Journal of Interactive Advertising, Volume 5, Number 2,
The recent arrival of interactive messaging/marketing units (IMUs) begs the question: Does interacting with an advertisement enhance its persuasive appeal? How does interactivity compare with other structural features of online ads such as animation and ad shape? A 3 (Interactivity: Low, Medium, High) x 2 (Animation: Animated, Static) x 2 (Ad Shape: Banner, Square) fully-crossed factorial within-participants experiment was conducted to explore these questions. All participants (N = 48) were exposed to 12 news-article Web pages, with one ad in each of them. Results show not only that the level of interactivity is positively associated with ad and product attitudes, but also that it interacts with animation and ad shape in complex ways to influence the persuasion process.
A growing body of research on the social psychology of interactivity, arguably the most distinctive aspect of computer-based media, has been documenting the persuasive influence of technologies that allow media users to interact with media interfaces. In the media equation literature, for example, interactivity is considered a primary reason for users’ social responses to computers (Reeves and Nass 1996). Interactivity serves the function of providing a humanlike cue in the context of human-computer interaction, thereby encouraging the categorization of computers as fellow social actors, and leading to mindless application of social rules and expectations to computers (Nass and Moon 2000). In effect, interactivity imbues the interface with agency, thus encouraging users to treat the computer as a source of communication and not merely as a medium, in contrast to their orientation toward traditional non-interactive communication technologies (such as radio and television) which tend to be treated solely as media or channels (Sundar and Nass 2000).
In the context of mass communication via the Web, interactivity seems to affect users’ perceptions of the visible or proximate source on a Website (Sundar and Nass 2001). Studies with political Web sites have shown that the greater the number of interactive features on the site, the higher the liking of the political candidate featured on the site (Ahern and Stromer-Galley 2000) and the stronger the psychological affinity felt by voters toward the candidate (Sundar et al. 1998), even after controlling for message variables such as level of informativeness. Other studies have shown that increased interactivity contributes to increased feelings of telepresence (Coyle and Thorson 2001), higher involvement with the site (Bucy 2003), and more positive attitudes toward the portal (Kalyanaraman and Sundar 2003), including higher credibility (Fogg 2003).
Clearly, there appears to be a persuasive component to interactivity effects that is largely tested with computers and Web sites, but not yet with online advertisements. Given that a primary, explicit purpose of advertising is persuasion, Web advertisements provide a real testing ground for realizing the persuasive effects of interactivity. However, a literature search yielded no empirical investigations that systematically vary levels of interactivity in online advertisements for experimentally determining their effects. Even though the name of this journal (Journal of Interactive Advertising; www.jiad.org) embodies the concept, its archive does not contain a single study that varies the level of interactivity in advertising. Instead, most of the advertising literature is concerned with users’ perceptions of interactivity (e.g., Liu and Shrum 2002; McMillan and Hwang 2002). This is partly because, as McMillan and Hwang (2002) suggest, the word “interactive” in “interactive advertising” refers to the medium, not the advertisements themselves, and partly because most online ads do not allow users to interact with them.
Although the Web medium is largely interactive, advertising in it is hardly so. For much of the short history of the internet, Web advertising has followed the traditional print-media model of placing ads in strategic locations in hopes of attracting user attention. The most that banner ads do by way of interactivity is invite Web surfers to click on them so that they can be transported to the advertiser’s site. If the click-through rates released by the advertising industry are any indication, such calls for interactivity are a failure. So much so that the Internet Advertising Bureau (www.iab.net) constantly updates guidelines for effective advertising by recommending new formats, including larger sizes, different shapes, and use of so-called rich-media, the “catch-all term for online advertising technologically enhanced by motion, sound, video, or some sort of interactive element” (Koegel 2003, p. 1). But, none of these suggested variations have captured the essence of the interactive nature of the medium as successfully as the interactive messaging/marketing units (IMUs). Launched in February, 2001 on CNET Networks’ News.com site, IMUs are veritable websites residing within an ad. Most IMUs have multiple layers of navigable content, accessible via clickable tabs, rollovers, and hyperlinks embedded in the ads themselves. Clicking on these ads does not simply lead the user to another site but instead refreshes content within the ad space, keeping the rest of the Web page constantly on the screen.
Industry research has touted the “stopping power” of these advertisements, and claimed marked improvements in their ability to promote brand awareness and purchase consideration (Anfuso 2002). But it is unclear which specific aspect(s) of these IMUs contribute to positive outcomes of advertising effectiveness. They use a wide variety of technologies ranging from Flash to DHTML, resulting in a number of aesthetically appealing innovations. As a result, we do not know if the positive reactions claimed by the industry are due to the interactivity of these ads or some other factor(s).
The present investigation is an effort to respond to this concern by systematically varying the levels of interactivity in IMUs, and tracking both the direct effects of interactivity on user attitudes and the combinatory effects of interactivity with animation and ad shape, the two most common variables in IMUs. The next section will describe the proposed relationship between the three independent variables (interactivity, animation, and ad shape) and user attitudes.
Dual process models in social psychology have historically proven useful for studying advertising effects (Chaiken and Trope 1999). The most famous among them in the communication literature, the Elaboration Likelihood Model (ELM), proposes two distinct “routes to persuasion” (Petty and Priester 1994). The “central route” is cognitively intensive in that the message receiver scrutinizes the issue-relevant arguments made by the advertisement. This route is usually followed when the receiver is motivated and able to process the message. Given that most recipients are not particularly motivated to review advertisements, they tend to be miserly with their cognitive resources and process ads through the “peripheral route.” When processed this way, certain cues in the persuasion context (such as the presence of an attractive message source) are likely to influence the persuasibility of the appeal. These cues are often referred to as “peripheral cues,” and advertisers are known to use them time and again to trigger positive, even if simplistic, associations with the advertised product, especially in cases where they lack strong message arguments. Indeed most ingredients of modern-day advertising outside of the central message argument (e.g., color, sound, sex appeal, humor, production quality) are considered peripheral cues because they are seen as encouraging simple associations that may lead to positive persuasive effects while being message-irrelevant.
