3

The influence of visual complexity on initial user impressions: testing the persuasive model of web design Andy J. King a, Allison J. Lazard b and Shawna R. Whitec

aGreenlee School of Journalism and Communication, Iowa State UniversityAmes, IA, USA; bSchool of Media and Journalism, University of North Carolina, Chapel Hill, NC, USA; cTexas School Safety Center, Texas State University, San Marcos, TX, USA

ABSTRACT Shortly after fixating on webpages, users form initial impressions. These initial impressions influence how much users will use and return to websites. Researchers have understudied how objective design features (e.g. visual complexity) influence subjective perceptions of website content and the favorability of initial user impressions. In the present study, the influence of two dimensions of visual complexity – feature complexity and design complexity – were tested within the boundaries of the persuasive model of web design. More specifically, the study examined how visual complexity influences perceptions of visual informativeness, cues for engagement, favourable initial impressions, and behavioural intentions in a sample of young adults (N = 277). Results suggest relationships for both dimensions of visual complexity on all outcome variables using ANOVA and OLS regression procedures and that perceptions of visual informativeness and cues for engagement mediate the relationship between visual complexity and favourable initial impressions and behavioural intentions. The study offers support for the utility of the persuasive model of web design for linking objective design features with subjective design perceptions to better understand favourable initial user impressions.

ARTICLE HISTORY Received 2 April 2018 Accepted 25 March 2019

KEYWORDS Web design; persuasive computing; human-machine interface; visual display; visual complexity; user psychology

Initial impressions of websites happen quickly (Bölte et al. 2017; Lindgaard et al. 2006; Tractinsky et al. 2006) and leaving a good impression can improve the likelihood of users’ cognitive engagement with a site (e.g. Miniukovich and De Angeli 2014; Tuch et al. 2009) even though first impression importance has been found to be related to task orientation (see Iten, Troendle, and Opwis 2018). Given the impor- tance of initial impressions, researchers have noted the need for more research on visual design in relation to online information (Lazard and Mackert 2015), with specific interest in the visual message features of infor- mation design (see, e.g. King et al. 2014). Website designers benefit from understanding what draws people toward visiting and engaging with information that might promote their well-being. Identifying visual design features that enhance site visitor perceptions of interface affordances, or cues for engagement, and informativeness should result in improved initial impressions and intentions to seek related information online. One visual message feature that could influence perceptions of cues for engagement, informativeness, and positive initial user impressions is visual complexity.

Research indicates visual complexity positively influ- ences judgments of strategic information design in web design (Lazard and Mackert 2014; Tuch et al. 2012) and advertising contexts (Geissler, Zinkhan, and Watson 2006; Pieters, Wedel, and Batra 2010). The conceptualis- ation of visual complexity for design purposes vary, sometimes focusing on subjective evaluations (e.g. Tuch et al. 2012) and other times emphasising objective operationalisation (e.g. Pieters, Wedel, and Batra 2010; Tuch et al. 2009). Objective operationalizations vary; some consider websites more visually dense (i.e. an increase in visual edges) as more complex, while others propose that the organisation of information, (e.g. asym- metric content), in addition to amount, defines what is more visually complex. The present study examined two dimensions of visual complexity – feature complex- ity (FC) and design complexity (DC) – that follow these differing operationalizations and have been examined in multiple contexts previously (e.g. Donderi 2006; Lazard et al. 2017; Pieters, Wedel, and Batra 2010).

Related to work on visual complexity and initial impressions is theorising on persuasive web design, which considers how web design features influence user impressions and behaviour through perceptual

© 2019 Informa UK Limited, trading as Taylor & Francis Group

CONTACT Andy J. King [email protected] Greenlee School of Journalism and Communication, Iowa State University, 613 Wallace Road, Hamilton Hall 101, Ames, IA 50011, USA

Supplemental data for this article can be accessed at doi:10.1080/0144929X.2019.1602167

BEHAVIOUR & INFORMATION TECHNOLOGY 2020, VOL. 39, NO. 5, 497–510 https://doi.org/10.1080/0144929X.2019.1602167

mechanisms. At present, no paper has worked to inte- grate research on visual complexity with theorising on persuasive web design. This study examines how varying visual complexity influences initial user impressions through two theoretical mechanisms: informativeness (e.g. how much information a user perceives to be avail- able on a site) and engagement (e.g. how many opportu- nities a user perceives to have to interact with a site). The study tests whether objective operationalisation of web- page visual complexity can increase perceptions of visual informativeness and cues for engagement to influence favourable initial impressions and intentions to seek information, consistent with the model of persuasive visual web design (Ibrahim, Shiratuddin, and Wong 2014), as well as other work interested in linking user impressions to website interactions (e.g. Thielsch, Blo- tenberg, and Jaron 2014) and connecting design aes- thetics to behavioural intentions (e.g. Hall and Hanna 2004). The present study offers an initial empirical test linking objective visual design features to factors within the model of persuasive visual web design.

Visual complexity and online information

Research on visual complexity evolves from previous general theorising of aesthetic complexity (see Berlyne 1974) and takes into consideration graphic design prin- ciples (e.g. Lazard and Mackert 2014), as well as algorith- mic considerations of the visual and interactive features on a site (e.g. Wu, Hu, and Shi 2013). Research on visual complexity and web design often focuses on user percep- tions of complexity, while some more recent research has examined user evaluations of codified and/or algorithmic definitions of visual complexity. This section provides: (1) an overview of visual complexity research relevant to online information, (2) a review of relevant theorising and research on visual web design, and (3) a discussion of how to integrate visual complexity theory with the persuasive web design model.

Visual complexity and online information presentation

Research on the visual complexity of web design can be found across a variety of disciplines including human– computer interaction, graphic design, communication, and psychology. Some research emphasises subjective perceptions of visual complexity (Michailidou, Harper, and Bechhofer 2008; Nadkarni and Gupta 2007; Rein- ecke et al. 2013; Sohn, Seegebarth, and Moritz 2017; Tuch et al. 2012), while others use objective design fea- tures to manipulate visual complexity in web interfaces (Deng and Poole 2010, 2012; Lazard and Mackert

2014, 2015; Seckler, Opwis, and Tuch 2015) or a combi- nation of objective/subjective operationalizations (e.g. Geissler, Zinkhan, and Watson 2001; see also Donderi 2006).

Theorising on visual complexity considers psycho- logical processing of, as well as the practical matters of designing, information for various populations. Research considering visual complexity theory has considered cognitive responses like initial user impressions or evalu- ations of attractiveness (e.g. Orth and Wirtz 2014), satis- faction (e.g. Tuch et al. 2012), and aesthetics (e.g. Reinecke et al. 2013). Work contributing to visual com- plexity theory has also considered emotional and arousal responses (e.g. Deng and Poole 2012). Increased visual complexity may encourage users to see online infor- mation or displays as more useful and visually informa- tive (Eytam, Tractinsky, and Lowengart 2017; Lazard and Mackert 2014), measured by perceived visual infor- mativeness or an assessment of the quality and con- gruency of the communication with visuals and text (King et al. 2014). Lastly, the amount of visual complex- ity can increase affordances or perceived cues for engage- ment, with the appearance of visual features that communicate the ability to engage with a website (Hart- son and Pyla 2012).

