What drives customer responsiveness to different evidence types on landing pages: An analysis based on conversion intention.
Leonardus Cornelis Elshof 10647155
22-06-2018 Final Version
MSc. in Business Administration – Digital Business Track Amsterdam Business School – University of Amsterdam First Supervisor: dr. U. Konuş
Statement of Originality
This document is written by Student Leonardus Cornelis Elshof, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
While spending on online advertising is increasing year by year, current knowledge about how to get the highest returns on these marketing expenditures is still not complete. Thanks to previous literature we now know that evidence types as persuasion tactic lead to an increase in conversion rate in search engine advertisements. This study builds on the call for more research into persuasion and seeks to find which evidence types have the highest positive impact on conversion intention and if this positive influence is stronger or weaker for different product categories, product involvement, shopping experience and internal cues.
We find that consumers are more likely to convert when shopping for fashion items compared to electronics and that the proposed effect of evidence types on conversion intention is moderated by affective internal cues and by prior online shopping experience. Affective internal cues moderated the effect of causal evidence on conversion intention, while prior online shopping experience moderated the effect of expert evidence on conversion intention. These findings help marketers to optimize their landing page copywriting texts. To be more specific, consumers who can be classified as impulsive should be shown causal evidence, while experienced shoppers and older can best be exposed to expert evidence.
Table of contents
1. Introduction 5
2. Literature Review 8
2.1. Empowerment in the Digital Age 9
2.2. Landing Pages 11
2.3. Evidence Types 16
2.4. Internal Cues and Product Involvement 18
2.5. Prior Online shopping Experience 20
2.6. Conversion rate 21
3. Conceptual Framework and hypothesis 22
3.1. Evidence Types 23
3.2. Internal Cues 24
3.3. Prior Online Shopping Experience 25
3.4. Category Involvement 26 3.5. Product Category 28 4. Research Design 29 4.1. Sampling method 30 4.2. Measures 30 4.2.1. Conversion Intention 30 4.2.2. Evidence Types 31 4.2.3. Product Category 31 4.2.4. Involvement 32
4.2.5. Prior Online Shopping Experience 32
4.2.6. Cognitive Internal Cues 32
4.2.7. Affective Internal Cues 33
4.3. Procedure 35
4.4. Analysis and Predictions 35
4.5. Data preparation 36
4.6. Sample 37
4.7. Reliability and Correlations 37
5. Results 39
6. Discussion and conclusion 42
7. Managerial Implications 44
8. Limitations and Future Research 46
Bibliography 47
1. Introduction
Businesses in the twenty-first century are ever evolving. First, businesses were shifting from traditional brick-and-mortar stores to bring their supplies to the increasing demand for e-commerce (In ‘t Zand, 2017). And after the switch to the online environment, technology evolved in such a way that people are becoming less important in companies. While 10 years ago people used to be at the forefront of making money, they are now rapidly being replaced by complex algorithms. These algorithms are written for applications such as search engines, and social networking sites. Search engine advertising (SEA) has been growing rapidly in the last decade (Haans, Raassens & van Hout, 2013) from a total market of $19.5 billion in 2012 (Atkinson, Driesener & Corkindale, 2014) to over $92 billion in 2017 (Statista, 2018b) and it nowadays makes up for about 50% of the total online marketing budget (Li & Kannan, 2014). Moreover, the global digital marketing spending is expected to grow from $162 billion in 2015 to $335 billion in 2020 (Statista, 2018a). This makes digital advertising one of the fastest growing businesses around (Ghose & Yang, 2009; Yao & Mela, 2011).
Even though digital advertising is one of the fastest growing businesses, research on how to maximize the returns on these advertisements can still be improved. And while printed and TV advertisements are bursting at the seams with persuasion techniques to make viewers turn into customers, for example, dentists who are suggesting to switch to a brand name toothbrush because it will remove 99 percent of plaque (Koh, 2012), it is not happening as much in the online marketing environment. This study seeks to find out what evidence types are most effective in persuading customers to do business with a webshop.
The Marketing Science Institute (2016) emphasized the most important marketing research priority is in the field of conversion attribution in the online environment. Conversion rate is an important marketing measure and oftentimes used as key performance indicator (Zenetti, Bijmolt, Leeflang & Klapper, 2014), because resource allocations on digital platforms
are often directly linked to the performance in terms of conversion rate. In this study, the focus will be on conversion intention, since conversion rate can only be measured well with big samples.
Research is missing on the last part of the online customer search journey, where the last actions before purchase are happening, the landing page. The landing page is the web page where consumers will be navigated to when they click on a search engine advertisement (White & Huang, 2010), or display advertisements. Extensive research has been done in the field of digital advertising but these researches are mainly focused on the relationships between the advertisements and the aforementioned conversion and click-through rates (Haans et al., 2013; Atkinson et al., 2014). Even though the last-click attribution metric, which is the metric that attributes conversions to the last click before purchase, is used most in practice (Li & Kannan, 2014), researches are focused mostly around the first step of a customer search journey. Therefore, additional research on how landing page content could influence conversion intention is needed.
Extensive research on how to influence the decision making in these last steps has brought the current literature to four main persuasive evidence types which are already used in day to day marketing activities. An evidence type can be seen as arguments that seek to sway people or take their doubts away when it comes to trying to choose between different alternatives (Hoeken & Hustinx, 2003). Think of distinguished sports brands who strike endorsement deals with successful athletes. These athletes then act as an expert and advocate the sports brand as the perfect brand for all athletes, which is a type of expert evidence.
As e-commerce websites oftentimes have the goal to convert visitors into purchasing customers, these shoppers need to be convinced that the product is superior to the alternatives and therefore the products should be purchased.
Despite the growing interest and importance of online marketing in research and practice, the incorporation of existing theories in online marketing studies is limited. Instead, existing online marketing literature focused on how to attribute conversions in a multi-touchpoint online journey (Kannan, Reinartz & Verhoef, 2016), on antecedents of buying behavior (Dawson & Kim, 2009) and issues with confidence in online retailers and shopping experience (Sharma, 2011; Ling, Chai & Piew, 2010).
On top of this, extensive research on how to influence decision making has brought the current literature to four main persuasive evidence types. As e-commerce websites oftentimes have the goal to convert visitors into purchasing customers, these shoppers need to be convinced that the product or webshop is superior to the alternatives and therefore the products should be purchased. This research will be contributing to the existing literature in two ways.
First, this research will contribute to the existing marketing literature and will incorporate existing marketing theories in the online environment. On top of this, this study will make a case for last-click attribution theory, where the importance of landing page copywriting will be underlined with regard to changing consumer behavior, instead of beating search engine algorithms.
