsing assignment; manipulations are made in the web shop.
Experimental design
The primary data of the experiment is collected through the browsing assignment, for which a web shop was built. The experimental web shop is created by making use of a platform that hosts templates of online retailers. To acquire more insight in how consumers behave online, the research of this study focuses on the in-store activities and the different phases participants pursue during their customer journey. This way, the most ideal conversion path will be revealed.
The web shop consists of a number of content elements (pages) that are linked to each other (Huizingh, 2000). The participant goes through all the pages of the web shop, and thus all the phases of the in-store customer journey, when buying a product from the experimental web shop. See Appendix D for a description of the in-store customer journey and screenshots of the different pages of the web shop.
By building a web shop, different web shop characteristics, such as the sizes of product assortments and product descriptions, can be manipulated. Two different product categories are made. To ensure an even level of interest of the participants in the two product categories, both categories are related to each other and are of interest to men and women. For a description of the products used in the web shop, see Appendix E. Participants are randomly assigned to the different conditions to ensure that the internal and external validity are not negatively influenced. Because participants were able to behave as naturally in the experimental web shop as they would in a real web shop, the external validity of the measurements done in the experiment can be seen as relatively high. Furthermore, potential extraneous variables could be controlled because of the experimental web shop, which is positive for the internal validity.
Construction of variables
Pre-testing the manipulated variables
By doing a pre-test, it can be decided that the proposed web shop manipulations are indeed effective afore sharing the study with the target population (Reynolds, Diamantopoulos & Schlegelmilch,
E. E. Neijzen
18
1993). Thus, the pre-test proves whether the manipulations of the anticipated effects of product assortment size and product description size are just. Participants of the pre-test visited the web shop and were asked to complete multiple tasks and questions. A description of the procedure, an overview of the survey, and the results of the pre-test can be found in Appendix F.
Consumer browsing behaviour & Purchase probability
This study knows two dependent variables: consumer browsing behaviour and purchase probability. The purchase probability can be derived from the percentage of customers buying from an online store using the term conversions (Li & Kannan, 2014). A purchase conversion is derived from the choice each participant has to make: do they purchase a product in the experimental web shop or not? The outcome of this variable is thus dichotomous (1 = conversion, 0 = no conversion). Consumer browsing behaviour is illustrated by clickstream data, which can attain all clicks made on the web shop and shows researchers how participants operate over time and how en when clicks were made (Lee, Podlaseck, Schonberg & Hoch, 2001; Moe & Fader, 2004). This way, the online behaviour of each visitor of the web shop can be collected into large datasets of clickstream data and converted into useful variables for further analyses. Furthermore, using clickstream data makes it possible to study actual purchase conversions instead of intentions, which are vulnerable to consideration biases (Young, DeSarbo & Morwitz, 1998). The possible effect of consumer cognitive style on consumer browsing behaviour is determined by looking at the number of clicks, browsing time, and the average time spent per page. The number of clicks and the browsing time of every participant can be
established by looking at the clickstream data, which can thereafter be used to determine the average time spent per page. Browsing time and time spent per page are continuous variables, and number of clicks is discrete, however, all three variables are numerical.
Product assortment size & Product description size
The two independent variables product assortment size and product description size are manipulated in the web shop by their two varying quantities: the small (6) or large (18) assortment size and the size of the product description given with each product. For an overview of the manipulations, see Table 3 and 4 in Appendix F. As discussed in the section describing the determinants of purchase probability, previous studies used differing sizes for assortment size: 2-6 for small and 14-36 for large assortments. When deciding on the sizes, the layout of the experimental web shop was taken into account; the same amount of products must be portrayed in each row to keep potential differing displays from
manipulating the results. Thus, when presenting 3 products per row for optimal visibility, the quantity of the varying assortment sizes is decided on 6 and 18 products for both product categories and each condition. This manipulation appears in the shopping page of the web shop by showing the participant either 6 or 18 products for both product categories.
