Keywords

1 Introduction

Credibility has been found to be a very important attribute not only in the field of human-to-human interaction but also in Human-Computer Interaction (HCI) as well. In online business, for example, it has been attributed to the success of several e-commerce websites [24]. Research [5] has shown that the perceived credibility of a given website has the power to change user attitudes towards the website and their behaviors. For example, with respect to user attitude, a website perceived as highly credible can make a user: (1) think positively about the site owner; (2) feel at ease interacting with the site; and (3) embrace the site owner’s viewpoint [5]. Moreover, with respect to user behavior, such a website can encourage the user to: (1) register with the website by willingly providing required personal information; (2) complete a financial transaction on the website; and (3) use the website more often [5].

Research has shown that visual aesthetics is one of the strongest determinants of the perceived credibility of websites, especially during users’ first contact with or visit to a website [13, 24]. However, hitherto, there is a lack of knowledge on which of the two theoretical dimensions of visual aesthetics (classical aesthetics and expressive aesthetics) [12] is the stronger determinant of the perceived credibility of a website. Answering this research question will help HCI designers creating certain types of website—hedonic or utilitarian—to understand which of the two dimensions of visual aesthetics resonate more with users when evaluating the credibility of a website. Hedonic websites are websites aimed at providing pleasure and enjoyment to their users (e.g., game-based websites [26]), while utilitarian websites are websites focused at providing instrumental or productive value to their users (e.g., search engines) [10, 30]. We hypothesize that the question whether classical aesthetics or expressive aesthetics is stronger in determining perceived credibility may depend on the type or function of a website. Hence, uncovering which of the two distinct dimensions of visual aesthetics—which have virtually opposing design objectives [12]—has a stronger impact on credibility will enable website designers at a finer-grain level to know which of the aesthetics dimensions to pay more attention to when designing hedonic or utilitarian websites.

We conducted an empirical study among participants from five different continents to investigate: (1) whether classical aesthetics or expressive aesthetics has a stronger influence on the perceived credibility of a website; and (2) how age and gender of the user moderate the interrelationships among the three design attributes. As a case study, we implemented the homepage of a utilitarian mobile website with four user interface (UI) versions, characterized by different levels of aesthetics, and evaluated them with 526 subjects from five continents. The data collected was analyzed using Structural Equation Modeling (SEM) [8]. The results of our SEM analysis show that, regardless of the levels of aesthetics of the mobile website, classical aesthetics influences perceived credibility more strongly than expressive aesthetics does. Moreover, our multigroup analysis shows that the finding generalizes across age, gender and different levels of UI aesthetics. Our finding underscores the need for designers of utilitarian websites who want to increase the perceived credibility of their websites to focus more on classical aesthetics (simplicity, clarity, orderliness, cleanness, etc. [12]) than expressive aesthetics (complexity, novelty, richness, sophistication, etc. [16]).

The paper is organized as follows: Sect. 2 defines the main concepts used in this work; Sect. 3 presents related work; Sect. 4 presents the research method and hypotheses of the study; Sect. 5 presents the results; and Sect. 6 discusses the results, their implications and limitations. Finally, Sect. 7 concludes the paper.

2 Background

In this section, we provide an overview of credibility, aesthetics, classical aesthetics and expressive aesthetics. For brevity, we will, in some cases, as we have done here, omit the qualifier “perceived” when referring to the design constructs.

2.1 Credibility

Credibility refers to the believability of a system, e.g., a website. It is viewed as a user-perceived quality and not a property that is resident in the system under evaluation. In HCI, Fogg and Tseng [7] theorize that perceived credibility is composed of two distinct dimensions: perceived expertise and perceived trustworthiness. Perceived expertise is defined as the degree to which a user perceives the designer of a given website to be competent, knowledgeable and experienced. Thus, it captures the perceived knowledge and skill of the source (i.e., the designer of the website). Moreover, perceived trustworthiness is defined as the degree to which a user perceives a website to be well-intentioned, truthful and free of bias [6]. It mainly captures the perceived goodness or morality of the source (i.e., the owner of the website). Fogg [4] asserts that, in the evaluation of the credibility of a website, both dimensions are perceived and judged simultaneously. As such, to judge a website as credible, the user must perceive it to be high in both trustworthiness and expertise [6].

