Content-based influence on website trustworthiness

(1)

1

Content-based influence on website

trustworthiness

Kubilay Keser 10610286

Bachelor thesis Bachelor Information Sciences

University of Amsterdam Faculty of Science

Science Park 904 1098 XH Amsterdam

Bachelor scriptie

Bachelor Opleiding Informatiekunde Universiteit van Amsterdam

Faculteit der Natuurwetenschappen, Wiskunde en Informatica Science Park 904 1098 XH Amsterdam Supervisor Jorg Brunner 2e_Examinator Dick Heinhuis

(2)

2

Abstract

The percentage of Dutch companies that have their own website has seen an increase (Statistics Netherland (CBS), 2015). This suggests the importance of the communication between customers and companies. It is important that websites look trustworthy, as an untrustworthy website can have negative consequences (Karimov, Brengman, & Hove, 2011). In this study, various trust factors were explored and two content-based factor groups were formed. The first factor group being social cues, which includes reviews, photos of humans and links to social media and the second factor group being content-design, which included Informativeness and e-assurances. Four images of websites were created, each incorporating a different combination of these factors groups. The trustworthiness of each website was asserted with the use of a questionnaire. The results suggest that the content-design factor group had a positive influence on the perceived trustworthiness of the website. The social cue factor group did not have any significant influence on the perceived trustworthiness. The results are discussed and suggestions for future research are outlined.

(3)

3

1. Introduction

Nowadays, almost every company has a website. Statistics Netherlands (CBS) suggests that in 2012, 84% of all Dutch companies had their own website. This percentage has jumped up to 90% in 2015 (Statistics Netherland (CBS), 2015). This increase in website ownership shows us the importance of communication between companies and visitors. In some cases, a visitor might only communicate with a company through their website without talking to any employees, as is often the case with e-commerce websites. Not having a website that is trustworthy can lead to a negative outcome, as a lack of trust towards the website is one of the main factors that deters consumers (Karimov, Brengman, & Hove, 2011).

Although most commerce sites prefer a physical store in shopping areas, close to 40% of e-commerce websites are situated in business areas (Bardoel, Schildkamp, & Weltevreden, 2014). As a result, not as many people walk by these stores. Therefore, most of the customers have their first contact with the company online instead.

Some e-commerce research has already been done on trustworthiness. For instance, a meta-analysis was done on literature concerning initial trust in a business-to-consumer (B2C) e-commerce setting (Karimov et al., 2011). Others like Lowry, Wilson and Haig (2014) have done research on the influence of website logos on website credibility. However, there is still a lack in research that focusses on website owner control on displayed content on e-commerce websites, even though the factors that fit into this category were found to have positive effects on trustworthiness (Karimov et al., 2011). In other words, the idea that there might be factors that can be directly and more easily implemented by website owners and that this property is the reason these factors are worth studying. The importance of this distinction is that smaller business owners might be able to relatively easily increase the perceived trustworthiness of their websites when lacking either the creative and technical knowledge to create a trustworthy design or the financial resources to hire a third party to create such a website for them.

This thesis will focus on content-based factors. In this case, design factors are not relevant, because the influence of company owners on the design choices are limited compared to the choice of content on their website. Content-based factors are all factors related to content published on the website directly by the company. This includes all photos and texts on the website. The goal of this thesis is examining content-based factors that might influence perceived trustworthiness of

commercial websites in a controlled setting. While doing this, this study will focus on the initial impressions users form after viewing a website for the first time, as opinions about a website can be assessed within 50ms (Lindgaard et al., 2006). This means that interactions will not be within the scope of this study. In this case, the term commercial website envelops web shops, as well as commercial websites that offer services.

(5)

5

2. Related works

In this study, the chosen factors are based on the meta-analysis done by Karimov and colleagues. However, not all factors will be included. The property the chosen factors have in common, is that the business owners have control over these factors. For example, an external company might build and design a website for business owners. However, the business owners themselves decide on what their policies are and what kind of images they upload onto their website. All these factors will be referred to as content-based factors. Content-based factors will be a term used to describe all features and content that business owners are able to upload to their website. Figure 1 provides an overview of the way the factors are grouped. This can range from social cue factors like links to social media pages to a content-design factors like a list of policies and guarantees the business owners offer.

Figure 1. factor overview

2.1 Trust factors

The idea that trust has an important place within e-commerce isn’t a new one. It is argued that e-commerce is about business and that businesses are built on relationships. These relationships are built on trust. Trust is said to be an essential ingredient for e-commerce and in creating loyal and satisfied customers. It is concluded that until the perceived risk of doing business electronically is reduced to an acceptable level, e-commerce will not reach its full potential (Ratnasingham, 1998). Nowadays, this important relation between trustworthiness and e-commerce is still relevant, as recent findings still suggest high overall trust has positive effects on purchase intentions online (Oliveira, 2017).