E. E. Neijzen
19
The effect on the manipulation of the product description size is examined by the use of a short product description with 3 features or a long description with 10 features. The quantity of information of the descriptions is determined by consulting previous research, as discussed in the theological section. Each product category has the same amount of products with short or long descriptions. The product description is shown on the product page, which participants access by selecting a product on the shopping page. For screenshots of the manipulations in the web shop, see Appendix G.
Consumer cognitive style
This research studies the possible moderating effect of consumer cognitive style on the relationships between the independent variables (product assortment size and product description size) and the dependent variables (consumer purchase probability and browsing behaviour). Ten scales of the system of Braun et al. (2009) are conducted to assess participants’ cognitive styles and thus, their web shop preferences. Participants can disclose their preference differing from “Strongly disagree to “Strongly agree” in the five-point Likert scale. By using these multi-item scales, it minimizes errors in the measurement of the cognitive style, which increases the reliability. Furthermore, it shows the heterogeneity of this construct, which improves the validity of the measurement. The system of Frederick (2005) is conducted at the three open questions to measure participants’ impulsive answer to an analytical dilemma. To classify participants as impulsive opposed to reflective, the answers should include 24 days, 10 cents or 100 minutes. A dummy variable is constructed to segregate answers (1 = impulsive, 0 = reflective). The cognitive styles conducted in this study are examined on internal consistency by a sample specific factor test, preceding the hypotheses testing. See Appendix H for a description of the checks made to verify the collective determinants and the underlying influences of the items measuring cognitive style.
Control variables
Four differing demographic and three other variables are included at the end of the survey to control for extraneous factors that can influence the effect of the independent variables on the dependent variables. See Appendix I for a description of the control variables.
E. E. Neijzen
20
Construction of dataset
To measure the effect of the independent variables on the dependent variables, the dataset made by the participants of this study is collected and cleaned to then interpret the results derived from the data. From the 13th of May to the 25th of May 2019, 335 participants have correctly completed the survey in
Qualtrics. During this timespan, participants together produced 3647 clicks in the experimental web shop. For detailed descriptions of the data gathering, the structure of the dataset, cleaning of the data, the descriptive statistics, and the validity of the dataset, see Appendix J.
E. E. Neijzen
21
Analyses and Results
Regression analyses are conducted to study the possible effect of quantity of product information on consumer purchase probability and browsing behaviour, by testing the size of product assortment and product description. Additionally, regression analyses are conducted to study the possible effects of consumer cognitive style on consumer purchase probability and browsing behaviour. These possible effects are studied by examining the possible moderating relationships between consumer cognitive style and both independent variables (product assortment size and product description size) on consumer purchase probability and browsing behaviour.
Testing the influence of information quantity on purchase probability
Binominal logistic regression analyses are conducted to study the possible effect of the quantity of product information on consumer purchase probability, by testing the size of the product assortment and the product description. See Appendix Kfor the binominal logistic regressions conducted to study the proposed relationships between the independent variables and the dependent variables.
The conducted regression analyses aid in comprehending the effect the independent variables have on the dependent variable and thus, aid in verifying the credibility of the hypotheses. The positive coefficients of both the product assortment size variable and the product description size variable suggests that an increase in the sizes of both the independent variables leads to an increase in the possibility of an occurring consumer purchase conversion. The addition of the control variables in the regression analyses shows that the effect of the payment dummies decreases the purchase
probability, whereas the more webpages visited, the higher the purchase probability. Odd ratios show the degree of influence variables have on the dependent variable. It indicates the increased chance of an occurring conversion when an independent variable increases or decreases by one unit (Mood, 2010).
The effect of product assortment size Results Hypothesis 1
The first hypothesis of this research predicts the effect of the product assortment size on consumer purchase probability:
» Hypothesis 1: Increasing the product assortment size, negatively influences purchase probability.
