Vector of the Visual: the Role of Color in Cross-media Advertising
Campaigns
Joshua Kreuger | 6054897
Master’s Thesis 2014 – 2015
Graduate School of Communication | Master’s Programme Communication Science Track – Persuasive Communication
University of Amsterdam
Supervisor |H.A.M. Voorveld
2 Abstract
The aim of the present study was to examine to what extent inconsistent coloring of cross-media advertising messages leads to enhanced ad recall and brand recall and brand recognition and less favorable attitudinal effects; in addition, it was examined if these effects were separately mediated by the processes of forward-encoding and processing-fluency. It was expected that inconsistent coloring would lead to enhanced ad recall, brand recall and brand recognition, yet less favorable attitude toward the ad and brand attitude in comparison to moderately and highly consistent colored conditions. Results of an experiment with three conditions (N = 79) indicated that there were no significant differences found between the conditions in terms of ad recall, brand recall, brand
recognition and attitude toward the ad and brand attitude. Further elaboration on findings and limitations are provided.
Keywords: Cross-media advertising, coloring consistency
1. Introduction
With the advent of Integrated Marketing Communications (IMC), marketers increasingly aim for coherence between the various marketing instruments at hand to persuade their audience. Though its definition is still far from settled, IMC’s vast adoption is often attributed to alignment of the different marketing instruments conceiving clarity and above all, synergy (Kliatshko, 2008). Synergy is often termed as an end result or effect that exceeds the sum of its parts and applies not only to effects generated between-, but
3 also within the various marketing instruments, as with cross-media advertising. Synergy is stated to be the main reason for the success of cross-media advertising campaigns (Voorveld & Valkenburg, 2015). In cross-media advertising campaigns, consistency between the composing advertising messages is one of the pillars that fund the IMC strategy; it is posited that all messages within an advertising campaign should speak with one unique voice, one message and one goal (Duncan & Caywood 1996).
In spite of the praise of IMC, more and more scholars have started to question the need for consistency sought in IMC (i.e. Voorveld 2011, 2015, Navarro-Bailón, 2012). It is argued that the exposure to a sequence of highly similar messages can yield a wear-out effect, where the viewer assesses follow up messages as ‘more of the same thing’ (Apple, 1971; Yoo, Bang & Kim, 2009). In the small body of literature regarding consistency in cross-media messages it has been concluded that repetition of messages reduces
recognition and recall of message information (Voorveld & Valkenburg, 2015). Yet, it has also been concluded that consistency in messages has a positive influence on
evaluative aspects, like brand attitude and attitude towards the ad (Navarro-Bailón, 2012). It seems two different mechanisms are at work here. Processing-fluency has been named as a possible mechanism favoring the attitudinal effects, as a consistent execution of messages enhances the fluency in which information can be processed. This fluent processing of information in turn leads to a pleasant processing experience (Fransen, Fennis & Pruyn, 2010). The process of forward-encoding, on the other hand, holds that an initial message evokes interest and expectations about a subsequent advertising message, leading to enhanced memory effects (Keller, 1996). Thorough attention for these mechanisms lacks as well, however. In all, it seems much is to be learned about the
4 conundrum that appears to exist between ‘consistency as a guiding principle’ in IMC on the one hand and the mere repetition of messages in cross-media advertising campaigns on the other.
Where previous research has found benefits of consistently executed ads over repeated ad executions in terms of brand attitude (e.g. Navarro-Bailón, 2012; Yoo, Bang & Kim, 2009: Voorveld & Valkenburg, 2015), others find detrimental effects of
consistency on recall and recognition (Voorveld & Valkenburg, 2015). However, these previous studies examined consistency mainly on the ad execution of textual content (e.g. Yoo, Bang & Kim, 2009; Navarro-Bailón, 2012) and brand associations (e.g. Torn & Dahlén, 2008). Visual elements as color, on the other hand, have been examined very scarcely while it is stated to be an essential part of advertising processing (Messaris, 1998). Color theme has been deemed as one of the most prominent visual aspects in advertising (McInnis, Moorman & Jaworski, 1991) and thus is an important domain for the examination of visual consistency.
The present paper will make an effort to gain knowledge on this topic, by
extending and contributing to prior research on IMC and message consistency in general (e.g. Yoo, Bang & Kim, 2009; Fill, 2001; Nordhielm, 2002; Navarro-Bailón, 2010) and in specific research by Voorveld and colleagues examining cue consistency (2011 & 2015) in cross-media advertising campaigns. As examined in these studies, effects of cue consistency (also termed ‘fit’) on evaluative and memory responses will be measured on the aspect of coloring (in)consistency. Above all, the present study will focus on the role of coloring (in)consistency in cross-media advertising campaigns. Consistency of the visual cue of color in cross-media advertising has (to my knowledge) not been studied
5 before. Above all, this study will regard to two possible mechanisms evoking the
contrasting findings in memory and attitudinal effects, to give insight in the underlying mechanisms; The proposed focus will allow for deeper understanding of cue
(in)consistency in cross-media advertising campaigns on evaluative and memory responses.
The results of this study may provide practitioners with knowledge on the right balance of visual cue (in)consistency to enhance ad and brand evaluations, while retaining optimum memory effects in order to maximize their advertising efforts. Findings can accordingly indicate what degree of cue (in)consistency is suitable in respect to the advertising objective (i.e. higher brand evaluations vs. higher recognition effects).
2. Theoretical Framework
2.1 The premises of consistency in IMC
Much turbulence has dictated the academic discourse regarding the phenomenon of Integrated Marketing Communications (IMC), with scholars predicating IMC to be the major communications development of the 20th century (i.e. Kitchen & Schultz, 1999) while others deemed it just another recurring marketing hype (i.e. Spotts & Lambert & Joyce, 1999). Though IMC has outgrown its infancy, having gained recognition among more and more scholars, its definition yet is far from solid (Kitchen, Brignell & Li, 2005). One of the earliest definitions is that of Schultz (1993), who has defined IMC as:
‘a concept of marketing communications planning that recognizes the added value of a comprehensive plan that evaluates the strategic roles of a variety of
6 communications disciplines (for example, general advertising, direct response, sales promotion, and public relations) and combines these disciplines to provide clarity, consistency, and maximum communications impact’ (p. 17).
