Industrial And Hazardous Waste Management

D o e s T r a d i t i o n a l A d v e r t i s i n g T h e o r y

A p p ly t o t h e D i g i t a l W o r ld ?

A Replication Analysis Questions the Relevance

Of the Elaboration Likelihood Model

GAYLE KERR

Queensland University of

Technology

gf.kerr@qut.edu.au

DON E. SCHULTZ

Northwestern University

dschultz@northwestern.

edu

PHILIP J. KITCHEN

ESC Rennes School of

Business

philip.kitchen@esc-rennes.

com

FRANK J. MULHERN

Northwestern University

fjm274@northwestern.edu

PARK BEEDE

Higher Colleges of

Technology

pbeede@hct.ac.ae

All theory is based on a set of seminal concepts and empirical research that are assumed

to be replicable and inviolate overtime. Recent changes in technology, consumer habits,

demographics, and marketplaces, however, have raised questions about the applicability

of advertising theory developed in a mass-media environment to today’s interactive

marketplace. The current study explores this idea by replicating the most-cited study in

advertising research, the elaboration likelihood model, of which just three of 27 findings

were replicated. The current results advocate further replication of historical studies to

verify their current value for ongoing scholarship.

INTRODUCTION

A dvertising researchers owe m uch to the halcyon d ay s of m ass m edia. T hat in clu d es the e n te r­ tain m en t of television series “I Love L ucy/’ the inform atio n -g ath erin g m achine of the BBC, and the po w er of television to build em otional brand connections. In W estern cultures, the m ass-m edia p erio d —roughly from 1950 to 1980—particularly w as fruitful, encouraging a new w ave of ad v e r­ tising research. As one scholar noted, “Some of the best research ever done on ad v ertisin g w as done d u rin g the early days of television” (Bogart, 1986, p. 13). A lm ost all of ad v ertisin g ‘s p rem ier

academ ic journ als w ere established after televi­ sion (one of the first being the Journal of Advertis­ ing Research in 1960).

The w o rld has changed radically since those days of mass-m edia dominance. And, advertising has changed as well. A simplistic w ay to measure this change is th rough advertising expenditures. Between 2013 and 2014, advertising expenditure

• grew in N o rth Am erica (+5.4 percent) and the United Kingdom (+7.2 percent);

• was flat in continental Europe, notably Germ any (+1.5 percent) and France (-2.1 percent); and

• To be truly a science—and of value to practitioners—seminal advertising theory, such as the

elaboration likelihood model (ELM), must be replicable across different cultures and periods.

• In addition to replication, advertising theory also should be validated through the documentation

and scrutiny of its practice by marketers.

• Practitioners should question planning frameworks th at use traditional advertising models such as

the ELM, as they likely do not reflect how consumers think in a digital world.

• Advertising is not always a rational or controllable process, and practitioners should embrace new

systems of consumer thinking in driving advertising strategy, tactics, and investment.

390 m m i D F H D U E R T I S in G R E S E A R C H December 2015 DOI: 10.2501/JAR-2015-001

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL WORLD?

• soared in the emerging markets of China (+12.5 percent), India (+14.2 percent), and Brazil (+14.7 percent).1

Digital and Demographic Shifts

A nother w ay to look at advertising change is by the div ersio n of th a t ex p en d itu re from traditional m ass m edia to online and digital channels.

• In A ustralia, online advertising expend­ itu re grew by 190 percen t in the year June 2012 to June 2013, exceeding free- to-air telev isio n e x p en d itu res for the very first tim e.2

• By the en d of 2014, in 11 o th er co u n ­ tries, including China, m arketers spent m ore on d ig ita l a d v e rtisin g th a n on television.3

• In tern et advertising sp en d in g has the h ig h est g ro w th rate of any m ed iu m globally (up 18.5 percent in 2014)1 and increasing 30.3 percent annually in the M iddle East and Africa and 20.6 percent in Latin America.4

C o n su m er m ed ia h ab its, like p u rc h a s ­ ing beh av io rs, also h av e ch an g ed since the last half of the tw entieth century. The co m b in atio n of an a b u n d a n c e of con­ sum er choice and consum ers’ increasing access to inform ation has created a cornu­ copia of alternatives.

For exam ple, a 2012 stu d y of shoppers ages 20 to 40 reported th at 65 percent of U.K. and 55 percent of U.S. p articipants searched for p ro d u c ts online and w en t in-store to inspect them before going back

1 “Ware International A d Forecast 2014/15.” Ware News, December, 2014. 2 “Australian mobile spend up.” Ware News, A u g u s t 13, 2013. 3 “China’s digital adspend to surpass TV.” Ware News, February 18, 2014. 4 Nielsen Global AdV iew Pulse Report. Retrieved Octo­ ber 21, 2013, fro m http://nielsen.com/us/en/reports/2013/ global-adview-pulse-lite—ql-2013.htm l)

online to m ake their purchases.5 A round one-third used their sm artphone to com­ pare prices in-store with alternative outlets. This so-called “show -room ing” approach is grow ing around the w orld (Earley, 2014; M cCauley and Donofrio, 2014). In India, consum ers used mobile phone photos to generate agreem ent on planned purchases from fam ily and friends in the U nited States and the United Kingdom (Jain and Pant, 2012).

In a d d itio n to th ese m a rk e tp la c e changes, fundam ental dem ographic shifts have occurred as well. For exam ple, in 2013, w om en accounted for tw o-thirds or $12 trillion of the $18 trillion total in global consum er spending.4 A nother exam ple of dem ographic shift is the grow ing m iddle class of sh o p p e rs in C hina. Because of their enthusiasm for online shopping and their enhanced financial position over the p a st few decades, C hina has overtaken the U nited States as th e w o rld ‘s leader in e-commerce.6

In sum m ary, advertisin g has evolved from a m ass-m edia m arketplace—dom i­ nated by the U nited States—to one driven by digital an d mobile m edia, buoyed by the grow th of em erging m arkets. This is n ot just the result of changing consum er m edia habits, decision m aking, and p u r­ chasing power, b u t it also appears to be part of the rise of a transform ative global society: M assive social, m arketing, and m edia changes clearly are reflected in advertising expenditure and allocation.