The variables under study in this investigation—interactivity, animation, and ad shape—all share one common trait in that they refer to non-message or non-argument aspects of the advertisements. If we consider these as peripheral cues, then the rest of the study is really about their ability to aid the persuasion function and the degree to which they can do it, both in isolation and in combination with each other. But first, these variables have to be appropriately defined for the purpose of empirical testing, followed by a directional prognostication of their main effects on attitudinal dependent variables underlying persuasion. We can then begin to propose interaction effects of two or all three of these structural variables.
Interactivity has been variously defined in the literature (see McMillan and Hwang 2002 for a listing of interactivity definitions), but almost all definitions emphasize the importance of interaction between user and system. Sundar, Kalyanaraman, and Brown (2003) review several definitions and classify them into two species — the “functional view” and the “contingency view.” The former is described as the “bells and whistles” approach in that the interface promises several functions (e.g., feedback forms, chat forums, downloads, etc.) that offer rich potential for dialogue or mutual discourse. These functions are specified in terms of particular features (such as audio and video), attributes (such as the presence of choice and control), processes (e.g., reciprocal communication), or outcomes (e.g., user satisfaction). An ad that incorporates many of these functions will, under the functional view, be considered more interactive than a comparable ad with a lesser number of these functions. Such an approach relies on a simple headcount of interactive elements available on the interface rather than considering the depth of interactive communication facilitated by each of the elements. Studies have shown that functional interactivity contributes to positive perceptions of site content due to simple association (e.g., Ahern and Stromer-Galley 2000). Consistent with ELM prediction, apathetic users, as opposed to involved ones, tend to rate Web site content as a direct positive function of site interactivity (Sundar et al. 1998). The higher the amount of interactivity, the greater its value as a peripheral cue it seems, especially in garnering users’ appreciation for its functional design.
The contingency view of interactivity is a more transactional conceptualization, emphasizing the behavioral nature of interaction between user and system. Under this view, interactivity is realized when messages are contingent upon previous messages (and those preceding them) in a threaded manner. Originally proposed by Rafaeli (1988) for conceptualizing interactivity between users in computer-mediated communication, this definition was recently operationalized for Web-based mass communication by Sundar, Kalyanaraman, and Brown (2003) by progressively fragmenting Web site content for different conditions. While the low-interactivity condition featured all the content on one scrollable page, the medium condition presented the same information by allowing four clickable hyperlink options on the main page. In the high-interactivity condition, the content was further fragmented within each of four branches into three clickable hyperlinks. This seemingly simple operationalization not only technically met Rafaeli’s message-dependency specification for interactive communication, but also passed the perceptual test. All three conditions were significantly differentiated in the expected direction when study participants were asked to globally assess the site’s interactivity. This is probably because the operationalization not only represents different levels of message contingency, but also most other attributes normally associated with the concept of interactivity. For example, the high condition gave the user considerably more control and choice, provided a richer sense of feedback and two-way communication, and imbued a greater feeling of system responsiveness and flow — all identified as defining characteristics by various researchers (see Liu and Shrum 2002; McMillan and Hwang 2002).
Fundamentally, both the functional and contingency approaches stress the role of interactivity as one of promoting user engagement with content. And, as Bucy (2003) points out, interactivity gives sites their “stickiness,” or continuing appeal beyond information content, which users have come to routinely expect. In fact, his study participants rated interactive news sites as significantly more “participatory” than non-interactive ones. Kalyanaraman and Sundar (2003) showed that interactivity is related to customization, the idea that each individual user is able to receive his or her own unique combination of online messages and experiences. They showed strong positive correlations between perceived interactivity and perceived relevance of — and involvement with — content, all of which positively predicted attitudes toward the site.
The same can be expected of interactive advertisements as well. To the extent an interactive ad possesses at least some of the formal properties specified by one or more definitions of interactivity, it should be perceived by users as interactive. Furthermore, it should be rated more positively than non-interactive ads due to the aforementioned favorable qualities of interactivity. This, in turn, should promote favorable attitudes toward the ad itself and the content within the ad. Therefore,
H1: The higher the level of ad interactivity, the greater it’s perceived interactivity.
H2: The higher the level of ad interactivity, the more positive the attitudes toward the ad (Aad).
H3: The higher the level of ad interactivity, the more positive the attitudes toward the product.
Research based on the limited-capacity theory of television viewing (e.g., Lang 1990; Lang 1995; Lang, Dhillon, and Dong 1995) also provides a framework for explaining why moving images or animation may elicit greater emotional arousal. According to this theory, certain structural features of television messages, such as cuts, pans, scene changes, pacing, and arousing content automatically elicit orienting responses (OR) in viewers and entail greater involuntary allocation of cognitive resources to encode the message. The end-result of this process typically includes a stronger skin conductance response and a slowing heart rate in viewers (Lang 1994). This heightened state of physiological response in turn mediates the processing as well as evaluation of media messages (Lang, Newhagen, and Reeves 1996; Mundorf and Zillmann 1991).
Another line of research on motion effects concerns the unique feature of animation as a potential attention-getting visual stimulus. According to the “distinctiveness theory,” animation or moving objects “pop-up” in the early stage of visual information processing because their unique and distinctive features distinguish them from the rest of the stimuli in the visual field (Gati and Tversky 1987; Nairne et al. 1997). Consistent with this proposition, researchers have demonstrated the cognitive effects of such distinctive features as color and moving objects (e.g., Beattie and Mitchell 1985; Taylor and Thompson 1982). In general, these features are believed to capture users’ immediate attention and therefore enhance memory of the stimulus objects (Li and Bukovac 1999).
A variant of the distinctiveness theory is the vividness effect. Previous research in psychology suggests that a variety of factors determine the availability of information in memory (e.g., Sherman and Corty 1984; Taylor 1982). Some researchers believe that vividness is the most important of these factors (see Nisbett and Ross 1980). Nisbett and Ross (1980) consider a message to be vivid if it is "emotionally interesting," "concrete and imagery provoking," and "proximate in a sensory, temporal, or spatial way" (p. 45). Vivid information is thus not only more appealing but also more likely to be stored and remembered than non-vivid or pallid information. Vividness is often identified as an important feature of online ads (e.g., Rodgers and Thorson 2000). In the context of the present study, an animated ad may be perceived as more emotionally interesting, imagery provoking, and inherently more appealing than a static ad because of the dynamism conveyed by constant change in content. At least one study has found a positive association between vividness and attitudes (Coyle and Thorson 2001).