This definition has an inside-out stance taking the business as the point of view (Kitchen, Brignell & Li, 2004). More recent definitions, however, point to the influence of
consumer-driven input. Duncan (2002), for instance, states that:
IMC is a process for managing the customer relationships that drive brand value. More specifically, it is a crossfunctional process for creating and nourishing profitable relationships with customers and other stakeholders by strategically controlling or influencing all messages sent to these groups and encouraging data-driven, purposeful dialogue with them. (p. 7)
Also Percy (2014) emphasizes the focus on the consumer and doubts Schultz’ definition. The author states:
…IMC is really all about planning in order to deliver a consistent message. Effective IMC should certainly encourage strong customer relationships, but it does that through effective planning in order to develop an integrated
communication program that will optimize certain specific communication objectives that lead to desired behaviors on the part of a target audience. (p. 6)
As it appears, consensus is yet to be reached. A rationale can be derived from the various definitions nonetheless.
7 orchestrate the communications of the various marketing tools (Public Relations,
advertising, sales forces etc.). The occupations practiced within each of the marketing tools are said to have diverged as a result of a specialization trend within the marketing occupations during the 20th century (Fill, 2001). Fill (2001) argues that specialization has subsequently led the various marketing occupations to internally but also externally communicate in their own and distinctive ways. In turn, consumers and stakeholders are left with dissonance regarding what an organization is about at large. IMC is said to have evolved in the urge to respond to this predicament; IMC counters the dissonance by implying consistency among the various marketing tools (Fill, 2001) and creating synergy (Voorveld, Neijens & Smit, 2011; Fill, 2001; Percy, 2014).
IMC’s aspiration for consistency also applies within the various marketing tools, though this does not imply that companies should repeatedly use the same type of
medium in the practice of, for instance, advertising. Research has shown that the effort of using different types of media for a single advertising campaign is more rewarding than the use of a single medium (Voorveld, Neijens & Smit, 2011; Bronner 2006).
Consistency between advertising messages in different types of media is often termed as fit, similarity, tactical integration or strategic consistency and defines the extent to which advertising messages within a cross-media campaign are similar and congruent (Bronner, 2006). Advertising messages thus have a high degree of fit when visual and verbal ad executions, also referred to as retrieval cues, are used recursively in the utilized media and facilitate a consistent and recognizable brand image (Keller, 1996).
8 In the domain of IMC, visual aspects of advertising seem especially worth
examining. Visual retrieval cues in IMC are likely to remain consistent across and within the marketing tools, whereas verbal and textual information logically alters according with the meaning expressed in any new message. In other words, visual cues are more solid than textual cues across cross-media advertising messages and are key identification aspects (Keller, 1987).
Unsurprisingly, color has been defined as one of IMC’s key visual cues that dictate consistency within advertising campaigns (Fill, 2001). MacInnis, Moorman and Jaworski (1991) argue that color is one of the most prominent visual cues used in advertising and enhances the motivation to pay attention to an advertising message; advertisers often make use of bright colors to attract attention. Yet, little academic attention has been granted to the ad execution of color and consistency of its use in
cross-media advertising.
2.3 Defining (in)consistency for cross-media campaigns and memory effects Since media campaigns are composed of more than one medium, cross-media campaigns demand a different approach to the definition of consistency than single-medium campaigns. Cross-media campaigns, unlike single-medium advertising campaigns, make use of different media modalities and paces of information flow (Dijkstra, Buijtels & van Raaij, 2005). A TV-commercial, for instance, generally uses both visual and auditory cues while information flow is externally paced (the sender controls the time of exposure to the content), while an Internet website generally uses visual and textual cues while information flow is internally paced (the viewer controls the time of exposure to the content). This inherently means that cross-media campaigns can
9 at most have a substantially high degree of fit of advertising cues due to the differences in modalities between the advertising messages, whereas with single-medium campaigns mere repetition of messages can be chosen. The latter is arguably an extreme or even perfect degree of fit or consistency.
Such an extreme degree of fit or consistency, though enhancing a consistent and recognizable image (Keller, 1996), is not necessarily beneficial for all the important advertising purposes. Within single medium advertising campaigns it has been shown that a wear-out effect can occur with the mere repetition of advertising messages, where every subsequent exposure is decreasingly effective in terms of recall and recognition (Moorman, Neijens & Smit, 2011; Nordhielm, 2002). This effect, also termed tedium, arises as repeated advertising messages lose their novelty to the viewer causing boredom and/or irritation. It has therefore been argued that ad executions should be slightly altered over multiple advertisements to induce some sort of novelty to enhance effectiveness (Burnkrant & Unnava, 1987).
Little is known about wear-out or tedium for cross-media campaigns, however. This lack of knowledge might be caused by the principle that mere repetition of
advertising messages in cross-media campaigns is less probable; cross-media campaigns are composed of advertisements in different types of media, which have different
characteristics or modalities. These modalities make that each different medium forces a different type of ad execution. These different type of ad executions, dictated by the media modalities, inherently induce some sort of novelty when a viewer is exposed to the different advertising messages in the various media used in a cross-media campaign (Fransen, Fennis & Pruyn, 2010). This may also possibly delay a wear-out effect or
10 tedium to occur and is probably one the key strengths of cross-media campaigns.
Empirical evidence on signs of wear-out in cross-media advertising is starting to grow, however. In recent research, it has been observed that advertising campaigns with a higher consistency indeed seem to be remembered less often than inconsistent advertising campaigns (Voorveld & Valkenburg, 2015). This observation suggests that inconsistency in advertising campaigns can be effective if recall or recognition of the campaign is the goal. If inconsistency is indeed effective for memory constructs, the underlying
mechanism may be found the forward-encoding process (Keller, 1987). This cognitive process holds that an initial exposure to an ad in one medium will set expectations and curiosity for a subsequent exposure to a related ad in another medium. When the
subsequent ad defies expectations (as it may do with inconsistent coloring) this may lead to deeper processing. Forward-encoding further posits that with the repetition of
messages, every subsequent exposure enhances a memory trace with a smaller amount than the preceding message, whereas differently executed messages leave multiple and thus stronger memory traces (Keller, 1987).
