PERSONALITY PREFERENCES IN E-‐COMMERCE: A CHOICE-‐BASED CONJOINT ANALYSIS ON TEXTUAL AND VISUAL ASPECTS OF WEBSITE DESIGN

(1)

PERSONALITY PREFERENCES IN E-‐COMMERCE: A

CHOICE-‐BASED CONJOINT ANALYSIS ON TEXTUAL AND

VISUAL ASPECTS OF WEBSITE DESIGN

Anonymized version

Lori Spruijt

(2)

(3)

PERSONALITY PREFERENCES IN E-‐COMMERCE: A

CHOICE-‐BASED CONJOINT ANALYSIS ON TEXTUAL AND

VISUAL ASPECTS OF WEBSITE DESIGN

Anonymized version

Lori Spruijt

University of Groningen Faculty of Economics and Business

Department of Marketing PO Box 800, 9700 AV Groningen (NL)

MSc. Thesis Marketing Intelligence & Management July 2016 C.H. Petersstraat 25a 9714 CH Groningen Telephone number: +31618990656 E-‐mail: l.spruijt@student.rug.nl Student number: S2325098 Supervisors

University of Groningen

First supervisor: Dr. J. T. (Jelle) Bouma Second supervisor: Dr. J. (Jenny) van Doorn

(4)

(5)

Management summary

Companies are trying to find effective criteria for online personalization bases. Recent research has found that personality of an individual is an effective criterion to base online personalization upon. An often-‐used personality concept in personality research and in practice is the Myers-‐Briggs Type Indicator (MBTI). To find out whether the MBTI offers potential for online personalization purposes, in this study, the MBTI is linked to preference for (e-‐commerce) website design. More specifically, it is researched whether Thinking and Feeling and Judging and Perceiving types prefer different types of website design in terms of textual and visual aspects. The main goal of this study is to find if there are differences in preferences for website design between the MBTI types and thus to find out whether the MBTI offers potential for online personalization.

(6)

(7)

Abstract

Online personalization has received much interest during the last years, as (e-‐commerce) companies are trying to find out what personalization criterion is most effective. This research investigates whether the Myers-‐Briggs Type Indicator (MBTI) offers potential to be an effective personalization criterion and investigates this on differences in preferences for textual and visual aspects of website design, as these are highly relevant in the online environment. In total, three attributes are elaborated upon. The textual aspects of website design that are investigated are information delivery type, which is composed of a content and design variable, and call-‐to-‐action. The visual aspect of website design that is investigated is the amount of visuals. Besides these attributes, product involvement is included as a moderator to see whether this influences the relationship between the MBTI type of an individual and his preference for the attributes of website design. Data for this research was collected through a survey that was distributed among customers from two financial services. The survey consisted of MBTI personality measures, product involvement measures and conjoint tasks. With the data that was obtained, a choice-‐based conjoint analysis was performed on aggregated level, to find out how important attributes of textual and visual aspects of website design are in general. Information delivery type was found to be most important, followed by amount of visuals. Call-‐to-‐action was found to be least important. Product involvement was found to strengthen the relationship of MBTI with information delivery type on aggregated level. Then, the analysis was performed on MBTI dimensions Thinking – Feeling and Judging – Perceiving level, to find out whether there are differences in preferences for attributes of textual and visual aspects of website design. Results show that there are no significant differences between MBTI Thinking and Feeling individuals. On the other hand, results show that there are significant differences between MBTI Judging and Perceiving individuals. Judging individuals have a stronger preference for bulleted factual texts than Perceiving individuals. Perceiving individuals, on the other hand, have a stronger preference for bulleted persuasive texts than Judging individuals. Judging and Perceiving individuals do not differ in their strength of preference for bulleted texts in general. Besides this, Judging individuals have a stronger preference for purchase-‐related call-‐to-‐actions than Perceiving individuals. This research has theoretical as well as managerial implications for (e-‐commerce) companies and offers bases for future research.

Key words: online personalization, personality, MBTI, website design, textual aspects, visual aspects, product involvement, choice-‐based conjoint analysis.

(8)

(9)

Preface

Dear reader,

In front of you lies my MSc Marketing Intelligence and Marketing Management thesis. This thesis was written in combination with an internship at a financial services company in LOCATION, the Netherlands.