A growing body of experimental research employing the theories reviewed above documents the psychological superiority of animated Web ads over static ones. Animated ads have been found to elicit stronger orienting responses (Lang et al. 2002), faster click-throughs (Li and Bukovac 1999), higher arousal (Heo and Sundar 2000a), better memory for ad content (Heo and Sundar 2000b; Lang et al. 2002; Li and Bukovac 1999), and more positive attitudes toward both the ads (Kalyanaraman and Oliver 2001) and the Website (Sundar et al. 1997). As a formal feature of the new medium, animation has proven to be a powerful tool for generating desirable advertising effects. Although much of the research on animation focuses on its obvious attention-getting potential, the theoretical arguments put forth for greater attention toward animated, compared to static, ads speak equally well to their ability to generate more positive perceptions amongst online users. This is because motion, distinctiveness, and vividness are all positive traits for an advertisement. Therefore, we propose
H4: Animated ads will be rated higher in Aad than static ads.
H5: Animated ads will generate more positive attitudes toward the product than static ads.
As indicated earlier, the banner format for online advertising has failed to realize the internet’s potential as a mass advertising vehicle, leading the Internet Advertising Bureau on a seemingly endless search for the right format. Therefore, online ads appear in numerous formats these days, but IMUs usually happen to be in one of two formats—banner and big-box. The latter is a large square ad that is typically placed in the middle of the Web page, either at the center or toward one side, but always with text wrapping around it (henceforth this will be referred to as the “square ad”).
The square ad is relatively novel and therefore more likely to be noticed. The distinctiveness theory discussed earlier may be applied here to posit that square ads stand out in comparison to banner ads and will therefore merit greater user attention and more positive evaluations. The novelty literature from consumer psychology (see Kalyanaraman and Sundar 2003 for a summary) may be invoked to propose that the relatively novel square ad would generate more positive Aad. On the other hand, given their placement on the page, square ads may be considered intrusive and lead to negative attitudes toward the ad and the product (Edwards, Li, and Lee 2002). The paucity of research investigating this variable prompts us to pose a research question instead of proposing a hypothesis:
RQ1: For IMU users, what is the relationship between the shape of the ad (banner vs. square) and user attitudes toward the ads as well as the products advertised?
In addition to investigating the main effects of interactivity, animation, and ad shape, a useful contribution that this study could make is to catalog the combined effects of these variables. As indicated earlier, dual process formulations are particularly helpful in this regard. If indeed interactivity and animation are peripheral cues as hypothesized, then they are likely to aid heuristic, rather than systematic, processing of ad information, to use the language of HSM, another dual-process model in social psychology (Chaiken 1980, 1987). Regardless of which heuristics (novelty, coolness, etc.) are triggered by a particular peripheral attribute, a cumulation of peripheral cues is likely to enhance the accessibility of those heuristics during evaluation of the ad and the product in it. Indeed, Kalyanaraman and Oliver (2001) have suggested, without empirically demonstrating, that multiple peripheral cues would be more powerful than single ones in promoting the persuasive ability of online ads. Given this, we propose the following general prediction for all interactions in this study.
H6: Interactivity, animation, and ad shape will have additive effects on user attitudes toward the ads as well as the products advertised.
All participants (N = 48) in a 3 (Interactivity: Low, Medium, High) X 2 (Animation: Static, Animated) X 2 (Ad Shape: Banner, Square) fully-crossed factorial within-participants experiment were exposed to 12 Web pages containing news articles, with each page containing a stimulus ad that represented a particular combination of values on the three independent variables. They saw one of three different samples of stimulus ads in one of four orders. After browsing through each Web page for a maximum of 90 seconds, they filled out a paper-and-pencil questionnaire eliciting their attitudes toward the ad and the product advertised in it.
Forty-eight undergraduate students (11 males and 37 females) enrolled in communication courses participated in this experiment for course credit. They were randomly assigned to one of three samples and then one of four orders. All participants signed an informed consent form prior to their participation in the study.
Given the 3 X 2 X 2 design, there were 12 different types of online ads to be assembled for the study. Ads were obtained from various sources on the Web and extensively pre-tested with undergraduates in the same institution, who evaluated all the ads using the manipulation-check measures described later in this section. The final set of ads was selected only after each ad met the requirements of a given cell in the factorial design. In order to minimize incidental confounds, we employed stimulus sampling such that we eventually had 3 ads for each of the 12 conditions. In all, 36 ads were assembled for the experiment, promoting a range of products such as computers, cosmetics, soda, a television channel, office equipment, and shoes. Each ad was embedded within a news story on a separate Web page. The ads were split into three samples, with each sample consisting of one instantiation each of all 12 types under study. Within each sample, the 12 Web pages with the ads were administered in one of four randomly determined orders.
Interactivity in the ads was determined by the contingency principle (Rafaeli 1988) discussed earlier, and operationalized in terms of the number of hierarchically hyperlinked layers or levels (Sundar, Kalyanaraman and Brown, 2003) using IMUs. While low-interactivity ads did not have any hyperlinks and therefore just one layer, medium ones had two layers that could be explored by clicking on hyperlinks, and high-interactivity ones featured three or more layers (see Appendix for an example of multiple layers in a high-interactive ad). In any given layer, there may be multiple parallel (not hierarchical) hyperlinks accessible via tabs. In some cases, the final layer consisted of a link to the advertiser’s own site which would pop up over the stimulus page. Half the ads had animation in them while the other half did not. The animated ads had either text or objects that moved or were flashing during or after loading (Sundar and Kalyanaraman 2004). Half the ads were banner ads placed near the top of the page while the other half were square ads placed in the middle of the Web page.
Although ontologically defined based on the hierarchical hyperlinking structure described above, an effort was made to ascertain participants’ perception of interactivity in the ad by embedding the following question in the questionnaire administered after every ad: “On a scale of 1 to 9, with 1 being non-interactive and 9 being very interactive, how interactive would you rate this advertisement?” A similar question was asked to verify the animation manipulation: “On a scale of 1 to 9, with 1 being non-animated and 9 being very animated, how animated would you rate this advertisement?”