Based on the foregoing reasoning, it can be expected that the inconsistent use of color enhances ad recall and subsequently brand recall and brand recognition. The
process of forward-encoding can be expected to mediate this process, based on theorizing by Keller (1987 & 1996). Hence the first three hypotheses are proposed:
H1: (a) A sequence of cross-media advertising messages that is inconsistently colored, yields significantly better ad recall than a sequence of advertising messages that is
11 H2: (a) A sequence of cross-media advertising messages that is inconsistently colored, yields significantly better brand recognition than a sequence of advertising messages that is consistently colored and (b) this effect is mediated through the forward-encoding process
H3: (a) A sequence of cross-media advertising messages that is inconsistently colored, yields significantly better brand recall than a sequence of advertising messages that is consistently colored and (b) this effect is mediated through the forward-encoding process
2.4 Effect of (in)consistency on ad and brand attitudes
Whereas inconsistency has been found to enhance recall and recognition,
attitudinal effects seem to suffer when inconsistency increases; ad evaluations have been shown to be less favorable when ad executions are inconsistent (Dahlén, 2005; Voorveld & Valkenburg, 2015; Navarro-Bailón, 2012). Though inconsistency thus seems to negatively influence attitudinal effects, the mere repetition of messages has also been found to be negatively influence attitudes as with memory effects. Similarities with (in)consistency in cross-media campaigns can be found in studies regarding context effects, where advertisements in regard to their media context (the medium or other advertisements enclosing another advertisement) are examined. In the studies of context effects, coherence between messages and its context is found to ease processing (Lord, Burnkrant & Unnava, 2001). Research on context effects further suggests that viewers this ease of processing positively influences evaluations of advertisements (Maoz & Tybout, 2002).
12 information processing-fluency. The theory of processing-fluency states that sequential information that is consistent with prior exposures is processed more easily. As a result, this more fluent processing of the information is experienced as pleasant and provides a hedonic cue to the message receiver (Fransen, Fennis & Pruyn, 2005). In turn, the
message will be assessed more favorably, as a result of this process (Lee & Aaker, 2004). Processing-fluency is thus essentially an affective mechanism, whereas forward-encoding is cognitive.
Color, being a prominent ad execution, is likely to be noticed in the processing of cross-media advertising messages. Therefore, it can be expected that inconsistent use of color can have a detrimental effect on processing-fluency and subsequently on ad evaluations. Moreover, inconsistent use of color, being a key visual cue, runs against IMC’s main principle of visual consistency. Based on this reasoning it can be expected that, as a result of hindering the processing-fluency, inconsistent use of color is
detrimental for ad evaluations and thus is a mediator in the relationship between (in)consistent coloring and the attitudinal constructs. Thus the following hypothesis is proposed:
H4: (a) A sequence of cross-media advertising messages that is inconsistently colored, yields significantly less favorable attitude toward the ad than a sequence of advertising messages that is consistently colored (b) and is mediated through processing-fluency
Although scarcely examined, brand attitude has also been found to be subject to inconsistency in ad executions (Navarro-Bailón, 2012). Though the academic ground for an expectation on this matter is small, attitude toward the ad has consistently been found
13 to positively influence brand attitude under various conditions (Spears & Singh, 2004; Gardner, 1985). Hence the following hypothesis is proposed:
H5: (a) A sequence of cross-media advertising messages that is inconsistently colored yields significantly less favorable brand attitude than a sequence of advertising messages that is consistently colored and (b) is mediated through processing-fluency.
Based on the hypotheses formulated a conceptual model is proposed (see figure 1).
Figure 1. Conceptual model
3. Method
14 In order to isolate the effect of (in)consistent color use in a cross-media
advertising campaign, an experiment was conducted. A three condition (highly consistent colored website, moderately consistent colored website and inconsistent colored
advertising messages) between-subject design was chosen. The cross-media condition was simulated with a TV-commercial followed by exposure to a combination of webpage advertisements related to the brand and product. A TV-commercial and a combination of webpage advertisements/information about the brand were chosen as these media are often utilized in cross-media campaigns and, above all, are composed by different media modalities (Fransen, Fennis & Pruyn, 2010). The TV-commercial is characterized by audiovisual, and to a lesser degree textual modalities. The webpage advertisement and information, on the other hand, are characterized by only visual and textual modalities (Moorman, Neijens & Smit, 2010). The website was manipulated to have highly consistent coloring, moderately consistent coloring and inconsistently coloring in regard to the TV-commercial.
3.2 Sample
A total of N = 98 respondents participated in the present study. Of these 98
respondents 79 respondents completed the questionnaire, of which the average age was M = 24.39 (SD = 4.82) and 57% were female. In terms of education the sample consisted mostly out of respondents that either attained University Bachelor’s degree (40.5%) or a University Master’s degree (39.5%), whilst a minority either finished college (12.7%) or high school (7.6%). None of the respondents were native English speakers. Random designation to conditions yielded N = 25 respondents for the highly consistent colored condition, N = 22 respondents for the inconsistent condition and N = 32 for the
15 moderately consistently colored condition. Respondents took around 8.5 minutes on average filling in the survey (M = 501.15 seconds, SD = 217.26), which is similar to what was expected based on pretesting.
3.3 Experimental stimuli
To simulate a cross-media advertising campaign, a colorful TV-commercial of a yogurt company not operating in the Netherlands was first selected, namely Yoplait. An unknown brand was chosen to forsake the possible influence of pre-experimental bias regarding the examined brand, as this might affect evaluative effects and the depth of processing (Blair and Innis, 1996). The product of yogurt was chosen as it is a common product found in many households. This diminishes the possibility for product novelty to affect the examined constructs (Rossiter, Percy & Donovan, 1992). Additional
requirements for the TV-commercial were that one color is prominently used to exclude the possibility that the brand was to be associated with multiple colors. Additionally, the color had to be present during most of the playtime to favor the chance the color was salient. The TV-commercial that met the requirements had a highly prominent use of purple throughout most of the advertisement and had duration of 30 seconds (see appendix A).