Second, the findings of this study will benefit online marketers and practitioners. While previous researches oftentimes focused on search engine advertisement content and the influence on conversion rate (e.g. Haans et al., 2013; Ghose & Yang, 2009; Atkinson et al., 2014), this study will contribute to the existing online marketing literature in a way that helps to understand the importance of landing page copywriting, and the incorporation of evidence types. It will help practitioners increase conversions on websites and help them understand more about the relative importance of incorporating evidence types in copywriting tactics. This can lead to a shift in budget allocation where more money can be spent on copywriting, and investments in retargeting campaigns can be decreased since conversion will go up.
In this research, we will first discuss the existing literature in the field of online marketing, landing pages, conversion rate and purchase intention and factors that affect these variables. Then we will develop a theoretical framework and a conceptual model. This will be followed by the methodology and a data analysis, and finally we will discuss our findings, will make suggestions for managers and give future research directions.
2. Literature Review
In this literature review, previous research which founded the bases for digital marketing will be discussed. First, digital marketing will be clearly defined and that will be followed by the importance of digital marketing nowadays, and the shifts that digital marketing brought with it. Then, the different types of digital marketing will be discussed. This will be followed by the antecedents of a successful digital marketing strategy, the determinants of e-commerce profitability and the need for persuasion in digital marketing. Finally, we will present the current research gap and our predicted managerial and theoretical contributions.
The term digital marketing has evolved over time, where it used to be a specific term which described product and service marketing through digital channels (Kannan & Li, 2017), but it involved towards where digital marketing is seen more as an umbrella term which creates, delivers and communicates value for customers and other stakeholders through digital technology-enabled activities, institutions, and processes (Kannan & Li, 2017). Kannan and Li introduced a more inclusive perspective and defined digital marketing as “an adaptive, technology-enabled process by which firms collaborate with customers and partners to jointly create, communicate, deliver and sustain value for all stakeholders” (2017, p. 23).
Digital marketing has become an increasingly important field and business component which companies need to incorporate in their strategy to remain competitive in the current environment (Kiang, Raghu & Shang, 2000). The digital technologies are changing the
environment rapidly (Kannan & Li, 2017). Consumer behavior is changing because consumers have access to a wider variety of choices and they are becoming more empowered because information asymmetry is being reduced by the internet (Kannan & Li, 2017).
Digital marketing or online marketing is an umbrella term for different marketing related activities, institutions, and processes. In the field of online marketing, a lot has been investigated. Ranging from how to attribute conversions in the online environment (Kannan & Li, 2014) to how to optimally design web pages (Ranganathan & Ganapathy, 2002). Table 1 will show an overview of the research that has been done thus far on the topics which are being discussed in this paper.
2.1. Empowerment in the Digital Age
It is becoming increasingly important to satisfy the consumers since their word-of-mouth power is no longer merely reaching their social inner circle consisting of friends and relatives, but digital technologies have expanded the reach of consumers’ electronic word-of-mouth to an almost unlimited audience with the emergence of social media platforms (Kannan & Li, 2017). This increase in knowledge by sharing reviews with other consumers is causing an increase in information-based power (Labrecque, Vor Dem Esche, Mathwick, Novak & Hofacker, 2013).
However, the digital era knife is cutting both ways. Firms are benefitting through the usage of cookies by consumers, which enables these firms to retarget consumers and give personalized recommendations (Lambrecht & Tucker, 2013).
Table 1
Important factors of e-commerce success
Area of Focus Author (year) Constructs Findings
Commerce Liao & Cheung (2001) Product offering, price Breadth of assortment (+) and price (-) are predictors of purchase intention
Service Baker et al. (2002) Customer Service When good customer service is provided, customers are more willing to do business.
Web page design Palmer (2002) Speed, navigation, content Web pages need to be fast, easy to use and information rich.
Ranganathan & Ganapathy (2002)
Web page information content, design, security, privacy
Security is most important to consumers on
e-commerce web pages, followed by privacy, design and information content.
Baker et al. (2002) Design
Gofman et al. (2009) Design & testing Prototype testing assures choosing the optimal website design.
Risk, trust & ease of use
Pavlou (2003) Perceived Risk, Perceived Usefulness
Increase in trust and usefulness also increases the willingness to do business again.
Gefen et al. (2003) Trust
Persuasion Haans et al. (2013) Expert, causal & statistical evidence
Expert and statistical leads to higher CTR, causal leads to higher conversion rate.
Click-through and Conversion Rate
Gupta & Mateen (2014); Cezar & Ögüt (2016)
Extensions, star ratings Both increased conversion rate on search engine advertisements.
Current Study Elshof (2018) Evidence types, description content, internal cues, product involvement.
Aim: What types of evidence have the highest impact on conversion on web pages? And how is conversion impacted by category involvement, internal cues and online shopping experience for different product categories?
Overall, there seems to be a balance between the increase in customer and company power in the increasingly digital age. Therefore, it is important for researchers and practitioners to gain insights on how to shift the balance towards the companies, and how to influence consumers’ decision-making process.
2.2. Landing Pages
In order to meet customer demand, companies try to optimize their e-commerce websites in order to rank as high as possible in search engines, this process is called search engine optimization, or SEO (Google, 2018a). But, when a company does not have a strong brand, it is recommended that they use a paid form, called search engine advertisement, since their organic search position, which is not strong enough to yield a high organic position (Li & Kannan, 2014).
These organic and paid search results are ordered by algorithms, and in these algorithms, the relevancy to the search query is an important factor if a company wants to end up high in the results (Google, 2018b). For search engine advertisement, it costs money to end up among the sponsored search results, and therefore it is important to know how to optimize your return on marketing investment by creating the optimal landing page.
Research on how to optimize conversions on landing pages can still be improved. Thanks to prior research, marketing practitioners know what elements to include in their landing pages and advertisement text. To be more concise, practitioners know how to influence search engine algorithms to increase visibility and get good advertisement ranks thanks to Google (Google, 2010), and other studies, but research is missing on how to influence the consumers’ mind in the decision-making process, so landing page texts would lead to more conversions.
Landing pages are an important factor for the effectiveness of a search engine advertisement. When a landing page experience is good, which is determined by factors such
as the usefulness and relevance of the information provided on the page, the quality score becomes higher (Google, 2018a). This quality score is computed by an algorithm that weighs these factors with regard to the search query of the user. The more relevant the content of your landing page is, the higher the quality score will be, and this will make it more likely that your advertisement will be shown in the query results (Google, 2018a). Therefore, online marketers are often seeking to design landing pages in such a way, that they achieve the highest quality score possible. Unfortunately, little research has been done on how a marketer can design a landing page in such a way that it maintains a high quality score and create more conversions. Researches that has been done on landing pages often are in the field of landing page optimization.