The dependent variable for H1, perceived interactivity, was operationalized in terms of three likert-type scaled questions derived from Liu and Shrum (2002): On a scale of 1 to 9, with “1” being “Strongly Disagree” and “9” being “Strongly Agree”, please circle the number that best represents your thought about this advertisement on each sentence (1) Enabled two-way communication; (2) Enabled synchronous communication; and (3) Enabled active control. The three items were averaged to form a “perceived interactivity” index, which was quite reliable (Cronbach’s alpha= .93).
Attitude toward the ad (Aad), a major dependent variable in this study (used for testing H2, H4, and H6), was assessed via a 9-point semantic-differential scale obtained from Burner (1998). The specific items were Like/Dislike, Dynamic/Dull, Interesting/Boring, Favorable/Unfavorable, Unappealing/Appealing, Persuasive/Not Persuasive, Not Eye-catching/Eye-catching, Uninformative/Informative, Pleasing/Irritating, Ordinary/Sophisticated, Well-structured/Badly-structured, Effective/Not effective, Enjoyable/Not Enjoyable, and Bad/Good. All 14 were summed to form the Aad index (Cronbach’s alpha= .97).
Attitude toward the product, used as the dependent variable for testing H3, H5, and H6, was operationalized in terms of three indices: perceived product knowledge, perceived product involvement, and purchase intention. Perceived product knowledge was a 3-item index derived by Li, Daugherty, and Biocca (2002) from the original 4-item scale by Smith and Park (1992) based on reliability considerations. Participants were asked to indicate their level of agreement with the following three assertions on a 9-point scale: (1) I feel very knowledgeable about this product; (2) If I had to purchase this product today, I would need to gather very little information in order to make a wise decision; (3) I feel very confident about my ability to tell the difference in quality among different brands of this product. Scores on these three items were averaged (Cronbach’s alpha = .83). Perceived product involvement was measured with the following six 9-point semantic differential items from Zaichkowsky (1985): Matters to me/Doesn’t matter, Relevant/Irrelevant, Unimportant/Important, Essential/Non-essential, Wanted/Unwanted, Useless/Useful. Responses to these six items were averaged (Cronbach’s alpha = .95). Purchase intention was assessed with a single measure asking, “How likely is it that you would consider buying this product if you had enough money?” followed by a 9-point scale anchored between “unlikely” and “very likely.”
The experiment was administered to participants individually in a laboratory that was equipped with a laptop computer. As participants arrived at the lab, they were greeted by the experimenter. The experimenter told the participants to have a seat in front of the monitor, asked them to read and sign the informed consent form, and directed them to study instructions on the computer screen. They were told that the study in which they were about to participate concerned browsing 12 different Web pages including 12 main advertisements, and that they would be given a maximum of 90 seconds to browse through each page, following which they would fill out a questionnaire before going onto the next Web page. If they finished a particular page before 90 seconds had elapsed, they were asked to notify the experimenter and allowed to move on to the questionnaire.
They were told to view each Web page as they would normally view a page on the internet, but that we were more interested in their reactions to the main advertisement in each page. Given that this study was not about the attention-getting potential of interactivity and animation, but rather about their ability to influence persuasion, we wanted to make sure that our study participants had seen the ads, hence the decision to specifically direct them to the ads. Written instructions requested them to “quickly take in the whole page at the beginning (without clicking on any of the links within the article text) and then proceed to explore the main advertisement on the page and spend most of your 90 seconds finding out as much as you can about the product or service that is advertised.” Participants were asked to feel free to click on the ads themselves and/or the links, if any, featured within those ads. In the event that a link in the ad transported them to a new Web page on a separate window, they were told not to go further. That is, they were asked to just look quickly at the page, then close the new window and return to the Web page with the article and the ad. Before viewing the first experimental page, all participants practiced the procedure with a test page.
A series of 3 (Interactivity) X 2 (Animation) X 2 (Ad Shape) x 3 (Stimulus Sample) x 4 (Order) fully-balanced mixed-design analyses of variance (ANOVA) showed, first of all, that the manipulations were very successful — F (2, 72) = 459.42, p < .0001, omega-squared = .64, for interactivity (with all three levels distinguished at p < .01 according to Tukey-Kramer HSD post-hoc test) and F (1, 36) = 160.54, p < .0001,omega-squared = .14 for animation — in the expected direction.
When the ANOVA was performed with perceived interactivity as the dependent variable, a significant main effect was observed for Animation and Interactivity. Participants reported higher levels of perceived interactivity for animated ads (M = 4.40) than for static ads (M = 3.99), F (1, 36) = 18.14, p < .001, omega-squared = .01. In addition, according to Tukey-Kramer HSD post hoc test, respondents showed a statistically significant difference in participants’ perceived interactivity as a function of the three levels of interactivity, F (2, 72) = 140.30, p < .001, omega-squared = .42. Low-interactive ads were rated the lowest on perceived interactivity (M = 2.09), whereas high-interactive ads were rated significantly higher (M = 5.96), with medium-interactive ones falling in between the two (M = 4.54) on this measure, with all pairs distinguished statistically. Therefore, H1 was supported.
In addition to the main effect, a significant Animation X Interactivity interaction, F (2, 72) = 10.18, p < .001, omega-squared = .01, revealed that while animation serves to enhance the perceived interactivity of a low- or medium-interactive ad, it does not have that effect in a high-interactive ad (see Figure 1).
Figure 1: Estimates of Perceived Interactivity as a Function of Animation and Interactivity
A significant Ad Shape X Animation X Interactivity three-way interaction, F (2, 72) = 3.44, p < .05, omega-squared = .002 showed that while the perceived interactivity of static ads was not affected by ad shape (banner vs. square ads) at all, animated ads tend to increase the level of perceived interactivity for square, rather than, banner ads in low and high interactivity conditions. However, in the medium interactivity condition, perceived interactivity was higher for animated banner ads than for animated square ads (see Figure 2).