E. E. Neijzen
22
The regression analyses show that the effect of product assortment size is significant when controlled for other variables (Wald X2(1, N=335)=5.898, p=0.014). The results prove that the possibility that a
purchase conversion takes place is 1.989 times higher with a large assortment than with a small assortment. Hypothesis 1 is rejected based on the fact that the effect of product assortment size on purchase probability proved to be reversed from the expected relationship.
The effect of product description size Results Hypothesis 2
The second hypothesis predicts the effect of product description size on consumer purchase probability:
» Hypothesis 2: Increasing the product description size, negatively influences purchase probability.
The regression analyses show that the effect of product description size is not significant (Wald X2(1,
N=335)=2.233, p=0.159). Because no significant effect of product description size on consumer purchase conversion was found, Hypothesis 2 is rejected.
The effects of the interaction between the sizes of product assortment & product description Results Hypothesis 3
The third hypothesis predicts the effect the interaction of the independent variables has on consumer purchase probability:
» Hypothesis 3: The interaction between the size of product assortment and product description significantly influences purchase probability.
The regression analyses prove that the interaction between the independent variables of product assortment size and product description size has no significant effect on consumer purchase probability (Wald X2(1, N=335)=0.479, p=0.646). Thus, the third hypothesis is rejected.
While analysing the results of the regression analyses, some interesting effects of the control variables on the dependent variable are shown. Results show that the webpage variable has a
significant effect on consumer purchase probability (Wald X2(1, N=335)=22.331, p=0.000): the
chance of a purchase conversion is 1.171 times higher when the number of visited pages increases with one. Furthermore, the payment dummies have a significant positive effect on consumer purchase conversion. This may have been due to the fact that only a small group of participants was paid the lowest amount of money, of which only half of the participants bought a product in the web shop.
E. E. Neijzen
23
Testing the influence of consumer cognitive style as a moderator
The moderating effect of consumer cognitive style
To be able to analyse the potential influence of consumer cognitive style on consumer purchase probability, first the moderating relationship between consumer cognitive style and both independent variables (product assortment size and product description size) and de dependent variable (consumer purchase probability) must be explored. These supposed moderating relationships are studied using the interaction between the product assortment size and product description size. See Appendix K for the results of the binominal logistic regression analyses conducted to examine the possible effects of the independent variables on the dependent variables and the moderating effect between these
relationships. After the studies on the effect on consumer purchase probability, the expected moderating effect of consumer cognitive style on consumer browsing behaviour can be explored.
Results Hypothesis 4
The conducted regression analyses serve to explain the effect of consumer cognitive style on the relationships between the other variables. The fourth hypothesis predicts the effect it has on consumer purchase probability:
» Hypothesis 4: Consumer cognitive style significantly influences purchase probability.
Sub hypotheses aiding in examining the fourth hypothesis:
» Hypothesis 4A: Consumer cognitive style moderates the relationship between product assortment size and purchase probability.
» Hypothesis 4B: Consumer cognitive style moderates the relationship between product description size and purchase probability.
The regression analyses show that the previous significant effect of assortment size on consumer purchase probability ceases to be when consumer cognitive styles are added. Thus, Hypothesis 4A is rejected. However, adding the variables of consumer cognitive styles and their interactions ensures that the effect of product description size turns slightly significant (Wald X2(1, N=335)=3.412,
p=0.065). Results show that participants who where bestowed upon with an extensive product description as to a short description, were 4.379 likely to make a purchase conversion. For examining
E. E. Neijzen
24
the possible moderating effect of consumer cognitive style on the relationship between this significant independent variable and purchase probability, the interacting variables created with the different cognitive styles must be studied. The outcomes show a slight significant effect of the interaction between product description size and the cognitive style of the Analytical visualizer-Reflective-Leader (Wald X2(1, N=335)=3.467, p=0.058). Thus, it can be concluded that Hypothesis 4B is marginally
supported. This brings us to the final sub hypothesis to explore the effect of consumer cognitive style:
» Hypothesis 4C: Purchase probability is higher for consumers with an analytical cognitive style than for consumers with an intuitive cognitive style, when product descriptions size increases.