Four years ago I started my BSc in Business Administration at University of Groningen. During this degree I discovered my great interest in marketing, which led to the choice for MSc Marketing. This thesis might infer that I am graduating after the completion of this thesis, which is actually not the case. I am a practical person and wanted to gain some practical experience before graduating. To make sure I would have gained this by the end of my studies, I spread my chances and applied for an internship already in the middle of my master’s. I am expecting to graduate in 2017.

Now that I have the opportunity to thank some people, I would like to do so.

First of all, I would like to thank my first (co-‐) supervisors, Jelle Bouma and Jenny van Doorn, for their support and useful feedback during the last five months. I have had quite a hard time writing this thesis; without their help I would not have been able to do this. Secondly, I would like to thank two financial services companies in general for allowing me to write my thesis in combination with an internship at these companies. I would also like to thank them for the internal possibilities I got. Thirdly, I would like to thank some people from the two financial services companies in specific. First of all, I would like to thank NAME, my supervisor, for his active involvement in my project. He always was enthusiastic about my ideas and willing to help me when I needed it. Furthermore I would like to thank NAME, NAME, NAME and NAME in general for everything the last five months. Thirdly, I would like to thank my fellow internship thesis students. We supported and helped each other when necessary. Last, but certainly not least, I would like to thank my family and friends for their great support during my whole study career.

I hope you enjoy reading my thesis.

(10)

1. Introduction

Personalization, or related terms such as customization, adaptation, individuation, consumer-‐centric, and one-‐to-‐one relationship (Fan & Poole, 2006) have existed as long as humanity exists. In essence, it pertains to offering ‘different things to different people in different fields’ (Fan & Poole, 2006, pp. 181). Already in the past, a tailor would offer different types of clothing to different people. In 1965, Smith proposed the idea of tailoring a specific offer to better suit a certain customer group (Smith, 1956). Personalization has existed for a long time; however, the arrival of the Internet has raised interest in this issue. Personalization has existed longer than the Internet and, thus, comprises more than this single technology. Besides this, the practice of personalization has advanced due to the rise of the Internet, more than any other technology has done (Montgomery & Smith, 2009).

Online personalization can be initiated by the customer, which is often referred to as customization (Montgomery & Smith, 2009). It is the process of the customer requesting to personalize certain aspects of a(n) (e-‐commerce) website, for example. However, companies are increasingly using personalization tools that incorporate techniques that discover patterns in the Internet usage data. In this way, user profiles are discovered and personalization is put into practice without the active interference of the customer (Cooley & Srivastava, 2000). Hauser, Urban, Liberali and Braun (2009) propose an approach of personalization management systems that is called morphing. It is the process of automatically matching website design to customers’ cognitive styles. They found that purchase intentions could be increased by 21% if one has perfect knowledge about the customers’ cognitive styles, due to optimal morphing. In today’s competitive e-‐commerce market, this is very interesting for companies that operate online.

In practice, customer personality profiles generated online can be used both online and offline. It can be used in personalizing the online environment, as well as for intelligent routing purposes. Ali (2011), for example, found that call center performance could increase with 15% if the employee’s psychometrical profile is coupled to that of the customer.

(13)

practicality, the focus of this study is online personalization based on the personality of the customer.

A model to describe personality is the Myers-‐Briggs Type Indicator (MBTI), which is used extensively in personality research as well as in practice (Choi, Deek & Im, 2008). Choi et al. (2008) argue that many business organizations currently profile their new employees by the use of this model. The MBTI is a model that describes the preferences of people on four bipolar psychological dimensions (Harrington & Loffredo, 2010). The dimensions MBTI uses are Introvert (I) – Extrovert (E), which is identical to one factor of the well-‐known Five Factor Model of Goldberg (1993), Sensing (S) – Intuition (N), Thinking (T) – Feeling (F) and Judging (J) – Perceiving (P) (Boyle, 1995). Each individual that has been assessed by the MBTI test receives a four-‐letter type that pertains to the individual’s preferences for each of the dimensions.

It has been proven that personalizing the online environment to the personality style of the customer is an effective strategy (Hauser et al., 2009), but it has not yet been quantitatively researched what aspects of a(n) (e-‐commerce) website design should be personalized based on personality traits of individuals by the use of morphing, and what aspects should not. Therefore it is very interesting and relevant to research what aspects of an e-‐commerce website design a company should consider personalizing based on the MBTI personality style of the customer. In this study, the term website design is used for e-‐commerce website design.

