Journal of Business Research 66 (2013) 499–505

Contents lists available at SciVerse ScienceDirect

Journal of Business Research

How captive is your audience? Defining overall advertising involvement

Nathalie Spielmann a,⁎, Marie-Odile Richard b,1

a Reims Management School, 59 rue Pierre Taittinger, BP 302, 51061 Reims Cedex, Franceb University of Montreal, Ecole d'Optométrie, Canada

⁎ Corresponding author. Tel.: +33 6 03 32 34 09; faxE-mail addresses: nathalie.spielmann@reims-ms.fr (

odile10@hotmail.com (M.-O. Richard).1 Tel.: +1 514 738 3520.

0148-2963/$ – see front matter © 2011 Elsevier Inc. Alldoi:10.1016/j.jbusres.2011.12.002

a b s t r a c t

a r t i c l e i n f o

Article history:Received 1 February 2011Received in revised form 1 November 2011Accepted 1 December 2011Available online 5 January 2012

Keywords:Overall advertising involvementMessage involvementMedia involvementCreative involvementScale development

This article aims to reconcile some inconsistencies on the three constructs of advertising involvement, advertis-ing relevance andmedia engagement. Then it develops a scale to holisticallymeasure overall advertising involve-ment. Three previously measured types of involvement (message, media, and creative) are regrouped into onemultidimensional structure with three correlated dimensions. The scale is then used to show that overall adver-tising involvement is capable of shaping attitudes leading to various consumer outcomes. Contributing tothe literature on advertising involvement, this research confirms that overall advertising involvement isboth situational and enduring. From a professional perspective, the research proposes a measurement tool bettersuited to understanding the scope of overall consumer involvement with advertising.

© 2011 Elsevier Inc. All rights reserved.

1. Introduction

In 1964, Marshall McLuhan stated: “The medium is the message”and mass media was not about “the size of the audience, but of thefact that everybody becomes involved” (p. 349). McLuhan meantthat consumers are not immune to advertising stimuli; rather, theyreact to various components of advertising, and media, message,and creative execution may all together work to influence consumers.For example, would Super Bowl ads for Doritos be as involving if theywere in print rather than on television, about nutritional facts ratherthan the pleasure of snacking during a great game, presented in abland rather than a humorous execution?

The advent of new formats, media convergence and diversificationof consumer target markets is forcing advertisers to reconsider andreconfigure campaign metrics in order to better understand why ad-vertising works, not just if it does (Rappaport, 2007; Woodard, 2006).Thus, the ad industry wants to define involvement with media, subject,and advertiser, and this has led to the desire to shift from impression-basedmodels towhat advertisers are calling engagement-basedmodels(Abdul-Ghani, Hyde, & Marshall, 2011; Mollen & Wilson, 2010; Skerik,2011). Unfortunately, advertisers have various definitions of whatare involvement, relevance, and/or engagement. The same is true ofresearchers, e.g. Wang (2006) considers it contextual relevancewhereas Heath (2009) considers it to be the result of stimulation ofemotions.

: +33 3 26 04 69 63.N. Spielmann),

rights reserved.

This article seeks first to review and reconcile some inconsis-tencies on the three constructs advertising involvement, relevanceand engagement, and to introduce the concept of overall advertisinginvolvement. Then, we develop a scale to holistically measure overalladvertising involvement. Three types of involvement (i.e., message,media, and creative) are regrouped into one second-order constructwith three correlated dimensions. The new measure is used to showthat overall advertising involvement is able to shape attitudes leadingto several consumer outcomes.

Contributing to the literature on advertising involvement, thisarticle confirms that overall advertising involvement is both situationaland enduring. From a managerial perspective, we propose a measurebetter suited to understanding overall advertising involvement.

2. Literature review

2.1. Defining advertising involvement

A literature review reveals that since 1960, the topic of involve-ment has been widely discussed. For Krugman (1965), consumerscan be in low involvement and change their attitudes after ad repeti-tions and only if their perceptions are influenced. Alternatively, con-sumers in high involvement experience rapid changes in cognitionsand beliefs post ad exposure, leading to attitudinal and behavioralchanges. Greenwald and Leavitt (1984) discuss four levels of audi-ence involvement: preattention, focal attention, comprehension, andelaboration. Ray et al. (1973) claim that different levels of involve-ment lead to different sequences of impacts on cognitions, affectand behavior. Zaichkowsky (1986) showed that different consumerscan be involved with an ad, i.e., more involved consumers respond

500 N. Spielmann, M.-O. Richard / Journal of Business Research 66 (2013) 499–505

more to ads, and consumers use different cues when evaluating adsdepending on the relevance of the topic. Since Krugman's (1965) dualintensity model of involvement, and in light of Mitchell's (1979) exper-imental manipulations of involvement, most researchers used (varioustypes of) involvement as a mediating or moderating variable by classi-fying respondents as either high or low involvement, and not basedon the type of involvement.

The terminology on advertising involvement is varied. Making dis-tinctions among involvement, relevance and engagement can be diffi-cult, as the words are often used interchangeably, even within thesamemanuscript. For example, for Laczniak andMuehling (1993), con-sumers who are highly involved attend to ad claims that they findrelevant. Similarly, for Batra and Ray (1986),message involvement occurswhen the message is deemed relevant. For Wang (2006), engagementresults from contextual relevance. Thus, relevance cannot be present ifconsumers are not involved and engagement is a result of perceived rel-evance. Thus, it appears that involvement, relevance and engagementare more synonyms than distinct constructs. The differences in use de-pend on the scope (i.e., academic or professional). For the rest of this ar-ticle, we define all forms of consumer involvement, engagement, andperceived relevance as dimensions of overall advertising involvement.