Is Traditional Advertising Theory

Still Relevant?

Given all these changes, the current study questions whether the foundational adver­ tising theories—constructed d u rin g the d ays of m ass m edia d om inance and a

5 “Consumers m ix channels in US and UK.” Ware News, November 20, 2012. 6 “China is biggest ecommerce m a r k e t Ware News, A u g u st 29, 2013.

United States-centric m arketplace—remain relevant today.

Although there is discussion (even dis­ quiet) ab o u t it am ong academ ics—and some empirical evidence to support these challenges—th e current article proposes that the best w ay to examine the relevance, rigor, and applicability of historic advertis­ ing theory is through empirical testing. In other w ords, if a d v ertisin g ‘s earlier so- called “sem inal research stu d ie s” w ere conducted again, the authors of the current study asked, w ould the original results be confirmed?

Thus, the position of the cu rren t a rti­ cle is sim ple: If one of the m ost-cited ad v ertisin g stu d ies could be replicated, som e of the grow ing concerns ab o u t the applicability of the historical advertising theory base in a changing w orld w ould be allayed. S ubstantial differences, if found betw een p ast studies and current replica­ tions, w ould

• lend su p p o rt to the cu rren t academ ic debate, and

• provide direction for subsequent inves­ tigations of the tra d itio n a l a d v e rtis­ ing fram ew orks th a t s u p p o rt cu rren t research approaches and guide ad v er­ tising practice.

B ecause c ita tio n s are th e a c c e p te d “c u rre n c y ” of a d v e rtisin g sch o larsh ip , the c u rre n t s tu d y te ste d one of the m o s t- c ite d s tre a m s of a d v e r tis in g research: th e lengthy, b ro a d , and deep w o rk co n d u c te d on the d e v e lo p m e n t, testing, and application of the elaboration likelihood m odel (ELM; Petty, Cacioppo, Schum ann, 1983).

Of all ad v e rtisin g th eo ry pillars, the ELM is the m ost frequently cited source of academ ic lite ra tu re b y a d v e rtis in g researchers (Pasadeos, P helps, an d Edi­ son, 2008; Kitchen el al, 2014). Further, it is considered to be “the m ost influential

D ecem ber 2 0 1 5 J D U R f l H L O F H D U E R T I S M G R E S E A R C H 391

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL WORLD?

theoretical contribution” (Beard, 2002, p. 72). Thus, the authors of the current study believe, a replication of that 1983 study would do much to allay the con­ cerns of current day researchers.

Such replication also w ould perm it the examination of the basic premises of advertising research, which clearly have changed over time. Traditional research from the 1950s through to the 1980s was based on the premise that “advertising is something one does to people” (Stewart, 1992, p. 15). The latter is a holdover from the “hypoderm ic” (or “magic-bullet”) approach that defined behaviorism in the 1930s (Berger, 1995) and was rooted in experiences of a rapidly growing market­ place—with few media options and lim­ ited consumer knowledge and choice.

Fast-forward to the digital age: Those concepts may no longer apply, as today’s empowered consumers have increasing control over most aspects of the adver­ tising process (Kerr and Schultz, 2010; Kitchen and Uzunoglu, 2015).

It is, therefore, important that advertis­ ing be explored in context—and across contexts—rather than in isolation. As one scholar noted, “A typical research para­ digm within the field uses relatively naive consumers, fictitious products, forced exposure to advertising for a single prod­ uct, and measures that are designed to identify incremental changes” (Stewart, 1992, p. 7). Such practice perhaps was an artifact of advertising research’s positivist traditions and borrowings from experi­ mental psychology (Bogart, 1986; Heath and Feldwick, 2008; Heath, 2012; Kerr and Schultz, 2010).

It is also a concern, however—one that was raised at the 2013 Wharton Confer­ ence on Empirical G eneralizations in Advertising. At that gathering, many del­ egates advocated that generalizability be explored by using multiple data sets across m ultiple contexts. “Rigor comes from

results that hold over and over, ideally when conducted by different researchers who use fully transparent processes, data, analyses, and results” (Wind, Sharp, and Nelson-Field, 2013, p. 178).

Finally, the current authors contend that their study is im portant from the practitioners’ perspective. Many agency­ planning models, which drive advertis­ ing strategy, tactics, and investment, are underpinned by models and theories from the 1970s and 1980s (Heath and Feldwick, 2008). A prime example is the linear, one­ way approach of the hierarchy of effects model, which still underpins most media planning today (Heath, 2012). There would appear to be substantial increases in adver­ tising efficiency and financial gain in using planning models that correctly reflect today’s consumer, media systems, and marketplace, rather than the standards of an earlier marketing ecosystem.

LITERATURE REVIEW

The ELM emerged from the maelstrom of conflicting literature, conceptual ambi­ guities, and methodological problems that had defined the field of persuasion and attitude change in the 1960s and 1970s (Fishbein and Ajzen, 1972; Petty and Cacioppo, 1983). ELM theorists pro­ vided a desperately needed, yet simple, concise framework that would include both cognitive argument quality and heu­ ristics (Schumann, Kotowski, Ahn, and Haugtvedt, 2012).

The resultant ELM advocates two basic routes to persuasion: the central and the peripheral, determined by the amount of cognitive effort a person used to process a message (Schumann et at., 2012).

• Central route to persuasion: When elaboration likelihood is high, informa­ tion processing will occur via the cen­ tral route. Attitude change will be more persistent (Haugtvedt and Petty, 1989)

and predictive of behavior (Petty and Cacioppo, 1983).

• Peripheral route to persuasion: When little cognitive effort is expended and elaboration is low, processing may occur via the peripheral route, relying upon cues such as source credibility and heuristics (Petty and Cacioppo, 1983) to enable the persuasion.