The three versions of the website were professionally created by a Dutch
advertising company and each comprised two webpages (see appendices B, C & D). The websites contained a webpage about the product and a webpage about the company. The websites were equal on all aspects, except the examined ad execution of color. The consistently colored website was prominently colored with the same color purple as used in the TV-commercial by having most of the background, product depiction and Yoplait
16 logo consistently colored accordingly. The neutral website was limited to purple in the Yoplait-logo, website menu and the depiction of the product. The inconsistent colored website made prominent use of color in the same degree as the consistent condition, albeit in the color yellow. Yellow was chosen, as this color is the direct opposite of the color purple in the traditional color scheme. Opposite colors are known to be vibrant when combined and are often used to make visual elements stand out (Nassau, 1997).
3.4 Procedure
Respondents were invited through social media and ‘request for respondent’ pages. Subsequently, the respondents were directed to the online survey platform of Qualtrics and were presented the informed consent to participate in the study. First the respondents were asked for demographics. Respondents were then shown the
TV-commercial and asked to go to the two page website afterwards and requested to surf the website as they would usually do. Afterwards, the respondents were asked to complete the survey. The respondents were subsequently exposed to the items that measured the constructs of ad recall, brand recall, brand recognition, attitude toward the ad and brand attitude respectively.
3.5 Measures
3.5.1 Ad recall
Ad recall was measured for the TV-commercial as well as for the information on the website. Respondents were asked whether they could indicate product advertised, the ingredient mentioned and what Yoplait did with the ingredient in an open question, in
17 accordance with recall methodology by Keller (1987). Each correct answer was given a score of 1 and a wrong answer a score of 0 so that a score of 3 would indicate a
maximum ad recall and a score of 0 no ad recall. Respondents scored an average of more than 2 correct answers, (M = 2.15, SD = 0.96).
3.5.2 Brand recall
Brand recall was measured by requesting respondents to name the brand of the ad they just were exposed to. A correct answer would be given a score of 1 while a wrong answer would yield a 0, making brand recall a dichotomous variable, in accordance with recall methodology by Keller (1987). Misspelling was tolerated to the degree that the pronunciation would still replicate the name ‘Yoplait’. Examples of ‘Yoplaid’ or
‘Yoplade’ were accounted as correct answers. Respondents recalled the brand correctly in 49.4% of the cases (N = 39).
3.5.3 Brand recognition
Brand recognition was measured by proposing a list of eight real brand names, to which the respondents could tick one answer, in accordance with brand recognition methodology by Chandon (2003). A correct answer would be given a score of 1 while a wrong answer would yield a 0, making it a dichotomous variable. All but one of the respondents correctly recognized the brand, (N = 78).
3.5.4 Attitude toward the ad
Attitude toward the ad was examined for both the TV-commercial and the website and measured using the same items as Voorveld, Neijens & Smit (2011) and Chang &
18 Thorson (2004). These included two times four (for the TV-commercial and the webpage advertisement) 7-point bipolar items, being likeable/not likeable, interesting/ not
interesting, good/bad and appealing/not appealing.
A principal component analysis of attitude toward the TV-commercial indicated that one component was extracted, with an eigen value of 3.41, explaining 85% of the variance. An additional reliability analysis showed that the four items composed a very reliable scale, Cronbach’s α = 0.94. Attitude toward the TV-commercial was reasonable on average (M = 4.68, SD = 1.42).
For the construct of attitude toward the webpage advertisement, a principal component analysis indicated that one component was extracted, with an eigen value of 3.40 and explaining 85% of the variance. The subsequent reliability analysis indicated that attitude toward the webpage advertisement was a very reliable scale with Cronbach’s α = 0.94. Attitude toward the webpage advertisement was reasonable on average (M = 4.13, SD = 1.32). The mean scores on both constructs were then combined to form the overarching attitude toward the ad construct. Attitude toward the ad was, logically reasonable on average (M = 4.40, SD = 1.16).
3.5.5 Brand attitude
Attitude toward the brand was measured using the same items applied for attitude toward the ad, adjusted to the brand attitude construct. A principal component analysis extracted one component with an eigen value of 3.37, explaining 83% of the variance. A subsequent reliability analysis indicated that the scale was very reliable with a
19 3.5.6 Forward-encoding process
The forward-encoding process was examined using the same items used by Voorveld, Neijens & Smit (2011), namely ‘I became more interested in the second ad after seeing the first ad’; ‘I wanted to know more about the brand after seeing the first ad’; and ‘I wanted more information about the brand after seeing the first ad’. Responses were measured on a five-point Likert scale, ranging from ‘totally disagree’ (1) to ‘totally agree’ (5). A principal component analysis indicated that one component was extracted with an eigen value of 1.64, explaining 82% of the variance. The subsequent reliability analysis showed that the construct was reliable with Cronbach’s α = 0.78. Forward-encoding was scored an M = 3.80 (SD = 1.59) on average.
3.5.7 Processing-fluency
Processing-fluency was measured using the two-item construct proposed by Lee and Aaker (2004) for both the TV-commercial and webpage advertisement separately. The construct is measured on a seven-point Likert scale and is comprised by the items ‘The information was-‘ with the answer category of (1) ‘difficult to process’ to ‘Easy to process’ (7) and the item ‘ the information was- (1) ‘easy to comprehend’ and (7) ‘difficult to comprehend’. For the TV commercial, a principal component analysis extracted one component, with an eigen value of 1.94, explaining 97% of the variance. A subsequent reliability test showed that the construct was very reliable, Cronbach’s α = 0.97. For the webpage advertisement, a principal component analysis yielded one item with an eigen value of 1.92, explaining 96% of the variance. A subsequent reliability analysis showed that the scale was very reliable, Cronbach’s α = 0.96. The mean scores
20 on both the constructs were combined subsequently to one an overarching construct measuring processing-fluency, which the respondents scored M = 4.95 (SD = 1.05) on average.
3.5.8 Control variables
Advertising skepticism is posed as a possible confounding variable and thus was tested in the present study. Advertising skepticism was measured using an 8-item scale proposed by Obermiller & Spangenberg (1998), which presented eight statements about advertising in general with a 7-point Likert scale for disagree/agree. Though a principal component analysis extracted two components, all of the items loaded higher than the rule of thumb item load of 0.45 on the first component, with an eigen value 5.13,
explaining 57% of the variance. Hence the full 8-item construct was applied. A reliability test showed that the scale was very reliable, Cronbach’s α = 0.90. Respondents scored an average of M = 4.59 (SD = 0.99) on advertising skepticism.