Across existing theorising on visual complexity, findings suggest an inverted-U influence of visual com- plexity on users – more specifically, low and high levels of complexity are viewed as being less effective than moderate levels due at least in part to variations in arou- sal levels (see Berlyne 1974). Too little information may be boring, while too much information may overwhelm viewers (Berlyne 1974). Not all research, however, is con- sistent with this expectation. Researchers have found that increased complexity decreases satisfaction linearly, where higher levels of complexity resulted in the lowest levels of user satisfaction (Tuch et al. 2012). Albeit differ- ent than conceptualisation of simplicity or classical aes- thetics, increases in visual complexity – the amount or amount and organisation of visual information – can be associated with perceptions that something is more comprehensible or less complicated (Lazard andMackert 2014; Mollerup 2006). Indeed, other research found that more website elements included on pages generally decreases user appeal (Bauerly and Liu 2008). Still other research found that low and moderate complexity levels, based on a task-oriented model, differed in effec- tiveness depending on if people were classified as goal- directed users or experiential users, respectively (Nad- karni and Gupta 2007).

One possible explanation for the inconsistent evalu- ations of complexity relate to largely fluid and hard-to- replicate classifications of websites as varying in

498 A. J. KING ET AL.

complexity. The present study focuses specifically on objective, codified levels of visual complexity, which has been considered as multi-dimensional in terms of what elements contribute to perceptions of complexity, but not operationally articulated in strategic communi- cation contexts (i.e. webpage design) until recently.

Relevant theorising suggests that the inverted-U influ- ence holds true for one type of visual complexity – fea- ture complexity. Feature complexity refers to the density of visual edges, which is consistent with similar conceptualizations of ‘visual richness’ (Deng and Poole 2010, 2012) that emphasise information detail (i.e. amount of visual information coded through pixel to pixel comparison) in webpages when studying visual complexity. This operationalisation of feature complex- ity accounts for the number of visual edges or detectable change in colour, saturation, or brightness between pix- els, regardless of organisation. For example, a plain white box has no variation and is without any edges (lowest feature complexity). Adding a black-and-white checker- board pattern to the box would increase the complexity, as each change in colour is an edge. If the squares of the checkerboard were reduced in size, the box would have even greater visual complexity with more edges or visual variation shown. Given these definitions of visual feature complexity, as well as previous work on the patterns of influence for consumer evaluations, the following hypotheses are posited:

H1: Moderate levels of feature complexity will result in more favorable initial user impressions of the webpage design compared to low feature complexity.

H2: Moderate levels of feature complexity will result in more favorable initial user impressions of the webpage design compared to high feature complexity.

H3: Moderate levels of feature complexity will result in more favorable behavioral intentions to seek site-rel- evant information compared to low feature complexity

H4: Moderate levels of feature complexity will result in more favorable behavioral intentions to seek site-rel- evant information compared to high feature complexity.

A second type of visual complexity – design complex- ity – has been found to improve user perception of infor- mation and the persuasiveness of said information in a linear manner (i.e. higher levels of design complexity improve attention, attitudes toward information, etc.; Pieters, Wedel, and Batra 2010). Complementing the operationalisation of feature complexity, design com- plexity refers to the amount and organisation of visual elements within some information unit, including six codifiable, objective design principles: object quantity, irregularity, dissimilarity, and detail, as well as the

asymmetry and irregularity of object arrangement (Pieters, Wedel, and Batra 2010). Design complexity operationalises complexity based on a higher level of information structure that accounts for number of objects, as well as their placement, which may be more akin to how viewers perceive websites. Viewers do not perceive visual displays pixel by pixel; rather, gradients, illustrations, and photographs are often viewed as whole objects. In this way, a photograph of an apple may be perceived as a single object, not as complex as a feature complexity algorithm would indicate (given the pixel-by-pixel changes in the image). Thus, design complexity is more consistent with ideas, like ‘visual diversity’ (Deng and Poole 2010, 2012) and ‘webpage complexity’ (Bucy, Lang, Potter, & Grabe, 1999), which consider the diversity of structural or informational design features like text, images, and links; however, design complexity does not account for interactivity, interdependence of features, or certainty in the user experience (Nadkarni and Gupta 2007). Given these definitions of visual design complexity, the following hypotheses are put forth:

H4: High levels of design complexity will result in more favorable user first impressions of the webpage design compared to low levels of design complexity.

H5: High levels of design complexity will result in more favorable behavioral intentions to seek site-relevant information compared to low levels of design complexity.

In addition to the hypotheses, the review of research in this area suggests a research question regarding the interaction of feature and design complexity. While there is considerable research on visual complexity in general, there is still limited evidence regarding if there is any type of interaction between specific visual com- plexity types. For the current study, an exploratory analysis was conducted on the data to examine the inter- action between feature and design complexities on initial user impressions and behavioural intentions.

With few exceptions, research on visual complexity focuses on informational outcomes without much consideration for how visual web designs might help persuade individuals to favourably process content and seek site-relevant information in the future. Visual complexity, when studied as an intrinsic (objective) feature of webpages, serves as a visual message feature that can be replicated, manipulated, and studied in a variety of informational and persua- sive contexts (see Jensen 2012; King 2014). One way to improve research on visual complexity is to integrate theorising on visual web design to improve user evalu- ations of webpages.

BEHAVIOUR & INFORMATION TECHNOLOGY 499

Persuasive web design theorising

The model of persuasive visual web design is founded in general research and theorising on persuasion (Cialdini 2007; Petty and Cacioppo 1986) and persuasive technol- ogies (Fogg 2003, 2009), as well as the dual structure model (see Herzberg 1968) adapted for online contexts (see Kim and Fesenmaier 2008; Zhang and von Dran 2000) related to factors supporting information-seeking processes of individuals. The two-factor model for web- site design and evaluation (Zhang and von Dran 2000) suggests there are factors that predict user satisfaction. Some of these factors, called hygiene factors, ensure that users are not dissatisfied with online design, while other factors, called motivating factors, result in added user satisfaction. Hygiene factors include informative- ness and usability, while motivating factors include con- cepts such as credibility, involvement (i.e. engagement), and reciprocity (see Kim and Fesenmaier 2008).

As an extension of the two-factor model for website design and evaluation, Ibrahim, Shiratuddin, and Wong (2014) proposed theirmodel of persuasive visualweb design to address the role of visual persuasionwithin online design. The goal of Ibrahim’s model is to predict positive user evaluations and behavioural intentions relevant to web con- tent. The model outlines various hygiene and motivating factors, with the most recent and tested iteration of the per- suasive visual web designmodel (Ibrahim, Shiratuddin, and Wong 2016), including informativeness as a hygiene factor and engagement as a motivating factor. For the present study, perceived visual informativeness (PVI), a concept that relates to evaluations of written and visual information (see King et al. 2014) compared to more general consider- ation of informativeness (see Cho & Boster, 2008), was used. Using the PVI construct allows for more nuanced evaluation of visually-intensive stimuli (see King et al. 2014) as the variable emphasises the interplay of visual and textual elements within mediated messages. Given pre- dictions of the persuasive model of web design, these two factors should predict favourable user impressions and behavioural intentions relevant to displayed information.

H7: User perceptions of (a) visual informativeness and (b) cues for engagement will be positively associated with user impressions of the website design.

H8: User perceptions of (a) visual informativeness and (b) cues for engagement will be positively associated with behavioral intentions to seek topic relevant information.