2.2 Aesthetics

The concept of “aesthetics” is closely related to the concept of “beauty,” which plays an important role in many aspects of human existence, including nature, art, culture, architecture, etc. Historically, aesthetics is defined as the branch of philosophy which deals with “the nature and appreciation of art, beauty and good taste” [15]. In the eighteen century, it was redefined by Baumgarten, a German philosopher, as the sensory pleasure or delight derived from the visual perception of an object. In other words, it refers to the gratification of the senses as a result of the visual experience of an object [11]. There are three schools of thought, which view aesthetics in three different ways. The first school of thought views aesthetics as an objective concept (a property resident in the object). The second school of thought views it as a subjective concept (resident in the eye of the beholder). Finally, the last school of thought (the interactionist position) adopts the middle ground: aesthetics being objective as well as subjective [12, 16]. While we may not disagree that aesthetics could be both objective (depending on the object or stimulus) and subjective (depending on the characteristics of the observer) as most current theories (such as the interactionist theory) hold, in our empirical study, we adopt the view of the second school of thought. In other words, our measurement of aesthetics is based on the subjective perception of the observer, regardless of the aesthetic property of the object from the designer’s perspective.

In human-computer interaction design, aesthetics (often known as “visual aesthetics”) refers to the visual appeal or pleasing appearance of a HCI artifact [29]. Research has shown that users are able to make stable aesthetic judgment within the first 50 ms of encountering the UI of a website [14]. Moreover, Lavie and Tractinsky [12], in a factorial analysis study, found that visual aesthetics is composed of two main dimensions: classical aesthetics and expressive aesthetics.

Classical Aesthetics.

Classical aesthetics reflects the traditional notion of aesthetics: orderliness, symmetry, proportion, clarity, harmony, etc. Its ultimate goal is to increase the understanding of UIs and reduce ambiguity in users’ interaction with them. Besides reflecting the objective notion of aesthetics, classical aesthetics has been found to be highly correlated with perceived usability [12, 22]. According to Lavie and Tractinsky [12], “the classical aesthetic dimension may serve as a linkage between usability and aesthetics, being both an aesthetic concept and a usability principle” (p. 290). Moreover, given that classical aesthetics induces pleasure, it is theorized “to have a calming effect on the senses” (p. 148) [23].

Expressive Aesthetics.

Expressive aesthetics reflects the expressive power of the designers, which borders on creativity, originality and novelty. Its expression in UI design is often described by words such as “creative,” “original,” “sophisticated,” “fascinating,” “uses special effect,” etc. The ultimate goal of expressive aesthetics is to increase arousal and user involvement [12]. Hassenzahl and Monk [9] have argued that expressive aesthetics is strongly associated with hedonic quality.

3 Related Work

Given the importance of credibility in the internet space, a great body of work has been dedicated to uncovering its determinants. Fogg et al. [3] conducted a study among 2,684 participants to investigate the design attributes that influence the judgment of web credibility the most. They found that design look (related to visual aesthetics) is the strongest determinant of the perceived credibility of a website, followed by information design/structure. Robins and Holmes [24] investigated users’ evaluation of the credibility of websites with high and low levels of aesthetic treatment. They found that the high-aesthetic websites were judged higher in credibility than the low-aesthetic websites. Alsudani and Casey [1] also found that the perceived aesthetics of a website influences users’ judgment of its credibility. In the mobile domain, Oyibo et al. [18,19,20,21] found that, irrespective of culture or gender, the perceived aesthetics of mobile websites influences their perceived credibility more strongly than perceived usability does. In general, most of the studies that have been conducted so far in the desktop and mobile domains have only examined visual aesthetics at a high level, i.e., as a composite construct predicting perceived credibility. However, in recent years, Lavie and Tractinsky [12] have found that visual aesthetics is composed of two main dimensions: classical aesthetics and expressive aesthetics. This makes it pertinent for researchers to investigate which of these two dimensions has a stronger influence on perceived credibility, especially with respect to the different types of websites. Answering this research question will help designers of HCI systems know which of the aesthetics dimensions they should focus more on when designing hedonic and utilitarian websites. Our study aims to bridge this gap in the extant literature using a utilitarian website as a case study.