Cunningham, Thach and Thompson (2007) made the linkage between the Myers-‐Briggs Type Indicator’s personality dimensions and preferences for different aspects of a website design (i.e. text, imagery, and navigation). Their study comprises literature research only and, therefore, lacks quantitative research. This study aims to quantify part of the literature results of the article mentioned. More specifically, this study focuses on textual and visual aspects of an e-‐commerce website. In this research, two aspects of website design are studied that are highly relevant in e-‐ commerce, namely textual and visual aspects. Both of these aspects have been researched amply, but have barely or not been related to personality, as will be addressed later on.

(14)

Visual aspects of web design are highly relevant nowadays, since a big part of the online content consist of visual information, e.g. images and graphics. Websites constantly change the amount of visuals, which makes the amount of visual information highly volatile (Smith & Chang, 1997). Previous research has already focused on the process preference of people for visual or verbal information with regard to their personality (Sojka & Giese, 2001), but the MBTI concept has not been used. In this study, the MBTI will be used to test the preference of different MBTI types for visual or verbal information, as well as the relative importance of visuals to purely textual aspects. Other than personality, there are many other factors that have a considerable influence on the way that people behave and make decisions in the online environment. Product involvement is one of them (Hong, 2015). Product involvement in this study is defined as ‘the extent to which an individual

consumer perceives a product to be important’ (Hong, 2015, pp. 322). Product involvement can have

considerable influence on the way that people make purchase decisions, from information search and evaluation of alternatives (Browne & Kaldenberg, 1997) to the time and effort they invest in the decision (Bian & Moutinho, 2008). As information search and time and effort spent all are related to aspects of website design, it is important to include product involvement in this study, as it is likely to moderate the relationship between MBTI personality and website design preference.

In short, this research aims to answer the following research question:

‘To what extent do personality differences affect the preference for textual and visual aspects of website design among people with different product involvement?’

In this research, data is gathered by means of a survey among customers from two financial services companies. The data will be analyzed by means of an on-‐the-‐MBTI-‐based test and by means of a choice-‐based conjoint analysis, to research how textual and visual aspects affect preference for website design on aggregated, and more importantly, the MBTI dimensions of interest; Thinking – Feeling and Judging -‐ Perceiving level.

(15)

This study has both theoretical and practical relevance. Much research is done on general Internet usage with regard to personality (e.g. Mark & Ganzach, 2014; Kosinki et al., 2014) or on content effectiveness of one aspect of a(n) (e-‐commerce) website design, with regard to personality (e.g. Moon, 2002; Hirsch, Kang & Bodenhausen, 2012; Schindler & Bickart; 2012) but little studies focus on the application of an individual’s personality on website design. More specifically, to the author’s knowledge, no study focuses on the application of an individual’s MBTI type and preference for textual and visual aspects of website design. Likewise, Noyes (2007) states that few website designs are personalized based on consumers’ personalities. Website design personalization based on consumer personality is an aspect of online personalization that has not been researched much, but theoretically offers much potential for future business applications (e.g. Hauser et al., 2009) and therefore should be quantitatively be researched. This study gives managerial insights about the potential of online personalization based on the Myers-‐Briggs Type Indicator. It shows some significant results on preference differences between MBTI profiles that could be implemented, but also offers potential for further future research.

The structure of this research is as follows. The first chapter is the research framework, in which the attributes of textual and visual aspects of website design, as well as the moderator are further specified and hypotheses are formed. After this, the methodology of this research is explained. In the results chapter of this study, the results of this research will be discussed on aggregate level as well as on MBTI dimension level. This paper ends with the discussion, in which theoretical and managerial implications as well as limitations and directions for future research are discussed.

(16)

2. Research Framework

The aim of this article is to find out to what extent an individual’s MBTI profile affects ones preference for website design. The aspects of website design that are included in this research are elaborated upon in this chapter. First, the textual aspects of website design are defined and elaborated upon. Secondly, the visual aspect of website design is discussed. Then, the MBTI is elaborated upon and discussed with regard to the online environment. Lastly, the moderator product involvement is discussed with regard to the relationship between MBTI and preference for website design. This chapter is finalized with the research model of this study.

2.1 Defining E-‐commerce Website Design Aspects

Before the use of the Internet, there were five key design elements that determined the usability of information systems (Nielsen, 1993). These key elements were consistency of the interface, response time, mapping and metaphors, interaction styles, and multimedia and audiovisual. With the rise of the Internet, the usability research started focusing on extending the basic usability elements into the online environment. In 2000, Nielsen (2000) extended the design principles to include navigation, response time, credibility and content. In today’s online world, the focus of website design has shifted from usability to user experience and has put emphasis on the aesthetics of the design (Deng and Poole, 2012).