Visual complexity and the persuasive visual web design model

Lastly, evidence from research on the persuasive model of web design suggests visual complexity may influence user

evaluations. Ibrahim, Wong, and Shiratuddin (2015) found more persuasive (visually complex) designs pro- duced greater ratings of informativeness, usability, engage- ment, satisfaction, credibility, and intentions. Consistent with other work on visual informativeness (Jensen et al. 2012; King et al. 2014) and engagement (Ibrahim, Wong, and Shiratuddin 2015), the hypothesis posited is that visual complexity’s influence on persuasion outcomes will be mediated by perceived visual informativeness and per- ceived cues for engagement. More specifically:

H9: Feature complexity will be associated with perceived (a) visual informativeness and (b) cues for engagement.

H10: Design complexity will be associated with per- ceived (a) visual informativeness and (b) cues for engagement.

H11: The relationship between feature complexity and favorable impressions of web design will be mediated by perceived (a) visual informativeness and (b) cues for engagement.

H12: The relationship between design complexity and favorable impressions of web design will be mediated by perceived (a) visual informativeness and (b) per- ceived cues for engagement.

H13: The relationship between feature complexity on favorable behavioral intentions will be mediated by per- ceived (a) visual informativeness and (b) cues for engagement.

H14: The relationship between design complexity on favorable behavioral intentions will be mediated by per- ceived (a) visual informativeness and (b) cues for engagement.

Methods

Participants and procedures

Emerging adults1 (N = 277) from college courses at a large public university in the Southwestern region of the United States participated in the study (see Sup- plemental Table 1 for demographic information). Par- ticipants volunteered for the study and completed the study in exchange for course credit. Once participants agreed to complete the study, they were provided a hyperlink directing them to an online survey platform (Qualtrics) that randomised them into one of the six study conditions. The Qualtrics online platform allows participants to complete the study on desktop compu- ters, laptop computers, or a mobile tablets/phones. The 2 × 3 between-subjects design (DC: high/low × FC: high/moderate/low) allowed us to test the hypotheses and research questions outlined above (see Figure 1 for images used for each condition).

500 A. J. KING ET AL.

Stimuli

Stimuli were created specifically for this study to adhere to the two dimensions (feature/design) of visual com- plexity (see Figure 1). More specifically, front pages for a website focused on fitness were modified from high- and low-ranking websites (see Lazard et al., 2014 for a similar approach). Using popular, existing sites as the foundation of the study stimuli allows for the evaluations to hold some ecological validity and ensure that the images displayed offer a prototypical-type design, which can influence user evaluations of sites and has been studied in the context of visual complexity pre- viously (see Tuch et al. 2012). Text and imagery were consistent across the conditions (albeit with more shown in some conditions), within a design structure that allowed for manipulations of visual complexity.

Consistent with previous work operationalising fea- ture and design complexity (Pieters, Wedel, and Batra 2010), high design complexity sites had: (1) more objects, (2) greater irregularity of object shapes, (3) dissimilar objects, (4) greater visual detail, (5) asymmetry of arrangement, and (6) irregular object arrangement. The high design complexity website stimuli used five of the six design complexity principles from Pieters, Wedel, and Batra (2010); low design complexity websites incor- porated only three principles.

Feature complexity was determined by compression ratio of jpeg file size in kilobytes (kb) to file size of raw image, an easy to access proxy based on the jpeg com- pression algorithms that eliminate redundant infor- mation (see Donderi and McFadden 2005; Pieters, Wedel, and Batra 2010; Tuch et al. 2009; Wallace 1992). High feature complexity leads to larger jpeg file sizes (and lower ratios). Thus, stimuli with low feature complexity had a ratio between 6.86:1 and 7:01:1, mid feature complexity from 5.51:1–5.63:1, and high feature complexity between 4.05:1 and 4.41:1.

The study’s dimensions of visual complexity represent a set of intrinsic, objective characteristics of website design (see O’Keefe, 2003). Participants were allowed to look at the webpage for as long as they wanted; but the survey displayed stimuli for at least 25 s during which time participants could not advance to the next screen. There were no statistically significant differences on time spent looking at the webpage by condition (all p values > .05; M = 47s; SD = 34s; Med = 37s, range 28s to 472s).2

Measures

In addition to the four variables outlined below, partici- pants completed demographic items, as well as items

related to another set of research questions and hypoth- eses reported elsewhere (Lazard & King, in press).

Perceived visual informativeness (PVI) Seven items evaluated the construct perceived visual informativeness (PVI), which assess the congruency and quality of visual/written information within mess- ages (see King et al. 2014). Items were modified from the original scale only regarding the target of the stimuli evaluation (i.e. ‘website’ for the current study instead of ‘pamphlet’ in King et al.). The items used for the study are as follows: ‘The images on the website contained essential information,’ ‘The visual information on the website was clear,’ ‘The images made other ideas easier to understand,’ ‘The images were large enough to see,’ ‘I found the images on the website informative,’ ‘Images on the website helped me understand the rest of the con- tent,’ and ‘I think the images on the website are worth remembering.’ Participants responded on a scale from strongly disagree (1) to strongly agree (5). Higher scores indicate greater perceptions of visual informativeness of the website. Overall, the scale demonstrated excellent reliability, Cronbach’s α = .87, M = 2.96, SD = .81, range = 1–5.

Perceived cues for engagement (PCE) Four items assessed participant perceptions of cues for engagement with the presented webpages. This variable is largely interested in measuring perceptions of visual affordances. Visual affordances generally refer to the appearance of cues that communicate the ability to engage with a website (i.e. links to click, interactive fea- tures of the website, etc.; see Hartson and Pyla 2012). There were no existing measures of perceived cues for engagement, so items unique to the present study were created. The four items included the following state- ments: ‘The website had lots of links I could click,’ ‘The website seemed to have lots of areas I could explore for more information,’ ‘The design of the website makes it likely I would engage with provided content,’ and ‘Visual cues on the website would’ve helped me explore the page.’ Participants responded to items again on a 1 (strongly disagree) to 5 (strongly agree) scale, with higher scores indicating greater perceptions of engagement cues. The measure demonstrated acceptable reliability, Cronbach’s α = .72, M = 3.39, SD = .75, range = 1–5.

Initial user impressions To assess participants’ initial user impressions of the webpages presented, a thought listing procedure was used (see Cacioppo and Petty 1981). The thought listing procedure requested that participants provide their thoughts during viewing the study stimuli into individual

BEHAVIOUR & INFORMATION TECHNOLOGY 501

boxes within the online questionnaire. Two of the authors coded the thought listing into positive and nega- tive thoughts. Both coders categorised all thoughts into

positive or negative codes and reached acceptable levels of reliability for both categories, Krippendorff’s α = .88 (positive) and α = .83 (negative). To determine a

Figure 1. Stimuli used in the study. Abbreviations: DC = design complexity; FC = feature complexity. (a) low DC/low FC, (b) low DC/ moderate FC, (c) low DC/high FC (d) high DC/low FC, (e) high DC/moderate FC, (f) high DC/high FC.