Criticisms of the ELM

Despite being heralded as one of the most influential advertising-research theories (Szczepanski, 2006), the ELM also has been one of the most criticized. This criticism includes fundamental constructs such as (Kitchen et al.r 2014):

• the dual-processing framework;

• the idea of a continuum of elaboration;

• the definition of the mediating variables and independent variables; and

• the fact that the model is descriptive, not analytical.

Instead of being explored in the current study, these criticisms were acknowl­ edged as issues that remain empirically unresolved. The current authors noted that these criticisms have not had an impact on the influence (or use of) the ELM by adver­ tising scholars.

Replication Attempts

Despite the pervasiveness—and continued criticism—of the ELM over the last three decades, very few studies have sought to replicate the original ELM experiment in its entirety. Instead, most studies have focused on trying to replicate a portion, variable, or construct of the ELM (Kang and Herr, 2006; Te’eni-Harari, Lampert, and Lehman-Wilzig, 2007; Trampe, Stapel, Siero, and Mulder, 2010).

3 9 2 J O U R n R L O F H D U E R T I S M G R E S E A R C H D ecem ber 2 0 1 5

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL WORLD?

On the one hand, those who did seek to replicate the original ELM study unani­ mously questioned the model’s validity. For example, scholars who closely repli­ cated the original model—using slightly different products—found little or no sup­ port for the ELM (Cole, Ettenson, Reinke, and Schrader, 1990). In a meta-analysis, there was concern that only researchers associated with the original researchers, Petty and Cacioppo, were able to generate results consistent with the ELM’s predic­ tions (Johnson and Eagly, 1989).

On the other hand, failure to replicate the results of the original study, most likely, was the result of modifications or exclusions of critical substantive features of the ELM, the original authors of the theory argued (Petty, Kasmer, Haugtvedt, and Cacioppo, 1987).

RESEARCH QUESTION

The current authors chose the seminal ELM study (Petty, Cacioppo, and Schu­ mann, 1983) for replication for a number of reasons:

• An in itia l s tu d y (P e tty a n d Cacioppo,1981) failed to provide any evidence of a peripheral route to persua­ sion (Petty and Cacioppo, 1983).

• The authors of the original study described the 1983 experiment as a “more sensitive test of the two routes to persua­ sion” (Petty and Cacioppo, 1983, p. 18).

• The 1983 study is the most republished of all of Petty, Cacioppo, and Schu­ mann’s work.

Guiding this replication, the research ques­ tion for the current study was:

RQ1: Does the ELM explain how to d a y ‘s consum ers process advertising and change attitudes through the central and periph­ eral routes to persuasion?

METHODOLOGY

The authors of the current study noted that they replicated the 1983 study faith­ fully, in its entirety and, for the first time, in three different countries: the United States (where the original was conducted), the United Kingdom, and Australia.

Like the original 1983 experiment, the replication used a 2 x 2 x 2 factorial design, manipulating the independent variables of message processing involvement (high/ low), argument strength (strong/weak), and source characteristics (high/low).

S a m p le

The 1983 experiment used a sample of 160 male and female undergraduate stu­ dents in a major M idwestern American university. In the current replication, the samples generally were larger and rep­ resented a larger global cross-section but still focused on a group of sample sub­ jects comparable to the original group of undergraduates:

• 218 in Australia, • 315 in the United Kingdom, and • 140 in the United States.

To ensure that the different results across the three countries did not reflect cultural differences, participants in Australia, the United Kingdom, and the United States studies were compared across the six dimensions of Hofstede’s cultural tool comparison. The three countries, rated from 0 to 100, scored almost identically on

• power distance (36, 35, 40); • individualism (90, 89, 91); • masculinity (61, 66, 62); and • indulgence (71, 69, 68).

Australia and the United States (51, 46) were stronger on uncertainty avoidance than the United Kingdom (35), although the United Kingdom (51) was far more

pragmatic compared to Australia and the United States (21, 26). Given the cultural similarities of the three countries, dif­ ferences were unlikely in cross-national responses to scales.

In d e p e n d e n t V ariab le s

The independent variables were v irtu ­ ally identical to those cited in the 1983 experiment:

• Involvement: Participants were given two booklets containing stimulus mate­ rial and a questionnaire. In the first book­ let, involvement was measured in the same two places—using the same two devices—as the original 1983 experiment.

• Endorsers (peripheral cues): Like the original experiment, the test material con­ tained both non-famous endorsers (who were unknow n and average-looking male and female models) and local celeb­ rities relevant to the market in which the advertisements were being tested (i.e. different sports stars from Australia, the United Kingdom, and the United States).

• A rgum ent strength: Like the 1983 experim ent, the current stu d y also contained different treatm ents using weak and strong arguments promoting disposable razors. Arguments in the original study, however, such as “floats in water with a minimum of rust” or “designed with the bathroom in mind” were not considered relevant or persua­ sive to today’s test groups. Copy points, therefore, were collected from the web­ sites of three leading disposable razor m anufacturers: Schick, W ilkinson- Sword, and Bic. They were evaluated by an expert panel and m atched as closely as possible with the original advertising claims, in terms of argu­ ment valence (logical or emotional) and strength (strong or weak).

D ecem ber 2 0 1 5 JQURI1RL OF RDUERTISIRG RESEARCH 3 9 3

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL WORLD?

Dependent Variables

The d e p e n d e n t variables from the 1983 s tu d y also w ere u se d in th e c u rre n t experim ent:

• A ttitu d es: W hat the 1983 s tu d y had defined as an ” a ttitu d e m e a su re ” or ” a ttitu d e in d ex ” w as re p re se n te d in the current stu d y as the average of the three scores—on a per-subject b asis— from the 9-point sem antic differential scales th a t m e a su re d overall im p res­ sion, expected satisfaction, and favora­ bleness of the Edge disposable razor.

• Purchase Intentions: This variable was rated on a 4-point scale.