Figure 2: Estimates of Perceived Interactivity as a Function of
Ad Shape, Animation, and Interactivity
On Aad, main effects were obtained for Ad Shape, Animation, and Interactivity, thus lending support to H2 and H4. Respondents showed a statistically significant difference in Aad as a function of the three levels of interactivity, F (2, 72) = 174.79, p < .001, omega-squared = .37. The Tukey-Kramer HSD post hoc test revealed that low-interactive ads were rated lowest on Aad (M = 52.65) and high-interactive ones rated significantly more positive (M = 95.19), with medium-interactive ads falling in between—and significantly differentiated from—the two (M = 72.86) on this measure. In addition, participants showed more positive attitudes toward animated ads (M = 77.82) than static ads (M = 69.31), F (1, 36) = 47.81, p < .001, omega-squared = .02, and toward square ads (M = 76.51) than banner ads (M = 70.62), F (1, 36) = 15.49, p < .001, omega-squared = .01.
In addition, a significant Ad Shape X Animation interaction was obtained, F (1, 36) = 14.02, p < .001, ?2 = .01, such that the effect of animation on attitudes toward square ads was stronger than its effect on promoting attitudes toward banners (see Figure 3).
Figure 3: Estimates of Attitude toward Ads as a Function of Ad Shape and Animation
Another two-way interaction, that between Animation X Interactivity, F (2, 72) = 10.18, p < .001, omega-squared = .02, showed a pattern that was identical to the one observed with the previous DV (perceived interactivity). While animation helps promote attitudes toward low- and medium-interactive ads, it does not help or hinder attitudes toward high-interactive ads (see Figure 4).
Figure 4: Estimates of Attitude toward Ads as a Function of Animation and Interactivity
All three independent factors showed significant main effects on perceived product knowledge as well, albeit in a different direction. Participants reported a higher level of perceived product knowledge for banner ads (M = 4.22) than for square ads (M = 3.88), F (1, 36) = 5.73, p < .05, and for static ads (M = 4.23) than for animated ads (M = 3.87), F (1, 36) = 6.40, p < .05. In addition, respondents showed a statistically significant difference in ratings of perceived product knowledge as a function of the three levels of interactivity, F (2, 72) = 75.09, p < .001. Participants reported that the low-interactive ads led to the least level of perceived product knowledge (M = 2.87), significantly lower than medium- (M = 4.21) and high-interactive ads (M = 5.08), with the latter two also distinguished significantly.
A significant Ad Shape X Animation X interactivity three-way interaction effect was observed for perceived product knowledge, F (2, 72) = 3.64, p < .05, omega-squared = .004. Participants did not show any big differences in perceived product knowledge as a function of animation in the low and medium interactivity condition. In the high interactive-ads condition, however, static ads were associated with higher levels of perceived product knowledge for banner ads than for square ads while animated ads did not lead to differential perception of product knowledge for banner versus square ads. That is, the difference between static and animated ads was pronounced only in high-interactive banners (see Figure 5).
Figure 5: Estimates of Perceived Product Knowledge as a Function of
Ad Shape, Animation, and Interactivity
ANOVAs with the product involvement index also yielded significant main effects for all three factors. Participants reported higher level of product involvement for square ads (M = 4.65) than for banner ads (M = 4.36), F (1, 36) = 4.60, p < .05, omega-squared = .003, and for static ads (M = 4.68) than for animated ads (M = 4.33), F (1, 36) = 4.20, p < .05, omega-squared = .004. In addition, respondents showed a statistically significant difference in ratings of product involvement by the three levels of interactivity, F (2, 72) = 53.59, p < .001, omega-squared = .12. Specifically, the Tukey-Kramer HSD post hoc test revealed that participants rated low-interactive ads lowest on product involvement (M = 3.44), high-interactive ads significantly higher (M = 5.31), with medium-interactive ads falling in between the two (M = 4.76), and significantly differentiated from both.
In addition, a significant Ad Shape X Animation interaction effect, F (1, 36) = 8.86, p < .01, omega-squared = .01, showed that participants reported higher levels of involvement with products advertised in square, rather than banner, ads, only when they were static and not when they were animated (see Figure 6).
Figure 6: Estimates of Product Involvement as a Function of Ad Shape and Animation
A significant Ad Shape X Animation X Interactivity three-way interaction effect, F (2, 72) = 3.48, p < .05, = .004 revealed that while participants showed higher product involvement for all static, over all animated, ads in low- and high-interactivity conditions, they did so only for square ads in the medium-interactivity condition, with banner ads going in the opposite direction (see Figure 7).
Figure 7: Estimates of Product Involvement as a Function of
Ad Shape, Animation, and Interactivity
Note: The means for both banner and square ads are virtually the same in the static and animated versions of high-interactive ads; hence the appearance of a single line, when in reality it represents two lines, one superimposed over the other.
Finally, on purchase intention, significant main effects were obtained for Animation and Interactivity. Participants reported higher levels of purchase intention as a result of viewing static ads (M = 4.61) than animated ads (M = 3.98), F (1, 36) = 8.74, p < .01, omega-squared = .01. In addition, low-interactive ads led to the lowest reported level of purchase intention (M = 2.87), while high-interactive ones contributed to a higher level (M = 5.49), with medium-interactive ads somewhere in between (M = 4.52) on this conative measure, F (2, 72) = 78.03, p < .001, omega-squared = .15, with all three interactivity conditions distinguished significantly according to Tukey-Kramer HSD test. There were no significant two-way or three-way interactions on this dependent variable.
Taken together, analyses with the three indices of attitudes toward the product (perceived product knowledge, product involvement, and purchase intention) showed strong support for H3, which predicted a direct, positive, linear effect of interactivity. However, they clearly disconfirmed H5 by showing significant effects for animation in a direction that was counter to prediction. On all three indices, attitudes toward the product were more positive when advertised through static, rather than animated, ads.
As for RQ1, Ad Shape significantly affected Aad such that square ads were rated more positively than banners. On attitudes toward the product, the results were mixed. Square was significantly higher on product involvement but lower on perceived product knowledge, with shape making no difference to purchase intention.