502 A. J. KING ET AL.

participant’s initial impression score, the negative thought count was subtracted from the positive count, M = 1.85, SD = 4.40, range =−10–13.3

Intentions to seek information Four items measured participants’ behavioural intentions to seek information related to the presented webpage, adapted from similar intentions items (see, e.g. Hovick, Kahlor, and Liang 2014). All items were presented with response options ranging from very unlikely (1) to very likely (5) for prompts including ‘I will look for information specific to the website I looked at in the near future,’ ‘I will probably read more about information specific to the web- site I looked at if I come across it in the future,’ ‘I will per- form a search online to gather more information related to the website design I viewed earlier,’ and ‘How likely is it that you will look for health information, specific to the website you looked at, in the near future?’ Overall, the measure again demonstrated high internal reliability, Cronbach’s α = .89, M = 2.33, SD = .97, range = 1–5.

Results

Prior to analysis, data were screened to ensure that par- ticipants randomised into conditions did not differ on demographic factors including age, sex, race, or ethni- city. The analyses suggested randomisation did occur as intended, as there were no differences found. Time with stimuli and time to complete the survey also did not differ by condition. A correlation matrix for study variables can be found in Table 1. For hypotheses 1 through 4, as well as 7 and 8, multivariate analysis of var- iance (MANOVA) procedures were used, using Tukey honestly significant difference (HSD) to test for feature complexity differences; only post hoc tests significant at p < .05 are reported in text (see Table 2 for full mean comparison results). For hypotheses 5 and 6, ordinary least squares (OLS) regression procedures were used. For hypotheses 9 through 12, PROCESS (v3.0) pro- cedures for SPSS developed by Hayes (2018) were used.

Using MANOVA procedures and entering all four study variables, results indicated significant multivariate effects for design complexity, Wilks’ Lambda = .87, F (4,268) = 9.75, p < .001, h2

p = .13 and feature complexity,

Wilks’ Lambda = .68, F(8,536) = 14.06, p < .001, h2 p

= .17, but not their interaction (p = .32). Univariate results related to this MANOVA are discussed below.

H1 through H4: feature complexity and user impressions and behavioural intentions

The first two hypotheses tested the prediction that mod- erate levels of feature complexity would be more influen- tial on participant impressions and behavioural intentions than low or high levels. Consistent with the MANOVA results, the ANOVA results found a main effect for feature complexity on favourable impressions, F(2, 271) = 10.35, p < .001, h2

p = .07. Tukey’s HSD post hoc tests found that moderate levels of complexity (M = 2.21, SD = 4.87) resulted in more positive impressions than low levels of feature complexity (M = .30, SD = 4.36) supporting H1 (Cohen’s d = .41). Results indicated no support for H2, however, as the mean for high levels of complexity was not statistically different, though numerically higher (M = 3.04, SD = 3.43), than moderate feature complexity; in other words, there was no statisti- cally significant difference and the mean was direction- ally inconsistent with the hypothesis.

Again, consistent with MANOVA results, the ANOVA results indicated a main effect for feature com- plexity on behavioural intentions, F(2,271) = 5.81, p = .003, h2

p = .04. Tukey’s HSD post hoc tests found that moderate levels of complexity (M = 2.26, SD = 1.00) did not result in higher behavioural intentions than low (M = 2.13, SD = .80) feature complexity. The only differ- ence between feature complexity types based on post hoc comparisons is that high levels of feature complexity (M = 2.60, SD = 1.03) resulted in greater behavioural inten- tions compared to either moderate or low levels. Taken together, the results find no support for H3 or H4, as the significant differences found were not consistent with the hypotheses.

H5 and H6: design complexity and user impressions and behavioural intentions

ANOVA results showed a main effect for design com- plexity on initial impressions, F(1, 271) = 7.54, p = .006, h2 p = .03. High design complexity resulted in more

favourable initial impressions (M = 2.51, SD = 4.14) than low design complexity (M = 1.12, SD = 4.58), which supports H5. There was no effect of design com- plexity on behavioural intentions (p = .29), which does not support H6. Additionally, the exploratory analysis of the interaction of the two complexity types did not find an interaction for either favourable initial impressions (p = .10) or behavioural intentions (p = .85).

Table 1. Correlation matrix of study variables. 1 2 3 4

1. PVI 1.00 2. PCE .54*** 1.00 3. Initial Impressions .52*** .43*** 1.00 4. Intentions .52*** .44*** .42*** 1.00

Abbreviations: PVI = perceived visual informativeness; PCE = perceived cues for engagement.

***p < .001, **p < .01, *p < .05.

BEHAVIOUR & INFORMATION TECHNOLOGY 503

H7 and H8: visual informativeness and cues for engagement influence on user initial impressions and behavioural intentions

Two separate OLS models were tested to determine if perceptions of visual informativeness (PVI) and cues for engagement (PCE) were associated with favourable initial impressions and behavioural intentions. Both PVI (B = 2.20, SE = .32, t = 6.79, p < .001) and PCE (B = 1.26, SE = .35, t = 3.62, p < .001) associated with favourable initial impressions, R2 = .31, F(2,274) = 60.04, p < .001, which supports H7.

H8 posited the same relationship would occur for behavioural intentions. Again, the OLS model offered results consistent with the hypothesis, as PVI (B = .47, SE = .07, t = 6.57, p < .001) and PCE (B = .30, SE = .08, t = 3.89, p < .001) associated with behavioural intentions, R2 = .31, F(2,274) = 60.21, p < .001. Thus, the results sup- port H8.

H9 and H10: visual complexity on visual informativeness and cues for engagement

The ANOVA results for feature complexity on visual informativeness demonstrated a linear pattern of means, F(2,271) = 35.32, p < .001, h2

p = .21, increasing as complexity increased, consistent with results from H1 through H4. Specifically, low levels of feature complexity resulted in the lowest scores for PVI (M = 2.60, SD = .72), with the levels increasing for moderate (M = 2.86, SD = .83, moderate vs. low PVI Cohen’s d = .33) and high (M = 3.44, SD = .70, high vs. low PVI Cohen’s d = 1.22)

feature complexity. All differences were significant, per Tukey’s HSD post hoc tests, at p < .05 (see Table 2 for full means and standard deviations). The results offer partial support for H9a.

The same pattern of results was found for perceived cues for engagement (PCE), F(2,271) = 35.80, p < .001, h2 p = .21, with means higher as feature complexity

moved from low (M = 2.92, SD = .70) to moderate (M = 3.55, SD = .69, moderate vs. low PCE Cohen’s d = .91) to high (M = 3.71, SD = .63, high vs. low PCE Cohen’s d = 1.19). Post hoc tests revealed that the mod- erate and high feature complexity scores were statistically significantly different than the low feature complexity scores, but there was no significant difference between moderate and high levels (see Table 2). As such, H9b received partial support.

Main effects presented for design complexity on both PVI and PCE. For PVI, high design complexity resulted in higher scores (M = 3.19, SD = .70) than low design complexity (M = 2.71, SD = .85), F(1,271) = 34.44, p < .001, h2

p = .11. Similarly, high design complexity led to higher PCE scores (M = 3.49, SD = .76) than low design complexity (M = 3.29, SD = .73), F(1,271) = 6.44, p = .01, h2

p = .02. Taken together, results support H10.