RESULTS PART 1

M anipulation Checks

In a m anipulation check of involvem ent,

• 75 percent of U.S. participants, 70 per­ cent of U.K. p articip an ts, an d 50 p e r­ ce n t of A u s tra lia n p a rtic ip a n ts in high-involvem ent conditions correctly recalled they w ere to select a b ran d of disposable razor.

• In low -involvem ent conditions, 79 per­ cent of U.S., 70 percen t of U.K., and 63 percen t of A u stralian p a rticip an ts c o rre c tly re c a lle d th e a lte r n a tiv e incentive.

• The foregoing re su lts com pare w ith 93 p ercen t for h ig h inv o lv em en t and 78 percent for low involvem ent in the original study.

In the endorser-m anipulation check, two q u e stio n s w ere a sk ed , re p lic a tin g the o rig in al study. The first q u estio n w as about recognition:

• 74 percent of U.K., 36 percent of A ustral­ ian, and 36 percent of U.S. participants

in d ic a te d reco g n itio n , c o m p a re d to 94 percent in the original study.

The se c o n d q u e s tio n co n c e rn e d th e resp o n d e n ts’ liking of the people in the advertisem ent:

• The celebrity w as liked m ore in the U nited States (5.36 com pared to 4.49 for an ordinary citizen) and in the original stu d y (6.06 com pared to 3.64).

• In the U nited K ingdom and A ustralia, there w as no difference in term s of the likeability of celebrities an d o rd in ary citizens.

In the o rig in a l s tu d y ‘s m a n ip u la tio n check for argum ent-persuasiveness, su b ­ jects exposed to strong arg u m en ts rated them significantly m ore persuasive (M = 5.46) than those exposed to w eak a rg u ­ m ents (M = 4.03).

This also w as th e case in the cu rren t stu d y w here, in th e U n ited K ingdom , strong arg u m en ts led to a h igher m ean score. In th e U n ited States a n d A u s­ tralia, strong argum ents w ere considered no m ore p e rsu a siv e th a n w eak a rg u ­ m ents. This is explored fu rth e r in the next section.

RESULTS PART 2

The results on the d ependent variables— attitudes and purchase intentions—from the three adm inistrations of the current stu d y (A ustralia, U nited States, and the U nited Kingdom) bore little resemblance to the original results from 1983 (See Tables 1 and 2).

In the replicated study, for the sam e d ependent variables, the m eans typically w ere close to the m id p o in t (zero) an d show ed m inim al differences betw een the high- an d lo w -involvem ent conditions for endorser and argum ent strength (See Table 1).

Attitudes and Involvement

In the original study, the attitu d e index w as h ig h e r for th e lo w -in v o lv e m e n t g ro u p (m ean score = 0.99) th an for the hig h -in v o lv em en t g ro u p (m ean score = 0.31).

A m ong the cu rren t s tu d y ‘s three re p ­ lications in the re-test, tw o of them , the U.K. and A ustralian respondents, show ed no significant difference in the m ean atti­ tu d e score across the involvem ent tre a t­ m ents. In the U.S. study, the difference in the attitude score approached significance (p = 0.064) b u t in the opposite direction of the 1983 study. That is, the attitu d e score w as h ig h er for th e h ig h er inv o lv em en t group than the low er involvem ent group (See Table 2).

Hence, the 1983 results w ere n ot con­ firm ed in any of the three replicated studies.

Attitudes and Endorsers

In term s of the im pact of the celebrity en d o rser on a ttitu d es tow ard the razor b ran d , the 1983 stu d y claim ed to find a m ain effect, indicating that advertisem ents featuring celebrity endorsers led to a more positive attitude score (0.86 for celebrity com pared to the non-celebrity m ean of 0.41). Notably, that conclusion was reached despite the p value being 0.09.

In th e th re e -s tu d y re p lic a tio n , the en d o rser effect w as significant only in the U.K. study w here the citizen endorser actually led to a higher attitude than the celebrity—the opposite of w h at the 1983 study claimed.

Attitudes and Argument Strength

The third m ain effect tested the im pact of strong versus w eak argum ents. The origi­ nal study found a m ean attitude score of 1.65 for the strong argum ent and a -0.35 for the w eak argum ent. That finding was replicated in the U.K. d ata (0.86 versus 0.35; p = 0.004).

3 9 4 JO U R N A L OF H D U E R T IS IflG R ESER RCH December 2015

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL WORLD?

TABLE 1 Means and Standard Deviations for Each Experimental Cell on the Attitude Index

Low Involvement High Involvement

W eak

Argument

Strong

Argument W eak

Argument Strong

Argument

AUS Citizen 1.26

(1.00) 0.58

(1.24) 0.99

(1.32) 0.94

(1.31)

Celebrity 0.87

(1.35) 0.89

(1.40) 0.78

(1.52) 0.40

(1.79)

UK Citizen 0.47 1.38 0.72 0.97

(1.42) (1.33) (1.61) (1.36)

Celebrity 0.16 0.85 0.04 0.27 (1.86) (1.58) (1.86) (1.27)

US Citizen 0.28 0.71 0.69 1.47

(1.94) (1.27) (1.43) (1.34)

Celebrity (0.02) 0.40 1.08 0.53 (1.66) (1.75) (1.68) (1.63)

PCS 1 9 8 3 Citizen -0.12 0.98 -1.1 1.98

(1.81) (1.52) (1.66) (1.25)

Celebrity 1.21 1.85 -1.36 1.80 (2.28) (1.59) (1.65) (1.07)

Overall, among the nine attempts to rep­ licate the 1983 study results for the impact of the three treatments on attitudes, this is the only one incident where the results rep­ licated the 1983 study.

Interaction Effects

The 1983 study reported the interaction to be significant (p = 0.02), and this finding was replicated in the United Kingdom (p = 0.006) but not in the United States (p = 0.6) or Australia (p = 0.2).

In the 1983 study, the high-involvement situation featured a large difference on the impact of a strong versus weak argu­ ment on the attitude score, while the low- involvement situation had no such effect

(See Table 3). In the three replications, the only significant result was in the United Kingdom, where exactly the opposite was found: The strength of argument mattered in the low-involvement condition but not the high-involvement one.