Results show a slight significant negative moderating effect of the Analytical visualizer-Reflective-Leader cognitive style on the relationship between the size of product description and consumer purchase probability. In other words: the positive effect on purchase probability because of the increase of information in product description diminishes when the consumer cognitive style is that of the Analytical visualizer-Reflective-Leader. Hypothesis 4C is rejected based on the fact that the effect of the consumer cognitive style on the relationship between product description size and consumer purchase probability proved to be reversed from the one that was expected.
The effect of consumer cognitive style on browsing behaviour
Multiple linear regression analyses are conducted to study the effect of consumer cognitive style on the variables: average duration per page, overall browsing time and the number of clicks. These variables are used in this research to study consumer browsing behaviour. See Appendix K for the results of the conducted regression analyses.
Results Hypothesis 5
The expected moderating effect of consumer cognitive style on consumer browsing behaviour is tested trough linear regressions. The hypothesis predicting the effect of consumer cognitive style on the browsing time in the web shop is as follows:
» Hypothesis 5: Consumer cognitive style significantly influences the browsing time in the web shop.
E. E. Neijzen
25
In particular the effect is studied of the reflective-impulse consumer cognitive style on the time consumers spent in a web shop:
» Hypothesis 5A: Time spent in the web shop is higher for people with a reflective cognitive style than for people with an impulsive cognitive style.
Results of the analysis show that the consumer cognitive styles of Analytical visualizer-Impulsive-Follower (t= -2.822, p= 0.006) and the Analytical visualizer-Impulsive-Leader (t=-2.534, p=0.013) to be significantly effective and the effect of the Analytical verbalizer-Impulsive-Follower to be slightly significant, when using the Intuitive verbalizer-Reflective-Leader as a baseline. These three cognitive styles, all including the impulsive cognitive style, lessen the browsing time of participants with -134.611, -107.879 and -94.112 seconds. Thus, results show that consumers having an impulsive cognitive style spent less time browsing in the web shop than consumers with a reflective cognitive style. Hence, the Hypothesis 5 is supported in its entirety.
Results Hypothesis 6
The influence of consumer cognitive style on the average time consumers spent per webpage is examined using the subsequent hypotheses:
» Hypothesis 6: Consumer cognitive style significantly influences the average time spent per webpage.
In particular the effect is studied of the reflective-impulse consumer cognitive style on the average time consumers spent per webpage:
» Hypothesis 6A: The average time spent per webpage is higher for people with a reflective cognitive style than for people with an impulsive cognitive style.
Results show that other than the Analytical visualizer-Reflective-Follower, all styles have a
significantly negative effect on the average time participants spent per webpage. The analyses show that participants with cognitive styles encompassing the reflective cognitive style, namely the Analytical visualizer-Reflective-Leader (t=-2.267, p=0.014) and the Intuitive verbalizer-Reflective-Follower (t=-1.899, p=0.024), have a significantly higher average time spent per page. These results show that Hypothesis 6 is wholly supported.
E. E. Neijzen
26
Results Hypothesis 7
Lastly, the hypothesis examining the effect of consumer cognitive styles on the number of clicks is:
» Hypothesis 7: Consumer cognitive style significantly influences the number of clicks.
In particular the effect is studied of the reflective-impulse consumer cognitive style on the number of clicks consumers make:
» Hypothesis 7A: The number of clicks is higher for people with a reflective cognitive style than for people with an impulsive cognitive style.
Also, the effect is studied of the analytical-intuitive consumer cognitive style on the number of clicks consumers make:
» Hypothesis 7B: The number of clicks is higher for people with an analytical cognitive style than for people with an intuitive cognitive style.