4 Method

This section presents our research objective and design, research hypotheses, measurement instruments and the demographics of participants.

4.1 Research Objective and Design

In this study on the relationship between aesthetics and credibility at a finer-grain level, we set out to answer the following research questions:

  1. 1.

    Which of the two dimensions of visual aesthetics (classical or expressive) has a stronger influence on perceived credibility in mobile website design?

  2. 2.

    Are the mutual relationships between classical aesthetics, expressive aesthetics and perceived credibility moderated by age and gender?

  3. 3.

    Are the relationships replicable across different levels of UI aesthetics?

To answer the above research questions, we came up with a hypothetical mobile websiteFootnote 1 (named “G-Ranch”) in the tourism domain. Figure 1 shows all four versions of the UI at different levels of aesthetics, which were systematically arrived at using a “UI transformation framework,” we called Artifact-Action framework [17].

Fig. 1.
figure 1

Systematically designed mobile websites using a UI transformation framework

Full size image

The artifacts are the UIs themselves (in the quadrants) and the actions are the aesthetic manipulations (on the axes) carried out on them to arrive at a new UI different from the previous one. For example, to transform version WA to WB, we apply the action “make gray and add icons.” Based on all four UIs, we designed a questionnaire and asked participants from five different continents in an online survey to respond to questions on classical aesthetics, expressive aesthetics and perceived credibility. We also requested participants to provide comments on the visual design. The study was based on a within-subject design in which each participant had to evaluate all four UIs and respond to the respective questions based on them. With respect to color scheme, we have three levels: (1) minimalist (WA); (2) moderate (WC and WD); and (3) multicolor (WA). With respect to content presentation, we have two layouts: (1) list layout (WA, WB and WC); and (2) grid layout (WD). We combined a color scheme with a layout to arrive at each of the four configurations (versions) shown in Fig. 1. However, to limit the experiment duration and to prevent participant fatigue, we could not investigate all of the possible UI configurations in our survey. Each of the four UI versions represents different levels of aesthetic treatment of the mobile website.

4.2 Research Hypotheses

Based on prior findings and theories in the literature [12, 13, 21, 23, 24, 28], we formulated three hypotheses as represented in Fig. 2. The hypotheses are stated as follows:

Fig. 2.
figure 2

Hypothesized research model (dashed arrows indicate explored moderated paths)

Full size image

  • H1: The classical aesthetics of a website user interface will significantly influence its perceived credibility.

  • H2: The expressive aesthetics of a website user interface will significantly influence its perceived credibility.

  • H3: The classical aesthetics of a website user interface will significantly influence its expressive aesthetics.

The first and second hypotheses (H1 and H2) are based on a prior finding in the desktop [6, 24] and mobile [20, 21] domains, in which visual aesthetics—as a composite construct—significantly influences the perceived credibility of the websites evaluated by users. Thus, we hypothesize that, as separate constructs, classical aesthetics (H1) and expressive aesthetics (H2) will significantly influence perceived credibility. The third hypothesis (H3) is based on the findings from previous studies by Lavie and Tractinsky [12] in the desktop domain and Oyibo and Vassileva [22] in the mobile domain. Both studies found that there is a significant relationship between both dimensions of visual aesthetics. Thus, in our study, we hypothesize that classical aesthetics will significantly influence expressive aesthetics (H3).