To examine the initial trust in a business-to-consumer (B2C) e-commerce setting, a meta-analysis was done. In this study, Karimov and colleagues reviewed literature based on different factors that could influence initial trust of consumers. These factors were gathered and grouped in three main dimensions: Visual design, Social cue design and Content design. Visual design

encompasses all graphical and structural qualities a website might have, while social cue design and content design focus on social/human characteristics and informative features respectively. The study found that their general hypothesis that these web design cues do enhance initial trust toward

(6)

6 unfamiliar online vendors. They suggest that e-commerce websites should consider implementing human-like cues into their interfaces. They also found that internal e-assurance structures, such as privacy policies and guarantees can be as effective as paid e-assurance mechanisms. The study also shows that that certain factors within the visual and social cue dimensions are still under-researched (Karimov, Brengman, & Hove, 2011). To gain insight into what kinds of factors might influence the consumers’ perceived trustworthiness in an e-commerce website, the next section will explore related works on a variety of trust factors.

2.2 Social cue factors

2.2.1 Images

The influence of images depicting sales personnel on consumer trust has been described to have a positive effect on trustworthiness. A laboratory-based experiment was conducted to assess the initial trust of consumers. Four e-commerce websites were used with a variety of images of sales personnel. The proposition that the stimulation of face-to-face contact builds initial trust was supported by the experiment’s findings. These findings suggested that the initial trust was enhanced for web sites including photographs and video clips when compared to web sites lacking such imagery (Aldiri, Hobbs, & Qahwaji, 2008).

To gain insight into how internet users perceive human images as an element of website design, an experiment with different types of images was conducted. This happened in a controlled setting where the participants browsed an e-commerce website. Three types of images were used. These consisted of images of humans with facial features could be seen, images of humans where the facial features couldn’t be seen, and a control condition with no images of humans. A paper-based questionnaire was used to gauge the participants’ experience, followed by an interview. Although the study did conclude that the human images overall had a positive effect on trust, no direct relationship was found between the human image condition and trust (Cyr, Head, Larios, & Pan, 2009).

2.2.2 Social Media cues

The inclusion of imagery is not the only way of implementing social and human-like features in a website. For example, the use of social media cues is a way to integrate these facets. Customer reviews and communities like Facebook and Twitter are examples of social media cues. Consumers’ online experiences can be improved by the inclusion of social media cues into e-commerce websites. These cues increase the perception of employee presence (Wang et al., 2007). Furthermore, reviews from customers who have already had experience with an e-commerce website also increase the social presence in a website (Mudambi & Schuff, 2010). This social presence, in turn, has a positive

(7)

7

2.3 Content design factors

Content design refers to the information components of a website, either textual or graphical (Wang & Emurian, 2005). Examples of these components are things like product and service information, company information and policy information (Hsiu-Fen, 2007).

2.3.1 Informativeness

Informativeness is described as the ability to inform customers about product alternatives. By providing information about products, the consumer is able to distinguish the product from others and make a proper decision. This includes up-to-date the information, correct information, useful

information, and completeness of information. To study whether Informativeness had an influence on consumers’ purchase behaviour in an e-commerce context, a survey website was built and

hyperlinked to each e-commerce website that was being researched. Informativeness can have a significant effect on e-commerce websites, as results of the survey indicate that information quality affect information satisfaction and relational benefit, that, in turn, are significantly related to each customer’s site commitment and actual purchase behaviour (Park & Kim, 2003).

2.3.2 e-Assurances

e-Asurances are defined as internally and externally provided assurance structures like company policies and third-party seals (Karimov et al., 2011). Two common seals often found on websites such as Alternate.nl and Coolblue.nl, are the Thuiswinkel Waarborg and the ICT Waarborg seals. Both the internally-provided and externally-provided assurances are implemented by e-commerce businesses to build trust amongst consumers by alleviating their concerns about the privacy and security of transactions on the e-commerce website (Bahmanziari et al., 2009). However, there have been some mixed results regarding the effect of such assurances. For example, research was done on the effects of the different combinations of three e-assurance functions and their effects on initial trust. The three e-assurances being security, privacy and transaction assurances. Detailed analyses show that there is a positive impact on consumers’ initial trust when the privacy insurance function was used, but only if the other two functions were absent. In contrast, these other two insurances, being security and transaction assurances, were found to only have a positive effect when the privacy insurance function was absent (Hu et al., 2010). Other studies have found e-assurance seals to be ineffective. For one experiment, multiple websites were created with each website having either no e-assurance seal, a TRUSTe seal, or a WebTrust seal. Participants visited one of the sites at random and completed a questionnaire afterwards. It was concluded that the seals had little influence on trusting beliefs (Fisher & Zoe Chu, 2009). Others also found that there is little association between the seals and trustworthiness (Kim, Steinfield, & Lai, 2008).