(17)

2.2 The Myers-‐Briggs Type Indicator

Jung (1921) proposed that people differ on basis of their preferences for different sources of data and for dealing with data differently and that people could be grouped on basis of that. He proposed that the different types are typed by three dichotomous dimensions; Introversion -‐ Extraversion, Sensing -‐ Intuition, and Thinking -‐ Feeling. Jung’s type theory was the basis for a number of concepts, among which the Myers-‐Briggs Type Indicator (MBTI). This instrument was developed by Kathryn Myers and Isabel Briggs-‐Myers (Lucas, 2007) to measure the three dimensions Jung proposes as well as a fourth dimension; Judging -‐ Perceiving. The MBTI, as well as Jung (1921), assumes that a person is typed as either one or the other end of the dimension. The dimensions are discontinuous, rather than continuous like the Five Factor model of personality traits of Goldberg (1993) for example. The Introversion -‐ Extraversion dimension pertains to the source of one’s energy. It measures a person’s preferred orientation to life (Carlyn, 1977). It relates to whether someone’s general attitude to other persons and objects is oriented internally or externally. Introvert types draw their energy from ‘the inner world of their thoughts and ideas’ (Harrington & Loffredo, 2010, pp. 90), whereas extravert types draw their energy from ‘the outer world of people and action’ (Harrington & Loffredo, 2010, pp. 90).

The Sensing -‐ Intuition dimension pertains to one’s perceptual style and the way one gathers and processes information. Sensing types gather information relying on their five senses. They tend to gather information in ‘a sequential fact-‐oriented manner’ (Harrington & Loffredo, 2010, pp. 90) and are focused on the present. Intuition involves ‘a more detached, insightful analysis of stimuli and

events’ (Boyle, 1995). Intuition refers to gathering information in a more random way with the ‘larger implications in mind’ (Harrington & Loffredo, 2010, pp. 90). Intuitive types are more focused on the

future.

(18)

The Judging – Perceiving dimension pertains to one’s lifestyle. More specifically, Judging types prefer to plan things in advance and have a structured and predictable way of life. Moreover, Judging people like to make decisions as fast as possible, possibly even before facts are available. Perceiving individuals like a flexible and open way of living. They prefer to wait for more information to base their decisions upon rather than deciding straight away. They take decisions less immediately (Harrington & Loffredo, 2010; Carlyn, 1977).

One’s preferences on the four dichotomous dimensions determines ones MBTI profile. As two preferences on each of the four dimensions are possible, sixteen different personality profiles are established, which are characterized by a letter code. An Introvert individual would be assigned an ‘I’, whereas an Extrovert individual would be assigned an ‘E’. The other dimensions; Sensing-‐Intuition, Thinking-‐Feeling and Judging-‐Perceiving, pertain to the letters ‘S’-‐‘N’, ‘T’-‐‘F’ and ‘J’-‐‘P’ respectively. A possible personality profile would be ENTP, which refers to an Extrovert, Intuitive, Thinking and Perceiving individual.

2.2.1 The Myers-‐Briggs Type Indicator in the online environment

In the previous section, it was stated that the Thinking – Feeling and Judging – Perceiving dimensions are related to decision-‐making. Thinking – Feeling relates to the rationality of the decisions made, whereas Judging – Perceiving relates to the time individuals take to make decisions. Decision-‐making in the online environment, and more specifically: the website design that aids most to the decision-‐ making process, is highly important to understand for companies operating online. As the Thinking -‐ Feeling and Judging -‐ Perceiving dimensions pertain to the way decisions are made, these dimensions will be the focus of this study.

2.3 Textual aspects of web design

Textual aspects of a website’s design, according to Powell (2002), refers to ‘any typed material within

a website’ (Cunningham, et al., 2007, pp.9). It includes text content as well as typographic variables.