The final interaction effect considered was the endorser by argument strength interaction as it affects the attitude score. This was unable to be replicated in any of the three studies (See Table 3).

Purchase Intentions

With respect to purchase intentions, the second dependent measure, the 1983 study found that strong arguments led to a mean attitude score of 2.23 compared to the

T h e t h ir d m a in e f f e c t

t e s t e d t h e im p a c t o f s tr o n g

v e r s u s w e a k a r g u m e n ts .

mean score of 1.68 for weak arguments (p < 0.001), indicating that strong arguments led to higher purchase intentions.

In the current study, none of the three country replications found a significant effect of argument strength on purchase intentions with the United States (p = 0.817), the United Kingdom (p = 0.255), and Australia (p = 0.97).

In addition to finding a main effect for argument strength on purchase inten­ tions, the 1983 study found that, in high- involvement conditions, the strength of the argument was more important than in low-involvement conditions. In the cur­ rent study, the same result was found in the United Kingdom, but the impact of strength of argument did not vary for the high- or low-involvement conditions in the two other countries.

The 1983 study reported a correlation that was higher for the high-involvement condition (0.59) th an for the low- involvement condition (0.36). In the current U.K. replication, however, the correlations were about the same for both high- and low-involvement conditions while, in the U.S. and Australian replications, the low- involvement conditions exhibited a higher correlation between attitude and purchase likelihood (See Table 4).

Again, the authors found little evi­ dence to confirm the findings of the 1983 paper.

DISCUSSION

The overall findings of the 1983 study— that attitudes formed via the central route

D ecem ber 2 0 1 5 JOUHIML OF HDUERTISinG RESERRCH 3 9 5

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL WORLD?

TABLE 2

Involvement, Endorser, and Argument Influence on Attitude Index AUS UK US PCS 1 9 8 3

Involvement Groups Means:

High Involvement 0.855 0.495 0.965 0.310

Low Involvement 0.849 0.717 0.319 0.990

N = 218 317 142 149

Test Statistics: F= 0.001 P= 0.973 F= 1.553 P= 0.214 F= 2.806 P = 0.064 F= 6.640 P = 0 .0 1

Endorser Groups Means:

Celebrity 0.793 0.331 0.471 0.860

Citizen 0.910 0.881 0.818 0.410

N = 218 317 142 149

Test Statistics: F= 0.386 P= 0.535 F= 9.731 P= 0.002* F= 0.831 P= 0.438 F= 2.910 P= 0.090

Argument Groups Means:

Strong 0.795 0.864 0.773 1.650

Weak 0.904 0.348 0.514 -0.35 0

N = 218 317 142 149

Test Statistics: F= 0.339 P= 0.561 F= 8.55 P= 0.004* F= 0.484 P= 0.617 F= 57.81 P= 0.0001*

TABLE 3

Involvement x Argument Interaction Impact on Attitude Index

F P

AUS Overall Model F( 1, 218) = 1.114 0.292

Low Involvement F ( l, 86) = 1.288 0.260

High Involvement F ( l, 131) = 0.500 0.481

UK Overall Model F( 1, 317) = 2.567 0.110

Low Involvement F ( l, 157) = 10.287 0.002*

High Involvement F ( l, 158) = 0.872 0.352

US Overall Model F (l, 142) = 0.172 0.679

Low Involvement F( 1, 69) = 1.043 0.311

High Involvement F ( l, 71) = 0.265 0.608

are more predictive of behavior than those formed via the peripheral route—could not be confirmed in the current study despite im plem enting the same treat­ ments and data-collection process in three different countries.

One important difference between the original study and the current replication is that, although the manipulations worked well in at least one (and sometimes two) of the three countries, they clearly were not as strong as in 1983.

This supports the contention that con­ sumers likely think differently in a faster, digital world.

As some of the manipulations worked quite well, however, it is further suggested that the mental processing of information did not work in the way the ELM purports. In other words, the current authors believe

TABLE 4

Correlation between Attitude Index and Purchase Likelihood

C o r r e la tio n S ig .

A U S

High Involvement 0.302 0.000*

Low Involvement 0.452 0.000*

UK High Involvement 0.445 0.000*

Low Involvement 0.437 0,000*

US High Involvement 0.329 0.005*

Low Involvement 0.526 0.000*

PCS 1 9 8 3 High Involvement 0.590 0.001*

Low Involvement 0.360 0.001*

3 9 6 JOURIIHL OF HDUERTISIflG RESEARCH D ecem ber 2 0 1 5

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL WORLD?

In 1 9 8 3 , a d v e r t is e m e n t s f e a t u r in g a c e le b r it y

e n d o r s e r le d t o m o r e p o s it iv e a t t i t u d e s c o r e s .

it m ay be incorrect to conclude that the fail­ ure to replicate w as simply a m atter of the m anipulations being weaker, even though they w ere replicated as closely as possible.

T herefore, it is im p o rta n t to look at alternative explanations in the divergent findings.

D iv e r g e n t F in d in g s

Among the findings that opposed the orig­ inal stu d y were the following :

• Involvem ent In th e o r ig in a l s tu d y , th e low – involvem ent group had a m ore positive a ttitu d e to w ard the object. In the U.S. replication, however, people in the low- involvem ent g ro u p w ere m ore sk ep ti­ cal and h ad a low er a ttitu d e score than those in the high-involvem ent group.

This is the reverse of the ELM’s predic­ tions, yet the findings support the social judg m en t theory (Sherif and H ovland, 1961), w hich suggests that uninvolved people will consider a w ider range of alternatives than those w ho are more highly involved w ith the object.