2.2. Types of involvement used in advertising research

In the advertising literature, while the discourse on ad involvementis wide, how to classify ad involvement and the results of studies usingad involvement do not always allow for a clear understanding of whatinvolvement entails (Day, Stafford, & Camacho, 1995). Several types ofinvolvement can be found: situational (Celsi & Olson, 1988); product-related (Zaichkowsky, 1994); enduring (Lumpkin, 1985); message(Lord & Burnkrant, 1993); purchase (Slama & Tashchian, 1985); andprogram (Levy & Nebenzahl, 2006) involvement, to name a few. Gener-ally, involvement types are regrouped into either enduring or situation-al (Day et al., 1995), even if some types can be classified as both,depending on the ad context and how they are manipulated. For exam-ple, political involvement can be situational when a campaign is rele-vant or interesting, or enduring when politics have a central role forconsumers. Thus, political involvement differs based on how consumersanswer two similar but distinct statements: “this political campaign isimportant to me” and “I am interested in politics.”

For Celsi and Olson (1988), situational involvement occurs whenindividuals are motivated to act upon their feelings. It is a more actionoriented aspect of involvement. What is relevant to consumers is out-lined as felt involvement (Celsi, Chow, Olson, & Walker, 1992). “Thisperspective explicitly recognizes that a consumer's perception or feel-ing of personal relevance for an object or event is an acute state thatonly occurs at certain times and in certain situations” (Celsi & Olson,1988, p. 211). Situational involvement is ephemeral and at times in-consistent, as what is relevant today may not be tomorrow. Thus, sit-uational involvement can be highly subjective and dependent on pastexperiences, memories, interpretations and ultimately, on how rele-vant the advertising is to the consumer well-being (Petty, Cacioppo,& Goldman, 1981).

In contrast, enduring involvement relates to interests and rele-vance. Howard and Sheth (1969) mention types of products or char-acteristics as reasons for involvement—stating that the relevance ofthese likely leads to purchase intent. Other examples are involvementwith politics (Lichtenstein, Netemeyer, & Burton, 1990), or productinvolvement (Zaichkowsky, 1994). Other than product, media canhave an influence if consumers feel an ad is relevant. For Krugman(1965), television is a low involvement medium, and print is a highinvolvement one, regardless of the content or the advertiser. DePelsmacker, Geuens, and Anckaert (2002) discuss media context as“an important situational factor. Media context is defined as thecharacteristics of the content of the medium in which an ad isinserted” (p. 49). Seemingly, they allude to media involvement as

being situational rather than enduring, in contrast to Krugman (1965).Yet they make the link between content and media and the relationshipbetween these two features of advertising working in tandem ratherthan independently.

In marketing, involvement is a state modified by personal character-istics, including motivation, and dependent on the personal relevance ofthe involving object (Day et al., 1995). As such, involvement can be ma-nipulated prior to advertising exposure and/or measured after advertis-ing exposure (Kamins, Assael, & Graham, 1990; Petty & Cacioppo,1979). In advertising, an involvement manipulation requires personalrelevance to be created via stimuli, or relevance to be supported by thestimuli using quality arguments and evaluative cues (Zaichkowsky,1986). It is argued that artificially manipulating involvement creates sit-uational involvement whereas classifying respondents based on theirexisting levels of involvement gauges enduring involvement (Laczniak& Muehling, 1993). As such rather than examining involvement as astate variable, researchers look at the antecedents of involvement as ameans to explain the outcomes of involvement (Bloch & Richins, 1983;Zaichkowsky, 1986).

2.3. Involvement as an antecedent: one or many dimensions?

Several studies show the value of involvement as a marketing var-iable. For example, Laczniak and Muehling (1993) compare brand re-lated beliefs across varying levels of message involvement. Otherssought to create measures for involvement types, for example prod-uct involvement (Zaichkowsky, 1985). While some proposed scalesof involvement, others caution against generalizing the concept of in-volvement (Laczniak & Muehling, 1990). Day et al. (1995) state: “Be-cause the object of involvement ranges from activities and issues toadvertisements and purchases, no single scale can measure all kindsof involvement” (p.72). Similarly to Zaichkowsky (1986), for Day etal. (1995) while it is known what types of involvement there areand how the intensity of involvement can impact marketing vari-ables, little is known about the combination of these involvementconstructs in advertising and what involves consumers overall. Itcould be that an advertisement is involving because the media formatin which it is presented, the product that is shown and the messagesin the ad are all together captivating. Few studies combine involve-ment with other variables, for example, product involvement andmedia format in order to examine consumer outcome behaviors (DePelsmacker et al., 2002). However these studies model involvementand context as opposing factors rather than as related dimensions ofan overall concept.

2.4. Is engagement the same as overall involvement?

The various involvement measures used in academic research canmake understanding what involvement means confusing and difficultto apply in professional contexts. Recently, both practitioners and aca-demics tried to understand how integrated marketing communicationswork (Calder & Malthouse, 2005) and examined a new concept calledengagement. Wang (2006) defines engagement as a “critical measure-ment of when consumers are strongly engaged in brands, brand mes-sages, and their surrounding environments” (p. 356). Rappaport(2007) defines engagement as brand relevance and an emotional rela-tionship between the consumer and the brand, all occurring within aquality context (see also, Geuens, De Pelsmacker, & Faseur, 2011). Brief-ly Wang (2006) makes the link between engagement and involvementby claiming that once contextual relevance is achieved, engagementdrives message involvement.