Landing page optimization (LPO) has become increasingly popular over time for improving website design. This approach uses statistical methods utilizing respondents who evaluate web pages (Gofman, Moskowitz & Mets, 2009). LPO is often executed by creating several prototypes and test them with consumers. Gofman et al. (2009) their focus was on the design part of the website and concluded that in order to find a winning landing page, multiple prototypes need to be tested in order to find the one that works the best. These winning landing pages are often the ones that score the highest on conversion rate, therefore in practice, LPO is also called conversion rate optimization.
Everard and Galletta (2005) studied how presentation flaws could affect the perceived site quality and ultimately the purchase intention. They found that website style, completeness of web pages and grammatical correctness were all significant predictors of perceived site quality, which is an antecedent of trust which ultimately predicts purchase intention. Everard and Galletta (2005) therefore stress that practitioners need to inspect websites thoroughly before launching them.
Constantidines (2004) divides the web experience building blocks into five elements which can be classified under three factors: the functionality factor, consisting of usability and interactivity; the psychology factor which consists of online trust; and the content factor, which consists of the aesthetics and marketing mix. An overview of these building blocks is given in Table 2.
Constantidines proposes that the buying process of a consumer on a website can be different in two ways (2004). First of all, the process and web experience can differ depending on the buying situation. Consumers who are likely to convert are more likely to weigh the transaction security and fulfillment process as more important than visitors who are just browsing the site for information (Constantidines, 2004). Besides the abovementioned difference, consumers are also evaluating the website differently when they are returning customers or not. Edelman and Singer (2015) concur with these findings and illustrate that returning customers are more likely to enter the loyalty loop, a path to purchase that skips the consideration phase and jumps to the purchase phase as a result of brand familiarity.
To summarize the findings of Constantidines (2004), online shoppers expect from a website that the pages load fast, it is easy to use and provides reliable results, it needs to have a fast and easy checkout process. Constantidines (2004) found that trust is an important factor that influences shopping behavior on websites and suggests that e-commerce websites enhance their trustworthiness by showing cues such as home-shopping awards and star-ratings. On top of this, customers expect an attractive web store with a pleasant appearance, similarly as an offline shopper would.
Lastly, Constantidines (2004) discusses some content factors with the marketing mix. He advocates that communication is a big part of the marketing mix, but rather as a provider of information about products instead of persuasion to buy the products. However, Hoeken and
Hustinx (2003) explored the relative persuasiveness of evidence types which has been incorporated into the marketing field by Haans et al. (2013).
Even though studies have been investigating the effects of website attributes on consumer behavior, research is missing on how the communication text on landing pages can affect conversions.
Table 2 Web experience building blocks (Source: Constantidines, 2004, p. 114)
Functionality factors Psychology factors Content factors
Usability Interactivity Trust Aesthetics Marketing Mix
Convenience Site navigation
Information Architecture Ordering / payment process Search facilities and process Site speed
Findability/Accessibility
Customer service / after sales Interaction with company personnel Customization
Network Effects
Transaction security Customer data safety
Uncertainty reducing elements Guarantees/return policies Design Presentation quality Design elements Style / atmosphere Communication Product Fulfillment Price Promotion Characteristics
2.3. Evidence Types
Evidence types have been studied extensively in the past by various researchers in different fields and situations ranging from healthcare issues to hypothetical claims in the hospitality industry (Kopfman, Smith, Ah Yun & Hodges. 1998; Hornikx, 2005; Hoeken & Hustinx, 2009; Massi Lindsey & Ah Yun, 2003). Hornikx (2005) gives a clear overview of all studies performed until then. Four main types of evidence are analyzed: expert evidence, statistical evidence, causal evidence, and anecdotal evidence. Table 3 will give an overview of performed researches on these four evidence types. The study of Hoeken and Hustinx (2003) is the only study who examined all four evidence types.
Expert evidence is an evidence type which is studied in the more recent years. It is often a restated claim provided by a person who is a leader in a particular field (Hoeken & Hustinx (2003). Haans et al. (2013) add to this definition that it is used to often enhance the credibility of the advertisement claims. Hoeken and Hustinx (2003) bring in the example that a much-used form of expert evidence is in academic research, where prior studies and researchers are quoted or referred to when building conceptual frameworks.
Statistical evidence is an evidence type which can be defined as “a numerical compacting of a series of instances” (Haans et al., 2013, p. 155). This means that these evidence types are shown as a summary of multiple cases. A statistical evidence message should be more persuasive following the norms of the argumentation theory, because such a message describes that a statistical computation has been made about a large sample size (Hornikx, 2005), and can yield statistical evidence such as odds and percentages (Hoeken & Hustinx, 2003; Kopfman et al., 1998).
The third evidence type is causal evidence. This evidence type can be defined as “an explanation of an occurrence or effect” (Haans et al., 2013, p. 155). Hoeken and Hustinx (2003) add to this that causal evidence contains the explanation of how a certain effect is explained,
following a cause and effect relationship. Slusher and Anderson (1996) found that causal is a mediator in the perseverance of social theories. This means that even though the evidence might not be true, the causal thinking steps, or the explanation availability, made it easier for people to assume something to be true.
Table 3.
Overview of performed studies on evidence types
Evidence type Reference
Expert Hoeken & Hustinx (2003); Hornikx & Hoeken (2007); Haans et al. (2013); Eisen & Tarrahi (2016)
Statistical Slusher & Anderson (1996); Hoeken (2001); Hoeken & Hustinx (2003)
Causal Slusher & Anderson (1996); Hoeken (2001);
Hoeken & Hustinx (2003)
Anecdotal Slater & Rouner (1996); Hoeken (2001); Hoeken & Hustinx (2003);
The fourth and final evidence type is anecdotal evidence. Anecdotal evidence can be described as evidence that uses illustrations or case stories to strengthen the argument quality (Haans et al., 2013). Although anecdotal messages tend to persuade people more than messages without anecdotal evidence (Slater & Rouner, 1996), it has the weakest effect on persuasion in comparison to the other three evidence types (Hoeken & Hustinx, 2009). Haans et al. (2013) did not include anecdotal evidence in their study as a variable because anecdotal evidence is found to be ineffective, especially when used in short texts.