(8)

8

2.4 Thesis question and hypotheses

Based on previously conducted research, 5 questions and hypotheses were formulated. The main research question for this study is: “Do content-based factors influence the perceived

trustworthiness of e-commerce websites?” To answer this question, two sub-questions were

formulated. The first sub-question is: “Do social cue factors influence the perceived trustworthiness of e-commerce websites?” and the second sub-question is: “Do content design factors influence the perceived trustworthiness of e-commerce websites?” Based on the previously conducted research described in sections 2.2 and 2.3, it is hypothesized that both social cue and content design factors will have a significant effect on the perceived trustworthiness of e-commerce websites. When this is the case for both sub-questions, the main question will be able to be answered. As it is possible that certain e-commerce websites have a specific target audience, it is interesting to know whether the influence of content-based factors is influenced by age or gender. This in mind, two extra

demographic sub-questions were formulated. These are: “is there a difference in perceived trustworthiness between male and female participants?” and “is there a difference in perceived trustworthiness between age groups?” For both questions the null-hypothesis is that there is no significant difference between gender or age groups.

3. Method

The hypothesized influences on perceived trustworthiness were investigated by means of empirical research. For both “Social cue” and “Content design” factors, two conditions were tested. Either the factor had a “present” condition or an “absent” condition, which led to a 2 x 2 design.

3.1 Participants

An online survey was conducted using Google Forms because of its ease of use and built-in spreadsheet capabilities. Google Forms made it possible to create a sharable link to the survey. This link was distributed through a public Facebook post containing the link, and by posting a message on webwinkelforum.nl. No exclusion or inclusion criteria were applied to the survey. A power analysis for ANOVA and t-test suggested that the minimum sample size of 40 was required for an effect size of 0.75 and a power of 0.99 (Baardewijk, 2011). All 63 questionnaires were complete. 42,9% of the participants were male and 57,1% of the participants were female.

3.2 Materials

3.2.1 Websites

Because of the 2 x 2 design, four full scale images of websites based on pre-existing websites were created for this survey using Adobe Photoshop (Appendix C). Images were used as the

participants didn’t need to interact with the website, because, as described before, this study focuses on initial trust. Another reason is that website details and influence factors could be manipulated more

(9)

9 efficiently. Furthermore, limiting interaction ensures a more equal experience across all participants. The images were hosted on a GitHub server and linked to during the task-phase of the survey. Every website had a different combination of influence factors, resulting in the combinations in Table 1. Table 1. the four websites and their factor conditions

Content design absent Content Design present Social Cues absent (A) Control website (C) Website Content design

Social Cues present (B) Website Social Cues (D) Website Social Cues and Content Design

Figure 2. Website A (Control)

Figure 3. Website B (Social Cues)

Figure 4. Website C (Content Design) Figure 5. Website D (Both)

The inclusion of social cue factors was done by equipping the website with icons and links to its Facebook and Twitter pages, as well as with a portrait of a company employee. The portrait included a small subtext introducing the person as an employee to participants, as this has been found

(10)

10 to increase trustworthiness (Steinbrück et al., 2002). The third and last social cue factor that was included was a window showing reviews of the website. The social cue factors are present in website B and D as can be seen in Figure 3 and Figure 5. For the content design section of the website, two e-assurance icons were added, namely the Thuiswinkel waarborg and the ICT waarborg. Texts were also included to inform the participants of the policies and general information about the company. This was done both by using plain text on the website or by using text on an image. The policies included guarantees, promises on delivery time and rules about returns. The content design factors are present in website C and D as can be seen in Figure 4 and Figure 5. Figure 2 displays the control website, which is the website without any content-based factors present, while Figure 5 displays the combination website, which is the website that has both social cue factors and content design factors.

3.2.2 Questionnaire

For this study, four anonymous and online questionnaires each containing a link to one of the four websites were used consisting of 12 questions (Appendix B). 2 of them being demographical questions, 1 question involving background information, and 9 questions on the perceived

trustworthiness of the website. To avoid incomplete questionnaire submissions, every question was made to be compulsory. This type of instrument was chosen because of the time and cost efficiency questionnaires provide. The participants are able to participate anonymously, which reduces social desirability (Joinson, 1999). First, demographic and background information was gathered. The demographic questions involved gender, age, while the background information involved whether the customer has had experience using e-commerce websites before. Next, instructions and a link to one of the four websites were provided. Lastly, after viewing the website, the participants answered the questions regarding the perceived trustworthiness of the website.