The analyses prove that the number of clicks made by participants is being influenced by their cognitive styles, and thus Hypothesis 7 is supported. Compared to the Intuitive verbalizer-Reflective-Leader cognitive style, cognitive styles 2, 3, 6 and 8 are significantly effective on the number of clicks and the effect of cognitive style 5 is slightly significant. Results show that Hypothesis 7A must be rejected, since it cannot be proven that consumers with a reflective cognitive style also have a higher number of clicks. However, results do show that participants with an analytical visualizer cognitive style have a higher number of clicks. Compared to participants with the Intuitive
verbalizer-Reflective-Leader cognitive style, participants with cognitive styles 2, 6 and 8 have 3.112, 2.971 and 5.256 more clicks. Thus, consumers with an analytical cognitive style have a higher number of clicks, supporting Hypothesis 7B.
Overview and visualization of the results
The results of the conducted regression analyses are shown in Table 16. Also, a conceptual framework visualizes the results of the tested hypotheses and can be seen in Figure 10.
E. E. Neijzen
27
E. E. Neijzen
28
Conclusion and Discussion
This study aims to find a solution for diminishing sales in online retail because of an abundance of information offered in online retail environments. By studying the use of manipulating online retail attributes to consumer preferences, this study tries to investigate the effect of the manipulated online retail attributes on consumer browsing behaviour and purchase probability. By testing manipulations of product information of leading variables in a web shop interface and the influence of cognitive styles of consumers, this study attempts to find a solution to the following research question:
“How can online retailers best match consumer preferences to avoid information overload using their size of product assortment and product description in order to influence consumer browsing
behaviour and purchase probability by looking at consumer cognitive style?”
Concluding the analyses on the influence of consumer preferences
By means of a browsing assignment in an experimental web shop, the effect of information quantity of the sizes of product assortment and product description on purchase probability was tested.
For the browsing assignment, the use of a laptop or computer was mandatory because the experimental web shop could not be opened on a mobile devise (e.g. tablet or mobile phone). This could have influenced the results of this study. Nowadays, a mobile device is frequently used for online shopping. Making use of a smaller display because of the smaller devise while shopping online can possibly result in a different consumer browsing behaviour.
Future research can prevent this by making sure that participants are also able to conduct the survey on a mobile devise.
When starting the browsing task, participants had to click on a link to be redirected to the web shop. The opening of the experimental web shop in a different screen happens automatically by using the given link. However, when finishing the browsing task, participants had to switch back to the survey themselves to start answering the survey questions. This could have caused a loss of
participants and thus, could have resulted in the exclusion of data from participants that did not finish the entire survey.
Also, when switching back to the survey, participants had to insert their unique user ID that was presented to them in the upper corner of the screen the entire time while conducting the browsing task. This requirement could have increased the difficulty in completing the survey correctly and ultimately could have caused loss of data.
Furthermore, with the use of Mechanical Turk 335 participants conducted a survey to study the possible moderating influence of consumer cognitive style on the effects of the sizes of product assortment and product description on consumer browsing behaviour and purchase probability.
E. E. Neijzen
29
Because Mechanical Turk is an Amazon featured crowdsourcing marketplace, the use of this American marketplace for the gathering of data caused the sample of participants to be quite homogenous in their nationality; a majority of the participants had an American nationality.
Future research should make use of a heterogeneous sample to be able to do cross-cultural comparisons for further generalization of future research in this field. Also, the use of Mechanical Turk could have caused a lack in motivation among the participants, because Mechanical Turk does not serve as a main source of incomes for most participants (Paolacci et al., 2010). This lack in motivation could have made the performance of participants to be in odds with their real life shopping performance and could have influenced the results used in the analyses of the study.