However, due to the paucity of research in this topic area, we adopted an exploratory approach in investigating: (1) which of classical aesthetics and expressive aesthetics has a stronger influence on perceived credibility; and (2) how gender and age moderate the interrelationships among all three design constructs. As a working rule, we pre-stated that, for any of our hypotheses to be fully validated, it has to be replicated across, at least, two of the UI designs characterized by different levels of aesthetics. Otherwise, if a relationship is validated with respect to one UI only, we conclude it is partially validated and requires further research.

4.3 Measurement Instruments

To measure the three HCI design constructs in our research model, we used validated instruments from prior literature. They are shown in Table 1. Classical aesthetics and expressive aesthetics are based on the 3-item scales developed by [12] as adapted by [26]. The measurement of perceived credibility is based on a single-item scale, which, previous research [2, 27] has shown, is equally reliable as a multi-item scale. All of the scales have been validated in previous studies (e.g., [27]); they ranged from “Strongly Disagree (1)” to “Strongly Agree (7).”

Table 1. Scales measuring all three mobile design constructs
Full size table

In the online questionnaire, prior to presenting the snapshot of each UI version to each study participant (in the order—WC, WA, WB and WD), the following opening statement (1) was presented, after which the question (2), which preceded all of the items shown in Table 1, was asked. Specifically, all six items in the classical aesthetics and expressive aesthetics scales were combined and randomized in a block. Finally, the questions (3 and 4) on perceived credibility and UI appeal, respectively, were asked.

  1. 1.

    Assume you were looking for a website on travels and tourism on your mobile phone, and you happened to open this webpage by clicking on one of the links returned by a search engine.

  2. 2.

    Please rate the website [label of website] on the following criteria based on your first impression.

  3. 3.

    Based on your first impression of the mobile webpage [label of website], please rate its credibility level.

  4. 4.

    Please kindly comment on the things that appeal to or interest, or annoy, you regarding the above mobile web page [label of website], if any.

4.4 Participants

The survey was submitted to and approved by the first author’s university’s Research Ethics Board. Thereafter, participants were recruited on social media, via email and on the first author’s university website to participate in the study. Over 540 participants took part in the study. After cleaning the data, a total of 526 were retained for the data analysis. Participants are spread across five continents (Africa, Asia, Europe, North America and South America). Table 2 shows the demographics of participants. Overall 53.6% of them were males, while 44.5% were females. Age-wise, 50.2% of the participants were between 18 and 24 years old, while 48.5% were above 24 years old.

Table 2. Participants’ demographics
Full size table

4.5 Data Analysis

Upon the completion of data gathering, we proceeded to data analysis. We carried out SEM analysis [8]. Specifically, we used the Partial Least Square Path Modeling (PLSPM) package (“plspm” [25]) in the R programming language to build and analyze our path models. All three design constructs were specified as reflective in our models. Two multigroup analyses were carried out. They were based on gender (male vs. female) and age (18–24 vs. above 24). The bootstrap test for statistical significance of the path coefficients (the strength of the relationship between two constructs) was based on 5000 samples as recommended by Hair et al. [8].

5 Result

In this section, we present the results of our SEM analysis, including the assessment of the measurement models, analysis of the structural models and the multigroup analyses.

5.1 Evaluation of Measurement Model

Prior to analyzing the structural models, we evaluated the respective measurement models to ensure the required preconditions are meant, both for the global and subgroup models. We report the respective criteria in the measurement models that were assessed.

Indicator Reliability.

All of the indicators in the respective measurement models have an outer loading greater than 0.7. Thus, the indicator reliability criterion is met, as the communality values for each indicator in the respective models are greater than 0.5 [8].

Internal Consistency.

We used the composite reliability criterion, DG.rho (ρ), to assess the internal consistency reliability of each multi-item construct in the respective models. The respective values for both aesthetics constructs are greater than 0.7 [8].

Convergent Validity.