The demographical questions were multiple choice questions, while the background information and the questions involving perceived trustworthiness were based on a 5-point Likert scale. To assess the perceived trustworthiness the questionnaire developed by Corritore and colleagues was used. The questionnaire involved measuring other aspects in addition to

trustworthiness, which meant that the sections “reputation”, “predictability” and “perceived ease of use” were removed from the original questionnaire. Using Cronbach alpha, the study concluded that the reliability was above 0,842 for all the factor measures. In addition, the convergent and

discriminant validity of the instrument were shown with the process of confirmatory factor analysis. This assured the unidimensionality of the instrument components (Corritore et al., 2005).

(11)

11

3.3 Procedure

As stated before, the link to the survey was distributed online via Facebook and a forum. To regulate the number of participants among the four questionnaires, a distribution page was created using Google Forms (Appendix A). Here the participants were appointed one of the four websites at random.

First, the participants viewed an introduction page at the top of the Google form. In this section, the researcher was introduced, the survey was explained and the participants answered some demographical questions. After this, the participants performed a short task. The instructions were given at the beginning of this phase which informed the participants about the task they were about to perform and what was expected of them. The purpose of the task was to get the participants engaged in the inspection, as they were about to view a website and share their perception on the

trustworthiness of the website. To avoid the risk of the participants being overly cautious or perhaps developing a disposition to distrust, the task did not involve the participants actively judging the website on its trustworthiness. Instead, the participants were given a scenario wherein their earphones were broken and they would be in the need of buying a new pair. Personal audio equipment like earphones were chosen, because they are popular items that are often seen on the homepage of sites such as Bol.com and they carry a relatively higher risk than cheaper items such as books

(Riegelsberger, Sasse & McCarthy, 2003). After the inspection of the website, the participants

returned to the questionnaire. Lastly, the participants answered questions on perceived trustworthiness and submitted the questionnaire. The method and the study’s procedure are both replicable and modifiable.

3.4 Data analysis

To gain insight into the data, statistical analyses were performed. ANOVA and the t-test were used to analyze the data. For both analyses an alpha level of 0,05 was set. The differences between the four conditions were analyzed using ANOVA. In this case the independent variables were the content-based factors, while the dependent variable was the trustworthiness. When using ANOVA four assumptions are made. Namely, normality, homoscedasticity, the independence of cases and the absence of outliers. The null-hypothesis was that there would be no significant difference between the different websites. The differences between age groups were also analyzed using ANOVA. The same assumptions were made. To measure the difference between male and female participants, a t-test was performed. When using the t-test three assumptions are made. Namely, normality, homoscedasticity and independency. The null-hypothesis was that there is no difference between male and female participants.

(12)

12

4. Results

The results regarding the main thesis question and the sub-questions will be shown in this section. This will be preceded by a data inspection that will further clarify and assert the data.

4.1 Participants

63 people participated in this study. All 63 questionnaires were complete. 42,9% of the participants were male and 57,1% of the participants were female. The number of participants that viewed website A (Figure 2), website B (Figure 3), and website D (Figure 5) was 16. The number of participants that viewed website C (Figure 4) was 15. The number of participants were not equally distributed among the age groups as depicted in Figure 6. The largest age group was the group 18-24, which composed 68.3% of

the total. Age groups <17, 55-64 and 65> had a frequency of 1. Figure 6. Age group frequencies

4.2 Data inspection

A data inspection was performed on every variable. Numeric values were inspected by looking at the means and standard deviations. This was done by using frequency tables and histograms. The extreme values were inspected and a test for normality was performed. For all variables, a minimum of 9 and a maximum score of 45 were possible. The control variable had a mean of 27.56, which was the lowest from the four. The social cue variable had a mean of 29.25, while the variable content design had the highest mean of 34.53. The variable containing both factors called combination had a mean of 32.81. These results are shown in Table 2. A Kolmogorov- Smirnov test of normality was performed and resulted in a significance value greater than 0.05 for all variables, confirming all variables were normally distributed. The Kolmogorov- Smirnov test was used because it’s a non-parametric test and thus, no assumptions need to be satisfied (Moore, Craig & McCabe, 2011). For all variables, the skewness and kurtosis values were divided by their respective standard deviations. This resulted in no value being greater than 3, meaning the variables were normally distributed (Tabachnick & Fidell, 2001). Using boxplots, two outliers were found in the Content Design variable. The outliers had a score of 15 and 20. The scores of these cases were changed to the next lowest value that was not extreme, which was 28. Changing the scores instead of removing the cases was done to avoid data loss (Kroonenberg, 2006).