When all four evidence types are compared, a clear distinction can be made between these strategies. Contrary to expert, causal and anecdotal evidence, statistical evidence is different in the way that this is the only evidence type which displays more than just one source
as a foundation for an argument, as it includes a statistical computation and sample size (Haans et al., 2013). It would be interesting to investigate if and how these evidence types can create better conversions on SEA landing pages.
2.4. Internal Cues and Product Involvement
Although purchase behavior can be influenced by factors such as online purchasing experience and evidence types, there are also internal factors that should be taken into account. Internal factors that can influence a person’s decision to buy or not can be described as internal cues.
These internal cues can be divided into cognitive and affective states. A cognitive state can be described best as a state where consumers take the time to understand, think about and interpret information (Dawson & Kim, 2009). An affective state can be recognized as a person’s self-feelings, emotional state, and mood (Dawson & Kim, 2009).
Dawson and Kim (2009) tested in their studies the effect of cognitive and affective state on impulse buying tendencies and found that when a consumer is in an affective state, it is more likely that they will buy products impulsively, while a cognitive state reduces the likeliness of impulse buying behavior.
Besides internal cues, also involvement is an important factor that affects purchase behavior. Product involvement has often been a subject in studies because of its importance in influencing the cognitive and behavioral responses of consumers (Dholakia, 2001). Involvement can be described as the intrinsic importance of a product to the buyer (Pavlou, Liang & Xue, 2007), measured by the amount of arousal, interest or drive evoked by a product class (Dholakia, 2001), or how interested the buyer is in a certain product (Koufaris, 2002). Product involvement is a consumer-defined concept instead of a product-defined concept, and
it essentially measures the consumers’ response to a product. An example of category involvement is that a consumer might have a lot of interest in fashion.
The level of involvement of a consumer indicates the amount of thought and motivation a consumer has when purchasing a product (Prendergast, Tsang & Chan, 2010). This means that high involved consumers should be more risk-averse when making buying decisions. It would be interesting to see if evidence types in copywrited texts can affect the perception of risk in buying products.
Current research could test if these internal cues also affect decision making in the online retail environment, and if copywrited texts can change the way people process their purchasing decision making.
To summarize, prior research surrounding search engine advertisement, click-through rates and conversions focused mainly on advertisement descriptions, advertisement extensions such as ratings and advertisement positions, but these studies failed to incorporate the effect of landing page texts and personal characteristics such as involvement, and other internal cues. On top of this, prior research was often limited to one or two product categories, due to real-time data from companies.
Although the possible existence of potential moderators has been mentioned by previous researchers (Haans et al., 2013), research which investigates potential moderators, such as involvement and product category, on these relationships is still missing. Therefore, this research will be novel in the way that it will focus on multiple product categories to see if there is a distinction in effectiveness of copywriting texts when tested around different product categories, and if there are moderating variables that can influence this relationship, the goal of this research is to find the possible effects of landing page copywriting strategies on the conversion intention.
This research is important because search engine advertising grew by 400% in the last six years to over $92 billion, and while these number will probably rise even further, research about how to maximize results are not completed yet. This research will contribute to the existing research surrounding landing page optimization and search engine optimization. This study will help copywriting managers to optimize web texts so, that conversions will be higher. Therefore, this study seeks to find an answer to the question: What drives customer responsiveness to different evidence types on landing pages in terms of conversion intention?
2.5. Prior Online Shopping Experience
While evidence type has an effect on purchase behavior, it is not the only factor that affects it. Future behavior, in this case purchasing a product, is determined strongly by prior experience (Ling et al., 2010). The decision whether a potential customer converts or not is dependent on three different aspects, context, stimulus and prior experiences (Helson 1963, in Ling et al., 2010). Shopping behavior is determined by factors such as ease of navigation, service, security, personalization, and risk involved (Ling et al., 2010).
Risk and trust are important predictors of online shopping behavior (Ling et al., 2010). According to Lu et al. (2017), a trustor will develop cognitive trust when good reasons to trust have been identified. When a customer trusts a brand or company, the perceived risk is lower, which ultimately would lead to a better shopping experience (Lee & Turban, 2001, in Ling et al., 2010). According to Ling et al. (2010), customers who have online purchase experience, are more likely to buy again through online channels than people who do not buy through online channels. When online shoppers do not know what to expect, they are more risk-averse than shoppers with online purchase experience (Lee & Tan, 2003).
If prior online purchases resulted in satisfactory outcomes, customers are likely to purchase again through online channels (Shim, Eastlick, Lotz & Warrington, 2001). This effect
also works vice versa, where negative outcomes cause people to be reluctant in engaging in online shopping (Ling et al., 2010).
2.6. Conversion rate
Conversion rate is beside click-through rate one of the two measurements that is most commonly used as online marketing metrics and as a key performance indicator (KPI). Previous authors have argued that the effectiveness of online advertising should be evaluated by their ability to create interest, measured by click-throughs, and other behavioral responses, such as sales measured by conversion rate (Lu, Chau & Chau, 2017). These two KPIs are also the two most used charging models in the advertising industry with the cost-per-click (CPC) and the cost-per-acquisition (CPA) method, where click-through rate and conversion rate respectfully are the main drivers for charging models for franchisees and budget allocation for marketing departments.
Yang, Lin, Carlson and Ross (2016), define conversion rate as a measure of sales return on an advertisement. However, there are many more ways for an online business to define conversions. Conversions also entail filling out a form (Nazerzadeh, Saberi & Vohra, 2008). In this paper, conversion rate will be defined as the percentage of consumers who take a desired action, examples are: subscribing to a newsletter; making a phone call through the action of pressing a particular button; or buying a product.
𝐶𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛 𝑟𝑎𝑡𝑒 =# 𝑑𝑒𝑠𝑖𝑟𝑒𝑑 𝑎𝑐𝑡𝑖𝑜𝑛𝑠 # (𝑝𝑎𝑔𝑒)𝑣𝑖𝑠𝑖𝑡𝑜𝑟𝑠
Conversion rate is at the base of the CPA model, where retailers only pay publishers if a conversion has actually occurred. This is a charging scheme which is most desirable for advertisers (Nazerzadeh et al., 2008), because they will only pay if they made a conversion.
Since most search engines work with a cost-per-click charging model, where advertisers pay money regardless of a conversion or not, it is becoming increasingly important for advertisers to optimize their landing pages, so conversion rates will be as high as possible and together with that, the return on marketing investment.