Empirically, engagement is rarely operationalized. In some de-signs, engagement is manipulated as with or without contextual rel-evance between a primary task and an online advertisement (Wang,2006). When engagement is tested it is identified at the focal atten-tion stage (Greenwald & Leavitt, 1984), and it is only possible to

501N. Spielmann, M.-O. Richard / Journal of Business Research 66 (2013) 499–505

estimate it if consumers are aware of the advertising context (Wang,2006). There is no scale to empirically test the level of advertising en-gagement. Media engagement was proposed as a multiple-dimensionalstructure and linked to purchase intentions, but the focus is limited tothe interaction between media format and behavioral outcomes, notoverall advertising engagement (Kilger & Romer, 2007). Thus the ex-pression of consumer engagement and the measurement of the con-struct are complicated and relatively untested (Skerik, 2011). Forexample, Heath (2009) states that media can fail to capture attention,even if they are involving and that media can be captivating but not atall engaging. Anecdotally, Siefert et al. (2009) show that Super Bowlads are emotionally engaging in a low captivity medium.

In terms of advertising content, consumers do not just interactwith media formats, but also with the visual and artistic aspects ofthe ads. Smith and Yang (2004) show that strong creative ads garnermore attention from consumers. Consumers relate to advertising cre-ative and become more involved as a consequence. Media format canalso be influential in creating media involvement in combination withproduct involvement (De Pelsmacker et al., 2002). Yet the questionstill remains: how can advertisers know if their advertising efforts,the combination of media format, product presentation, and creativeexecution lead to relationships between their brand and consumers?And what is the quality of the relationships created when consumersare entirely involved?

3. Conceptual framework

Based on this literature, it is clear that advertisers and academicsuse different names for advertising involvement. It appears thatmost researchers agree on the notion that involvement relates tothe active participation (or not) of an individual when faced with ad-vertising (Greenwald & Leavitt, 1984). However, understanding whatinvolves consumers when faced with advertising and why this is soremains somewhat unclear. Furthermore, it is unknown how levelsof involvement depend on multiple personal, situational, and objectcharacteristics (Zaichkowsky, 1986). How can advertisers know ifthey are advertising at the right place, at the right time, to the rightconsumer, and if these actions lead to overall involvement withadvertising?

Beyond the linear models of communication (i.e., sender-message-receiver),modern advertising contexts are farmore complex and usual-ly include considerations such as the media format (De Pelsmacker etal., 2002), and the variable nature of the consumers (Stern, 1994). Con-sumers interact with media, product and advertising characteristics si-multaneously when faced with advertising and the measurement oftheir involvement should take the overall ad context into consideration(Laczniak & Muehling, 1990).

Clearly there is a link among media format, creative and content in-volvement, even if together these have not yet been tested. Smith, Chen,and Yang (2008) found that creative ads are perceived as beingmore in-teresting and that creative has the power to influence cognitions and at-titudes. Also, certain media formats are more creative, more visual anddynamic (e.g., television) whereas in others it is easier to detail content(e.g., print). Selecting certain types of media engages higher involve-ment (i.e., central processing) or lower involvement (i.e., peripheralprocessing) (Petty & Cacioppo, 1986). Likewise, the type of presentationin an ad can involve consumers: they more likely process informationand direct more attention to verbal cues than to non-verbal cues(MacInnis, Moorman, & Jaworski, 1991). Maheswaran and Sternthal(1990) state that increased attention to a message leads to more de-tailed consideration of its content. The more involving a context is, themore likely a consumer is to invest resources and ismotivated to attendto the context (Celsi & Olson, 1988).

Combining these results regarding involvement with a creative, or amedia format, or a topic, it is possible to considermultiple outcomes. Forexample, combining television with strong creative about a product

germane to consumers may make the ad overall much more involving.As such, we propose that overall advertising involvement is a three di-mensional construct:

H1. Overall advertising involvement is a second order construct com-posed of message involvement, media involvement, and creativeinvolvement.

The literature on involvement shows clear links with cognitions andaffect; involvement can either modify beliefs or perceptions (Krugman,1965). In general, models of involvement stipulate that motivation toprocess will result in outcome behaviors (Batra & Ray, 1986). The in-volvement construct is applicable to advertising because it determinesthe way in which information is processed and which component ismost likely to be relied on for attitude formation (Kover, 1995; Petty,Cacioppo, & Schumann, 1983). For an advertiser, being able to controlthe quality of brand attitudes is important. Attitude is an affective reac-tion that occurs as a result of cognitions (Lazarus, 1984; Roseman, 1984;Scherer, 1988). Thus, advertisers want to ensure that their ads instigatebrand attitudes leading to favorable behaviors. Beyond just media con-texts (De Pelsmacker et al., 2002), it is overall advertising (i.e., the com-bination of message, creative and media) involvement that leads tooutcome behaviors.

H2. Compared with consumers who are overall lowly involved with ad-vertising (i.e., message, creative, media), those who are overall highlyinvolved have more positive brand attitude and higher outcomebehaviors.