H11 through H14: mediation of visual complexity effects on user impressions and behavioural intentions

Mediation analyses were conducted using PROCESS for SPSS (Hayes 2018). Table 3 presents full results for the multiple mediation analyses. For this analysis, the two

Table 2. Means for study variables by condition. Low DC High DC DC Total

M SD n M SD n M SD n

PVI Low FC 2.34 .72 44 2.84 .64 48 2.60a .72 92 Moderate FC 2.51 .83 44 3.16 .71 50 2.86b .83 94 High FC 3.27 .71 44 3.59 .56 47 3.44c .65 91 Total FC 2.71a .85 132 3.19b .70 145 2.96 .81 277

PCE Low FC 2.90 .69 44 2.94 .72 48 2.92a .70 92 Moderate FC 3.38 .70 44 3.69 .65 50 3.55b .69 94 High FC 3.58 .66 44 3.84 .57 47 3.71b .63 91 Total FC 3.29a .73 132 3.49b .76 145 3.39 .75 277

Initial Impress Low FC −.43 4.45 44 .98 4.21 48 .30a 4.36 92 Moderate FC .77 5.13 44 3.48 4.29 50 2.21b 4.87 94 High FC 3.02 3.39 44 3.06 3.49 47 3.04b 3.43 91 Total FC 1.12a 4.58 132 2.52b 4.14 145 1.85 4.40 277

Intention Low FC 2.11 .75 44 2.16 .84 48 2.13a .80 92 Moderate FC 2.15 .95 44 2.36 1.04 50 2.26a 1.00 94 High FC 2.54 .97 44 2.65 1.09 47 2.60b 1.03 91 Total FC 2.27a .91 132 2.38a 1.01 145 2.33 .97 277

Notes: There was no interaction between design complexity (DC) and feature complexity (FC). As such, the only comparisons accounted for in the table above are those of the main study factors (FC and DC). Within each variable, the means located in the DC columns that do not share a superscript within each row for “Total FC” are different at p < .05 (using Tukey’s Honestly Significant Difference post hoc test within MANOVA procedures as described in text). Within each variable, the FC rows that do not share a superscript within the “DC Total” column are different at p < .05, using the comparisons within univariate ANOVAs provided through the MANOVA procedures in SPSS.

Abbreviations: FC = feature complexity, DC = design complexity, PVI = perceived visual informativeness; PCE = perceived cues for engagement; Initial Impress = initial impressions; Intention = behavioral intentions to seek website information.

504 A. J. KING ET AL.

types of visual complexity were dummy coded so that lowest level of feature and design complexity served as the comparison conditions. The interaction term was not included as there was no interaction of the two com- plexity types on any outcome. Mediation results indicate full support for hypotheses 11 through 14, as perceptions of visual informativeness and cues for engagement mediated the relationship of each visual complexity type on the two outcomes relevant to favourable initial impressions (see Table 3 for coefficients and confidence intervals).

Discussion

The goal of the present study was to determine if the effects of two conceptualizations of visual complexity on initial user impressions and behavioural intentions could be explained by mechanisms put forth by the per- suasive model of web design, namely informativeness and engagement. Results largely supported past work on visual complexity and the persuasive model of web design, with a few caveats. The present study found sup- port for most hypotheses, although the curvilinear relationship of feature complexity on user evaluation outcomes did not emerge in the current study. Overall, visual complexity predicted user perceptions of the infor- mativeness and the opportunities for engagement of websites. In turn, those perceptions influenced user evaluations and intentions related to the website. The overall model tested suggests that objectively-modified features, such as complexity, can have both a direct and indirect impact on user evaluations and information seeking related to website content. Theoretically, results provide evidence in support of propositions posited by the persuasive model of web design. Effects sizes were small to moderate in size and directionally consistent with the theory.

The first six hypotheses were interested in the relationship of visual complexity (feature and design) on initial impressions of webpages. Previous theorising suggested a linear relationship for visual design complex- ity and a curvilinear relationship for visual feature com- plexity on user evaluations. Results supported the design complexity, but not the feature complexity, predictions; across both conceptualizations, increased complexity led to more positive impressions. One explanation for this might be that the current operationalizations of fea- ture complexity, while relatively distinct, did not overall reach the necessary threshold to be considered high fea- ture complexity.

More specifically, the webpage stimuli were designed to be examples of realistic, quality overall designs instead of webpages simply found online that offer variations of

complexity. Of course, visual complexity typically rep- resents a relative term in any sample of webpages. Per- haps, for example, while the visual feature complexity of the stimuli was high relative to other stimuli, some even more visually complex design being included would shift the linear pattern of findings. Indeed, con- sistent with Berlyne’s (1974) argument, complexity increases the potential of arousal to a point of diminish- ing returns (i.e. the inverted-U figure) in the current visually rich digital environment, so perhaps the stimuli – while relatively high in feature complexity – failed to hit the point of diminishing returns. Modern discussion of complexity (e.g. Norman 2010) discuss concepts like appropriate complexity. Perhaps current, active users (e.g. young adults as used as participants in the present study) perceive more visually complex stimuli as being appropriate and not otherwise notably. Again, the dis- cussion is limited to visual complexity and initial impressions, not other complexity considerations like task orientation (see Nadkarni and Gupta 2007). More careful conceptualizations of visual complexity will benefit understanding of the effect of said complexity. The results support other recent work that complicates previously accepted findings related to the inverted-U shape pattern of influence related to complexity (e.g. Nadkarni and Gupta 2007).

The next hypotheses suggested relationships between visual complexity and perceptual variables (e.g. informa- tiveness and cues for engagement), as well as those per- ceptual variables’ associations with user evaluations. The largest effects (differences) produced by the visual com- plexity manipulations related to these perceptual vari- ables. Consistent with the hypotheses derived from the persuasive model of web design, enhanced perceptions of visual informativeness and cues for engagement were associated with more favourable user evaluations. Additionally, visual complexity influenced perceptions in the same manner as found for the influence of visual complexity on user evaluations. Taken together, the results relevant to these hypotheses suggest that visual complexity – and perhaps other objective web design fea- tures – can be meaningfully integrated into theorising on persuasive web design. Users should perceive that the design of the site contains visual information of interest and affords users opportunities to revisit and use site content.

The final hypotheses of the study were interested in a theoretical test of the complete model, illustrated in Figure 2, to determine if the direct relationship found between complexity and user evaluations could be explained by the persuasive model of web design pre- dictions. The theory received support in the present study.

BEHAVIOUR & INFORMATION TECHNOLOGY 505

The study findings support that adding visual infor- mation (feature complexity) and organising this infor- mation in diverse ways (design complexity) will lead to favourable perceptions, impressions, and intentions to seek information online. While many people have rec- ommended incorporating more persuasion research into web design research (e.g. Lazard and Mackert 2014), few studies have attempted to link these different research literatures and consider objective design’s influ- ence on subjective evaluations. The study results suggest that greater perceptions of informativeness and engage- ment based on increases in a website’s visual complexity, consistent with the persuasive web design model, are strongly associated with positive initial impressions of

a webpage. The present study is one of the first empirical tests of the persuasive web design model (Ibrahim, Shir- atuddin, and Wong 2014; Ibrahim, Shiratuddin, and Wong 2016) that links specific web design characteristics with model components (i.e. mediating perceptions of visual informativeness and cues for engagement) and outcomes (i.e. impressions and intentions). The results provide a foundation for future inquiry attempting to link specific web design characteristics with user evalu- ations and positive initial impressions.