In summary, while ad interactivity uniformly positively affects attitudes, animation effects are not so uniform. Animated ads are rated more interactive and promote better Aad, but they tend to hinder product-related attitudes. Square ads also lead to more positive Aad and greater product involvement than banners, but they detract from perceived product knowledge. The two-way interactions between animation and ad shape reveal additive consequences of these main effects, with animated square ads being most successful in contributing to positive Aad and static square ads doing the same for product involvement. The two-way interactions between animation and interactivity showed that the relative advantage enjoyed by animated over static ads in promoting Aad and perceived interactivity of low- and medium-interactive ads is lost in the case of high-interactive ads. The three-way interaction effect on perceived interactivity (Figure 2) appears to be due to the anomalous performance of medium-interactive animated banners, which lead to higher perceptions of interactivity than medium-interactive animated square ads, reversing the trend found with low- and high-interactive animated ads. The three-way interaction effect on perceived product knowledge (Figure 5) is the result of high-interactive static banners showing significantly higher effect than all the other 11 types of ads. The three-way interaction effect on product involvement (Figure 7) is caused by medium-interactive animated banners showing a stronger effect than their static counterparts, which is in the opposite direction of the animation effect for the remaining five comparisons.
Main effect results overwhelmingly show that interactivity, when attended upon, is a strong cue aiding the persuasive function of online ads, indicating that the operational procedure of classifying interactivity is quite powerful psychologically. This means the contingency-based conceptualization of interactivity is applicable not just to Web sites (as demonstrated by Sundar, Kalyanaraman, and Brown 2003), but also to Web advertisements.
Findings also confirm animation to be an important cue that positively influences attitudes toward the ad but negatively affects product involvement (as well as product knowledge and purchase intention). The latter finding, while unexpected, speaks to the distractive potential of animation. While study participants seemed positively impressed with the ad itself and its design, they apparently felt unable to recall enough product information to pronounce a judgment about it. The perceptual bandwidth argument proposed by Reeves and Nass (2000) may be used to understand this finding. According to this argument, psychologically significant aspects of the interface may result in sensations (leading to perceptions), which compete for the same finite amount of mental effort as the cognitive effort to encode the information presented (see also Lang 2000). Therefore, the experience of having viewed a flashy Web ad is memorable but it comes at the cost of actual memory for product information contained in the ad.
Somewhat the same mechanism appears to be underlying the effect of ad shape. While overall, square seems to be favored over banner ads (lending support to the novelty and distinctiveness arguments over the intrusiveness proposition), the positive result for product involvement alongside the negative finding for product knowledge implies that the square shape is better than the banner format to engage the user but not as effective in conveying product information, again revealing a trade-off between memorability and memory.
The two-way interactions between animation and ad shape on Aad and product involvement echo the directions in the main effect, but more importantly, demonstrate the additive effect hypothesized by H6. However, the interactions between animation and interactivity show that the effects of peripheral cues are not always additive. High interactivity appears to be an overpowering peripheral cue that negates the relative advantage of animated over static ads. This implies the possibility of a threshold or sufficiency, meaning an upper limit beyond which peripheral cues are unlikely to cumulate in their impression-formation effects.
Another possibility is that when combined, the presence of one cue serves to alter user’s perception of another. For example, the three-way interaction with perceived interactivity shows that animation and square shape together enhance user perceptions of interactivity in low-interactive ads, but their contribution to perceived interactivity diminishes progressively in more interactive ads.
User perceptions, especially about the product, are also likely to be influenced by the relative (in)accessibility of ad content at the time of evaluation. For example, the general finding in the three-way interaction with perceived product knowledge is that while animated ads fare generally poorer than their static counterparts, this difference is most pronounced in the case of high-interactive banner ads. It may be argued that this condition is least distracting (while being quite informative) in terms of formal features interfering with the process of acquiring product knowledge from the ad. That is, it strikes a better balance than all other conditions in maximizing information transmission and minimizing distraction.
Indeed, the three-way interaction with product involvement indicates that static ads are better than animated ads for promoting product involvement five times out of six. The only exception is in the case of medium-interactive banner ads wherein animation is positively related to product involvement. This exception is probably due to a familiarity effect, the exact opposite of the novelty effect. Most online ads these days are medium-interactive banners which, when clicked upon, lead the user to the advertiser’s site. What this implies is that particular combinations of peripheral cues may, because of overuse on the Web, show persuasive effects that are distinct from the general pattern and hence are anomalous.
However, when this anomalously positive effect of medium-interactive animated banners on product involvement is pooled together with a similar finding on the dependent variable of perceived interactivity, it is suggestive of the theoretical mechanism by which interactivity influences persuasion. It appears that ad interactivity enhances user involvement with product and hence leads to more positive evaluations. Under the ELM perspective, higher involvement would mean more careful scrutiny of central message arguments. These arguments are typically embedded in the copy of the ad, especially in traditional media. However, one could make the case that the interactive design of the ad is itself a message argument because of the message-based contingency model of interactivity operationalized in this study. Therefore, highly interactive ads not only provide more product information, but also create the necessary flow to enhance user involvement with the product, and thereby positively influence persuasion. Ultimately, the transmission of more information at progressively higher levels of interactivity might be the key behind interactive ads’ ability to show a positive correlation between product involvement and perceived product knowledge. This ability stands in striking contrast to most other formal features (such as animation and ad shape), which tend to be, at least perceptually, more cosmetic and less informational.
The study’s findings should be interpreted with some caveats about design: Ad shape, as a variable, was confounded by position, with banners always appearing on the top and square ads in the middle of the Web page. Although we employed message replication by having three ads representing each condition, we cannot completely rule out incidental confounds in our operationalizations of interactivity and animation. Not all renditions of animations or interactive devices are equal or even similar, so differences in them could be driving some of the results. Next, instructions given to participants preclude us from studying the attention drawn by interactive or animated ads because exposure to ads was imposed rather than assumed or measured in this study. Interactive ads may not be persuasive in reality because they may go unnoticed by users. Another external-validity limitation is the small, college-student sample used in the experiment.
Future academic as well as industry research may strive to resolve some of these limitations on its way to documenting stable attitudinal effects of non-content technological variables such as interactivity. Even though the current study showed uniformly positive effects for interactivity, evidence in the literature suggests that the effect of interactivity upon user perceptions is not always monotonically positive (Bucy 2003). Perhaps this study did not operationalize interactivity to a level that is high enough to cross the threshold beyond which it would have negative effects. Perhaps the advertising context changes the nature of interactivity effects. Given the patently persuasive objective of ads, future research would do well to distinguish situations under which interactivity would be considered a peripheral cue from those under which it would be considered a message argument (or at least the promoter of central processing). One technique by which we may be able to achieve this is by manipulating level of user involvement. Another is to vary the nature and amount of interactivity in the advertisements themselves. Regardless of the method used, it seems important to situate the status of interactivity in dual-process terms—not just because it may help identify what kinds of interactivity under which circumstances promote ad effectiveness, but because it has the potential to enrich academic understanding of theoretical mechanisms governing the role played by interactivity in the process of persuasion.