Extending product assortment size increases purchase probability
Previous research states that providing consumers with too much information can lessen purchase probability (Rosen & Purinton, 2004). Retailers can try to remedy this by manipulating their website to match consumer preferences on the amount of information offered. Researchers found that the expanding choice that comes with an increase in assortment size leads to a reduction in consumer purchase probability (Iyengar & Lepper, 2000; Chernev, 2006).
However, binominal logistic regression analyses proved an opposite effect to be true: an increase in assortment size causes a positively significant effect on purchase conversions. Results of this study can be due to the possibility that because offline stores offer a wide variety of products, consumers are accustomed to the amount of choices offered and find the availability of only a small assortment a lack in product choice (Chernev, 2013).Thus, an increase in assortment size raises the purchase probability.
This shows us that previous conclusions of earlier literature alter from the conclusion drawn in this study and thus, conclusions made in previous research cannot be blindly consulted to real retail environments in online retail and that more research on this field should be conducted. The dissension may be due to the fact that previous research was mostly conducted in offline retail (Iyengar & Lepper, 2000; Chernev, 2006). Because this research experimented in a self-constructed online web shop, more insight is given in consumer preference of the structuring of information attributes of online retail environments.
No direct effect of product description size
On the basis of previous conducted research it was concluded that when the amount of product description increases, it causes an overload of information forcing consumers to selectively process information and thus, decreases consumer purchase conversion (Malhorta, 1982; Wells, Valachis & Hess, 2011; Dvir & Gafni, 2018).
E. E. Neijzen
30
When testing the influence of product description size on purchase conversions of the participants, the regression model was not significant. This result can be caused by the quantity of product information for the two product description sizes tested in this study versus previous studies. Preceding research tested an overload of information present for product descriptions including fifteen of more product characteristics, while this study tested the effect of a maximum of eight product characteristics present in the product description (Malhorta, 1982).
Future research can study solutions for the increasing overload of information by product description size including fifteen or more product characteristics. An example of a solution could be the pull down function with extra product descriptions that online retail environments nowadays offer to consumers. This function would allow online retail environments to thus not be with or without vast amounts of product description and to empower the consumer to get more information if they feel the need for it. Future research should investigate the influence of such solutions on the negative effect of information overload on consumers and if it does indeed increases the purchase probability.
The moderating effect of the analytical visualizer cognitive style on the preferred size of product descriptions
Earlier studies show the influence consumer cognitive styles have on the assessment of various web shop attributes and the intention to purchase (Braun et al., 2009; Hauser et al. 2009; Urban et al., 2013).
When testing the effect of cognitive style on the relationship between product description size and purchase conversions, a binominal logistic regression analysis proved that enlarging the product description causes a significant positive effect on consumer purchase probability. The conducted regression analyses proved one of the cognitive styles as a significant moderator: having an analytical visualizer cognitive style negatively influences the positive effect that a large product description has on consumer purchase probability.
This result opposes the claim stated in previous research that consumers with an analytical cognitive style, in comparison to consumers with an intuitive cognitive style, appreciate a greater amount product characteristics described while shopping (Soane et al., 2015). The contradiction can be explained by the fact that the addition of product characteristics was limited to only verbal
descriptions and no pictured additions, while analytical persons prefer visualized over textual
descriptions. Future research should prove if consumers with different cognitive styles disagree in the way they prefer textual or visual presentation of information and if this has consequences for purchase probability.
The results provide online retailers with the answer that consumer cognitive style does indeed have an effect on consumer preferences and that retailers should alter various web shop characteristics into the most favourable for consumers and thus, the most profitable for them. It is probable that an
E. E. Neijzen
31
abundance of information increases the difficulty of making decisions. Although offering more products can be profitable as results of this study on product assortment size showed, retailers must be mindful in designing their web shop and the quantity of information presented to the possible different preferences of consumers. Manipulating the visual presentation of product assortments in the web shop can reduce this phenomenon. Because of the online environment of web shops, online retailers can easily manipulate the presentation of product assortments. Further research should be done to examine the possible solutions.