Practical implications

The present study adds evidence that visual complexity influences user evaluations of webpages to existing findings, as well as provides support for the integration of visual interface features like complexity into research interested in persuasive online design. Visual complexity clearly influences favourable impressions of webpage design. Given that users judge websites (generally ‘good’ or ‘bad’) in a fraction of a second (Lindgaard et al. 2006), the findings provide evidence for objective design strategies to increase the likelihood that web designs will function as gateways to critical information, instead of barriers. The study offers insights by account- ing for a persuasion perspective of user impressions of

Table 3. PROCESS Model Coefficients for Indirect Effects Analyses.

Paths

Dependent Variables

Initial Impressions Behavioral Intentions

B p LLCI ULCI B p LLCI ULCI

PVI Mod FC .252* .014 .051 .453 High FC .837*** <.001 .635 1.039 High DC .492*** <.001 .327 .656

PCE Mod FC .625*** <.001 .432 .817 High FC .797*** <.001 .603 .991 High DC .204* .012 .046 .362

DV PVI 2.294*** <.001 1.590 3.00 .527 <.001 .374 .680 PCE 1.170** .002 .435 1.905 .350 <.001 .191 .510 Mod FC .585 .314 −.558 1.729 −.228 .071 −.476 .020 High FC −.105 .868 −1.347 1.137 −.256 .062 −.526 .013 High DC .033 .944 −.896 .961 −.208* .043 −.410 −.007

Indirect Effects Through PVI Mod FC .578 – .096 1.060 .133 – .022 .256 High FC 1.920 – 1.272 2.627 .441 – .300 .593 High DC 1.128 – .654 1.678 .259 – .151 .382

Through PCE Mod FC .731 – .269 1.267 .219 – .110 .351 High FC .932 – .330 1.618 .279 – .150 .440 High DC .238 – .034 .524 .071 – .014 .139

Notes: Indirect effects for which the confidence intervals do not overlap zero are shaded light gray. The coefficients for the paths to PVI and PCE are entered for favorable initial impressions, but not behavioral intentions, because the coefficients are identical. The indirect effects coefficients provided are unstandardized relative indirect effects.

Abbreviations: PVI = perceived visual informativeness; PCE = perceived cues for engagement; FC = feature complexity; DC = design complexity; Mod = moderate; LLCI = lower level for confidence interval; ULCI = upper level for confidence interval; DV = the dependent variable in each model, so for the “Initial Impressions” column the coefficients represent the paths to that variable; same for the “Behavioral Intentions” column.

***p < .001, **p < .01, *p < .05

Figure 2. Conceptual diagram of PROCESS models.

506 A. J. KING ET AL.

intrinsic visual characteristics of design that could be easily used by web designers in the creation of user pre- ferred webpages.

Given past work suggesting complexity correlates with web traffic (Bucy et al., 1999), the influence of fea- ture and design complexity may function differently in web design contexts. Regular internet use likely allows user to form mental models or expectations about proto- typical designs to orient themselves to information quickly (Roth et al. 2010; Roth et al. 2013). In this way, high feature complexity might not deter people as pre- dicted (i.e. is not perceived as overwhelming); users may be savvy enough to see the structure of websites, even if never encountered before, and are not put off by having more visual detail.

Research in web design will likely change rapidly as user familiarity and savviness increases over time. Younger users who have more familiarity navigating online information sources may be less cognitively over- whelmed by the information detail (i.e. amount of visual stuff) included within a webpage. Perhaps younger users are more adept at navigating visual clutter of web designs, so judgments at first glance hinge less on a page’s design having a moderate amount of complexity.

Broadly, to increase visual complexity, researchers and practitioners should consider the following strategies in line with definitions and distinctions made by Pieters, Wedel, and Batra (2010): (1) increase the number of unique objects on the pages, including images with rich detail (i.e. photographs or artistic illustrations) and dis- tinct content sections; (2) when possible, use irregular shaped objects, such as image cut outs instead of stan- dard rectangular borders, to increase appeal; (3) vary the shape and size of design elements on the webpage (i.e. object dissimilarity) so users can easily distinguish unique features and functions at a glance; and (4) while information should be clearly organised, placing some design elements or objects ‘off grid’ so they increase the overall asymmetry of the pages can help elements stand out and create more visual interest for users in their scan of the design.

Limitations

The study had limitations that warrant mention. First, the study used a small pool of stimuli compared to many others in this area of work. The creation of stimuli unique to the different levels of complexity was a strength of the study, but future work should test mul- tiple webpages. Additionally, the sample of emerging adults included far more women than men. A more evenly distributed sample would be preferred, although post hoc analyses demonstrated no differences on any

study variable related to sex. Future work might consider pursuing scaling studies to determine what level of objec- tive design/feature complexity represent thresholds that designers could consider when trying to optimise com- plexity within digital design interfaces.4 Measurement of study variables could have been improved, particularly for a measure related to perceived cues for engagement. Related, future work would benefit by assessing per- ceived cues for engagement and then allowing partici- pants to interact with the website. The lack of participant interaction with the webpages in the present study is a limitation that necessitates caution in inter- preting how the results would affect user behaviour specific to the pages in question (e.g. click behaviour and information seeking specific to the design).

Finally, results should be interpreted with caution given the lack of validated measures used in the present study. Future studies could validate measures of per- ceived cues for engagement or perceived visual affor- dances, which would improve persuasive model of web design, as well as validate other complexity-related measures. Future research would also benefit from con- sidering the moderating role of variables not included in the present study including measures related to eye- sight, visual processing preferences, device used to com- plete the survey, and interest in presented content.

Future research

Future research should incorporate more visual persua- sion research (see Messaris 1997; King et al. 2014), test- ing other design strategies to appeal to users for favourable website impressions. Given recent work on how imagery can influence online sharing of content (see Keib et al. 2018), including such work to improve evaluations that happen quickly (see Lindgaard et al. 2006), investigating a variety visual appeals could improve the likelihood that users spend time engaging with site content immediately or in the future. Addition- ally, researchers should attempt to replicate the findings from this study and other related work across various contexts, expanding beyond the context of health and fitness studied herein, to verify the scope and usefulness of the formulated recommendations. It is probable that an individual searching online for general health and fitness information comes to a site with a different frame of mind than someone searching for information pertaining to purchasing a car, for example.

Conclusion

Overall, the study offers evidence about specific mechan- isms visual complexity influences that, in turn, improve

BEHAVIOUR & INFORMATION TECHNOLOGY 507

users’ initial impressions of webpages. These findings should contribute to the evidence base on the influence of visual complexity, as well as on theorising related to the persuasive web design model. Improving research and theorising in this area of human–computer inter- action research through the inclusion of principles of persuasion will enhance the quality of user web experi- ence and could improve the appeal of informational interfaces in the future.

Notes

1. Initially, 301 people were recruited for the study. The target was 300, but one additional case was collected unintentionally. Of those 301 people, 24 were removed for analysis due to unrealistic completion times (± 3 SDs of average time; n = 17) and participants indicating they did not take their participation seriously (n = 7).

2. The intent of the study procedure was to ensure partici- pants spent at least some time evaluating the webpage presented and not simply click through the survey. As such, our study examines initial user impressions of websites consistent with past work in web design, but exposed participants to the webpage more than research that looks at ‘first impressions’ exclusively. Research on first impressions often limits exposure to a short period of time (e.g., 500ms, see Tractinsky et al., 2006). Hence, we use the term ‘initial user impressions’ to study evalu- ation similar to other work interested in website design features and user evaluations that includes longer exposure times for participants to formulate an initial impression (e.g., Lazard et al., 2017).