Respondents were examined for ‘color ‘affection’, on which respondents were asked to rate their liking of the eight main colors in the color scheme (Nassau, 1997) on a 7-point Likert scale. Color affection was examined to examine the possibility if the examined memory and attitudinal constructs correlated significantly with affection for the colors used in the experimental stimuli.
4. Results
421 Prior to the main hypothesis testing, the examined constructs of ad recall, brand recall, brand recognition, attitude toward the ad and brand attitude were examined for possible correlation with possible confounding factors advertising skepticism and color affection. The one and only covariation to be found was that between the construct of advertising skepticism and attitude toward the ad, which will be obtained as a covariate in the testing of the H4 that regards attitude towards the ad, r = -0.287, p = 0.01. Above all, pre-experimental awareness of the brand Yoplait did not correlate significantly with brand attitude. Brand awareness was therefore not included as a covariate in H5 that measures brand attitude.
4.2 Hypothesis testing
4.2.1.1 H1a - Ad recall
The first hypothesis held that respondents in the inconsistently colored condition would have a significantly higher ad recall than respondents in the consistently or very consistently colored condition. To examine whether any significant difference existed between the conditions an ANOVA was conducted, in which the (in)consistent coloring conditions were the independent variable and ad recall the dependent variable. Levene’s test of homogeneity of variances indicated that the variances between groups could be regarded as equal across conditions, F (2, 75) = 0.303, p = 0.739. Counter to the
hypothesis no significant difference was found between conditions, F (2, 75) = 0.979, p = 0.380; respondents in the inconsistent condition (M = 1.95, SD = 1.05) did not score significantly different from respondents in the highly consistent condition (M = 2.12, SD
22 = 0.93) or the moderately consistently colored condition (M = 2.32, SD = 0.91) on ad recall; H1a was therefore rejected.
4.2.1.2 H1b - Mediation of ad recall through forward-encoding
For hypothesis 1b, posing the expected mediation role of forward-encoding between the (in)consistent coloring and ad recall, the regression-based mediation path analysis MEDIATE developed by Hayes & Preacher (2014) was applied. Indirect is one of several mediation macro’s developed by Hayes to examine indirect effects. For the regression purposes of the model, the highly consistent coloring was set as a reference variable, categorizing the moderately consistent and inconsistent condition as two
dummy variables. The indirect effect of forward-encoding was found to not be significant for either of the dummy variables in reference to the highly consistent colored reference condition, ß = 0.10, p = 0.983, (95% CI {-0.0835, 0.0786} with 5000 resamples) for the inconsistently colored condition and ß = 0.169, p = 0.709 (95% CI {-0.0352, 0.1312} with 5000 resamples) for the moderately consistent colored condition. Forward-encoding could thus be concluded not to mediate ad recall in this study; therefore, H2b was
rejected.
4.2.2 H2 - Brand recognition and mediation through forward-encoding
The second hypothesis regarded (in)consistent coloring in relation to brand recognition. All but one the respondents recognized the correct brand (N = 78), thereby excluding the possibility for any effect to be found. Further analysis of H2a and H2b were omitted accordingly.
23 4.2.3.1 H3a - Brand recall
For the examination of H3a and H3b, a binary logistic regression analysis was performed as brand recognition was measured on a correct/incorrect answer category, making the dependent variable dichotomous. The brand was recognized by around half (N = 39) of all respondents (N = 79). The regression model, however, was observed not to be significant, χ2(2) = 1.032, p = 0.597, indicating that (in)consistent coloring had no significant effect on brand recognition in this study. Hence, hypothesis 3a was rejected.
4.2.3.2 H3b - Mediation of brand recall through forward-encoding
A suited mediation analysis named INDIRECT was chosen (Preacher & Hayes, 2008), which allows for dichotomous dependent variable outcomes. The highly consistent condition was set as the category variable. The INDIRECT macro indicated that both the mediating path from the independent variable of (in)consistent coloring to the mediating variable of forward-encoding and the mediating path from the latter construct to the outcome variable of ad recall were not significant, ß = 0.078, p = 0.717 and ß = 0.136, p = 0.364 respectively, (95% CI of total indirect effect {-0.0441, 0.1851} with 5000 resamples). Hence, H3b was also rejected.
4.2.4.1 H4a - Attitude toward the ad
The fourth hypothesis posed that respondents in the inconsistently colored
condition would have a significantly lower attitude toward the ad than respondents in the consistent and highly consistently conditions. As advertising skepticism was found to be a covariate to attitude toward the ad, this variable was included as a covariate in the
24 ANCOVA performed. Levene’s test for homogeneity indicated that the variances could be assumed equal across conditions, F (2, 76) = 0.128, p = 0.880. The ANCOVA indicated that no significant differences were observed between the conditions.
Respondents in the highly consistent colored condition (N = 25) scored lowest on attitude toward the ad (M = 4.18, SD = 0.23), whilst the inconsistently colored condition (N = 22) scored highest on attitude toward the ad (M = 4.55, SD = 0.24). Further examination indicated that, while controlling for advertising skepticism, the (in)consistent coloring of the advertising messages did not yield a significant effect on attitude toward the ad, F (2, 76) = 0.487, p = 0.617. Thus, H4a was rejected.
4.2.4.2 H4b - Mediation of attitude toward the ad through processing-fluency
In similar conduct as H1, the mediation analysis method MEDIATE was used to examine the expected mediating effect of processing-fluency between (in)consistent coloring and attitude towards the ad. As advertising skepticism was found to correlate significantly with attitude toward the ad, the variable was included as a covariate in the analysis. The MEDIATE analysis indicated that the reference condition of highly
consistent coloring did not significantly differ in terms of processing-fluency’s mediation in comparison to the inconsistent and moderately consistent conditions, ß = -0.27, p = 0.391 (95% CI {-0.4944, 0.1660} with 5000 resamples) and ß = 0.111, p = 0.707 (95% CI {-0.2833, 0.4228} with 5000 resamples) respectively. H4b was therefore rejected.