Studies that focused on enhancing click-through and conversion rate often times limited their research to advertisement description content (Haans et al., 2013), advertisement extensions (Gupta & Mateen, 2014), advertisement star ratings and advertisement positions in search listings (Cezar & Ögüt, 2016). Studies focused on the communication in landing page texts are missing in the field of online marketing. Therefore, this study is focused on how the incorporation of evidence types in the communication on landing pages can benefit the process of conversion optimization.
3. Conceptual framework and hypotheses.
In order to conduct a quantitative research with a deductive approach, a conceptual model needs to be developed. This model will be established through the generation of
hypotheses which are based on previous theories regarding purchase intention and its antecedents: conversion rate and evidence types.
3.1. Evidence Types
While purchase intention seems to be a self-evident antecedent of conversion rate, the opposite can be the case. Consumers who interact with a webshop must define their tasks in the decision-making process (Gudigantala, Bicen & Eom, 2014). They need to search the needed information and generate alternatives, after that the alternatives need to be compared and a choice need to be made before a purchase can be made (Kotler, 2003). Gudigantala et al. (2014) argue that consumers who are in a decision-making process, face difficulty while going through those decision-making steps. It is possible that consumers abandon their shopping cart or the website and proceed to do their purchase on another website.
The five-stage model of the consumer buying process shows that consumers are recognizing a problem and are searching for information to solve this problem followed by an evaluation of all alternative products (Kotler & Keller, 2012). This means that consumers need to be persuaded by the seller before the consumer will actually purchase a good or service.
A number of studies have been performed to examine the persuasiveness of different evidence types (Hornikx, 2005), i.e., statistical, anecdotal, causal and expert. These first three evidence types are comparable to the founding research methods being respectively a survey, case study and experiment, while expert evidence can be seen as referring to prior studies in scientific articles (Hoeken & Hustinx, 2003).
Outcomes of the relative persuasiveness of evidence types compared to each other varied among the different studies, but except for Koballa (1986), all other researchers found that anecdotal was weakest in relative persuasiveness compared to the other evidence types. Reasons for this is that accepting a claim which is supported by anecdotal evidence would be a
“fallacy of hasty generalization” (Hoeken & Hustinx, 2009, p. 494), which implies that one cannot accept a claim which is supported by a single instance.
Haans et al. (2013) argued in their study that consumers who have a goal when they are browsing the internet (e.g. purchase or search for a specific product), are taking the central route. They are planning to shop and are processing the content of an advertisement more actively. Therefore, they are more likely to respond better to a causal advertisement, because this is regarded as a strong quality argument (Haans et al., 2013).
Consumers who are conducting their initial search, are more likely to process advertisements heuristically (Haans et al., 2013), meaning they want to learn themselves. Source credibility and verifiability of information will result in higher interest in terms of CTR (Haans et al., 2013), but because these consumers are in their initial search phase, it is less likely that they will convert.
Hoeken and Hustinx (2003) also found that anecdotal evidence had the weakest persuasion power compared to the other evidence types even though the claim its understandability score was the same as the others. Therefore, we propose
H1: Causal evidence will lead to the highest conversion intention, followed by expert and anecdotal evidence, respectfully.
3.2. Internal Cues
There are a lot of studies that focused on buying behavior and impulse buying (e.g. Haans et al., 2013; Dawson & Kim, 2009; Ling et al., 2010). Antecedents of these buying behaviors are among others, the internal factors of the individual, also called the internal cues (Dawson & Kim, 2009). These internal factors are factors that involve some of their personality traits which can predict impulse buying behavior. Studies have explained that the act of buying impulsively is an outcome of the imbalance between willpower and the urge to possess a certain
product, where the urge is bigger than the willpower to resist (Vohs & Faber, 2007). This also highlights the two separate aspects of this buying behavior, which are the desire to buy, the affective state, and the willpower to resist, the cognitive state.
The affective aspect is about how people feel, what their emotional state is and what mood they are in (Dawson & Kim, 2009). When a consumer is in an affective state, internal stimuli are processed affectively which can result in feeling the urge to buy products, which leads to impulse purchases (Dawson & Kim, 2009). Retailers try to trigger the affective state by placing products in a physical proximity (Vohs & Faber, 2007), but Verplanken and Herabadi (2001) showed that affective behavior can be part of a person’s personality and that extravert people are more likely to act in an affective way.
The cognitive aspect of internal cues entails the amount that people think about and interpret information before they make decisions. When thoughtful consideration is out of the picture and a lack of cognitive deliberation is present, consumers are more likely to act and buy impulsively (Vohs & Faber, 2007; Verhagen & Van Dolen, 2011). When a consumer has a high level of self-control or willpower, they are less likely to buy impulsively (Vohs & Faber, 2007).
Since impulse buying behavior would lead to easier sales, we expect that internal cues can affect conversion intention. Therefore, we hypothesize that:
H2: The positive effect of evidence types on conversion intention is moderated by internal cues so that this effect is a) stronger for higher values of affective state, and b) weaker for higher cognitive states.
3.3. Prior Online Shopping Experience
Just as we do with personal relationships with humans, we prefer to have contact and relationships with those who we trust. Trust can best be defined as the willingness to rely on a person or entity to perform as it was stated (Laroche, Habibi, Richard & Sankaranarayanan,
2012). In other words, it has to do with if the consumer is willing to take the risk that a brand or company is not going to live up to the expectations.
Consumers with less experience or prior knowledge are more risk-averse since their behavioral choice is mostly dependent on their perceived chance for success (Ling et al., 2010). In the study performed by Ling et al. (2010), prior online purchase experience was found to have the strongest impact on customer online conversion intention. This is likely the case because consumers with prior online experience are less risk-averse than inexperienced users because they have had satisfying prior online shopping experiences. Therefore, we hypothesize the following:
H3: There is a positive effect of Prior Online Shopping Experience on conversion intention.
Wiener and Mowen (1986) found that trust has an impact on the persuasiveness of claims. They claim that trust causes an attitude change in the perception of the persuasiveness of messages. Since consumers with higher online shopping experience are more likely to trust websites and e-commerce retailers (Ling et al., 2010), the following is hypothesized:
H4: The positive effect of evidence types on conversion intention is moderated by prior online shopping experience, so that this effect is stronger for higher values of prior online shopping experience.
3.4. Category Involvement
In order to predict conversion, it is good to know how involved a person is when they buy a product or service. Product involvement has often been a subject in studies because of its importance in influencing the cognitive and behavioral responses of consumers (Dholakia, 2001). Involvement can be described as the intrinsic importance of a product to the buyer
a product class (Dholakia, 2001), or how interested the buyer is in a certain product (Koufaris, 2002). Product involvement is a consumer-defined concept instead of a product-defined concept, and it essentially measures the consumers’ response to a product. An example of product involvement is that a consumer might have a lot of interest in fashion.