3. The greatest available range of scores would have been from −20 to 20, as there were 20 boxes provided to par- ticipants for the thought listing task.

4. As noted earlier, the study operationalisation of com- plexity, while following guidelines from previous research, may have engaged only relatively low, med- ium, and high levels of complexity. That is, there is not at present a pre-defined threshold to determine each level of complexity, which could explain the failure to find the inverted-U relationship for feature complex- ity initially hypothesized.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Andy J. King http://orcid.org/0000-0002-2789-2550 Allison J. Lazard http://orcid.org/0000-0002-2502-2850

References

Bauerly, M., and Y. Liu. 2008. “Effects of Symmetry and Number of Compositional Elements on Interface and

Design Aesthetics.” International Journal of Human- Computer Interaction 24: 275–287.

Berlyne, D. E., ed. 1974. Studies in the new Experimental Aesthetics: Steps Toward an Objective Psychology of Aesthetic Appreciation. Oxford, UK: Hemisphere.

Bölte, J., T. Hösker, G. Hirschfeld, and M. Thielsch. 2017. “Electrophysiological Correlates of Aesthetic Processing of Webpages: A Comparison of Experts and Laypersons.” PeerJ 5: e3440. doi:10.7717/peerj.3440.

Bucy, E. P., A. Lang, R. F. Potter, and M. E. Grabe. 1999. “Formal Features of Cyberspace: Relationships Between Web Page Complexity and Site Traffic.” Journal of the Association for Information Science and Technology 50 (1999): 1246 –1256. doi:10.1002/(SICI)1097-4571 (1999)50:133.0.CO;2-E.

Cacioppo, J. T., and R. E. Petty. 1981. “Social Psychological Procedures for Cognitive Response Assessment: The Thought-Listing Technique.” In Cognitive Assessment, edi- ted by T. V. Merluzzi, C. R. Glass, and M. Genest, 309– 342. New York: Guilford Press.

Cho, H., and F. J. Boster. 2008. “Effects of Gain Versus Loss Frame Antidrug Ads on Adolescents.” Journal of Communication 58: 428–446. doi:10.1111/j.1460-2466. 2008.00393.x.

Cialdini, R. B. 2007. Influence: The Psychology of Persuasion, Revised Edition. New York: HarperCollins.

Deng, L., and M. S. Poole. 2010. “Affect in web Interfaces: A Study of the Impacts of web Page Visual Complexity and Order.” MIS Quarterly 34: 711–730.

Deng, L., and M. S. Poole. 2012. “Aesthetic Design of e- Commerce web Pages—Website Complexity, Order and Preference.” Electronic Commerce Research & Applications 11: 420–440. doi:j.elerap.2012.06.004.

Donderi, D. C. 2006. “Visual Complexity: A Review.” Psychological Bulletin 132: 73–97. doi:10.1037/0033-2909. 132.1.73.

Donderi, D. C., and S. McFadden. 2005. “Compressed File Length Predicts Search Time and Errors on Visual Displays.” Displays 26: 71–78. doi:10.1016/j.displa.2005.02. 002.

Eytam, E., N. Tractinsky, and O. Lowengart. 2017. “The Paradox of Simplicity: Effects of Role on the Preference and Choice of Product Visual Simplicity Level.” International Journal of Human-Computer Studies 105: 43–55. doi:10.1016/j.ijhcs.2017.04.001.

Fogg, B. J. 2003. Persuasive Technology: Using Computers to Change What we Think and do. San Francisco, CA: Morgan Kaufmann.

Fogg, B. J. 2009. “A Behavior Model for Persuasive Design.” Persuasive ‘09: Proceedings of the 4th International Conference on Persuasive Technology (article 40). doi:10.1145/1541948.1541999.

Geissler, G., G. Zinkhan, and R. T. Watson. 2001. “Web Home Page Complexity and Communication Effectiveness.” Journal of the Association for Information Systems 2: 1.

Geissler, G. L., G. M. Zinkhan, and R. T. Watson. 2006. “The Influence of Home Page Complexity on Consumer Attention, Attitudes, and Purchase Intent.” Journal of Advertising 35 (2): 69–80. doi:10.1080/00913367.2006. 10639232.

Hall, R. H., and P. Hanna. 2004. “The Impact of web Page Text-Background Colour Combinations on Readability,

508 A. J. KING ET AL.

Retention, Aesthetics and Behavioural Intention.” Behaviour & Information Technology 23: 183–195. doi:10. 1080/01449290410001669932.

Hartson, R., and P. S. Pyla. 2012. The UX Book: Process and Guidelines for Ensuring a Quality User Experience. Waltham, MA: Morgan Kaufmann.

Hayes, A. F. 2018. Introduction to Mediation, Moderation, and Conditional Process Analysis. 2nd ed. New York: Guilford Press.

Herzberg, F. 1968. “One More Time: How do you Motivate Employees?” Harvard Business Review 46: 53–62.

Hovick, S. R., L. Kahlor, and M.-C. Liang. 2014. “Personal Cancer Knowledge and Information Seeing Through PRISM: The Planned Risk Information Seeking Model.” Journal of Health Communication 19: 511–527. doi:10. 1080/10810730.2013.821556.

Ibrahim, N., M. F. Shiratuddin, and K. W. Wong. 2014. “Proposed Model of Persuasive Visual Design for Web Design.” Proceedings of the 25th Australasian Conference on Information Systems. http://hdl.handle. net/10292/8044.

Ibrahim, N., M. F. Shiratuddin, and K. W. Wong. 2016. “The Impact of Online Visual on Users’ Motivation and Behavioural Intention: A Comparison Between Persuasive and non-Persuasive Visuals.” AIP Conference Proceedings 1761: 020053. doi:10.1063/1.4960893.

Ibrahim, N., K. W. Wong, and M. F. Shiratuddin. 2015. “Persuasive Impact of Online Media: Investigating the Influence of Visual Persuasion.” Proceedings of the Asia Pacific Conference on Multimedia and Broadcasting, doi:10.1109/APMediaCast.2015.7210269.

Iten, G. H., A. Troendle, and K. Opwis. 2018. “Aesthetics in Context: The Role of Aesthetics and Usage Mode for a Website’s Success.” Interacting with Computers 30: 133– 149. doi:10.1093/iwc/iwy002.

Jensen, J. D. 2012. “Addressing Health Literacy in the Design of Health Messages.” In Health Communication Message Design: Theory and Practice, edited by H. Cho, 171–189. Thousand Oaks, CA: Sage.

Jensen, J. D., A. J. King, N. Carcioppolo, and L. Davis. 2012. “Why are Tailored Messages More Effective? A Multiple Mediation Analysis of a Breast Cancer Screening Intervention.” Journal of Communication 62: 851–868. doi:10.1111/j.1460-2466.2012.01668.x.

Keib, K., C. Espina, Y.-I. Lee, B. W. Wojdynski, D. Choi, and H. Bang. 2018. “Picture This: The Influence of Emotionally Valenced Images on Attention, Selection, and Sharing of Social Media News.” Media Psychology 21: 202–221. doi:10.1080/15213269.2017.1378108.

Kim, H., and D. R. Fesenmaier. 2008. “Persuasive Design of Destination web Sites: An Analysis of First Impression.” Journal of Travel Research 47: 3–13. doi:10.1177/ 0047287507312405.