The Average Variance Extracted (AVE) was used to assess convergent validity of each multi-item construct in the respective models. The AVE for each construct in the respective models is greater than the threshold value of 0.5 [8].

Discriminant Validity.

This criterion was assessed using the crossloading of each construct on the other. No indicator of a given construct loaded higher on any other construct than the one it was meant to measure [8].

5.2 Analysis of Structural Model

Figure 3 shows the global models of the four UIs for the general population. The global models characterize the following parameters: (1) path coefficients (β), which indicates the strength of the direct effect exerted by an exogenous construct (e.g., classical aesthetics) on an endogenous construct (e.g., perceived credibility); (2) the coefficients of determination (R2), which indicates the amount of variance of an endogenous construct explained by its exogenous constructs; and (3) the goodness of fit (GOF), which indicates the predictive power of the model, i.e., how well the model fits its data [25].

Fig. 3.
figure 3

Data-driven global models for all of the mobile website user interfaces

Full size image

Overall, the path models for all four webpages have a moderate to high GOF. A moderate GOF ranges between 60% and 70%, while a high GOF is 70% and above. Thus, the WA model has a high GOF, while the rest three models (WB, WC and WD) have a moderate GOF, which are nearly high and considered acceptable [25]. Furthermore, classical aesthetics explains between 63% and 69% of the variance of expressive aesthetics. Based on the 60% benchmark [25], R2 values greater than 60% indicates high predictive accuracy. Thus, expressive aesthetics is well explained or predicted by classical aesthetics for all four webpages given that all of the R2 values are greater than 60%. Moreover, perceived credibility is highly well-explained or predicted by classical aesthetics and expressive aesthetics in the WA model (69%) and WB model (64%), while it is moderately well-explained or predicted in the WC model (50%) and WD model (59%).

Finally, all of the interrelationships among the three design constructs are statistically significant at p < 0.0001, with classical aesthetics having the strongest effect on expressive aesthetics (β = 0.79 to 0.83, p < 0.0001), followed by classical aesthetics on perceived credibility (β = 0.50 to 0.64, p < 0.0001), and expressive aesthetics on perceived credibility (β = 0.19 to 0.25, p < 0.0001). Overall, classical aesthetics has a stronger effect on perceived credibility than expressive aesthetics does. For example, in the WB and WC models, the direct effects of classical aesthetics on perceived credibility are (β = 0.64 and 0.50, respectively) at p < 0.0001, while the corresponding direct effects of expressive aesthetics on perceived credibility in both models are (β = 0.19 and 0.24, respectively) at p < 0.0001.

5.3 Subgroup Models Based on Multigroup Analysis

Figure 4 shows two pairs of submodels based on the gender-based and age-based multigroup analyses. Overall, the significant interrelationships among all three design constructs in each pair of submodels replicate those in the global model. However, the multigroup analyses show that the two subgroups in each pair of submodels significantly differ at p < 0.05.

Fig. 4.
figure 4

(a) Male and female subgroup models (highlighted indicates where both groups differ). (b) Younger and older subgroup models (highlighted indicates where both groups differ)

Full size image

The gender-based multigroup analysis (see Fig. 4a) shows that males and females significantly differ at p < 0.05 with respect to the influence of classical aesthetics on expressive aesthetics and the influence of classical aesthetics on perceived credibility. In the WC model, the direct effect of classical aesthetics on expressive aesthetics is stronger for males (β = 0.83, p < 0.0001) than for females (β = 0.78, p < 0.0001). Similarly, in the WA model, the direct effect of classical aesthetics on perceived credibility is stronger for males (β = 0.66, p < 0.0001) than for females (β = 0.52, p < 0.0001).

Moreover, the age-based multigroup analysis (see Fig. 4b) shows that younger and older people significantly differ at p < 0.05 with respect to the influence of classical aesthetics on expressive aesthetics. In the WB and WC submodels, the direct effect between both constructs is stronger for younger people (β = 0.82 and 0.85, p < 0.0001, respectively) than for older people (β = 0.76 and 0.77, p < 0.0001, respectively).