It also is w orthw hile to consider that the a ttitu d e to w a rd the object m ay n o t be th e only d eterm in an t. As the theory of p la n n e d beh av io r suggests, h o w ev er (Ajzen an d Fishbein, 1991), attitu d e to w ard buy in g also could be im p o rta n t—a co nsideration th a t m ay be even m ore im portant in today’s digi­ tal environm ent w ith greater access to product inform ation, m ore w ays to buy, an d m ore em p o w ered skeptics look­ ing tow ard custom er reviews as a more trusted source than m arketing inform a­ tion (K rishnam urthy and Dou, 2008). In such instances, people m ay be not so

m uch “involved” in the product as they are “connected” to information.

• E ndorser In 1983, a d v ertisem en ts fe atu rin g a celebrity e n d o rse r led to m ore p o si­ tive attitude scores. In the U.K. replica­ tion, however, the opposite effect was found. A dvertisem ents featuring citizen endorsers h ad a higher attitu d e than celebrity advertisem ents.

A gain, this m ig h t be an artifact of the grow ing belief in citizens as more reliab le sources of in fo rm a tio n an d th e acceleratio n of electronic w o rd of m o u th (K rish n am u rth y an d Dou, 2008). Such credibility also is evid en t in the escalation of “reality” television show s, w here the average citizen is the celebrity.

• Interaction Effects In the original study, strength of arg u ­ m en t w as im p o rta n t in high- b u t not in low -involvem ent conditions. In the current study, the U.K. results show ed the opposite. A rg u m en t stren g th w as significant for low, ra th e r th a n h ig h involvem ent.

The notion that “if you buy something you m ust like it,” as suggested by the self perception theory (Bern, 1972), could be applied to the high-involvem ent group. This also is su p p o rte d by K rugm an’s (1965; 1966-1967) notion that behavior sometimes comes before attitude.

Equally, the stre n g th of arg u m en t being significant in low -involvem ent conditions is supported by social judg­ m ent theory (Sherif and H ovland, 1961), which suggests the uninvolved typically consider a w ider range of alternatives.

This is am plified in the n o tion th a t “because I am n ot involved, I need to be convinced.” More than anything else, this show s th a t con ten t—rath er than co ntent m a n ip u la tio n —som etim es is m ore im p o rta n t for low -involvem ent conditions, disagreeing w ith the essen­ tial premise of the ELM.

• C orrelations In the original study, there w as a sig­ nificant p ositive correlation betw een attitude tow ard the pro d u ct and likeli­ hood to purchase in both the high- and low -involvem ent conditions (although s tro n g e r in th e h ig h -in v o lv e m e n t condition).

In the current study, in Australia and the U nited States, a m ore positive atti­ tu d e tow ard the object w as associated w ith greater likelihood to purchase in lo w -in v o lv em en t con d itio n s, w ith a lower correlation for high-involvem ent conditions. Perhaps, the authors of the current stu d y suggest, sim ply “liking” an advertisem ent, rath er than consid­ erin g the elab o ratio n of co n sid ered arg u m en t, leads to p urchase in low- involvem ent conditions.

This result also could be explained by new er m odels of thinking, such as “Thinking Fast and Slow” (Kahneman, 2011).

-y- Thinking fast (or “System 1 thinking”) is typical of low -involvem ent condi­ tions, w here thin k in g is autom atic, and the em otion w here “som ething h ap p en s to you” produces an au to ­ m atic response, free from v o luntary control. In the case of these findings, autom atic thinking generates in ten ­ tion to purchase.

More effortful or slow thinking—per­ h aps akin to high elaboration—only is activated w hen System 1 thinking

Decem ber 2 0 1 5 J D U e n B L O F H D U E R T I S in G R E S E A R C H 39 7

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL WORLD?

does not have an answer or when its model of the world is violated.

Low attention has been the focus of much scholarly work (Heath, 2012). It suggests that television advertis­ ing is not processed systematically, but rather like System 1, it is auto­ matically processed in response to stimuli.

Advertisements high in emotional content generally received 20 per­ cent less attention (Heath et ai, 2009). Lower attention could reduce counter­ argument and, therefore, increase like­ lihood of purchase.

In summary, the results of this three- study replication diverge from the prem­ ise of the ELM model. In all instances, the respondents went through an evalu­ ation process, albeit through two differ­ ent pathways. However, the findings do support the contention of more recent research that there can be learning (and even persuasion) as a result of subcon­ scious processing of advertising expo­ sure, suggesting exposure may be more important than processing (Heath, 2012; Kahneman, 2011).

IMPLICATIONS

The current authors believe that the cur­ rent study has a number of implications for both academics and practitioners:

Replication should be an inherent and

ongoing part of theory validation.

As an objective akin to finding a way to “world peace,” revisiting and replicat­ ing advertising theory is an overwhelm­ ing task. It is likely that such efforts will upset a number of academicians who have built their entire careers on following the dictates of “the literature.” The results of the current study and the directives of a num ber of academics, however—among them, many of the participants at Wharton

Conference on Empirical Generalizations in Advertising—validate the urgent need to take on this task.

Journal editors and reviewers should

lead the way.

As guardians of research quality, editors and reviewers have an obligation to ques­ tion the rigor and the appropriate use of theory in research. Hence, many academic journals and associations have champi­ oned research quality.

• Kent Monroe, then editor of the Journal of Consumer Research, was a lone voice for replication in the 1990s, promoting a clear editorial policy of encourag­ ing and accepting replication research for publication.

• The Journal of Advertising Research has encouraged debate with its “New Mod­ els for a New Age of Research” issue (Vol. 51, Issue 2) and “Future of Market Research” (Vol. 51, Issue 1; 2011)

• Charles Taylor, International Journal of Advertising editor, confirmed the jour­ nal’s commitment to research involving replication, publishing a call for stronger theory development and more relevant research for advertising professionals (Taylor, 2011).

Academic associations must work

together.

The American Academy of A dvertis­ ing and European Advertising Academy both have considered the topic of research quality w orthy enough to feature it in their keynote addresses. Action must fol­ low awareness, however: If the agenda is to revisit advertising theory—and if edi­ tors and reviewers are the guardians of research quality—academic associations should provide the necessary leadership to support that view.