4.2.5.1 H5a - Brand attitude
The fifth and final hypothesis regarded brand attitude. Again, an ANOVA was conducted to examine whether any significant difference was found between the
25 conditions. Levene’s test for the homogeneity indicated that the variances across
conditions could be assumed equal F (2, 76) = 2.878, p = 0.062. The ANOVA conducted yielded no significant difference in brand attitude across conditions, F (2, 76) = 0.983, p = 0.379. Respondents in the highly consistent colored condition (N = 25) were found to have the lowest brand attitude (M = 4,60, SD = 1,37), whilst respondents in the
moderately consistent condition (N = 32) were observed to score highest on brand attitude (M = 4,99, SD = 1,16). Additionally, respondents the inconsistently colored condition (N = 22) scored similar to the moderately consistent condition (M = 4,97, SD = 0,89), yet these differences were, as stated, not significant. (In)consistent coloring of advertising messages was not found to be of effect on brand attitude in this study. Thus, hypothesis H5a was rejected.
4.2.5.2 H5b – Mediation of brand attitude through processing fluency
Hypothesis 5b proposed that brand attitude would be mediated through the variable of processing-fluency as a result of (in)consistent coloring in advertising
messages. Again the MEDIATE method (Hayes, 2014) was applied as brand attitude was measured on a 7-points Likert scale, allowing it to be assumed continuous. With the highly consistent condition used as a reference variable, processing-fluency was found not to significantly mediate the effect between brand attitude than in comparison to the (in)consistent condition, ß = -0.283, p = 0.363 (95% CI {-0.4736, 0.1064} with 5000 resamples) and moderately consistent condition ß = 0.123, p = 0.676 (95% CI {-0.2008, 0.3804} with 5000 resamples) respectively. Hypothesis 5b was therefore rejected.
26 5.1 Conclusion
The aim of the current study was to examine the role of (in)consistent color use in cross-media advertising messages by means of an experiment. More specifically, it was thought that inconsistent color use in cross-media advertising messages would result in enhanced memory effects while leading to lower evaluative effects in comparison to consistently colored sequences of advertising messages. For the cognitive constructs of memory, the reasoning was that an inconsistently colored sequence of cross-media advertisements would lead to a more prominent and attention attracting combination as a result of visual incongruence, leaving stronger memory traces. This enhancement of the memory effects was expected to work through the underlying mechanism forward-encoding process, which holds that attention to the first message will enhance attention for the second message (Keller, 1996). Memory effects were not found to be different for respondents in the inconsistently colored condition in comparison to the consistently colored conditions, indicating that the coloring did not have significant effects on
memory. It was further observed that the mechanism of forward encoding did not explain differences in memory across conditions.
In addition affective, attitudinal effects were expected to suffer from
inconsistency in coloring of cross-media advertising messages, being explained by the fluency in which information is processed (Fransen, Fennis & Pruyn, 2010). It was that thought that disruption in information, was caused by inconsistent coloring, would lead to less favorable attitudes toward the brand and advertisements, yet this was not the case as in the present study.
27 These observations do not affirm to previous findings suggesting a relationship between inconsistency in advertising messages and memory constructs (Stammerjohan et al., 2005; Voorveld & Valkenburg, 2015). The reasoning that an inconsistently colored sequence of advertising messages is more prominent and subsequently leads to stronger recall and recognition than consistently colored is not found the present study. In addition, the expected mediating effect of forward-encoding was also not observed in the present research, though this hypothesis was of a rather exploratory nature and thus does not
counter previous literature.
Inconsistent color use was also not found to yield significantly lower attitudinal effects. Specifically, brand attitude and attitude toward the ad were not significantly different in the consistently colored conditions. This finding is not in line with previous studies (Navarro-Bailón, 2012; Fransen, Fennis & Pruyn, 2010; Voorveld & Valkenburg, 2015), which suggest that one of pillars for the formation of synergistic effects and subsequent positive attitudinal effects is the overlap between message execution cues. In this study, such overlap did not seem to favor attitudinal constructs though this might be related to the setup of the experiment (see discussion in limitations section 5.3).
The results might also be an indication that visual and textual consistency are not extensively alike. The expectations in the present study where mostly based on
consistency related literature related to textual consistency across messages. As visual cues are deemed as primary identification aspects for consumers (MacInnis, Moorman & Jaworski, 1991), it could well be that severe changes in these visual might even lead to a consumer not even paying attention as the inconsistently colored advertisement is not even recognized. Consistent use of visual cues could well form the threshold for a
28 consumer to recognize a brand and pay attention and the textual cues, where IMC’s ‘speaking with one voice’ can continue to conduct consistency through textual cues.
Otherwise, the possibility exists that the concept of IMC is not as effective as some scholars may suggest. IMC’s focus on consistency only holds to the extent that consumers encounter a company on multiple occasions before consistency across messages can be perceived. IMC thus might only be applicable for companies that have (very) regular contact with their clientele or stakeholders. Companies that have few contacts per client or stakeholders might not find benefits in the forming alignment of their marketing tools and could possibly make relentless costs and effort trying to do so.
It thus appears that the conundrum that seems to exist between consistency as an IMC principle versus the mere repetition of messages is a field that is defined by a fine line that demands further enquiry; it is still in the abyss what an optimum level of (visual) consistency is in cross-media advertising theory or even if such a level exists, which leaves plenty of opportunity for future research.
5.3 Limitations and Future Research
The present study has its limitations, which may contribute to the results found. First is the manipulation of color. Though color was not examined in isolation in previous research, it was expected that inconsistent coloring would lead to stronger recall than consistent coloring, as it was sought to make it a prominent aspect in the advertising messages used in the experiment. The applied colors in this study, purple and yellow, are direct opposites in the traditional color scheme. Colors that are direct opposites are not uncommon to be used in combination to make a vibrant color setting (Nassau, 1997), allowing the possibility that respondents did not experience the combination of purple
29 and yellow as a misfit. Future research may seek to study other color combinations that are not directly related or directly opposing in the traditional color scheme (see Nassau, 1997). Also, a manipulation check for the (in)consistent coloring was not examined in the present study. It could thus not be concluded that the respondents experienced the
coloring as different.