(13)

13 Table 2. Descriptive statistics of numeric values

Variable M* SD* ZSkewnes* ZKurtosis* Missing Outliers

Trustworthiness 30.98 .93 -2.34 .66 0 0

Control 27.56 2.23 0.9 -.40 0 0

Social Cues 29.25 1.92 -.12 .24 0 0

Content Design 34.53 1.23 .29 -.56 0 2

Both 32.81 1.45 1.07 .77 0 0

* Values in the table are the values after the removal of outliers

4.3 Difference in perceived trustworthiness between male and female participants

The first the sub-question was whether there was a difference between the perceived trustworthiness between male and female participants. The null-hypothesis was that there is no difference between gender. A t-test was used to determine whether this was the case. The assumptions for this test were satisfied. No significant difference was found in the perceived trustworthiness between male and female participants (t(61)=-1.34, p = .186) as can be seen in Table 3. Generally, female participants (M = 32.06, SD = 7.12) were found to have higher perceived trustworthiness than male participants (M = 29.56, SD = 7.61).

Table 3. T-test male and female participants

t df p (2-tailed) Mean difference Std. Error Difference 95% Confidence interval of the difference Lower Upper Trustworthiness Equal Variances assumed -1.337 61 .186 -2.500 1.869 -6.238 1.238

4.4 Difference in perceived trustworthiness between age groups

The second sub-question was whether there was a difference between the perceived

trustworthiness between age groups. The age groups were compared using a one-way ANOVA. The null-hypothesis was that there is no difference between age groups. No significant difference was found in the perceived trustworthiness between age groups (F(6,56) = .60, p = .733).

4.5 Difference between websites

The main question was whether content-based factors influence an e-commerce site’s perceived trustworthiness. The two main factor groups were social cue factors and content design factors. The social cue factors were images and social media cues. The content design factors were

(14)

14 informativeness, and e-assurances. The trustworthiness scores between the control website, social cues website, content design website and combination website were compared using a one-way ANOVA. It was hypothesized that content-based factors would have a positive effect on the perceived trustworthiness. A significant difference was found in the perceived trustworthiness between the four websites (F(3,59) = 3.23, p = .029). There was a significant difference between the control website (M = 27.56, SD = 8.92) and the content design website (M = 34.53, SD = 4.76), the control website and the combination website (M = 32.81, SD = 5.81), and between the social cues website (M = 29.25, SD = 1.92) and the content design website (M = 34.53, SD = 4.76). The mean difference was the most significant between the control website and the content design website, which was 0.008. The second most significant mean difference was between the control website and the combination website. The third most significant mean difference was between the social cues website and the content design website. The mean differences and significance scores are shown in Table 4.

Table 4. Post-Hoc test. Mean comparison between the control website, social cues website, content design website and the combination website. An asterisk indicates there was a significant difference at .05

Compared websites Mean Difference SE p

Control and Social Cues -1.688 2.482 .499

Control and Content Design -6.971* 2.523 .008

Control and Combination -5.250* 2.482 .039

Social Cues and Content Design -5.283* 2.523 .041

Social Cues and Combination -3.563 2.482 .156

Content Design and Combination 1.721 2.523 .498

5. Conclusion

The purpose of this study was to research the relation between content-based factors and perceived trustworthiness. Some research has already been done on the individual factors and their contribution to trustworthiness. There have also been some studies that have researched these claims about the influence of the factors on trustworthiness (Karimov et al., 2011). Using these studies, the content-based factors to be used in this study were defined. In this study, it was hypothesized that these factors would have a significant positive influence on perceived trustworthiness.

The first sub-question was whether social cue factors had an influence on perceived

trustworthiness. It was hypothesized that social cue factors would have a positive effect. However, it was found that this was not the case. Although the mean trustworthiness of the social cues website was higher than that of the control website, the difference was not significant. The second sub-question was whether content design factors had an influence on perceived trustworthiness. It was hypothesized that content design factors would have a positive effect on perceived trustworthiness. In

(15)

15 accordance with the hypothesis, it was found that content design website was perceived to be more trustworthy and that the content design factors had a significant positive effect on perceived

trustworthiness. The difference between age groups and gender groups were also researched to ensure that there would not be any interference by the gender and/or age of the participant on the perceived trustworthiness. The results suggest that there is no difference between the different gender and age groups.

The main question was whether content-based factors influence the perceived trustworthiness of e-commerce websites. It was hypothesized that content-based factors would have a positive effect. However, this hypothesis could not be fully confirmed. Despite that the content-design factors were found to have a significant positive effect on perceived trustworthiness, Social Cue factors did not have the same result.