A c a d e m ic a s s o c ia t io n s

m u s t w o r k t o g e t h e r .

Practitioners should document the

practice of theory.

It is contingent upon practitioners—the implementers of advertising theory—to document conditions under which theory works and those conditions that oppose it. Their findings should be published in peer-reviewed journals, where practition­ ers and academics can learn from the prac­ tice of theory.

Advertising is not always a rational

process.

Practitioners should not be constrained by an organizational view that sees advertising as a manageable, informational resource for rational consumers (Heath, 2012). They should embrace new technology (such as neuroscience) and new thinking (like Thinking, Fast and Slow [Kahneman, 2011]) or even more emotion-centric ideas (like implicit communication or low attention). These all are concepts more challenging than a central route to persuasion but per­ haps better reflective of today’s consumer and today’s marketplace.

CONCLUSION

To question the relevance of advertis­ ing theory, the current study empirically tested its most cited work, the ELM (Petty e ta l, 1983).

What those scholars found in 1983 could not be replicated today in any of the three countries in which the current study was conducted. This global inability to replicate one of the most fundamental experiments from advertising’s halcyon mass-media days suggests advertising scholars need to re-think the assumptions and foundations of what they call “advertising theory.”

398 J O U R O H L O F R D U E R T I S If l G R E S E A R C H D e c e m b e r 2 0 1 5

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL WORLD?

Just because it has been cited a number of times and “everyone” believes it to be true does not necessarily mean a theory is relevant or even empirically generaliz- able given the massive changes that have occurred in the marketplace.

The onus is on the marketing-research industry and academia to question adver­ tising theory: When everything around it has changed, why should any particular theory stay the same? And if advertis­ ing theory is not questioned, subsequent advertising research will become increas­ ingly irrelevant.

ABOUT THE AUTHORS

G ayle K err is a professor a t Queensland University of

Technology School o f Advertising, M arketing, and Public

Relations in Brisbane, Australia, where she teaches

advertising and integrated-m arketing com m unications

(IMC). Her advertising research interests include

consum er em powerm ent in digital and social spaces,

ad vertising self-regulation, ethics, and management.

Her IMC research has focused on integration and

measurem ent, and her stud ies have been published in

a nu m ber o f research journals, including the European

Journal o f Marketing, International Marketing

Review, International Journal o f Advertising, Journal

o f Advertising Research, and Journal o f Marketing

Communications.

D on E. S chultz is professor (Emeritus-in-Service) of

integrated marketing com munications, The Medill School,

Northwestern University, Evanston, IL , and president

o f Agora, Inc., a global marketing, communication, and

branding consulting firm. He consults, lectures, and

holds sem inars on integrated marketing communication,

marketing, branding, advertising, sales promotion, and

com m unication management worldwide. He is th e a u th o r/

co-author o f 28 books and more than 1 5 0 academic and

professional articles.

P hilip J. K itchen is research professor of m arketing a t

ESC Rennes School o f Business, Rennes, France, He

specializes in m arketing and corporate com m unications

and m arketing theory. He has published papers on

these and related topics in th e Journal o f Advertising

Research, Journal o f Business Research, Journal

o f Marketing Education, Journal o f Marketing

Management, Journal o f Business Ethics, and

European Journal o f Marketing, am ong other journals.

He has also published more than 2 0 books in these

areas including co-editorship o f Word o f Mouth and

Social Media (Routledge, 20 15) and Integrated

Communications in the Postmodern Era (Palgrave-

M acmillan, 2015).

F rank J. M ulhern is the Hamad bin Khalafi Al-Thani

professor o f integrated marketing com m unications and

is associate dean a t th e Medill School, Northwestern

University. He specializes in technology, promotion

marketing, quantitative analysis of advertising media

effects, and integrating internal and external marketing

com munications. M ulhern’s articles have appeared in

scholarly journals including the Journal o f Marketing,

Journal o f Retailing, Journal o f Advertising, International

Journal o f Research in Marketing, Journal o f Interactive

Marketing, and Journal o f Business Research. He is

co-author of the textbook Marketing Communications:

Integrated Theory, Strategy and Tactics (Pentagram

Publishing, 1999).

P ark B eede is chair o f graduate business programs at

th e Higher Colleges o f Technology in the United Arab

Emirates. His research activities include advertising,

branding, and entrepreneurial marketing, in addition

to academ ic roles, he has held senior professional

positions in leading advertising agency and corporate

client organizations, including Colenso/BBDO, Euro

RSCG, and Fonterra.

REFERENCES

A zjen, I. “The Theory of Planned Behavior.”

Organizational Behaviour and Human Decision

Processes 80 (1991): 179-211.

Beard, F. “Peer Evaluation and Readership of

Influential Contributions to the Advertising

Literature.” Journal of Advertising 31, 4 (2002): 65-75.

Berger, A. Essentials of Mass Communication The­

ory. London, UK: Sage Publications, 1995.

Bern, D. “Self-Perception Theory.” In Advances

in Experimental Social Psychology, vol. 6,

L. Berkowitz, ed. New York: Academic Press,

1972.

Bogart, L. “Progress in Advertising Research?”

Journal of Advertising Research June/July (1986):

11-15.

C ole, C., R. Ettenson, S. R einke, and T.

Schrader. “The Elaboration Likelihood Model

(ELM): Replications, Extensions, and Some Con­

flicting Findings.” Advances in Consumer Research

17 (1990): 231-236.

Earley. S. “Mobile Commerce: A Broader Per­

spective.” IEEE Computer Society 16, May-June

(2014): 61-65.

F ishbein, M., and 1. A jzen. “Attitudes and Opin­

ions.” Annual Review of Psychology 23 (1972):

487-544.

H augtvedt, C., and R. P etty. “Need for Cogni­

tion and Attitude Persistence.” Advances in Con­

sumer Research 16 (1989): 33-36.

H eath, R. Seducing the Subconscious: The Psychol­

ogy of Emotional Influence in Advertising. West

Sussex, UK: Wiley-Blackwell, 2012.