The effect of the reflective and intuitive cognitive styles on browsing behaviour
In addition to testing cognitive styles as a moderating variable between the two quantities of product information (product assortment size and product description size) and consumer purchase probability, the influence of cognitive styles on consumer browsing behaviour is also tested by the variables browsing time, average time spent per webpage and number of clicks. The outcomes prove that consumer cognitive style does influence the browsing time, average time spent per webpage and the number of clicks. Multiple linear regression analyses prove that browsing time is negatively influenced (meaning that the browsing time lessened) for participants with the three cognitive styles of the Analytical visualizer-Impulsive-Follower, the Analytical visualizer-Impulsive-Leader and the Analytical verbalizer-Impulsive-Follower.
This shows that consumers with an impulsive cognitive style are inclined to spent less time while shopping than people with a reflective cognitive style, which is in line with a previous study on the effect of consumer cognitive style on browsing time that showed that people with a deliberative cognitive style are inclined to spent more time while fulfilling an assignment (Frederick, 2005).
The multiples linear regression analyses present that consumer cognitive style also has a significant effect on the average time consumers spent per webpage: consumers with a reflective cognitive style spent on average more time per webpage, opposed to consumers with an impulsive cognitive style.
Additionally, consumer cognitive style has an effect on how many clicks someone makes in a web shop. The outcomes show that, opposed to people with an intuitive cognitive style, people with an analytical cognitive style make more clicks and thus, view more pages of a web shop. This result shows us that consumers with different cognitive styles disagree in their willingness to take time for browsing a web shop. Retails should take into account the different preferences of their consumers. The results of this study provide insights into the influence of consumer preferences on the success of online retail environments. With the continuous software developments for gathering experimental data nowadays, it grants researchers the opportunity to acquire more knowledge on the influence of manipulating different web shop characteristics, presentation and content.
E. E. Neijzen
32
Future research should further investigate the effect of different retail environmental attributes and the influence of consumer cognitive style on browsing behaviour and thus, create guidelines for retailers on how to alter various web shop characteristics into the most favourable for consumers and the most profitable for business.
E. E. Neijzen
33
References
Abutabenjeh, S., & Jaradat, R. (2018). Clarification of research design, research methods, and research methodology: A guide for public administration researchers and practitioners. Teaching Public
Administration, 36(3), 237–258.
Allinson, C., & Hayes, J. (2012). The cognitive style index: Technical manual and user guide.
Boatwright, P., & Nunes, J.C. (2001). Reducing assortment: An attribute-based approach. Journal of
marketing, 65(3), 50-63.
Bown, N.J., Read, D., & Summers, B. (2003). ‘The lure of choice’. Journal of Behavioral Decision
Making, 16, 297–308.
Braun, M., Lee, C., Urban, G., & Hauser, J.R. (2009). Does Matching Website Characteristics to
Cognitive Styles Increase Online Sales? Retrieved from
http://web.mit.edu/~hauser/www/Papers/Braun-Lee-Urban-Hauser%20Web%20Style%20Match%2012-03-09.pdf
Broniarczyk, S.M., Hoyer, W.D., & McAlister, L. (1998). Consumers’ perceptions of the assortment offered in a grocery category: The impact of item reduction. Journal of Marketing Research, 35, 166–176.
Bucklin, R.E., & Sismeiro, C. (2009). Click here for Internet insight: Advances in clickstream data analysis in marketing. Journal of Interactive Marketing, 23(1), 35-48.
Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon's Mechanical Turk: A new source of inexpensive, yet high-quality, data? Perspectives on psychological science, 6(1), 3-5.
Chernev, A. (2006). Decision focus and consumer choice among assortments. Journal of Consumer
Research, 33(1), 50-59.
Childers, T.L., Houston, M.J., & Heckler, S.E. (1985). Measurement of individual differences in visual versus verbal information processing. Journal of Consumer Research, 12(2), 125- 134.