4. Methodology

4.1. Developing a scale for overall advertising involvement

4.1.1. QuestionnaireA questionnaire was added to a professional online study examin-

ing the media habits and perceptions of North American students. Thefirst page was an invitation to begin the survey. The second page fea-tured the stimulus ad with instructions to scroll to the bottom of thepage when they were ready to answer questions. No indication wasgiven to review the ad specifically. The use of the creative (a genericLevi's Jeans ad) was selected because it did not have a gender refer-ence and is used by both genders and individuals of all ages. Afterquestions regarding their campus and media habits, respondentswere asked if they recalled seeing an ad in the survey and if so,which one. Subsequently, they were shown the ad and asked abouttheir attitude toward the ad. Respondents then answered questionsregarding their post exposure behaviors.

4.1.2. MeasuresPreviously developed scales for message, media and creative in-

volvement, were included in order to define an overall advertising in-volvement measure. All measures were presented on five point Likertscales (1 = not at all, 5 = very much). For message involvement, twomeasures were combined into a 15-item measure: Zaichkowsky's(1994) product involvement scale and Baker and Lutz (2000) ad mes-sage involvement scale. The ad was once again displayed and respon-dents were asked to answer the 15-item scale while considering thead. The same style of presentation was used for the media involve-ment and the creative involvement measures. Media involvementwas an 8-item scale constructed from the ad message involvementmeasure from Lee (2000) and the ad preference index from Bruneland Nelson (2001). Creative involvement incorporated motivationto process items (Petty & Cacioppo, 1986) and ad involvement fromLaczniak, Muehling, and Grossbart (1989) to form an 8-item scale.The attitude toward the ad used the Sengupta and Johar (2002) mea-sure for attitudes toward the brand which include the three followingitems: “I think the brand is a very good brand”, “I think the brand is a

Table 1Measurement model for overall advertising involvement (N=641).

Latent factors Standard coefficient(t value)

Averagevarianceextracted

Cronbachalpha

Message involvement .71 .96When looking at the ad, you findwhat is advertised to beImportant .721 (16.0)Of concern to you .806a

Relevant .764 (17.3)Meaning a lot to you .893 (21.7)Valuable .881 (21.3)Beneficial .886 (21.4)Mattering to you .921 (22.8)Essential .795 (18.3)Significant to you .902 (22.0)Motivating .814 (18.9)

Media involvement .72 .94When thinking of the ad, did you findyourself doing any of the followingPaying attention to the content .809a

Concentrating on the content .909 (22.1)Thinking about the content .851 (20.)Focusing on the content .912 (22.2)Spending effort looking at thecontent

.817 (18.8)

Carefully reading the content .765 (17.2)Creative involvement .64 .87When thinking of the ad, did you findyourself doing any of the followingTaking note of the visual aspectsof the ad

.769a

Focusing on the colors and/orimages of the ad

.860 (17.4)

Noting some specific colors orimages in the ad

.836 (16.9)

Paying close attention to the ad as apiece of art

.716 (14.2)

Fit indicesChi square (df=167) 494.96CFI .983RMSEA .055

a Fixed.

502 N. Spielmann, M.-O. Richard / Journal of Business Research 66 (2013) 499–505

very useful brand” and “Myopinion of the brand is very favorable”. Singleitem measures were used for the various behaviors (see Table 2 for thestatements) as per Bergkvist and Rossiter (2007) since all the behaviorswere concrete and easily imagined by respondents.

4.1.3. SampleRespondents came from North American campuses, polled from

the database of a scholarship information website. Of the 643 com-pleted questionnaires (86.3% completion rate), 73.1% were women,87.5% were between the ages of 18 and 24, 95.6% were full time stu-dents and 85.7% were undergraduates.

4.1.4. Scale developmentSince the scales for message, media, and creative involvement

were constructed by combining previously developed measures, it isreasonable that some items may have to be dropped during thescale development process due to redundancy. Furthermore, crossvalidation is recommended for such measure purification processesto minimize error probability and capitalization on chance. Accord-ingly, responses were randomly split into two halves (Group 1:n=317; Group 2: n=326) to perform an exploratory factor analysiswith one set and confirm the validity of the scales with the second set.The sample from Group 1 was used to uncover the constructs under-lying overall advertising involvement in an exploratory factor analy-sis using principal component extraction and oblimin rotation. Afteran iterative purification process where low loading indicators andcross-loading items were removed (Churchill, 1979), a solution withthree factors encompassing ten items for message, six items formedia and four items for creative was retained with all factor load-ings greater than .70. The total variance explained by this structurewas 74% and the variance per factor was 51.4% for message, 15.4%for media, and 7.2% for creative. The Cronbach alphas for each factorwere, respectively, .96, .93, and .88.

Following the two-step procedure recommended by Andersonand Gerbing (1988), we estimated and respecified the measurementmodel prior to incorporating the structural paths. The next step in-volved a confirmatory factor analysis of the purified 20-item mea-surement model using EQS with ERLS estimation. The three factormeasurement model produced a satisfactory fit (χ2=331, df=167,χ2/df=1.98, CFI=.983, and RMSEA=.056). Next, we tested thismodel on the second split sample, after removing two outliers. Theresulting fit indices indicated that the measurement model had alsoa good fit to the data (χ2=360, df=167, χ2/df=2.16, CFI=.981,RMSEA=.060). A more stringent test of measurement equivalencein a multi-sample confirmatory factor analysis where all factor load-ings and covariances were constrained to equality between the twogroups showed a metric invariant model (χ2=704, df=354, χ2/df=1.99, CFI=.982, RMSEA=.056). The model replicated satisfacto-rily over the two independent samples and it was deemed appropri-ate to combine them for further analyses. The measurement modelfor the combined sample (n=641) showed an even better fit to thedata in terms of CFI and RMSEA (χ2=495, df=167, χ2/df=2.96,CFI=.983, RMSEA=.055). All items loaded significantly on their re-spective constructs with the lowest t-value being 14.2, and the aver-age variance extracted by each of the constructs greater than .5,indicating that the variance captured by the construct is greaterthan the variance due to measurement error, providing evidence ofconvergent validity (Fornell & Larcker, 1981). The results of the mea-surement model appear in Table 1.