6. Discussion

6.1 Factor groups

Informativeness and e-assurances, which compose content design factors, were found to have a significant positive effect on perceived trustworthiness. Contrary to previous studies (Aldiri, Hobbs, & Qahwaji, 2008; Cyr, Head, Larios, & Pan, 2009; Wang et al., 2007), this was not the same for social cue factors which was comprised of images and social media cues. Although the mean was higher than that of the control website, the difference was not significant. This might be the result of the grouping of factors. The social cue factors are comprised of three factors, namely links to social media, reviews, and human imagery. In this case the image of the human was a supposed employee. It could be that some of the factors might have contributed to a higher perceived trustworthiness, while others had very little or even a negative effect. This would result in factors counteracting, meaning the overall effect would be negatively affected. The factor that could have caused this is the human image. One study suggested that photographs should be used carefully, because of the wide variety of customer reactions. While some customers might have a positive reaction towards photographs of employees, other customers might be deterred by the same thing (Riegelsberger & Sasse, 2002).

6.2 Limitations

One of the limitations was the lack of diversity in age within the participants. As was discussed in the results section, the majority of the participants were between the age of 18 and 24. The smallest three age groups had a mere frequency of 1. This means that the group of participants might not accurately represent the real-world e-commerce website user population. To be able to achieve this, participants should be recruited in a way that is more random than spreading a

questionnaire via social media. This way, the group of participants will not reflect the userbase of the social media that was used to spread the questionnaire, but rather the general population. In addition, a pre-research analysis could be performed to ascertain the intended target audience of a particular type of website. By doing this, the process of recruiting participants could be tailored towards the target

(16)

16 population of the website.

Using a questionnaire certainly had its benefits in this study. However, the use of a questionnaire had its limitations as well. Using a questionnaire meant that most of the participants took part in the questionnaire without any supervision. Although the absence of supervision provides anonymity to the participant, which in turn reduced socially desirable answers (Joinson, 1999), this also reduces ability to optimally instruct the participants. The instructions were provided in the form of text in the questionnaire itself (Appendix B), but there was nothing to prevent participants from skimming the text and continuing the questionnaire.

The questionnaire was done on any computer the participants had access to at the time. This meant that not every screen was the same size. To view the image of the provided website in full resolution, the image needed to be clicked on. This information was clearly provided in the

instructions, but without a lab environment or at least proper supervision, a chance remains that the participants did not view the website in full resolution. This would mean that not all participants would have gotten the experience intended by the study.

The results showed that the control website, which was supposed to be the least trustworthy, still had a relatively high mean score of 27.56 with the lowest possible score being 9 and the highest possible score being 45. Although the results do indicate a significant difference when some trust factors are introduced, this raises the question whether these trust factors would have had the same effect if the control website had been less trustworthy to begin with.

6.3 Implications

This study provided insight into perceived website trustworthiness and factors that contribute to a trustworthy website. This study confirmed previous studies on the influence of factors on

trustworthiness and acknowledged the sub-group of trust factors called content-based factors. It was confirmed that content-design factors had a positive influence on perceived website trustworthiness and that this is not the case for social cue factors. Based on study’s findings, it is suggested that website owners try to inform their website visitors as good as they can in the form of product

descriptions, policy descriptions and company information. It is also recommended to inform website visitors of the e-assurances that the website or a third party provides in the form of a seal. Website owners should be aware of the possible risks images may have on perceived trustworthiness as it is not clear yet as to whether images provide a positive or negative influence, if any influence at all.

6.4 Future research

This study forms a basis for further research. The limitation mentioned before can be further improved upon to get more trustworthy results. For the participants to be able to represent the general population, it is recommended to use a more random approach when recruiting participants or to investigate the intended target audience beforehand and create a similar participant pool. No

(17)

17 interesting step to take would be to research the difference between male and female groups within each website group. This was attempted in this study, but because of the limited sample size, it did not yield any significant results.

To improve the testing procedure, it is recommended to instruct the participants clearly and to make sure that the website experience is the same for all participants. To do this, a lab environment could be created where only one type of computer is used to make sure all viewing experiences are identical. This also makes it possible for the participants to be instructed clearly beforehand and still partake in the questionnaire anonymously. Another way to improve the testing procedure is to minimize the amount of instructions the participants need to complete the survey. This way, the chance of missing or not noticing instructions is minimized.

Grouping the influence factors aided in improving test and website creation efficiency. However, it was suspected that because of this, some effects on the perceived trustworthiness were obstructed. Although the results suggest that Social Cue factors do not improve trustworthiness, this might be because of the counteracting factors within the Social Cue group as mentioned before. Further research could explore these factors in more detail by creating combinations based on individual factors and not factor groups.

(18)

18

7. References

Aldiri, K., D. Hobbs, and R. Qahwaji, “The Human Face of E-Business: Engendering Consume Initial Trust Through the Use of Images of Sales Personnel on E-Commerce Web Sites,” International Journal of EBusiness Research, Vol. 4, No. 4: 58-78, 2008.

Baardewijk, J. V. (2011). Sample Size and Power Analysis for a 2 x 2 ANOVA design.