(19)

skipping others (Masson, 1985) and to scanning through text to find specific information (Robeck & Wallace, 1990). Due to the rise of the Internet, the amount of material we can read from screen has increased enormously. Websites, in particular, are most of the time skimmed instead of read in detail (Horton, Taylor, Ignacio, & Hoft, 1996). It therefore is very important to display the type of text that your customer prefers to read and thus content should be personalized. Previous research on content personalization has for example concluded that tailoring content with regard to the personality of the receiver can be effective if the content is aligned with the to the trait preferred motivational system (Hirsch, et al., 2012)

In this research, two attributes of textual aspects of a website will be explored, namely information delivery type, which consists of content information delivery method and information design, and call-‐to-‐action.

2.3.1 Information delivery type

In this study, the concept of information delivery type is developed and refers to the combination of content information delivery method, i.e. factual versus persuasive texts, and information design, i.e. bulleted versus paraphrased texts.

2.3.1.1 Content information delivery method

Text content is related to writing style (Cunningham et al., 2007) and can be based on persuasive information delivery, which appeal to an individual’s emotions, or factual information delivery (Reinard, 1988), which does not aim to appeal to emotions.

To make a clear distinction between persuasive information delivery and factual information delivery, it is important to know what makes a text persuasive. The persuasiveness of texts has been researched for a long time by over thousands of studies and concepts. Little research has touched what makes a particular argument persuasive in isolation. Petty and Cacioppo (1986) found that arguments in general increase the effect of the message, regardless of the strength of the argument, but that ‘persuasion cues, simple heuristics, or other relatively thoughtless means’ can also persuade individuals (Edwards, 2003, pp. 91).

(20)

persuasive text has been read. People process text differently (Edwards, 2003) and therefore it is likely that persuasive information delivery or factual information delivery is processed differently by individuals. More specifically, message-‐person congruence effects have been researched with regard to several psychological characteristics (Dijkstra, 2008) but have not been related to a comprehensive personality model. This would ‘advance the message-‐framing literature by opening the door to

exploring new ways to make persuasive message more personalized and effective’ (Hirsch et al.,

2012, pp. 1).

An often-‐used method in persuasive information delivery is testimonial assertion. Testimonials are stories, opinions or experiences by individuals, also known as social proof (Hirsch, Kang & Bodenhausen, 2012). Typically, a person would tell her personal positive experience with a product or service and encourages the reader to follow the example (Braverman, 2008). According to her, an example would be: ‘I began managing my eating habits four years ago. To me, the most rewarding

thing is that I was able to lose weight and keep it off since my first semester’ (Braverman, 2008, pp.

1). Brosius and Bathelt (1994) argued that testimonials are more persuasive than factual information, which is an alternative information delivery method (Reinard, 1988). In factual information delivery, facts about the product or service are told.

No study has yet focused on the relationship between MBTI and preference for persuasive or factual information delivery, to the author’s knowledge. However, as Thinking individuals tend to use objective reasoning when making decisions and Feeling individuals tend to base decisions upon non-‐ rational criteria as interpersonal factors (Loffredo & Harrington, 2010), it is likely that Thinking individuals have a stronger preference for factual texts and Feeling individuals have a stronger preference for persuasive texts. Individuals scoring high on the Agreeableness trait of the Big Five are found to be influenced by persuasive online communication that includes the social proof principle (Heise, 2003). As social proof, or testimonial assertion is related to persuasive texts in this study, it is likely that individuals scoring high on Agreeableness have a stronger preference for persuasive texts than individuals scoring low on Agreeableness. Agreeableness is found to correlate positively with the Thinking – Feeling dimension of MBTI (Furnham, 1996). Moreover, Thinking individuals tend to use a directive decision making style, which refers to rationality and low tolerance for ambiguity (Myers & McCaulley, 1985; Ahmed, Hasnain & Venkatesan, 2012). Feeling individuals tend to use a behavioral decision making style, in which acceptance of others is important. Therefore the following hypotheses are formed:

(21)

H1b: Feeling individuals have a stronger preference for persuasive texts than Thinking individuals.

2.3.1.2 Information design

Information design literature focuses on how to best present text. This can for example be continuous prose, i.e. a paragraph, or a displayed list, e.g. by the use of bullet points (Kozak & Hartley, 2011). Hartley (1992) found that the design of a text affects the recall. Kozak and Hartley (2011) argue that lists using bullet points are easier to read, easier to search within and are easier to remember than a continuous text. Regarding text design, text length can also vary. Textual aspects of a website design can vary from one sentence to a long paragraph or from a few words to many words for example. Combining design of a text and text length, text can be displayed in a bulleted text, i.e. bullets without much detailed information, i.e. less detailed, or paraphrased text, i.e. a paragraph with more detailed information.