Discriminant validity was assessed by constraining the estimatedcorrelation parameter between two scales to 1 and comparing theresulting chi-square statistics to that obtained when the correlation be-tween pairs of scales was unconstrained (Anderson & Gerbing, 1988;Bagozzi, 1981). Results of all chi-square difference tests found theunconstrained models to be statistically better than the constrainedones, providing evidence of discriminant validity. Discriminant validity

was also assessed by Fornell and Larcker's (1981) procedure in whichdiscriminant validity is established if the average variance extracted islarger than the squared correlation coefficients between factors. In allcases, this criterion was met across all pairs of factors (squared correla-tions between message and media=.30; message and creative=.18,and media and creative=.45).

4.2. The overall advertising involvement scale

A review of the dimensions and their relationships showed thatthe correlations between message and media was .55, message andcreative was .42, and media and creative was .67. These high correla-tions indicate that overall advertising involvement can be defined byinterrelated dimensions which can be modeled as a second order fac-tor. A second order CFA (χ2=495, df=167, χ2/df=2.96, CFI=.983,RMSEA=.055) showed that the media dimension of overall advertis-ing involvement accounts for more variance in the construct(β=.934) than the creative (β=.714) and message (β=.589) di-mensions. All three dimensions are statistically significant (pb .01).Therefore H1 is supported.

Overall advertising involvement is postulated to shape brand atti-tude leading to behaviors and the relationship between the secondorder measure of advertising involvement and brand attitude wastested in a structural CFA model where a path from advertising in-volvement to attitude was added. In a similar manner, the overall

Table 2Nomological validity tests for the advertising involvement construct.

Standard coefficient(t-value)

χ2 df CFI RMSEA

Brand attitude .619 (11.1) 638.58 226 .983 .053Behaviors

Be tempted to purchasethis product

.638 (12.3) 613.29 186 .980 .060

Make an effort to seekout more information

.541 (10.2) 585.40 186 .981 .058

Tell a friend about thebrand/product

.558 (10.6) 558.28 186 .982 .056

Call the company/go ontheir web site

.536 (10.1) 555.01 186 .982 .056

Direct someone you knowto the website

.508 (9.42) 563.77 186 .981 .056

503N. Spielmann, M.-O. Richard / Journal of Business Research 66 (2013) 499–505

advertising involvement effect on behaviors was tested. However,prior to testing the above relationships, the items of the brand atti-tude scale were added to the measurement model. The Cronbachalpha for this measure was .88. This full measurement model had aχ2=741, df=246, χ2/df=3.01, CFI=.981, RMSEA=.056, howeverone item in the attitude scale (“informative”) had a high standardizedresidual. Removing this item produced a significant fit improvementin the full measurement model (χ2=615, df=224, χ2/df=2.75,CFI=.984, RMSEA=.052). The three items used as indicators of atti-tudes had significant paths from the construct and the average vari-ance extracted was .65. Discriminant validity tests showed evidenceof discriminant validity between all pairs of constructs in the fullmeasurement model (intercorrelations among the measures were at-titudes and message=.258, attitudes and media=.205, and attitudesand creative=.249). Results of the structural model in Table 2 showthat the second order overall advertising involvement construct issignificantly related to attitudes and to behaviors, establishing thusthe nomological validity of the overall advertising involvement con-struct. Therefore, H2 is also supported.

Overall advertising involvement, brand attitude and the variousbehaviors were modeled with attitudes as a mediator between over-all advertising involvement and behaviors. Single item measureswere used for behavior as per Bergkvist and Rossiter (2007) sinceall the behaviors were concrete and easily imagined by respondents.The CFA results for this model indicated a good fit for each of thefive behavioral variables but the LMχ2 modification index suggesteda model where a path from message to behavior could be added foreach of the five behaviors. Overall advertising involvement, brand at-titude and the various behaviors were thus modeled as outlined inFig. 1. The model fit statistics for each of the behaviors are includedin Table 3. Descriptive statistics appear in Table 4. For each of thesemodels, all parameter estimates were significant (pb .01).

Message involvement

Mediainvolvement

Creativeinvolvement

Overalladvertising

involvement

Fig. 1. Structur

5. Discussion

The findings show that overall advertising involvement as a mea-sure encompassing message, media, and creative involvement signif-icantly shapes brand attitude and leads consumers to engage in moreaction-oriented marketing behaviors. Additionally, the interest gen-erated by the message contributes to favorably influence the behav-iors directly.

The findings confirm the relationships among overall advertisinginvolvement, brand attitude, and behaviors. They also show thatoverall advertising involvement is not well represented using a one-dimensional approach such as situational (Celsi & Olson, 1988), prod-uct (Zaichkowsky, 1985), or message (Petty et al., 1983) in isolation.It is a combination of the three types of involvement that results ina more powerful representation of how the overall involvement ofconsumers actually shapes their brand attitude leading to strongermarketing responses.