Jiang, Chan, Tan and Chua (2010) state that involved consumers are more motivated to search for information that can help them with making their purchase decision. Therefore, they are more likely to find valuable product information from a website which can help them ease their decision-making. High involved shoppers are often satisfied with products or companies (Park, Lennon & Stoel, 2005), and are therefore more likely to spend more time in the store, which increases conversion intention (Jiang et al., 2010). Therefore, we hypothesize that:
H5: Product involvement will have a positive effect on conversion intention.
Haans et al. (2013) argue that there are different ways for people to process information, the peripheral route, and the central route. The peripheral route is taken by those users who are in the first stages of searching for product information (Singh & Dalal, 1999). These users typically score low on involvement and are more likely to respond favorably to simple acceptance and rejection cues, which are cues that expert and statistical evidence provide (Haans et al., 2013).
The central route, however, is typically taken by consumers with a particular goal in mind, for example, purchasing a product (Singh & Dalal, 1999). This group is small compared to the consumers who take the peripheral route. The level of involvement of a consumer indicates the amount of thought and motivation a consumer has when purchasing a product (Prendergast, Tsang & Chan, 2010). The shoppers who take the central route are planning to shop and are motivated more to process the content of messages. These consumers see causal
evidence as a strong argument (Slusher & Anderson, 1996), and therefore are more likely to respond more favorably to this evidence type. Therefore, we hypothesize that:
H6: Product involvement will influence the positive effect of evidence types on conversion intention in such a way, that a) low product involved users are more likely to convert with expert evidence, and b) high product involved users are more likely to convert with causal evidence.
3.5. Product Category
When product categories are discussed, it is important to take some characteristics of these products into account. When purchasing online, a consumer can have a higher need to touch and experience products (Lynch, Kent & Srinivasan, 2001). This is especially the case for tangible product types. According to Lynch et al. (2001), consumers will have a preference for offline channels for purchasing tangible products. Zeng and Reinartz (2003) dispute this and argue that it is not solely dependent on tangibility since books are sold very often through online channels. Also, the argument from Lynch et al. is most likely outdated, since e-retailers are becoming more flexible in their return policies and ease of returns.
Stone & Grønhaug (1993) found that perceived financial risk has a strong positive impact on overall risk, which could decrease the conversion intention. Liao & Cheung (2001) studied multiple determinants of willingness to e-shop on the internet and found that product price indeed has a negative effect on the customers’ willingness to e-shop. Since financial risks are higher for products that are more expensive, and consumer electronics are oftentimes more expensive than clothing we propose the following:
H7: Product category has an impact on the effect of evidence types on conversion intention in such a way, that the effect will be stronger for fashion compared to electronics.
Table 4 will show an overview of all independent and dependent variables in this study and shows our proposed directions for the effects.
Table 4
Variables with the proposed direction for the effects
IV DV Direction
Causal Evidence Conversion intention +
Expert Evidence Conversion intention +
Anecdotal Evidence Conversion intention +
Prior Online Shopping Experience (POSE) Conversion intention +
Product Category Involvement Conversion intention +
Affective State × Anecdotal Evidence Conversion intention + Affective State × Causal Evidence Conversion intention + Cognitive State × Expert Evidence Conversion intention +
POSE × Evidence Types Conversion intention +
Causal Evidence × Product Category Involvement Conversion intention + Product Category × Evidence Types
Fashion Electronics Conversion intention Stronger Weaker 4. Research Design
Because this study will test theory, a quantitative deductive approach is used to test our hypotheses and the conceptual model. Constructs are measured through a cross-sectional, online experimental survey, where the responsiveness to different evidence types and other influential factors will be tested. The variables will be tested numerically and they will be analyzed using a range of statistical techniques. This design has been chosen due to the time constraint in which this research needs to be performed. This study can also be performed with real-time data, but to avoid possible idle time and the risk of not being able to access data, a survey is chosen.
The choice for a quantitative experimental survey was simple, this method makes it possible to test our hypotheses and claim causality between the independent and the dependent variable. We will use a between-subjects design, so it allows us to compare the results between the experimental groups and the control groups (Saunders, Lewis & Thornhill, 2012). The between-subjects design helps us avoid a potential contrast effect between the different evidence types.
4.1. Sampling method
Since there is no sampling frame, a non-probability sampling method will be used, and because of the distribution through social media channels, a self-selected sample will be most likely the outcome. Most respondents will be recruited through the distribution of the survey on different social media channels, with hyperlinks referring to a Qualtrics web survey. In this case, a self-selection bias will be unlikely to influence the research outcomes since commonly detected biases, such as overrepresentation of socioeconomic statuses or competent participants (Krokoff, 1990), do not affect conversion intention.
4.2. Measures
This study will use pre-defined constructs with multiple items measuring our variables from our conceptual framework. These constructs will be already proven to be valid from prior studies.
4.2.1 Conversion Intention
To measure the probability of conversion, we measure the willingness to buy through an adapted scale consisting of four items developed by Dodds, Monroe and Grewal (1991), which showed a good internal consistency (α = 0.92) in the study performed by Grewal,
Krishnan, Baker and Borin (1998). An example item is: “The probability that I would consider buying at this store is high”. All statements are answered on a Likert scale ranging from strongly disagree (1) to strongly agree (7). A high score on these statements means that the respondent is probable to buy the product on this website.
4.2.2 Evidence Type
In this survey experiment, respondents are exposed to four different e-commerce web pages, where anecdotal, expert and causal evidence types, accompanied with one control group will be used to persuade the respondent to buy the product. Evidence types will be incorporated in the text according to the examples given by Haans et al. (2013), but expert and statistical evidence are merged into one since most advertisements use expert evidence in a way where experts sum up statistics and graphs from previous research (Koh, 2011).
This study chooses to incorporate the anecdotal evidence type since there is no limit on the number of words which can be used on the web page, while Haans et al. (2013) chose to not use anecdotal evidence since there is a word limit in SEA description texts of 80 (back then 70) characters (Google, 2018a; Haans et al., 2013), and anecdotal evidence is less persuasive in short texts. Table 5 shows the final operationalization of the texts.
4.2.3 Product Category
Product category will be measured on a nominal scale. This study will use two product categories previously used by Konuş et al. (2008), which are consumer electronics, clothing/apparel.