Bahmanziari, T., Odom, M. D., & Ugrin, J. C. (2009). An experimental evaluation of the effects of internal and external e-assurance on initial trust formation in B2C e-commerce. International

Journal of Accounting Information Systems, 10(3), 152-170.

Bardoel, J., Schildkamp, B., & Weltevreden, J. (2014). Vestigingsvoorkeuren van webwinkels: een onderzoek naar de behoefte van webwinkels aan winkelruimte in Nederland.

Corritore, C. L., Marble, R. P., Wiedenbeck, S., Kracher, B., & Chandran, A. (2005). Measuring online trust of websites: Credibility, perceived ease of use, and risk. AMCIS 2005

Proceedings, 370.

Cyr, D., Hassanein, K., Head, M., & Ivanov, A. (2007). The role of social presence in establishing loyalty in e-service environments. Interacting with computers, 19(1), 43-56.

Cyr, D., Head, M., Larios, H., & Pan, B. (2009). Exploring human images in website design: a multi method approach. MIS quarterly, 539-566.

Everard, A., & Galletta, D. F. (2005). How presentation flaws affect perceived site quality, trust, and intention to purchase from an online store. Journal of Management Information

Systems, 22(3), 56-95.

Fisher, R., & Zoe Chu, S. (2009). Initial online trust formation: the role of company location and web assurance. Managerial Auditing Journal, 24(6), 542-563.

Hu, X., Wu, G., Wu, Y., & Zhang, H. (2010). The effects of Web assurance seals on consumers' initial trust in an online vendor: A functional perspective. Decision support systems, 48(2), 407-418.

Joinson, A. (1999). Social desirability, anonymity, and Internet-based questionnaires. Behavior

Research Methods, Instruments, & Computers, 31(3), 433-438.

Karimov, F. P., Brengman, M., & Van Hove, L. (2011). The effect of website design dimensions on initial trust: a synthesis of the empirical literature. Journal of Electronic Commerce

Research, 12(4), 272.

Karimov, F. P., & Brengman, M. (2013). Adoption of Social Media by Online Retailers: Assessment of Current Practices. Modern Entrepreneurship and E-Business Innovations, 41.

Kim, D. J., Steinfield, C., & Lai, Y. J. (2008). Revisiting the role of web assurance seals in business to-consumer electronic commerce. Decision Support Systems, 44(4), 1000-1015.

Kroonenberg, P. M. (2006). Data inspection for students. Child & Family Studies and Data Theory Leiden University, 1-20.

(19)

19

Li, Y. M., & Yeh, Y. S. (2010). Increasing trust in mobile commerce through design aesthetics. Computers in Human Behavior, 26(4), 673-684.

Lin, Hsiu-Fen. "The impact of website quality dimensions on customer satisfaction in the B2C e commerce context." Total Quality Management and Business Excellence 18.4 (2007): 363 378.

Lindgaard, G., Fernandes, G., Dudek, C., & Brown, J. (2006). Attention web designers: You have 50 milliseconds to make a good first impression!. Behaviour & information technology, 25(2), 115-126.

Lowry, P. B., Wilson, D. W., & Haig, W. L. (2014). A picture is worth a thousand words: Source credibility theory applied to logo and website design for heightened credibility and consumer trust. International Journal of Human-Computer Interaction, 30(1), 63-93.

Moore, D. S., Craig, B. A., & McCabe, G. P. (2011). Introduction to the practice of statistics (7th ed.). W. H. Freeman & Amp.

Mudambi, S. M., & Schuff, D. (2010). What makes a helpful review? A study of customer reviews on Amazon. com.

Oliveira, T., Alhinho, M., Rita, P., & Dhillon, G. (2017). Modelling and testing consumer trust dimensions in e-commerce. Computers in Human Behavior, 71, 153-164.

Park, C. H., & Kim, Y. G. (2003). Identifying key factors affecting consumer purchase behavior in an online shopping context. International Journal of Retail & Distribution Management, 31(1),

16 29.

Ratnasingham, P. (1998). Trust in web-based electronic commerce security. Information Management

& computer security, 6(4), 162-166.

Riegelsberger, J., & Sasse, M. A. (2002, April). Face it-photos don't make a web site trustworthy. In CHI'02 Extended Abstracts on Human Factors in Computing Systems (pp. 742 743). ACM. Riegelsberger, J., Sasse, M. A., & McCarthy, J. D. (2003, April). Shiny happy people building

trust?: photos on e-commerce websites and consumer trust. In Proceedings of the SIGCHI conference on Human factors in computing systems (pp. 121-128). ACM.