King, A. J. 2014. “Visual Messaging and Risk Communication.” In Sage Handbook of Risk Communication, edited by H. Cho, T. Reimer, and K. A. McComas, 193–205. Thousand Oaks, CA: Sage.

King, A. J., J. D. Jensen, L. A. Davis, and N. Carcioppolo. 2014. “Perceived Visual Informativeness (PVI): Construct and Scale Development to Assess Visual Information in Printed Materials.” Journal of Health Communication 19: 1099–1115. doi:10.1080/10810730.2013.878004.

Lazard, A. J., A.Dudo, T.Dennis,M.G. Ewald, andB. Love. 2017. “Making a Visual Impression (or not): Current Design Practices of Nutritional Websites.” Health Communication 32: 470–482. doi:10.1080/10410236.2016.1140267.

Lazard, A. J., and A. J. King. in press. “Objective Design to Subjective Evaluations: Connecting Visual Complexity to Aesthetic and Usability Assessments of eHealth.” International Journal of Human-Computer Interaction.

Lazard, A., and M. Mackert. 2014. “User Evaluations of Design Complexity: The Impact of Visual Perceptions for Effective Online Health Communication.” International Journal of Medical Informatics 83: 726–735. doi:10.1016/j.ijmedinf. 2014.06.010.

Lazard, A. J., and M. S. Mackert. 2015. “E-health First Impressions and Visual Evaluations: Key Design Principles for Attention and Appeal.” Communication Design Quarterly 3 (4): 25–34. doi:10.1145/2826972.2826975.

Lindgaard, G., G. Fernandes, C. Dudek, and J. Brown. 2006. “Attention web Designers: You Have 50 Milliseconds to Make a Good First Impression!.” Behaviour & Information Technology 25: 115–126. doi:10.1080/01449290500330448.

Messaris, P. 1997. Visual Persuasion: The Role of Images in Advertising. Thousand Oaks, CA: Sage.

Michailidou, E., S. Harper, and S. Bechhofer. 2008. “Visual Complexity and Aesthetic Perception of Web Pages.” SIGDOC ‘08: Proceedings of the 26th Annual ACM International Conference on Design of Communication (pp. 215–224). doi:10.1145/1456536.1456851.

Miniukovich, A., and A. De Angeli. 2014. “Visual Impression of Mobile App Interfaces.” ACM Proceedings of the 8th Nordic Conference on Human-Computer Interaction: Fun, Fast, Foundational (pp. 31–40). doi:10.1145/ 2639189.2641219.

Mollerup, P. 2006. “Simplicity.” Design Research Quarterly 2: 7–15.

Nadkarni, S., and R. Gupta. 2007. “A Task-Based Model of Perceived Website Complexity.” MIS Quarterly 31: 501– 534. doi:10.2307/25148805.

Norman, D. A. 2010. Living with Complexity. Cambridge, MA: MIT Press.

Orth, U. R., and J. Wirtz. 2014. “Consumer Processing of Interior Service Environments: The Interplay among Visual Complexity, Processing Fluency, and Attractiveness.” Journal of Service Research 17: 296–309. doi:10.1177/1094670514529606.

Petty, R. E., and J. T. Cacioppo. 1986. “The Elaboration Likelihood Model of Persuasion.” In Advances in Experimental Social Psychology, edited by L. Berkowitz, Vol. 19, 123–204. San Diego, CA: Academic Press.

Pieters, R., M. Wedel, and R. Batra. 2010. “The Stopping Power of Advertising: Measures and Effects of Visual Complexity.” Journal of Marketing 74 (5): 48–60. doi:10.1509/jmkg.74.5.48.

Reinecke, K., T. Yeh, L. Miratrix, R. Mardiko, Y. Zhao, J. Liu, and K. Z. Gajos. 2013. “Predicting Users’ First Impressions of Website Aesthetics with a Quantification of Perceived Visual Complexity and Colorfulness.” CHI ‘13: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 2049-2058). doi:10.1145/2470654.2481281.

Roth, S. P., P. Schmutz, S. L. Pauwels, J. A. Bargas-Avila, and K. Opwis. 2010. “Mental Models for web Objects: Where do Users Expect to Find the Most Frequent Objects in Online Shops, News Portals, and Company web Pages?”

BEHAVIOUR & INFORMATION TECHNOLOGY 509

Interacting with Computers 22: 140–152. doi:10.1016/j. intcom.2009.10.004.

Roth, S. P., A. N. Tuch, E. D. Mekler, J. A. Bargas-Avila, and K. Opwis. 2013. “Location Matters, Especially for non-Salient Features–An eye-Tracking Study on the Effects of web Object Placement on Different Types of Websites.” International Journal of Human-Computer Studies 71 (3): 228–235. doi:10.1016/j.ijhcs.2012.09.001.

Seckler, M., K. Opwis, and A. N. Tuch. 2015. “Linking Objective Design Factors with Subjective Aesthetics: An Experimental Study on how Structure and Color of Websites Affect the Facets of Users’ Visual Aesthetic Perception.” Computers in Human Behavior 49: 375–389. doi:10.1016/j.chb.2015.02.056.

Sohn, S., B. Seegebarth, and M. Moritz. 2017. “The Impact of Perceived Visual Complexity of Mobile Online Shops on User’s Satisfaction.” Psychology & Marketing 34: 195–214. doi:10.1002/mar.20983.

Thielsch, M. T., I. Blotenberg, and R. Jaron. 2014. “User Evaluation of Websites: From First Impression to Recommendation.” Interacting with Computers 26: 89– 102. doi:10.1093/iwc/iwt033.

Tractinsky, N., A. Cokhavi, M. Kirschenbaum, and T. Sharfi. 2006. “Evaluating the Consistency of Immediate Aesthetic Perceptions of web Pages.” International Journal of

Human-Computer Studies 64: 1071–1083. doi:10.1016/j. ijhcs.2006.06.009.

Tuch, A. N., J. A. Bargas-Avila, K. Opwis, and F. H. Wilhelm. 2009. “Visual Complexity of Websites: Effects on Users’ Experience, Physiology, Performance, and Memory.” International Journal of Human-Computer Studies 67: 703–715. doi:10.1016/j.ijhcs.2009.04.002.

Tuch, A. N., E. E. Presslaber, M. Stocklin, K. Opwis, and J. A. Bargas-Avila. 2012. “The Role of Visual Complexity and Prototypicality Regarding First Impression of Websites: Working Towards Understanding Aesthetic Judgments.” International Journal of Human-Computer Studies 70: 794–811. doi:10.1016/j.ijhcs.2012.06.003.

Wallace, G. K. 1992. “The JPEG Still Picture Compression Standard.” IEEE Transactions on Consumer Electronics 38: xviii–xxxiv. doi:10.1109/30.125072.

Wu, O., W. Hu, and L. Shi. 2013. “Measuring the Visual Complexities of web Pages.” ACM Transactions on the Web (TWEB) 7 (1): 1. doi:10.1145/2435215.2435216.

Zhang, P., and G. von Dran. 2000. “Satisfiers and Dissatisfiers: A two-Factor Model for Website Design and Evaluation.” Journal of the American Society for Information Science 51: 1253–1268. doi:10.1002/1097-4571(2000)9999:9999<:: AID-ASI1039>3.0.CO;2-O.

510 A. J. KING ET AL.

Copyright of Behaviour & Information Technology is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.