Before forming brand attitude, consumers engage in cognitionsand establish the overall relevance of an advertisement. The rela-tionship between cognitions and affect is supported by the structuralmodel. The findings demonstrate a more coherent picture of the me-diating role of brand attitude rather than their primary role ininfluencing behavior. The model also demonstrates that brand atti-tude is not an independent variable capable of influencingmarketingbehaviors, but rather that in most cases, brand attitude is first shapedby overall advertising involvement.

The overall advertising involvement construct (composed of mes-sage, media, and creative involvement) shows the interrelated natureof the various types of involvement and confirms the necessity tostudy them together when testing consumer involvement with ads.The correlations among the three dimensions of the overall advertis-ing involvement scale indicate that involvement with advertising isindeed multidimensional. The relationships between the dimensionsare independently shown in the current literature (De Pelsmackeret al., 2002; Siefert et al., 2009), but they had not been tested simul-taneously until now. We show that the quality of the message de-pends on the medium used to communicate it at the same time asthe quality of the media also impacts the perceived level of creativity.As posited by Houston and Rothschild (1978), response involvement,or conceptualized here as advertising involvement, incorporates bothsituational and enduring involvement. The overall context of the ad-vertising leads to overall involvement.

Media involvement has more variance than the other two dimen-sions, confirming what Kilger and Romer (2007) highlight with theirstudy: media vehicle choice matters. However this research goes fur-ther and highlights that media with a relevant message is more pow-erful then media alone, echoing McLuhan (1964).

Interestingly, while media was the strongest dimension, it did notrelate directly to behavior, as previously posited (Kilger & Romer,2007). Rather, it is message involvement which can directly orient

Marketing behavior

Brand attitudes

al model.

Table 3Structural model fit statistics.

Standard coefficient (t-value) χ2

(df=246)CFI RMSEA

Overall ad involvement→brandattitude

Brand attitude→behavior

Message involvement→behavior

BehaviorsBe tempted to purchase this product .619 (11.2) .288 (6.38) .486 (10.6) 661.39 .984 .051Make an effort to seek out moreinformation

.620 (11.2) .219 (4.58) .464 (9.58) 669.19 .983 .052

Tell a friend about the brand/product .622 (11.2) .311 (6.43) .390 (8.17) 678.75 .983 .052Call the company/go on their web site .623 (11.1) .238 (4.83) .416 (8.48) 660.43 .984 .051

.621 (11.1) .198 (4.02) .446 (9.01) 656.15 .984 .051

504 N. Spielmann, M.-O. Richard / Journal of Business Research 66 (2013) 499–505

consumer behavior. Consumers can have very strong beliefs about amessage or an object and relevance of an object to one's values andneeds may be intransigent (Zaichkowsky, 1985). Message involve-ment can be enduring, and less likely to change in light of circum-stances, making it more of a trait in certain cases (Kassarjian, 1981).The consequences of message involvement include search behaviors,information processing and persuasion (Andrews, Durvasulam, &Akhter, 1990), thus they are capable of directly influencing outcomebehaviors. As such, strong message claims do not necessarily rein-force brand attitude but confirm them, leading to immediate behav-iors rather than appreciation first.

6. Managerial implications

The emotional relationship between a brand and a consumer canbe considered as overall involvement with message, creative andmedia. If involvement is perceived as the harmonization of an adwithin a context and with a consumer, then advertising involvementmay be an effective method of measuring the occurrence of engage-ment. Essentially, “engagement” as discussed in professional settingscan be operationalized as the behavioral outcomes observed whenconsumers find ads involving at every level, since these behavioraloutcomes represent a successful relationship between brand adver-tising and consumers.

Using this scale, advertisers may see the levels and types of relation-ships between the brand and the consumer that the advertising effortshave created. Furthermore, “engagement represents the quintessenceof what we ultimately want from advertising metrics” (Woodard,2006, p.353). Using our overall advertising involvement measure mayenable advertisers to gauge the relevance of an ad thus explaining sub-sequent brand attitude and behaviors. It is more revealing to measureoverall advertising involvement in order to distinguish which dimen-sions, if not all, are most involving to consumers when they formbrand attitude.

In order to use this scale, advertisers can incorporate the itemsretained in post-campaign surveys, as they do with other metrics suchas recall and awareness. Afterwards itmay bepossible to classify respon-dents based on their overall and dimensional levels of involvement.

Table 4Descriptive statistics.

Variables Mean Std. dev.

Message involvement 1.55 .74Media involvement 2.01 .99Creative involvement 2.43 1.11Brand attitude 2.25 1.06Behaviors

Be tempted to purchase this product 1.46 .83Make an effort to seek out more information 1.42 .82Tell a friend about the brand/product 1.43 .85Call the company/go on their website 1.41 .85Direct someone you know to the website 1.31 .72

Alternatively, our proposed measure can also be used in pre-campaigntesting to ensure that ads are adequately involving prior to launch.Using the overall advertising involvement scale in pre-campaign studiesmay limit the expense of inefficient advertising.

7. Limitations and future research

Using a student sample for this study makes the results somewhatyouth oriented. However, the sizes of the samples as well as thestrengths of the relationships obtained with them show that the ef-fects are not negligible. Additionally, the use of one type of advertis-ing medium also makes these results more specific to this type ofmedia format. Future research should attempt to replicate this re-search with the general population and with alternative advertisingexecutions.