To the author’s knowledge, information design has not yet been related to personality preferences or more specifically, the MBTI. However, as Judging individuals feel the urge to make decisions as fast as possible, possibly even before all facts are available (Harrington & Loffredo, 2010), they are likely to be attracted to bulleted lists, as these provide the most important information, without getting into details. Perceiving individuals, on the other hand, postpone their decision until all, or more, information is available (Harrington & Loffredo, 2010). They, therefore, are likely to be attracted to paraphrased texts, which displays more details. In conclusion, the following hypotheses are formed:

H2a: Judging individuals have a stronger preference for bulleted texts than Perceiving individuals.

H2b: Perceiving individuals have a stronger preference for paraphrased texts than Judging individuals.

2.3.2 Call-‐to-‐action

(22)

information or making a purchase on an e-‐commerce website (McMillan, 2004). Examples of call-‐to-‐ action buttons of an e-‐commerce site are ‘add to chart’, ‘sign up’, ‘get started’, ‘contact us’, ‘continue’, ‘more information’ and many more. Call-‐to-‐action buttons are continuously tested by means of A/B tests. The text in the call-‐to-‐action is tested, but also colors and shapes (Eisenberg & Eisenberg, 2005). No research, however, has yet focused on the difference in preference for call-‐to-‐ action by different MBTI profiles.

As was stated in the previous part, individuals process textual aspects differently and are attracted to different types of texts. Judging individuals tend to make decisions fast (Harrington & Loffredo, 2010) and therefore are likely to be attracted to call-‐to-‐action buttons that provide in this tendency. In website design this would be purchase-‐related call-‐to-‐action buttons. Perceiving individuals, contrarily, postpone their decision until all information is available (Harrington & Loffredo, 2010). They are likely to be attracted to call-‐to-‐action buttons that will direct them to more information. In e-‐commerce website design this would be information-‐related call-‐to-‐action buttons. This leads to the following hypotheses:

H3a: Judging individuals have a stronger preference for purchase-‐related call-‐to-‐actions than Perceiving individuals.

H3b: Perceiving individuals have a stronger preference for information-‐related call-‐to-‐actions than Judging individuals.

2.4 Visual Aspects of Website Design

According to Powell (2002), the so-‐called imagery aspect of web design refers to ‘any visual elements

of a website’s design such as photography and images’ (Cunningham et al., 2007, pp.9).

2.4.1 Amount of visuals

(23)

found that only 10 percent of the written information was remembered, but almost 65 percent of the visual information was remembered after three days. In essence, most studies conclude that visuals are processed faster by the brain than textual aspects and agree on the fact that we think in images.

Sojka and Giese (2001) have found that people differ on their process preference for visual or verbal information. They argue that visual information can impact the decision process of an individual in the same way in which verbal information can impact this process. The difference in process preference, however, can be explained by personality trait theory. They use the theoretical concepts need for cognition and need for affect for this. Need for cognition refers to an individual’s tendency to enjoy thinking, whereas need for affect refers to the opposite; an individual’s tendency to enjoy processing feelings. They found that individuals with a low need for cognition but high need for affect (labeled as ‘feelers’) prefer visual information whereas individuals with a high need for cognition but a low need for affect (labeled as ‘thinkers’) prefer verbal information. Even though need for cognition and need for affect and its references to thinking and feeling imply that need for cognition and need for affect are related to the MBTI dimension Thinking – Feeling. This is not the case in the article, as they simply labeled individuals with certain scores accordingly. However, previous research (Bolton & Capella, 1995) found that there is indeed a significant positive correlation between need for cognition and MBTI Thinking – Feeling dimension and a significant negative correlation between need for affect and MBTI Thinking – Feeling dimension (Myer, 2008). It therefore is likely that MBTI Feeling types prefer visual information and MBTI Thinking types prefer verbal information. The following hypotheses are formed:

H4a: Thinking individuals have a stronger preference for verbal information than Feeling individuals.

H4b: Feeling individuals have a stronger preference for visual information than Thinking individuals.

2.5 Product Involvement

As stated in the introduction chapter, product involvement is defined as ‘the extent to which a

(24)

product to an individual and to the motivation in decision-‐making in and interest for a certain product class.