H eath, R., and P. Feldwick. “Fifty Years Using

the Wrong Model of Advertising.” International

Journal cf Market Research 50,1 (2008): 29-59.

H ofstede, G. “The Hofstede Centre.” Retrieved

April 14,2014, from http://geert-hofstede.com/

australia.html

Jain, V., and S. P ant. “Navigating Generation

Y for Effective Mobile Marketing in India: A

Conceptual Framework.” International Journal cf

Mobile Marketing 7,3 (2012): 17-26.

D ecem ber 2 0 1 5 J o i n O F H D U E R T I S If l G R E S E A R C H 3 9 9

 

 

DOES TRADITIONAL ADVERTISING THEORY APPLY TO THE DIGITAL W ORLD?

Johnson, B., and A. Eagly. “Effects of Involve­ ment on Persuasion: A Meta-Analysis.” Psycho­

logical Bulletin 106, 2 (1989): 290-314.

K ahnem an, D. Thinking, Fast and Slow. New York: Farrar, Straus and Girous, 2011.

K a n g , Y., and P. H err. “Beauty and the Beholder: Toward an Integrative Model of Com­

munication Source Effects.” Journal of Consumer Research 33 (2006): 123-130.

K err, G., and D. Schultz. “Maintenance Man

or Architect? The Role of Academic Advertising Research in Building Better Understanding.”

International Journal of Advertising 29, 4 (2010): 547-568.

K itchen, P. J., and E. U zunoglu (Eds.). Inte­

grated Communications in the Postmodern Era. Basingstoke: Palgrave-Macmillan, 2015.

Kitchen, P. J., G. K err, D. Schultz, R. M cC oll, and H. P als. “The Elaboration Likelihood Model: Review, Critique and Research Agenda.”

European Journal of Marketing 48, 11-12 (2014): 2033-2050.

Krishnamurthya, S., and W. Dou. “Advertising

with User-Generated Content: A Framework and Research Agenda.” Journal of Interactive Advertising 8, 2 (2008): 1-4.

K rugman, H. E. “The Impact of Television Advertising: Learning Without Involvement.”

Public Opinion Quarterly 29 (1965): 349-356.

K rugman, H. E. “The Measurement of Adver­

tising Involvement.” Public Opinion Quarterly 30 (1966-1967): 583-596.

M cC auley, E., and T. D o nofrio. “A ppara­

tus, System and Methods for Stimulating and

Securing Retail Transactions.” Patent Application

Publication, 43, August 7 (2014): 783-791.

M onroe, K. “Editorial—On Replications in

Consumer Research: Part 2.” journal of Consumer

Research 19, September (1992): Preface.

O ‘K eefe, D. Persuasion: Theory and Research.

Newbury Park, CA: Sage Publications, 1990.

P asadeos, Y., J. P helps, and A. Edison. “Search­ ing for Our ‘Own Theory’ in Advertising: An

U pdate of Research Networks.” Journalism

and Mass Communication Quarterly 85, 4 (2008): 785-806.

P etty, R., and J. C acioppo. Attitudes and Persua­ sion: Classic and Contemporary Approaches. Dubu­ que, IA: Wm. C. Brown Company Publishers, 1981.

P etty, R., and J. C acioppo. “Central and Periph­ eral Routes to Persuasion: Application to Adver­ tising.” In Advertising and Consumer Psychology,

L. Percy and A. G. Woodside, eds. Lexington,

MA: D.C. Heath and Company, 1983.

P etty, R., J. C acioppo, and D. Sch u m a n n . “Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involve­

ment.” Journal of Consumer Research 10 (1983): 135-146.

P etty, R., J. K asmer, C. H augtvedt, and J.

C a c io ppo . “Source and Message Factors in

Persuasion: A Reply to Stiff’s Critique of the Elaboration Likelihood Model.” Communication Monographs 54 (1987): 233-249.

Sc h um a nn, D ., M. K otowski, H. A h n , and C. H augtvedt. “The Elaboration Likelihood

Model: A Thirty Year Review.” In Advertising

Theory, S. Rodgers and E. Thorson, eds. New York: Routledge, 2012.

Sherif, M., and C. H ovland. Social Judgment:

Assimilation and Contract Effects in Communica­ tion and Attitude Change. Philadelphia, PA: W. B. Saunders, 1961.

Stewart, D. “Speculations on the Future of

Advertising Research.” Journal of Advertising 21, 3 (1992): 1-18.

Szczepanski, C. General and Special Interest

Magazine Advertising and the Elaboration Likeli­ hood Model: A Comparative Content Analysis and

Investigation of the Effects of Differential Route

Processing Execution Strategies (doctoral disser­ tation). Retrieved from ABN/INFORM Global (3203946), 2006.

T aylor, C. “Making Academic Research on Advertising More Managerially Relevant.” Inter­

national Journal of Advertising 30,5 (2011): 739-740.

Te’ eni-H arari, T., S. Lampert, and S. Lehman- W ilzig. “Information Processing of Advertising

among Young People: The Elaboration Likeli­

hood Model as Applied to Youth.” Journal of Advertising Research 47,3 (2007): 326-340.

T rampe, D., D. Stapel, F. Siero, and H. M ulder.

“Beauty as a Tool: The Effect of Model Attrac­ tiveness, Product Relevance, and Elaboration Likelihood on Advertising Effectiveness.” Psy­

chology & Marketing 27, 12 (2010): 1101-1121.

W in d , Y., B. Sh a rp, and K. N elson-F ield. “Empirical Generalizations: New Laws for

Digital Marketing. How Advertising Research Must Change.” Journal of Advertising Research 53,

2 (2013): 175-180.

400 JO U R R B L OF R O U E R T IS IR G RESER R C H D e c e m b e r 2 0 1 5

 

 

Copyright of Journal of Advertising Research is the property of Warc LTD and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder’s express written permission. However, users may print, download, or email articles for individual use.