Ahern, R. Kirkland and Jennifer Stromer-Galley (2000), "The Interaction Effect: An Experimental Study of High and Low Interactivity Political Websites," Paper presented at the meeting of the International Communication Association, Acapulco, June.
Anfuso, Dawn (2002), “Research Shows IMUs Are Working” iMedia Connection (April 4) <http://www.imediaconnection.com/news/560.asp> (accessed on 10/20/2003).
Beattie, Ann E. and Andrew A. Mitchell (1985), "The Relationship between Advertising Recall and Persuasion: An Experimental Investigation," in Psychological Processes and Advertising Effects: Theory, Research, and Application, Linda F. Alwitt and Andrew A. Mitchell, eds,. Hillsdale, NJ: Erlbaum, 129-155.
Bucy, Erik P. (2003), "The Interactivity Paradox: Closer to the News but Confused," in Media Access: Social and Psychological Dimensions of New Technology Use, Erik P. Bucy and John E. Newhagen, eds., Mahwah, NJ: Erlbaum, 47-72.
Burner, Gordon C. (1998), "Standardization and Justification: Do Aad Scales Measure Up?," Journal of Current Issues & Research in Advertising, 20 (1), 1-18.
Chaiken, Shelley (1980), "Heuristic versus Systematic Information Processing and the Use of Source versus Message Cues in Persuasion," Journal of Personality and Social Psychology, 39 (5), 752-766.
——— (1987), "The Heuristic Model of Persuasion," in Social influence: The Ontario Symposium, Vol. 5, Mark. P. Zanna, James. M. Olson, and C. Peter Herman, eds., Hillsdale, NJ: Erlbaum, 3-39.
——— and Yaacov Trope (1999), Dual-Process Theories in Social Psychology. New York: Guildford.
Coyle, James R. and Esther Thorson (2001), "The Effects of Progressive Levels of Interactivity and Vividness in Web Marketing Sites," Journal of Advertising, 30 (3), 65-77.
Detenber, Benjamin H. and Byron Reeves (1996), "A Bio-Informational Theory of Emotion: Motion and Image Size Effects on Viewers," Journal of Communication, 46 (3), 66-84.
Detenber, Benjamin H., Robert F. Simons, and Gary G. Bennett Jr. (1998), "Roll ’em!: The Effects of Picture Motion on Emotional Responses," Journal of Broadcasting & Electronic Media, 42 (1), 113-127.
Edwards, Steven M., Hairong Li, and Joo-Hyun Lee (2002), "Forced Exposure and Psychological Reactance: Antecedents and Consequences of the Perceived Intrusiveness of Pop-up Ads," Journal of Advertising, 21 (3), 83-95.
Fogg, Brian J. (2003), Persuasive Technology: Using Computers to Change What We Think and Do. Boston: Morgan Kaufmann.
Gati, Itamar and Amos Tversky (1987), "Recall of Common and Distinctive Features of Verbal and Pictorial Stimuli," Memory & Cognition, 15 (2), 97-100.
Heo, Nokon and S. Shyam Sundar (2000a), "Emotional Responses to Web Advertising," Paper presented at the meeting of the Association for Education in Journalism and Mass Communication, Phoenix, August.
——— and ——— (2000b), "Visual Orientation and Memory for Web Advertising: A Study of Animation and Position Effects," Paper presented at the meeting of the International Communication Association, Acapulco, June.
Kalyanaraman, Sriram and Mary Beth Oliver (2001), "Technology or Tradition: Exploring Relative Persuasive Appeals of Animation, Endorser Credibility, and Argument Strength in Web Advertising," Paper presented at the meeting of the Association for Education and Journalism and Mass Communication, Washington, DC., August.
Kalyanaraman, Sriram and S. Shyam Sundar (2003), "The Psychological Appeal of Personalized Online Content: An Experimental Investigation of Customized Web Portals," Paper presented at the meeting of the International Communication Association, San Diego, May.
Koegel, Kathryn (2003), “Rich Media: What? Where? Why? A DoubleClick® White Paper” <http://www.doubleclick.com/us/knowledge_central/documents/research/dc_richmedia_0307.pdf> (accessed on 02/05/2005).
Lang, Annie (1990), "Involuntary Attention and Physiological Arousal Evoked by Structural Features and Emotional Content in TV Commercials," Communication Research, 17 (3), 275-299.
——— (1994), "What Can the Heart Tell Us about Thinking?," in Measuring Psychological Responses to Media, Annie Lang, ed., Hillsdale, NJ: Erlbaum, 99-111.
——— (1995), "Defining Audio/Video Redundancy from a Limited-Capacity Information Processing Perspective," Communication Research, 22 (1), 86-115.
——— (2000), "The Limited Capacity Model of Mediated Message Processing," Journal of Communication, 50 (3), 46-67.
———, Jennifer Borse, Kevin Wise, and Prabu David (2002), "Captured by the World Wide Web: Orienting to Structural and Content Features of Computer-Presented Information," Communication Research, 29 (3), 215-245.
Lang, Annie, Kuljinder Dhillon, and Qingwen Dong (1995), "The Effects of Emotional Arousal and Valence on Television Viewers’ Cognitive Capacity and Memory," Journal of Broadcasting & Electronic Media, 39 (3), 313-327.
Lang, Annie, John E. Newhagen, and Byron Reeves (1996), " Negative Video as Structure: Emotion, Attention, Capacity, and Memory," Journal of Broadcasting & Electronic Media, 40 (4), 460-477.
Li, Hairong and Janice L. Bukovac (1999), "Cognitive Impact of Banner Ad Characteristics: An Experimental Study," Journalism & Mass Communication Quarterly, 76 (2), 341-353.