A likelier explanation for the lack of significant results may be that the present study aimed to simulate a cross-media advertising campaign by using just two exposures in two different media. When respondents are exposed new to a brand, as performed in the present study, they may not yet have gained the experience with the look and feel of the brand necessary to assess subsequent advertising messages as incongruent or to experience tedium (boredom) as a result of too many exposures. It may take stronger awareness of the visual characteristics of a brand for inconsistency to be perceived than can be generated by exposure to two advertising messages. Future research could thus aim to use a familiar brand to examine inconsistently colored advertising messages, at the cost of possible confounding pre-experimental attitudes toward the brand or by
preselecting respondents with the same level of attitude toward an examined brand. Another limitation is that respondents were asked to go on the webpage
advertisement as if they were casually surfing the web. As discussed, Internet is a ‘hot medium’, which is (in textual form at least) internally paced and thus the viewer decides what to read. Above all, the viewing of webpage is directed by ones own enquiry, whilst encountering an ad in magazine could happen more ‘coincidental’. This makes that the exposure to the webpage is inherently less natural. Future research may seek to examine combinations other then TV-commercial/webpage ads.
30 5.4 Managerial Implications
The present findings have implications managers affiliated with marketing and advertising may find useful. The findings that it (in)consistent coloring did not affect the memory and evaluative effects for the respondents indicates that consistency is not per se a necessary regiment. Especially a company that have little contact with their clientele thus seem not to be bound by the principle of consistency. The fact that the present study used an unknown brand and thus simulated only very few moments of contact with the respondents make that the present findings are arguably also relevant for new brands.
Also, the findings indicate that managers do not face a dilemma when weighing memory versus attitudinal purposes in a cross-media advertising decision. Managers can, according to the present findings, use consistent or inconsistent coloring without being afraid of negatively influencing attitudinal or memory effects.
Literature
Apple, V. (1971) On advertising wear-out. Journal of Advertising Research, 11 (2), 11– 13.
Bronner, A. E. (2007). Multimediasynergie in reclamecampagnes.
Burnkrant, R. E., & Unnava, H. R. (1987). Effects of variation in message execution on the learning of repeated brand information. Advances in Consumer
31 Chandon, P. (2003). Note on measuring brand awareness, brand image, brand equity and
brand value (pp. 1-12). Insead.
Dahlén, M. (2005). The medium as a contextual cue: Effects of creative media choice. Journal of Advertising, 34(3), 89-98.
Dijkstra, M., Buijtels, H. E., & van Raaij, W. F. (2005). Separate and joint effects of medium type on consumer responses: a comparison of television, print, and the Internet. Journal of Business Research, 58(3), 377-386.
Duncan, T. R. (2002). IMC: Using advertising and promotion to build brands. McGraw-Hill Companies.
Duncan, T., & Caywood, C. (1996). The concept, process, and evolution of integrated marketing communication. Integrated communication: Synergy of persuasive voices, 13-34.
Fill, C. (2000). Essentially a matter of consistency: integrated marketing communications. The Marketing Review, 1(4), 409-425.
Fransen, M. L., Fennis, B. M., & Pruyn, A. T. H. (2010). Matching communication modalities: The effects of modality congruence and processing style on brand evaluation and brand choice. Communication Research, 37(4), 576-598.
Gardner, M. P. (1985). Does attitude toward the ad affect brand attitude under a brand evaluation set?. Journal of Marketing Research, 192-198.
32 Hayes, A. F., & Preacher, K. J. (2014). Statistical mediation analysis with a
multicategorical independent variable. British Journal of Mathematical and Statistical Psychology, 67(3), 451-470.
Keller, K. L. (1987). Memory factors in advertising: The effect of advertising retrieval cues on brand evaluations. Journal of Consumer Research, 316-333.
Keller, K. L. (1996). Brand equity and integrated communication. Integrated
communication: Synergy of persuasive voices.
Kitchen, P. J., & Schultz, D. E. (1999). A multi-country comparison of the drive for IMC. Journal of Advertising Research, 39(1), 21-38.
Kitchen, P. J., Brignell, J., Li, T., & Jones, G. S. (2004). The emergence of IMC: A theoretical perspective. Journal of advertising research, 44(01), 19-30.
Kliatchko, J. (2008). Revisiting the IMC construct: A revised definition and four pillars. International Journal of Advertising, 27(1), 133-160.
Lord, K. R., Burnkrant, R. E., & Unnava, H. R. (2001). The effects of program-induced mood states on memory for commercial information. Journal of Current Issues & Research in Advertising, 23(1), 1-15.
MacInnis, D. J., Moorman, C., & Jaworski, B. J. (1991). Enhancing and measuring consumers' motivation, opportunity, and ability to process brand information from ads. The Journal of Marketing, 32-53.
33 Messaris, P. (1998). Visual aspects of media literacy. Journal of communication, 48(1),
70-80.
Moorman, M., Neijens, P. C., & Smit, E. G. (2010). Advertising Media Selection and Planning. Wiley
Nassau, K. (Ed.). (1997). Color for science, art and technology (Vol. 1). Elsevier.
Navarro-Bailón, M. Á. (2012). Strategic consistent messages in cross-tool campaigns: effects on brand image and brand attitude. Journal of Marketing
Communications, 18(3), 189-202.
Nordhielm, C. L. (2002). The influence of level of processing on advertising repetition effects. Journal of consumer research, 29(3), 371-382.
Obermiller, C., & Spangenberg, E. R. (1998). Development of a scale to measure c onsumer skepticism toward advertising. Journal of consumer psychology, 7(2), 159-186.
Percy, L. (2014). Strategic Integrated Marketing Communications 2e. Routledge.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models.Behavior research methods, 40(3), 879-891.
Spears, N., & Singh, S. N. (2004). Measuring attitude toward the brand and purchase intentions. Journal of Current Issues & Research in Advertising, 26(2), 53-66.
34 Spotts, H. E., Lambert, D. R., & Joyce, M. L. (1998). Marketing déjà vu: The discovery
of integrated marketing communications. Journal of Marketing Education, 20(3), 210-218.
Stammerjohan, C., Wood, C. M., Chang, Y., & Thorson, E. (2005). An empirical
investigation of the interaction between publicity, advertising, and previous brand attitudes and knowledge. Journal of Advertising, 34(4), 55-67.
Törn, F., & Dahlén, M. (2008). Effects of brand incongruent advertising in competitive settings. European Advances in Consumer Research, 8, 234-239.