6 Discussion

The perceived credibility of the UI design of an e-commerce website is critical to its success, so is the perceived aesthetics—a strong determinant of perceived credibility. In this paper, we have presented a path model based on four UI versions of a hypothetical mobile website to uncover which of the two dimensions of visual aesthetics (classical and expressive) has a stronger influence on the perceived credibility of a utilitarian website. We have also presented the results of multigroup analyses to uncover how age and gender moderate the interrelationships among all three design constructs. In the subsequent subsections, we discuss the validation of our hypotheses and the summary of our findings vis-à-vis our research questions.

6.1 Validation of Hypotheses

Our SEM analysis shows that, irrespective of age and gender, all of the three hypothesized relationships in our research model are significant at different levels of UI aesthetics, with classical aesthetics having a stronger influence on perceived credibility than expressive aesthetics does. Table 3 shows the summary of our main findings in respect of our hypotheses and exploratory analysis.

Table 3. Summary of supported hypotheses
Full size table

Our SEM analysis shows that our first hypothesis (H1: the classical aesthetics of a website user interface will significantly influence its perceived credibility) is supported and replicated across all four UI designs. This suggests that a classically aesthetic utilitarian website is more likely to be perceived credible than a less classically aesthetic website. This is evident in some of the participants’ comments. For example, participant 283 comments on webpage WC thus: “credible[,] information are obtain[ed] eas[i]ly.” For this participant, webpage WC is credible because information can be easily accessed. This suggests that this participant must have based his/her judgment of the credibility of WC on the perceived classical aesthetics features, such as clarity and orderliness of the webpage, which generally facilitate access to sought information on the web.

Secondly, our SEM analysis shows that our second hypothesis (H2: the expressive aesthetics of a website user interface will significantly influence the perceived credibility) is supported and replicated across all four UI designs. This suggests that an expressively aesthetic website is more likely to be perceived credible than a less expressively aesthetic website. This is evident in some of the comments from participants. For example, participant 236 comments on webpage WD thus: “looks clearer and more professional with the grid layout and better colour coordination.” The participant seems to have made his/her credibility-related judgment on the site’s professionalism partly based on the perceived “colour coordination” of webpage WD, which hints at expressive aesthetics. On the flip side, unfavorable expressive aesthetics perception can harm credibility perception as well as evident in participant 119’s comment on webpage WA: “too many different colours to seem credible.” This suggests that the multicolor scheme in webpage WA may have taken away from its perceived credibility. This clearly demonstrates how an unfavorable perception of expressive aesthetics can negatively influence the perception of credibility.

Moreover, by replicating the validation of H1 and H2 across all four different UI versions, we have particularly demonstrated that, irrespective of the level of UI aesthetics, characterized by layout (a classical aesthetics feature) and color (an expressive aesthetics feature), the perceptions of both dimensions of visual aesthetics significantly influence the perception of web credibility. However, our results reveal that the influence of classical aesthetics on perceived credibility is stronger than the influence of expressive aesthetics on perceived credibility. One possible reason why classical aesthetics influences perceived credibility more than expressive aesthetics does is that the website we investigated is a utilitarian system aimed at accomplishing a productive task: search and retrieval of information. According to [30], a utilitarian system addresses a productive task. As such, users’ motivation in using a utilitarian system is driven by the expectation of a reward or benefit that is external to the user interaction with the system. In our case, the hypothetical G-Ranch website helps the user to find a desired place, hotel, event, etc., for tourism purpose. Thus, being able to get the right information as quickly and easily as possible becomes the user’s reward or benefit. Our utilitarian website contrasts hedonic systems (e.g., games), which sole aim is to provide users with entertainment by satisfying their intrinsic need. Consequently, in our model, we see classical aesthetics (operationalized by “clean,” “clear,” “orderly,” etc. [12])—which is closely linked to perceived usability (operationalized by “easy to navigate,” “easy to use,” etc.) [22]—having a stronger influence on perceived credibility than expressive aesthetics (operationalized by arousal-based items such as “fascinating,” “sophisticated,” “creative,” etc.) does. This finding is consistent with Lavie and Tractinsky’s [12] submission that “the classical aesthetic dimension may serve as a linkage between usability and aesthetics, being both an aesthetic concept and a usability principle” (p. 290), as confirmed by Oyibo and Vassileva [22]. Based on this finding, summarized in our fourth hypothesis (H4: the influence of classical aesthetics on perceived credibility is stronger than the influence of expressive aesthetics on perceived credibility), we make the following recommendation to website designers:

When designing websites with utilitarian value or benefit, such as a tourism website aimed at finding tourist attractions via search, designers should focus on enhancing classical aesthetics features (cleanness, orderliness, clarity and simplicity) more than expressive aesthetics features (richness, novelty and sophistication) to increase the chances of their sites being perceived credible and eventually used by first-time users.

Finally, given that both classical aesthetics and expressive aesthetics influence perceived credibility, this suggests, by extension, that both dimensions of visual aesthetics have a significant relationship between them. This is validated by our third hypothesis (H3: the classical aesthetics of a website user interface will significantly influence the expressive aesthetics), which holds true irrespective of the level of aesthetic treatment of the UI, age and gender of the user. This suggests that a classically aesthetic website is more likely to be perceived as expressively aesthetic by users than a less classically aesthetic website. The strong relationship between classical aesthetics and expressive aesthetics can be likened to the aesthetic-usability halo effect, where the perceived aesthetics of a UI spills to influencing its perceived usability [12, 21]. Specifically, in our case, the relationship between classical aesthetics and expressive aesthetics could be regarded as the “classical-expressive aesthetic” halo effect, in which the perception of classical aesthetics of the UI spills to influencing the expressive aesthetics dimension.

6.2 Gender and Age Differences in the Evaluation of Website UIs

Our multigroup analysis shows that there are gender and age differences in the interrelationship between classical aesthetics and the other two design constructs in the path model. With respect to gender, the result of the multigroup analysis shows that the influence of classical aesthetics on expressive aesthetics (for version WC) is stronger for males than for females (see H5 in Table 3). This finding suggests that males are more likely to judge a website as expressively aesthetic based on the perception of classical aesthetics. However, given the non-replication of this finding for the other website versions (WA, WB and WD), it requires further investigation. The same applies to the sixth hypothesis (H6: the influence of classical aesthetics on perceived credibility is stronger for male than for female), which only holds true for version WA.

The result of the age-based multigroup analysis shows that younger people are more likely than older people to base the evaluation of the expressive aesthetics of a website on the perception of classical aesthetics (see H7 in Table 3). Specifically, this finding holds true in participants’ evaluation of version WB and WC (see Fig. 4b). This suggests that, in web design, younger people are less likely to differentiate between the traditional notion of visual aesthetics (orderliness, clarity and pleasantness) and the expressive power (creativity, originality and novelty) of the designer. Put differently, younger people are more likely than older people to judge a classically aesthetic (minimalist) design, such as WB, as expressively aesthetic. To verify this assertion, we analyzed the participants’ ratings of versions WB and WC, where both age groups differ, in terms of classical aesthetics and expressive aesthetics. Aesthetically, both WB and WC differ in terms of color scheme. While version WB (relative to WC and the other UI versions) is classically aesthetic given its minimalist (gray) color scheme, version WC (relative to WB) is expressively aesthetic given its blue color scheme.