4.2.4 Involvement
Involvement is measured by a seven-point semantic differential scale consisting of ten items with a Cronbach’s alpha of .95 developed by Zaichkowski (1985). These ten items are covering how interested someone is in a certain product or service. An example item is: “Please use the series of descriptive words listed below to indicate your level of interest in (product): unimportant - important”. All statements are measured on a seven-point bipolar scale. The scale consists of 3 counter-indicative statements to flag careless responses. A high score on these involvement statements would imply that the respondent cares about that particular product.
4.2.5 Prior Online Shopping Experience
Prior online shopping experience is measured by three items used by Khalifa and Liu (2007). These developed items all have a high loading of at least 0.8 and a demonstrated convergent validity at the 1% level (Khalifa & Liu, 2007). An example statement is: “I shop online frequently”. All statements are answered on a Likert scale ranging from strongly disagree (1) to strongly agree (7). A high score on these statements means that the respondent is a very experienced online shopper.
4.2.6 Cognitive State
The cognitive state of internal cues is measured by the validated ten-item scale with a Cronbach’s alpha of .91 developed by Verplanken and Herabadi (2001). An example item is: “I usually think carefully before I buy something”. All statements are answered on a Likert scale ranging from strongly disagree (1) to strongly agree (7). A high score on these statements would imply that the respondent often plans a purchase and thinks about it before purchasing it. The cognitive scale also consisted of three counter-indicative items, an example item is: “I often buy things without thinking”.
4.2.7 Affective State
The affective state of internal cues is measured by the validated ten-item scale with a Cronbach’s alpha of .83 which is also developed by Verplanken and Herabadi (2001). An example item is: “It is a struggle to leave nice things I see in a shop”. All statements are answered on a Likert scale ranging from strongly disagree (1) to strongly agree (7). A high score on these statements would imply that the respondent often makes purchases decisions without deliberation. The affective scale also consisted of one counter-indicative item, this is: “I’m not the kind of person who ‘falls in love at first sight’ with things I see in shops”.
Table 5
Evidence types with body textsa
Product Type Web page evidence type Body text
Electronics Causal Evidence Buy your gadgets and electronics at SenaMoino and enjoy hassle-free shopping with a great return policy and a not satisfied, money back guarantee.
Expert Evidence Tech reporter NRC: “By choosing SenaMoino, you’re choosing for service which is chosen by over 100,000 satisfied customers. 97% of them would recommend SenaMoino to their friends and colleagues.”
Anecdotal Evidence Bram and Sara, both living in Amsterdam, are prior customers. Because they chose SenaMoino, they enjoy the ease of using the newest electronics, and that helps them in their daily lives.
Control Condition Buy your gadgets and electronics at SenaMoino.
Fashion Causal Evidence Buy your fashion at SenaMoino and enjoy hassle-free shopping with a great return policy and a not satisfied, money back guarantee.
Expert Evidence Fashion Stylist Fred van Leer: “By choosing SenaMoino, you’re choosing for service which is chosen by over 100,000 satisfied customers. 97% of them would recommend SenaMoino to their friends and colleagues.”
Anecdotal Evidence Bram and Sara, both living in Amsterdam, are prior customers. Because they chose SenaMoino, they walk the streets with better style, and that helps them in their daily lives. Control Condition Buy your fashion at SenaMoino.
4.3. Procedure
First, a pilot test has been conducted with five participants, to see how long it takes on average to fill in the survey and to remove any grammatical errors and reformulate poorly stated questions. This pilot test was successful and led to a few reformulated questions or additional comments on how to properly answer particular questions. Also, the pilot test showed that the test would take between six to ten minutes, depending on the device used, when filling in the survey.
The survey experiment was conducted online through Qualtrics Survey Software. Respondents were gathered by distributing an anonymous survey link through social media platforms such as Facebook, Instagram, LinkedIn, and WhatsApp. The response rate is therefore not calculated since it is impossible to see how many times the links are viewed.
On the first page of the survey experiment, a small notice has been said given about the nature of the survey experiment, and that the survey can be abandoned anytime. Then, contact details were given of the experiment leader, and the informed consent box had to be checked in order to proceed with the survey experiment.
At the following page of the survey experiment, the participants are shown a random mobile web homepage consisting of a big picture congruent with the clothing/apparel or consumer electronics category, together with one of the four evidence conditions. Then, participants need to indicate their intentions to shop with the website and how involved they are with the shown product category. After this, additional information is asked about the other variables. The full survey flow can be found in the appendix.
4.4. Analysis and predictions
According to Field (2013), this study can be best analyzed by a multiple regression since the outcome variable and the multiple predictors are all continuous. In order for the evidence
types to be analyzed through a linear regression, the evidence types needed to be recoded into three dummy variables; expert, causal and anecdotal evidence and a control condition. Then, a multiple regression analysis will be performed based on the methods used by Baron and Kenny (1986). A regression analysis will be performed first, testing the individual effects of evidence types, prior online shopping experience, and product involvement. It is expected that from these analyses we can say that online purchase experience and product involvement are causing an increase in conversion intention. Also, it is expected that causal evidence will have the strongest positive effect on the conversion intention.
After testing the direct effects, tests will be conducted to test the moderating effects of internal cues, involvement, and prior online shopping experience. It is predicted that the affective internal cues, involvement, and prior online shopping experience will strengthen the influence of evidence types on conversion intention. The cognitive internal cues will weaken the aforementioned influence. Also, it is expected that the fashion category will cause a stronger effect for evidence types and electronics will have a weaker effect on the influence of evidence types on conversion intention.
4.5. Data Preparation
Before analyzing the data, the dataset needed to be cleaned and reversely coded items needed to be computed into a new variable with the right coding. Out of the complete list of 316 respondents, 104 responses were deleted because of incomplete responses.
Three involvement items, three cognitive and one affective internal cue items needed to be reversely coded so that a high score on the Likert scale (e.g. a score of seven) would imply for example, that the respondent has a high cognitive state when making purchases.
4.6. Sample
The sample consisted of 212 respondents, all gathered through the online distribution of the survey experiment. The respondents were aged between 13 and 75 years with an average age of 35 (SD = 14.0). 56 percent of all respondents were female and the majority of the respondents have studied at a university of applied science (36.3 %). Also, an overwhelming majority of 64.6 percent filled the survey in through their smartphones.
4.7. Reliability and correlations
The measurements were reliable for affective internal cues (α = .820), cognitive internal cues (α = .873), product involvement (α = .895), online shopping experience (α = .925) and conversion intention (α = 0.938). In the product involvement scale and the affective internal cues scale, one item needed to be deleted for each scale, since the corrected item-total correlations for these items were insufficient and scored lower than .30 (Nurosis, 1994; as cited in Cristobal, Flavián & Guinalíu, 2007).