Voorveld, H. A., & Valkenburg, S. M. (2014). The Fit Factor: The Role of Fit Between Ads in Understanding Cross-Media Synergy. Journal of Advertising, (ahead-of-print), 1-11.
Voorveld, H. A., Neijens, P. C., & Smit, E. G. (2011). Opening the black box:
Understanding cross-media effects. Journal of Marketing Communications,17(02), 69-85.
Yoo, C., Bang, H. K., & Kim, Y. (2009). The effects of a consistent ad series on consumer evaluations: A test of the repetition–variation hypothesis in a South Korean context. International Journal of Advertising, 28(1), 105-123.
35 Appendix A
36 Appendix B
Screenshot of highly consistent webpage ads
37 Appendix C
Screenshot of inconsistent webpage ads
38 Appendix D
Screenshot of moderately consistent webpage ads
39 Appendix D
Online experiment
Master Thesis Joshua Kreuger Dear participant,
First, I want to express my gratitude for your willingness to participate in my research for my Communication master thesis at the University of Amsterdam. In this research I study advertisements and need your help to do so. In the following minutes you will be shown a TV-commercial and an additional website and be asked to answer questions afterwards. Participating will take around 8 minutes and is very valuable to me as a master student aiming to graduate. As this research is conducted under the authority of ASCoR and the University of Amsterdam, you are ensured that:
1. Your participation is anonymous and your answers will in not in any case be shared with third parties.
2. You can abort the research at anytime you please. You can also retract from your willingness to participate within 24 hours after participating, upon which your answers will be deleted.
3. That the research will impose no considerable discomfort to you and that no deceptive information will be provided, as well as content that you might experience as shocking. 4. After a maximum of 5 months you can, if you please, attain the research report in which the findings of the research will be elicited.
40 If you have any questions regarding this research feel free to contact me and I will gladly answer them. Should you have any complaints about the process or conduct of this research, you can contact the representative of the Ethical Committee of ASCoR on the adres: ASCoR secretariaat, Commissie Ethiek, Universiteit van Amsterdam, Nieuwe achtergracht 166, 1018 WV Amsterdam, 020 525 3680; ascor-secr-fmg@uva.nl. Confidential treatment of your complaint is guaranteedJoshua Kreugere-mail: joshua.kreuger@student.uva.nltel:
06 45 222 333 University of Amsterdam
Supervisor: H. Voorveld
41 Q1 What is your age?
Q2 What is your nationality?
Q3 What is your gender? Male (1)
Female (2)
Q4 What is your highest attained education level? (provides Dutch equivalent educational level)
Primary school (basisschool) (1) High school (middelbare school) (2)
Intermediate Vocational Education (MBO) (3) College (HBO) (4)
University BSc (Bachelor universiteit) (5) University MSc (Master) (6)
What is your highest attained education level? High school (middelbare school) Is Selected
Q5 Where did you attain your high school degree? In the Netherlands (1)
42 Q6 What high school degree did you attain in the Netherlands?
VMBO (1) HAVO (2) VWO (3) Gymnasium (4)
Q7 Next you will be shown a commercial and a website. Please watch the TV-commercial once and read the information on the website as if you were casually surfing the internet. The website contains two webpages.
Q8 In the following section you will be asked several questions about the advertisements you have just seen. If you do not know the answer to what is asked, please state 'I don't know'
Q9 What was the main ingredient discussed in the advertisements?
Q10 What was mentioned about the ingredient in the advertisements?
Q11 What was the name of the producer of the yoghurts?
Q12 Please try to remember which of the following brands was shown in the advertisements. The name of the yoghurt producer was:
43 Fage (1) Chobani (2) Yoplait (3) Dannon (4) YoCrunch (5) La Yogurt (6) YoBaby (7) I don't know (9)
Q13 Next a series of questions will be asked on a two word scale that runs from negative to positive. Please indicate to what you thought about the advertisements and to what extent
Q14 The TV-commercial was
1 (1) 2 (2) 3 (3) 4 (4) 5 (5) 6 (6) 7 (7) not likable:likable (1) not interesting:interesting (2) bad:good (3) not appealing:appealing (4)
44 Q15 Yoplait's website was
1 (1) 2 (2) 3 (3) 4 (4) 5 (5) 6 (6) 7 (7) not likable:likable (1) not interesting:interesting (2) bad:good (3) not appealing:appealing (4)
45 Q16 The website was
1 (1) 2 (2) 3 (3) 4 (4) 5 (5) 6 (6) 7 (7) not likable:Likable (1) not interesting:interesting (2) bad:good (3) not appealing:appealing (4)
Q17 I became more interested in the second advertisement after seeing the first advertisement 1 (1) 2 (2) 3 (3) 4 (4) 5 (5) 6 (6) 7 (7) totally disagree:totally agree (1)
46 Q18 I wanted to know more about the brand (Yoplait) after seeing the first advertisement
1 (1) 2 (2) 3 (3) 4 (4) 5 (5) 6 (6) 7 (7)
totally disagree:totally
agree (1)
Q19 The information in the advertisements was
1 (1) 2 (2) 3 (3) 4 (4) 5 (5) 6 (6) 7 (7) difficult to process:easy to process (1) difficult to comprehend:easy to comprehend (2)
Q20 Were you familiar with Yoplait before seeing these advertisements o No (1)
o Yes (2)
47 Q21 Indicate to what extent you agree with the following statements
Totally disagree (1) Firmly disagree (2) Somewhat disagree (3) Don't agree/don't disagree (4) Somewhat agree (5) Firmly agree (6) Totally agree (7) We can depend on getting the truth in most advertising (1) Advertising's aim is to inform the consumer (2) I believe advertising is informative (3) Advertising is generally truthful (4) Advertising is a reliable
48 source of information about the performance and quality of products (5) Advertising is truth well told
(6) In general, advertising presents a true picture of the product being advertised (7)
I feel I've been accurately informed after viewing most advertisements (8) Most advertising provides
49 consumers
with essential information
(9)
Q22 Please indicate to what extent you like the following colors:
Totally dislike (1) Firmly dislike (2) Dislike (3) Don't dislike/don't like (4) Firmly like (5) Totally like (6) Yellow (1) Blue (2) Purple (3) Grey (4) Black (5) Orange (6) Red (7) White (8)