Having used an online survey may make the results more general-izable across content-free static media formats rather than dynamiccontent-based media such as television. Future research could incor-porate dynamic media as well as static media formats in order tosee how they perform in terms of the overall advertising involvementconstruct as well as see how the various behaviors are influenceddepending on the media format. Are some media formats more orless engaging and are some product categories more or less involv-ing? These should also be factored into future research designs incor-porating the overall advertising involvement measure in order touncover the strength of these relationships.

Finally, the means for the variables and behaviors are relativelylow. However the results are significant, showing that the proposedrelationships clearly do exist.

Acknowledgment

The authors sincerely thank Isabelle Miodek and Laurie Babin(University of Louisiana at Monroe) for their help with this manu-script. The authors would also like to thank Zoom Media for theirhelp with the data collection.

References

Abdul-Ghani E, Hyde KF, Marshall R. Emic and etic interpretations of engagement witha consumer-to-consumer online auction site. Journal of Business Research 2011;64(10):1060–6.

Anderson JC, Gerbing DW. Structural equation modeling in practice: a review andrecommended two-step approach. Psychological Bulletin 1988;103(3):411–23.

Andrews JC, Durvasulam S, Akhter SH. A framework for conceptualizing and measuringthe involvement construct in advertising research. Journal of Advertising 1990;19(4):27–40.

Bagozzi RP. Evaluating structural equation models with unobservable variables andmeasurement error. Journal of Marketing Research 1981;18(3):375–81.

Baker WE, Lutz RJ. An empirical test of an updated relevance-accessibility model ofadvertising effectiveness. Journal of Advertising 2000;29(1):1-14.

Batra R, Ray ML. Situational effects of advertising repetition: the moderating influenceof motivation, ability, and opportunity to respond. Journal of Consumer Research1986;12(4):432–45.

505N. Spielmann, M.-O. Richard / Journal of Business Research 66 (2013) 499–505

Bergkvist L, Rossiter JR. The predictive validity of multiple-item versus single-itemmeasures of the same constructs. Journal of Marketing Research 2007;44(2):175–84.

Bloch PH, RichinsML. A theoretical model for the study of product importance perceptions.Journal of Marketing 1983;47(3):69–82.

Brunel FF, Nelson MR. Explaining gendered responses to ‘help-self’ and ‘help-others’charity ad appeals: the mediating role of world-views. Journal of Advertising2001;29(3):15–28.

Calder BJ,MalthouseEC.Managingmedia and advertising changewith integratedmarketing.Journal of Advanced Research 2005;45(4):356–61.

Celsi RL, Olson JR. The role of involvement in attention and comprehension processes.Journal of Consumer Research 1988;15(2):210–24.

Celsi RL, Chow S, Olson JC,Walker BA. The construct validity of intrinsic sources of personalrelevance: an intra-individual source of felt involvement. Journal of Business Research1992;25(2):165–85.

Churchill GA. A paradigm for developing better measures of marketing constructs.Journal of Marketing Research 1979;16(1):64–73.

Day E, StaffordMR, CamachoA. Opportunities for involvement research: a scale developmentapproach. Journal of Advertising 1995;24(3):69–75.

De Pelsmacker P, Geuens M, Anckaert P. Media context and advertising effectiveness:the role of context appreciation and context/ad similarity. Journal of Advertising2002;31(2):49–61.

Fornell C, Larcker DF. Evaluating structural equation models with unobservable variablesand measurement error. Journal of Marketing Research 1981;18(2):39–50.

Geuens M, De Pelsmacker P, Faseur T. Emotional advertising: revisiting the role ofproduct category. Journal of Business Research 2011;64(4):418–26.

Greenwald AG, Leavitt C. Audience involvement in advertising: four levels. Journal ofConsumer Research 1984;11(1):581–92.

Heath R. Emotional engagement: how television build big brands at low attention.Journal of Advanced Research 2009;49(1):62–73.

Houston MJ, Rothschild ML. Conceptual and methodological perspectives in involve-ment. In: Jain S, editor. Research Frontiers in Marketing: Dialogues and Directions.Chicago: American Marketing Association; 1978. p. 184–7.

Howard JA, Sheth JN. The theory of buyer behavior. New York: John Wiley; 1969.Kamins MA, Assael H, Graham JL. Cognitive response involvement model of the process

of product evaluation through advertising exposure and trial. Journal of BusinessResearch 1990;20(3):191–215.

Kassarjian HH. Low involvement: a second look. In: Monroe KB, editor. Advances inconsumer research, 8. Ann Arbor, MI: Association for Consumer Research; 1981.p. 31–4.

Kilger M, Romer E. Do measures of media engagement correlate with product purchaselikelihood? Journal of Advanced Research 2007;47(3):313–25.

Kover AJ. Copywriters' implicit theories of communication: an exploration. Journal ofConsumer Research 1995;21(4):596–611.

Krugman HE. The impact of television advertising: learning without involvement. PublicOpinion Quarterly 1965;29(3):349–56.

Laczniak RN, Muehling DD. Delayed effects of advertising moderated by involvement.Journal of Business Research 1990;20(3):263–77.

Laczniak RN, Muehling DD. Toward a better understanding of the role of advertisingmessage involvement in ad processing. Psychology and Marketing 1993;10(4):301–19.