E. E. Neijzen
34
Danaher, P. J., Wilson, I.W., & Davis, R.A. (2003). A comparison of online and offline consumer brand loyalty. Marketing Science, 22(4), 461-476.
Degeratu, A. M., Rangaswamy, A., & Wu, J. (2000). Consumer choice behavior in online and traditional supermarkets: The effects of brand name, price, and other search attributes.
International Journal of research in Marketing, 17(1), 55-78.
Dvir, N., & Gafni, R. (2018). When Less Is More: Empirical Study of the Relation Between Consumer Behavior and Information Provision on Commercial Landing Pages. Informing Science: The
International Journal of an Emerging Transdiscipline, 21, 019-039.
Entwistle, N. (1981). Styles of Learning and Teaching. New York: John Wiley & Sons.
Frederick, S. (2002). Automated choice heuristics. New York: Cambridge University Press, 548- 58.
Frederick, S. (2005). Cognitive reflection and decision making. Journal of Economic perspectives,
19(4), 25-42.
Goodenough, D. R. (1986). History of the field dependence construct. In Field dependence in
psychological theory, research, and application-two symposia in memory of Herman A. Witkin,
5-13.
Groves, R. M., Fowler Jr, F. J., Couper, M. P., Lepkowski, J. M., Singer, E., & Tourangeau, R. (2011). Survey methodology, 561. John Wiley & Sons.
Hayes, J., & Allinson, C.W. (1994). Cognitive style and its relevance for management practice. British
journal of management, 5(1), 53-71.
Hayes, J., & Allinson, C.W. (1998). Cognitive style and the theory and practice of individual and collective learning in organizations. Human Relations, 51.
Hayes, J., Allinson, C.W., Hudson, R., & Keasey, K. (2003). Further reflections on the nature of intuition-analysis and the construct validity of the Cognitive Style Index. Journal of Occupational
and Organizational Psychology, 76(2), 269-278.
Hauser, J.R., Urban, G.L., Liberali, G., & Braun, M. (2009). Website morphing. Marketing Science,
E. E. Neijzen
35
Hodgkinson, G. P., & Sadler‐Smith, E. (2003). Complex or unitary? A critique and empirical re‐ assessment of the Allinson‐Hayes Cognitive Style Index. Journal of Occupational and
Organizational Psychology, 76(2), 243-268.
Huizingh, E. K. (2000). The content and design of web sites: an empirical study. Information &
Management, 37(3), 123-134.
Iyengar, S.S., & Lepper, M.R. (2000). When choice is demotivating: Can one desire too much of a good thing?. Journal of personality and social psychology, 79(6), 995.
James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An introduction to statistical learning (Vol. 112). New York: Springer.
Karimi, S., Papamichail, K.N., & Holland, C.P. (2015). The effect of prior knowledge and decision-making style on the online purchase decision-decision-making process: A typology of consumer shopping behaviour. Decision Support Systems, 77, 137-147. doi:10.1016/j.dss.2015.06.004
Lee, J., Podlaseck, M., Schonberg, E., & Hoch, R. (2001). Visualization and analysis of clickstream data of online stores for understanding web merchandising. Data Mining and Knowledge
Discovery, 5(1/2), 59-84.
Li, H., & Kannan, P.K. (2014). Attributing conversions in a multichannel online marketing
environment: An empirical model and a field experiment. Journal of Marketing Research, 51(1), 40-56.
Malhotra, N.K. (1982). Information load and consumer decision making. Journal of consumer
research, 8(4), 419-430.
Mason, W., & Suri, S. (2012). Conducting behavioral research on Amazon’s Mechanical Turk.
Behavior research methods, 44(1), 1-23.
Messick, S., & Kogan, N. (1963). Differentiation and compartmentalization in object-sorting measures of categorizing style. Perceptual and Motor Skills, 16(1), 47-51.