Product involvement has often been researched in relation to the decision-‐making process. More specifically, product involvement was found to influence both observational and behavioral responses of customers, including attention and information processing (Kwon & Chung, 2010). Browne and Kaldenberg (1997) argue that product involvement influences the decision-‐making process length. They state that highly involved individuals go through an ‘extended decision-‐making

process’ (Hong, 2015, pp. 323). They intensively focus on information search and the evaluation of

alternatives. Moreover, an involved individual is motivated to spend time and effort on the decision-‐ making process (Bian & Moutinho, 2008).

Some research has focused on the effect that product involvement has on the amount of persuasion an individual manifests. For persuasive information delivery, it was found that particularly low or high product involvement can hinder the persuasive process of an individual (Buehl, Alexander, Murphy & Sperl, 2001). Another approach is by Petty and Cacioppo (1986) who state that an individual who scores low on product involvement is not motivated to think about the quality of arguments in a message. For this individual, the more arguments, the better, sounds acceptable. In conclusion, product involvement is likely to weaken the relationship between an individuals and his preference for factual or persuasive information delivery.

H5a: Product involvement weakens the relationship between an individual and his preference for factual or persuasive texts.

To the author’s knowledge, no research has touched the moderating effect of product involvement and the other hypothesized aspects of website design. Nevertheless, product involvement with regard to the relationship between personality and preference for the other aspects of website design is expected to have a weakening moderating effect. That is, highly involved individuals are likely to search for more information (Bian & Moutinho, 2008) and do not value the personalized website as important as other individuals might do. They do not use one webpage to base their decision upon, but instead are more likely to click-‐through to find the information they need. This means that their preference for a certain website design is likely to be less strong.

H5b: Product involvement weakens the relationship between an individual and his preference for bulleted or paraphrased texts.

(25)

H5d: Product involvement weakens the relationship between an individual and his preference for amount of visuals.

2.6 Research Model

Based on the hypotheses that were stated in the research framework, a research model is created. The research model can be found in Figure 1.

Figure 1: Research model

(26)

3. Methodology

In this section, the methodology of this research will be discussed. First, the chosen method for this research is explained, from which the attribute and levels, MBTI, moderator and other measurements scales are highlighted. Then, the choice design of this research is explained. After this, the data collection is elaborated upon. This chapter finalizes with the plan of analysis.

3.1 Method

The method that is used to analyze website design preference is conjoint analysis. Conjoint analysis is a multivariate analysis technique that allows researchers to understand how respondents’ preferences for any type of object (products, services, or ideas) (Hair, Black, Babin & Anderson, 2010, pp. 409) are developed. It ‘portrays consumers’ decisions realistically as trade-‐offs among multi-‐

attribute’ objects (Hair et al., pp. 404). With conjoint analysis, a respondent evaluates a set of

different ‘object profiles’. The object profiles consist of attributes, which are specified in terms of levels and by this, the market researcher is able to assess the importance of the attributes and its levels. For this study, conjoint analysis is very appropriate to assess the importance of attributes of textual and visual aspects of website design.

With conjoint analysis it is possible to estimate preferences on aggregated level, but it is also possible to estimate segments and test differences in preferences among segments. As this study aims to find differences in preferences among different personalities, segments of different personalities will be estimated. Therefore, conjoint analysis is appropriate for this research.

3.1.1 Attributes and levels

(27)

For the textual aspect of website design, two attributes are researched. The first attribute that is researched is information delivery type. As can be seen in Figure 2 and 3, information delivery type combines content information delivery method and information design that have been defined in the research framework. Content and design variables are components of a text (Cober, Brown, Levy, Keeping & Cober, 2003) and therefore are combined in this research into the variable information delivery type.

Regarding content information delivery method, in this design, a factual text is displayed by means of a product description with factual text. Next to this, there is a section for a specific fact of the product. This section is displayed in green in Figure 2 and 3. A persuasive text in this design is the factual text plus some persuasion cues, simple heuristics, and other relatively thoughtless means (Edwards, 2003; p. 91). Instead of the section with a specific fact of the product, there is a section that includes a testimonial. A testimonial is a persuasive information method (Braverman, 2008) and therefore this increases the persuasiveness of the text.