Laczniak RN, Muehling DD, Grossbart S. Manipulating message involvement in adver-tising research. Journal of Advertising 1989;18(2):28–38.

Lazarus RS. On the primacy of cognition. The American Psychologist 1984;39(2):124–9.Lee YH. Manipulating ad message involvement through information expectancy:

effects on attitude evaluation and confidence. Journal of Advertising 2000;29(2):29–43.

Levy S, Nebenzahl ID. Program involvement and interactive behavior in interactivetelevision. International Journal of Advertising 2006;25(3):309–32.

Lichtenstein DR, Netemeyer RG, Burton S. Distinguishing coupon proneness from valueconsciousness: an acquisition–transaction utility theory perspective. Journal ofMarketing 1990;54(3):54–67.

Lord KR, Burnkrant RE. Attention versus Distraction: the interactive effect of programinvolvement and attentional devices on commercial processing. Journal of Adver-tising 1993;22:47–60.

Lumpkin JR. Shopping orientation segmentation of the elderly consumer. Journal ofAcademy of Marketing Science 1985;13(2):271–89.

MacInnis DJ, Moorman C, Jaworski BJ. Enhancing and measuring consumers' motivation,opportunity, and ability to process brand information from ads. Journal of Marketing1991;55(4):32–53.

Maheswaran D, Sternthal B. The effects of knowledge, motivation, and type of messageon ad processing and product judgments. Journal of Consumer Research 1990;17(1):66–74.

McLuhanM. Understandingmedia: the extensions ofman. Cambridge,MA:MIT Press; 1964.Mitchell AA. Involvement: a potentially important mediator of consumer behavior. In:

Wilkie WH, editor. Advances in consumer research, 6. Ann Arbor, MI: Associationfor Consumer Research; 1979. p. 191–6.

Mollen A, Wilson H. Engagement, telepresence and interactivity in online consumerexperience: reconciling scholastic and managerial perspectives. Journal of BusinessResearch 2010;63(9–10):919–25.

Petty RE, Cacioppo JT. Issue involvement can increase or decrease persuasion by enhancingmessage relevant cognitive responses. Journal of Personality and Social Psychology1979;37(10):1915–26.

Petty RE, Cacioppo JT. Communication and persuasion: central and peripheral routes toattitude change. New York: Springer-Verlag; 1986.

Petty RE, Cacioppo JT, Goldman R. Personal involvement as a determinant of argument-based persuasion. Journal of Personality and Social Psychology 1981;41(5):847–55.

Petty RE, Cacioppo JT, Schumann D. Central and peripheral routes to advertising effec-tiveness: the moderating role of involvement. Journal of Consumer Research1983;10(2):135–47.

Rappaport SD. Lessons from online practice: new advertising models. Journal of AdvancedResearch 2007;47(2):135–41.

RayML, Sawyer AG, RothschildML,Heeler RM, Strong EC, Reed JB.Marketing communicationand the hierarchy of effects. In: Clarke P, editor. Newmodels formass communicationresearch, 5. Beverly Hills, CA: Sage; 1973. p. 147–76.

Roseman IJ. Cognitive determinants of emotion. In: Shaver P, editor. Review of person-ality and social psychology. Emotions, relationships and healthBeverly Hills, CA:Sage; 1984. p. 11–36.

Scherer KR. Criteria for emotion-antecedent appraisal: a review. In: Hamilton V, BowerGH, Frijda NH, editors. Cognitive perspectives on emotion and cognition. Boston,MA: Kluwer Academic; 1988. p. 89-126.

Sengupta J, Johar V. Effects of inconsistent attribute information on the predictive value ofproduct attitudes: toward a resolution of opposing perspectives. Journal of ConsumerResearch 2002;29(1):39–56.

Siefert CJ, Kohturi R, Jacobs DB, Levine B, Plummer J, Marci CD. Winning the super buzzbowl: how biometrically-based emotional engagement correlated with onlineviews and comments for super bowl advertisement. Journal of Advanced Research2009;49(3):293–303.

Skerik S. The death of the impression & scaling true engagement. accessed online at:PRnews wire: beyond PR; 2011 http://blog.prnewswire.com/2011/05/19/the-death-of-the-impression-scaling-true-engagement/. on August 23rd 2011.

Slama ME, Tashchian A. Selected socioeconomic and demographic characteristics asso-ciated with purchasing involvement. Journal of Marketing 1985;49(Winter):72–82.

Smith RE, Yang X. Toward a general theory of creativity in advertising: examining therole of divergence. Mark Theory 2004;4(1/2):29–55.

Smith RE, Chen J, Yang X. The impact of advertising creativity on the hierarchy ofeffects. Journal of Advertising 2008;37(4):47–61.

Stern BB. A revised communication model for advertising: multiple dimensions of thesource, the message, and the recipient. Journal of Advertising 1994;23(2):5-15.

Wang A. Advertising engagement: a driver of message involvement on message effects.Journal of Advanced Research 2006;46(4):355–68.

Woodard B. Building ‘engagement’ one brick at a time. Journal of Advanced Research2006;46(4):353–4.

Zaichkowsky JL. Measuring the involvement construct. Journal of Consumer Research1985;12(3):341.

Zaichkowsky JL. Conceptualizing involvement. Journal of Advertising 1986;15(2):4-14.Zaichkowsky JL. The personal involvement inventory: reduction, revision, and application

to advertising. Journal of Advertising 1994;23(4):59–70.