The product involvement scale has a high reliability (α = .895) after the deletion of one item due to a low corrected item-total correlation, which indicated that the item had an insufficient correlation with the total score of the scale (.275). None of the remaining items would substantially affect reliability if they were deleted ( < .10).
The measurement scale for affective internal cues is reliable (α = .820). One item was deleted, this item had a corrected item-total correlation of .045 which indicates that this item has an insufficient correlation with the total score of the scale.
The correlation table in Appendix A shows some interesting relations between the analyzed variables. For instance, it shows that there is a small negative effect of age on conversion intention r (212) = -.212, p < .01 and affective cues r (212) = -.212, p < .01. This
means that when someone gets older, they are less likely to purchase and are less likely to be led by impulsive emotional cues. Age also has a moderate negative effect on involvement r (212) = -.423, p < .01 which means that when someone gets older, he or she gets less involved with certain products. Lastly, age has a small positive effect on cognitive cues r (212) = -.198, p < .05, which makes sense, older shoppers are led less by impulsive emotional cues, and more by well-thought about decision-making.
The correlation table also shows that there are small positive relations between gender and yearly expenditure r (212) = .170, p < .05, involvement r (212) = .193, p < .01, cognitive cues r (212) = .197, p < .01 and a small negative relation between gender and affective cues (r 212) = -.156, p < .05. This would mean that men are less likely to be led by impulsive emotional cues and more by well-thought about decision-making. It also means that men are more involved with electronics than women and spend more on this on average per year.
Lastly, conversion intention and affective cues have a small positive relation r (212) = .222, p < .01, which means that when people are led by impulsive emotional cues, they are more likely to shop at a store. The full correlation table, with means, standard deviations and reliabilities can be seen in the appendix.
5. Results Table 6
Results of Multiple regression analysis in the fashion category
Control Variables All effects
Dependent variable Conversion intention Conversion Intention
Coefficient SE Beta Coefficient SE Beta
Constant 3.937*** .280 3.397** .965 Expenditure 4.21 E-5 .000 .050 4.71 E-5 .000 .056 Gender .094 .192 .034 .108 .199 .039 Age -.007 .007 -.070 -.006 .008 -.064 Expert Evidence 1.010 1.407 .320 Causal Evidence -.193 1.293 -.060 Anecdotal Evidence .649 1.139 .207 POSE -.032 .138 -.038 Product Involvement .165 .095 .132# Anecdotal Affective Cues -.141 .197 -.153
Causal Affective Cues -.017 .176 -.019
Expert Cognitive Cues -.133 .193 -.194 Causal Product Involvement -.259 .209 -.399 Anecdotal POSE -.074 .181 -.127 Expert POSE -.103 .192 -.181 Causal POSE .249 .175 .432 R2 .008 .058 Note: N = 212, *p<.05, **p<.01 Table 7
Results of Multiple regression analysis in the electronics category
Control Variables All effects
Dependent variable Conversion intention Conversion Intention
Coefficient SE Beta Coefficient SE Beta
Constant 3.949*** .272 4.462*** .892 Expenditure 4.59 E-5 .000 .052 6.72 E-5 .000 .077 Gender .143 .191 .051 .262 .202 .094 Age -.021 .007 -.211* -.022 .008 -.220* Expert Evidence -1.768 1.273 -.540 Causal Evidence -.798 1.131 -.252 Anecdotal Evidence -1.715 1.075 -.544
POSE -.097 .113 -.114
Product Involvement -.035 .115 -.028
Anecdotal Affective
Cues .341 .187 .390
#
Causal Affective Cues .286 .166 .340#
Expert Cognitive Cues .108 .170 .154 Causal Product Involvement -.009 .192 -.013 Anecdotal POSE .077 .174 .139 Expert POSE .301 .174 .506# Causal POSE .022 .157 .037 R2 .051 .116 Note: N = 212, *p<.05, ***p<.001, #p<.10
A hierarchical multiple regression analysis has been performed to test the effects of evidence types, internal cues, prior online shopping experience, product category and product involvement on the conversion intention at a website, after controlling for gender, age and expenditure. The results of the two regressions, one for each product category, can be seen in Table 6 and Table 7.
In the first step of the analysis the average yearly expenditures, age, and gender were entered as control variables. This model was statistically insignificant for the fashion category F (3, 208) = .54; p = ns. For the electronics category, the model was statistically significant F (3, 208) = 3.56; p < .05 and explained 5.1 % of variance in conversion intention.
In the second step, all direct and moderation hypotheses were tested. In the fashion category, the second model was again insignificant F (15, 196) = .805; p = ns. In the gadget category, the model was significant after the entry of prior online shopping experience, product category involvement, and all hypothesized interactions, the model explained 11.6 % variance in conversion intention F (15, 196) = 1.717; p < .05.
In the fashion category, none of the proposed effects were significant at the .05 level. One effect was marginally significant, this was the positive effect of product involvement (β =
.165, p = .083), which suggests that product involvement would have a positive impact on conversion intention.
In the electronics category, however, one out of fifteen predictors had a significant impact on conversion intention, being age (β = -.220, p < .01), our control variable. In other words, when someone gets older, he or she gets less interested in purchasing products at the electronics website.
Also, three effects were marginally significant with p-values below the 10%. There is a marginally significant positive interaction effect between anecdotal evidence and affective state (β = .390, p = .069) and also between causal evidence and affective state (β = .340, p = .087). This means that consumers who are led by affective internal cues are more likely to respond positively to anecdotal and causal evidence in terms of conversion intention. Lastly, the interaction between prior online shopping experience and expert evidence was marginally significant (β = .506, p < .085), which suggests that consumers with online shopping experience respond more positive to expert evidence.
The Wilcoxon Signed-Ranks test (Wilcoxon, 1945) is used as a non-parametric substitute of the paired-samples t-test to test hypothesis seven. The output showed that the conversion intention on the fashion landing page had a statistically significantly higher score than the electronics category (Z = -4.33, p < .01). These results will be discussed extensively in the discussion section.
Table 8
Overview of all hypotheses
H1 Causal evidence will lead to the highest conversion intention,
followed by expert and anecdotal evidence, respectfully. Rejected H2a The positive effect of evidence types on conversion intention
is moderated by internal cues so that this effect is stronger for higher values of affective state.
Partially Accepted H2b The positive effect of evidence types on conversion intention
is moderated by internal cues so that this effect is weaker for higher cognitive states.