Li, Hairong, Terry Daugherty, and Frank Biocca (2002), "Impact of 3-D Advertising on Product Knowledge, Brand Attitude, and Purchase Intention: The Mediating Role of Presence," Journal of Advertising, 31 (3), 43-57.
Liu, Yuping and L. J. Shrum (2002), "What is Interactivity and is It Always Such a Good Thing? Implications of Definition, Person, and Situation for the Influence of Interactivity on Advertising Effectiveness," Journal of Advertising, 31 (4), 53-64.
McMillan, Sally J. and Jang Sun Hwang (2002), "Measures of Perceived Interactivity: An Exploration of the Role of Direction of Communication, User Control, and Time in Shaping Perceptions of Interactivity," Journal of Advertising, 31 (3), 29-42.
Mundorf, Norbert and Dolf Zillmann (1991), "Effects of Story Sequencing on Affective Reactions to Broadcast News," Journal of Broadcasting & Electronic Media, 35 (2), 197-211.
Nairne, James S., Ian Neath, Matt Serra, and Eunhee Byun (1997), "Positional Distinctiveness and Ratio Rule in Free Recall," Journal of Memory & Language, 37 (3), 155-166.
Nass, Clifford and Youngme Moon (2000), "Machines and Mindlessness: Social Responses to Computers," Journal of Social Issues, 56 (1), 81-103.
Nisbett, Richard E. and Lee Ross (1980), Human Inference: Strategies and Shortcomings of Social Judgment. Englewood Cliffs, NJ: Prentice-Hall.
Petty, Richard E. and Joseph R. Priester (1994), "Mass Media Attitude Change: Implications of the Elaboration Likelihood Model of Persuasion," in Media effects: Advances in Theory and Research, Jennings Bryant and Dolf Zillmann, eds., Hillsdale, NJ: Erlbaum, 91-122.
Rafaeli, Sheizaf (1988), "Interactivity: From New Media to Communication," in Advancing Communication Science: Merging Mass and Interpersonal Processes, Robert Hawkins, John M. Weimann and Suzanne Pingree, eds., Newbury Park, CA: Sage, 124-181.
Reeves, Byron, Annie Lang, Eun Young Kim, and Debroah Tatar (1999), "The Effects of Screen Size and Message Content on Attention and Arousal," Media Psychology, 1 (1), 49-68.
Reeves, Byron and Clifford Nass (1996), The Media Equation: How People Treat Computers, Television, and New Media Like Real People and Places. Palo Alto, CA: Center for the Study of Language and Information.
——— and ——— (2000), "Perceptual bandwidth," Communications of the ACM, 43 (3), 65-70.
Reeves, Byron, Esther Thorson, Michael L. Rothschild, Daniel McDonald, Judith Hirsch, and R. Goldstein (1985), "Attention to Television: Intrastimulus Effects of Movement and Scene Changes on Alpha Variation over Time," International Journal of Neuroscience, 27 (3), 241-255.
Rodgers, Shelly and Esther Thorson (2000), "The Interactive Advertising Model: How people Perceive and Process Interactive ads," Journal of Interactive Advertising 1 (1), <http://www.jiad.org/vol1/no1/rodgers/index.htm> (accessed on 10/10/2004).
Sherman, Steven J. and Eric Corty (1984), "Cognitive Heuristics," in Handbook of Social Cognition, Robert S. Wyer and Thomas K. Srull, eds., Hillsdale, NJ: Erlbaum, 189-286.
Smith, Daniel C. and C. Whan Park (1992), "The Effects of Brand Extensions on Market Share and Advertising Efficiency," Journal of Marketing Research, 29 (3), 296-313.
Sundar, S. Shyam, Kenneth M. Hesser, Sriram Kalyanaraman, and Justin Brown (1998), "The Effect of Website Interactivity on Political Persuasion," Paper presented at the meeting of the International Association for Media and Communication Research, Glasgow, UK, July.
Sundar, S. Shyam and Sriram Kalyanaraman (2004), “Arousal, Memory, and Impression Formation Effects of Animation Speed in Web Advertising,” The Journal of Advertising, 33 (1), 7-17.
———, ———, and Justin Brown (2003), "Explicating Website Interactivity: Impression-Formation Effects in Political Campaign Sites," Communication Research, 30 (1), 30-59.
Sundar, S. Shyam and Clifford Nass (2000), "Source Orientation in Human-Computer Interaction: Programmer, Networker, or Independent Social Actor?," Communication Research, 27 (6), 683-703.
——— and ——— (2001), "Conceptualizing Sources in Online News," Journal of Communication, 51 (1), 52-72.
Sundar, S. Shyam, George Otto, Lisa Pisciotta, and Karen Schlag (1997), "Animation and Priming Effects in Online Advertising," Paper presented at the meeting of the Association for Education in Journalism and Mass Communication, Chicago, August.
Taylor, Shelley E. (1982), "The Availability Bias in Social Perception and Interaction," in Judgment under Uncertainty: Heuristics and Biases, Daniel Kahneman, Paul Slovic and Amos Tversky, eds., New York: Cambridge University Press, 190-200.
——— and Suzanne C. Thompson (1982), "Stalking the Elusive Vividness Effect," Psychological Review, 89 (2), 155-81.
Zaichkowsky, Judith Lynne (1985), "Measuring the Involvement Construct," Journal of Consumer Research, 12 (3), 341-52.
First Layer (This is the main layer of the IMU, which the user will see without clicking on the ad)
Second Layer (Upon clicking on the second of the four tabs, the screen refreshes to produce four sub-tabs under the “Find the Right Car” tab)
Third Layer (Upon clicking “Safety” sub-tab under the “Find the Right Car” tab, the screen refreshes to display safety-related information)
S. Shyam Sundar (Ph.D., Stanford University), is associate professor and co-director of the Media Effects Research Laboratory at the College of Communications in The Pennsylvania State University (http://www.psu.edu/dept/medialab). His research focuses on psychological aspects of technological elements in new media. His recent work investigates the effects of interactivity, navigability, multi-modality, and source attribution (agency) on user responses to Web-based mass communication. Email: [email protected]
Jinhee Kim (MA, University of Alabama), is a Ph.D. candidate at the College of Communications in The Pennsylvania State University.
Copyright © 2005 Journal of Interactive Advertising