Statistics Netherland (CBS) (2015, October 28). Bedrijven met website, 2012-2015. Retrieved April 24, 2017, from https://www.cbs.nl/nl-nl/maatwerk/2015/44/bedrijven-met-website 2012- 2015

Steinbrück, U., Schaumburg, H., Duda, S., & Krüger, T. (2002, April). A picture says more than a thousand words: photographs as trust builders in e-commerce websites. In CHI'02 extended abstracts on Human factors in computing systems (pp. 748-749). ACM.

Tabachnick, B. G., Fidell, L. S., & Osterlind, S. J. (2001). Using multivariate statistics. Wang, L. C., Baker, J., Wagner, J. A., & Wakefield, K. (2007). Can a retail web site be

(20)

20

8. Appendix

8.1 Appendix A

Doorverwijzing

Klik alstublieft op de bovenste link. De onderste drie kunt u negeren. De enquête moet op de computer worden ingevuld.

1. https://goo.gl/forms/hjJ0h5vOImJYB0963 2. https://goo.gl/forms/ooI3EsoBJZB2W75j1 3. https://goo.gl/forms/p9KH1ZnkZrELEOYv2 4. https://goo.gl/forms/V0Kq9oApRIPbEkdk2

8.2 Appendix B

Onderzoek naar webshops

Mijn naam is Kubilay Keser en ik studeer informatiekunde aan de Universiteit

van Amsterdam. Voor mijn scriptie over website ervaringen voer ik een

onderzoek uit door middel van een enquête. Tijdens het onderzoek zult u kort

een foto van een website bekijken en hier vervolgens vragen over

beantwoorden. Er zijn geen juiste of onjuiste antwoorden mogelijk; het

onderzoek gaat uitsluitend over uw ervaring. Het onderzoek duurt vijf tot tien

minuten. Deelnemen is vrijwillig en anoniem.

Deze enquête kan alleen worden ingevuld op de computer.

Algemeen

1. Uw ingevulde gegevens en antwoorden zullen worden gebruikt voor het onderzoek en de

statistische analyse. (U kunt alleen door e=met het onderzoek als u akkoord geeft).

Ik geef toestemming,

2. Bent u een vrouw of een man?

Vrouw

Man

3. In welke leeftijdscategorie valt u?

17 jaar en jonger

18-24 jaar oud

25-34 jaar oud

35-44 jaar oud

45-54 jaar oud

(21)

21

55-64 jaar oud

64 jaar en ouder

4. Bestelt u weleens spullen online?

Nooit 1 2 3 4 5 Bijna altijd

De Opdracht

Het scenario: uw oordopjes zijn kapot en u wilt nieuwe bestellen.

De opdracht: u krijgt zo meteen een afbeelding van een website te zien. Kijk gerust op uw

eigen tempo naar de website. Als u denkt de website goed te hebben bekeken kunt u weer

terug naar de vragenlijst. U hoeft de website niet af te sluiten.

Hieronder vindt u de link naar de website.

HET KAN ZIJN DAT DE AFBEELDING NIET SCHERP IS. U MOET DAN EEN KEER

KLIKKEN OP DE AFBEELDING.

<Link naar website>

Vragenlijst

Hieronder vindt u een aantal stellingen met betrekking tot de webstie die u zojuist heeft

gezien. U kun aangeven in hoeverre u het eens bent met elke stelling. Nogmaals: er zijn geen

juiste of onjuiste antwoorden mogelijk; het onderzoek gaat uitsluitend over uw ervaring.

Deelnemen is vrijwillig en anoniem.

5. Deze website toot waarheidsgetrouwe informatie.

Helemaal oneens 1 2 3 4 5 Helemaal eens

6. De informatie op de website is geloofwaardig.

Helemaal oneens 1 2 3 4 5 Helemaal eens

7. De inhoud van de website straalt bekwaamheid uit.

Helemaal oneens 1 2 3 4 5 Helemaal eens

8. De inhoud van de website straat expertise uit.

Helemaal oneens 1 2 3 4 5 Helemaal eens

9. Ik ben een gok aan het wagen wanneer ik gebruik maak van deze website.

Helemaal oneens 1 2 3 4 5 Helemaal eens

10. Ik denk dat het onveilig is om gebruik te maken van deze site.

Helemaal oneens 1 2 3 4 5 Helemaal eens

(22)

22

Helemaal oneens 1 2 3 4 5 Helemaal eens

12. Ik denk dat deze website betrouwbaar is.

Helemaal oneens 1 2 3 4 5 Helemaal eens

13. Ik vertrouw deze website.

Helemaal oneens 1 2 3 4 5 Helemaal eens

8.3 Appendix C

Website A: Control

(23)

23 Website B: Social Cues

(24)

24 Website C: Content Design

(25)

25 Website D: Social Cues and Content Design

Content-based influence on website trustworthiness