Our one-way analysis of variance on the overall mean values of classical aesthetics and expressive aesthetics for WC shows that there is no significant difference between younger people’s ratings (4.75 and 4.06, respectively) and older people (4.72 and 3.90, respectively) ratings. Similarly, with respect to the overall mean values of classical aesthetics for WB (the minimalist webpage), there is no significant difference between younger people’s rating (4.74) and older people’s rating (4.64). However, with respect to the overall mean values of expressive aesthetics for WB, younger people’s rating (4.15) is significantly higher than older people’s rating (3.85) at p < 0.05. This shows that the perception of classical aesthetics (virtually the same for both age groups) resulted in higher perception of expressive aesthetics for younger people (4.15) than it does for older people (3.85). This confirms the SEM-based multigroup analysis finding that classical aesthetics is more likely to influence expressive aesthetics for younger people than for older people (H7). The design implication of this finding is as follows:

When designing websites, for example, to entertain people, the designer will have to be more creative to implement designs that older users will perceive novel, fascinating, sophisticated, creative, etc. The reason is that older people are less likely to be subject to the “classical-expressive aesthetic” halo effect than younger people. Moreover, older people are more likely to be critical in their judgment of expressive aesthetics than younger people.

6.3 Summary of Findings and Contributions

In this paper, we presented a path model of the interrelationships between the two dimensions of visual aesthetics and the perceived credibility of a utilitarian website. Our main findings vis-à-vis our three research questions can be summarized as follows:

  1. 1.

    Classical aesthetics influences perceived credibility more strongly than expressive aesthetics does.

  2. 2.

    Irrespective of the age and gender of the user, the interrelationships among classical aesthetics, expressive aesthetics and perceived credibility hold true.

  3. 3.

    Irrespective of the level of UI aesthetics, the interrelationships among classical aesthetics, expressive aesthetics and perceived credibility hold true.

Our main contribution to the body of knowledge in the field of HCI is that we have shown at a finer-grain level that, with respect to utilitarian websites, the classical aesthetics dimension contributes more strongly than the expressive aesthetics dimension in the aesthetics-credibility relationship. We specifically showed that this finding holds true, irrespective of the level of aesthetic treatment of the target website, age and gender of the user. Furthermore, as a second contribution, we have shown in a replicated fashion that the influence of classical aesthetics on expressive aesthetics is stronger for younger people than for older people. Our work, to the best of our knowledge, is the first in HCI research to uncover the relationship between perceived aesthetics and perceived credibility at a finer-grain level in a replicated fashion.

6.4 Limitations and Future Work

Our study has a number of limitations. The first and foremost limitation is that our findings are based on users’ (subjective) perception and evaluation of a hypothetical website interface (with four versions) and not on the outcome of an actual interaction with a live or experimental website. Consequently, our findings may not generalize to a real-life application setting. The second limitation of our study is that we did not investigate the moderating effect of other user characteristics than age and gender. User characteristics, such as culture, education level, internet experience, etc., could also moderate the strengths of the various interrelationships among the three design constructs we investigated in our path model. Consequently, we recommend that, in future research efforts, researchers can address these limitations. Moreover, with respect to our research question on which of the two dimensions of aesthetics has a stronger influence on perceived credibility, future work can investigate hedonic systems.

7 Conclusion

In this paper, we presented the results of an empirical study among 526 participants, which investigated, at a finer-grain level, which of the two dimensions of visual aesthetics (classical and expressive) has a stronger influence on perceived credibility in the evaluation of websites. Using a mobile website as a case study, we showed that, for utilitarian websites aimed at providing instrumental values to users, classical aesthetics influences perceived credibility more strongly than expressive aesthetics does. This finding holds true irrespective of the level of aesthetic treatment of the target website, the age and gender of the user. Moreover, we showed that the relationship between classical aesthetics and expressive aesthetics is stronger for younger people than for older people. The broad implication of our findings is that designers of utilitarian websites should focus on designing user interfaces which possess classical aesthetics features (e.g., clarity, cleanness, orderliness, pleasantness, etc. [12]) rather than expressive aesthetics features (e.g., novelty, richness, special effects, etc.) to increase the overall perceived credibility of their websites. A classically pleasing website is more likely to be perceived credible than an expressively arousing website.