Regarding information design, in this design, a paraphrased text is displayed by means of a paraphrased text. A paraphrased text includes that there are sentences that form a paragraph. The text includes the product specification plus more detailed information that elaborates on this main idea. A bulleted text in this design is the text displayed by means of bullets. The bulleted text includes less detailed information than the paraphrased text. The text that is displayed in bullets includes the main idea of a product specification only, whereas the paraphrased text elaborates on this main idea. The combination of the content and design variable leads to four types of information delivery. The overview of the levels can be found in Table 1. In Figure 2 and 3 all levels of information delivery type are displayed in the way the respondents are exposed to information delivery type. In Figure 2 the left picture represents level 1: Paraphrased factual text. The left picture in Figure 2 represents level 2: Bulleted factual text. In Figure 3 the left picture represents level 3: Paraphrased persuasive text. The left picture in Figure 3 represents level 4: Bulleted persuasive text.

Figure 2: Information delivery type, level 1 and 2 (respectively)

DELETED FOR THIS ANONYMIZED VERSION

(28)

Figure 3: Information delivery type, level 3 and 4 (respectively)

The second attribute of the textual aspect of website design is call-‐to-‐action. For call-‐to-‐action two levels are specified. The first level is purchase-‐related. In this study’s design this level is displayed by means of a button with the text ‘bereken uw premie’ (‘calculate your premium’), which refers to making a purchase. The second level is information-‐related. This level is displayed by means of a button with the text ‘meer information’ (‘more information’) and therefore is information-‐related. In Figure 4 the two levels of call-‐to-‐action are displayed.

Figure 4: Call-‐to-‐action, level 1 and 2

For the visual aspect of website design, the amount of visuals is researched. The first level is verbal information. This level equals the standard amount of visuals. The second level is called visual information. This level pertains to more visuals, as information is displayed visually instead of verbally. In this design, the first level is a text about how customers rate the product. It says ‘Klanten beoordelen deze verzekering gemiddeld met een score van 4 uit 5!’ (‘Customers rate this FINANCIAL

SERVICE on average with a score of 4 out of 5’) The second level displays the same information, but is

displayed by means of visuals. The visuals are displayed by means of stars. In this level the text says ‘Klanten beoordelen deze verzekering gemiddeld met’ (‘Customers rate this FINANCIAL SERVICE on average with). Then, the score of 4 out of 5 is displayed by means of a visual. In Figure 5, the two levels of amount of visuals are displayed.

Figure 5: Amount of visuals, level 1 and 2

(29)

Table 1 provides an overview of the attributes and levels that have been discussed in this section.

Table 1: Attributes and levels

Attributes Level 1 Level 2 Level 3 Level 4

Textual aspects Information delivery type Paraphrased factual text Bulleted factual text Paraphrased persuasive text Bulleted persuasive text Call-‐to-‐action Purchase-‐ related Information-‐ related -‐ -‐ Visual aspects

Amount of visuals Verbal information

Visual information

-‐ -‐

3.1.2 MBTI measurement scales

PERSONALITY PREFERENCES IN E-­‐COMMERCE: A CHOICE-­‐BASED CONJOINT ANALYSIS ON TEXTUAL AND VISUAL ASPECTS OF WEBSITE DESIGN

PERSONALITY PREFERENCES IN E-­‐COMMERCE: A

CHOICE-­‐BASED CONJOINT ANALYSIS ON TEXTUAL AND

VISUAL ASPECTS OF WEBSITE DESIGN

Anonymized version

PERSONALITY PREFERENCES IN E-­‐COMMERCE: A

CHOICE-­‐BASED CONJOINT ANALYSIS ON TEXTUAL AND

VISUAL ASPECTS OF WEBSITE DESIGN

Anonymized version

Management summary

Abstract

Preface

Table of contents

1. Introduction

2. Research Framework

2.1 Defining E-­‐commerce Website Design Aspects

2.2 The Myers-­‐Briggs Type Indicator

2.3 Textual aspects of web design

2.4 Visual Aspects of Website Design

2.5 Product Involvement

2.6 Research Model

3. Methodology

3.1 Method

PERSONALITY PREFERENCES IN E-‐COMMERCE: A CHOICE-‐BASED CONJOINT ANALYSIS ON TEXTUAL AND VISUAL ASPECTS OF WEBSITE DESIGN

PERSONALITY PREFERENCES IN E-‐COMMERCE: A

CHOICE-‐BASED CONJOINT ANALYSIS ON TEXTUAL AND

PERSONALITY PREFERENCES IN E-‐COMMERCE: A

CHOICE-‐BASED CONJOINT ANALYSIS ON TEXTUAL AND

2.1 Defining E-‐commerce Website Design Aspects

2.2 The Myers-‐Briggs Type Indicator