Using Drama to Persuade

JOHN DEIGHTONDANIEL ROMERJOSH MCQUEEN*

Television ads can be classified as either arguments or dramas or hybrids of thesetorms. We claim that fomi dimension influences how ads are processed. An argu-ment backs its claims with appeals to objectivity and is processed evaluatively. Adrama appeals more to subjective criteria and is processed empathically. A studyis reported in which 40 television commercials were classified on a dramatizationscale. They were shown to 1,215 people, and measures of evaluative and empathicprocessing were taken. The measures were found to be weighted differently forarguments and dramas, supporting the contention that fomi influences processing.

S tudies of persuasion in many fields distinguishbetween reasoned argument on the one hand and

story, narrative, or drama on the other. The distinc-tion is found in psychology (Bruner 1986), literature(Booth 1974;Chatman I978;Scholes 1981), theology(Goldberg 1982), law (Bennett and Feldman 1981),communication (Fisher 1984), history (White 1981),and economics (McCloskey 1985). Wells (1988) is thefirst to apply it to advertising.

The distinction between argument and drama mat-ters to the study of advertising because, we argue,different forms of advertising lead consumers to pro-cess claims in different ways. These distinctions needto be preserved in systematic research on persuasionand in diagnostic testing of commercials.

This article explores the topic conceptually andempirically. It defines the concept of drama in adver-tising, proposes a measure, and tests whether drama-tization influences how advertising claims are pro-cessed. Four indicators of process are proposed: ex-pressions of belief, counterargument, expressions offeeling, and judgments of verisimilitude or plausibil-ity. Effective drama is hypothesized to influence be-liefs by a path that evokes more expression of feelingand verisimilitude, less counterargument, and less di-rect elicitation of belief than occurs with effective ar-gument.

'John Deighton is Assistant Professor of Marketing, GraduateSchool of Business, University of Chicago, Chicago, IL 60637,Daniel Romer is Associate Research Director and Josh McQueenis Executive Vice President and Director of Research, both at LeoBurnett, Inc.. 35 West Wacker Drive, Chicago, IL 60601. The au-thors thank Steve Hoch and Robert Schtndler for opinions on anearlier draft of this article, and the first author acknowledges sup-port from the University of Chicago Graduate School of Business'smarketing research fund.

335

ADVERTISING FORMAND FUNCTION

Advertising Forms

Form is a popular theme in advertising texts. Ray(1982) classifies commercials by "format," e.g.,warmth, testimony, refutation, repetition, and fear.Aaker and Myers (1987) use the term "message fac-tors." Rothschild (1987) refers to classes of creativeappeal (rational versus emotional, product focus ver-sus consumer focus) and execution style (slice of life,product comparison, problem/solution, music, sex,and humor).

Following Wells (1988), we argue for a differentform dimension—the extent to which the advertisingis dramatized. Speculation and empirical research ina wide variety of disciplines suggest that dramas workin quite different ways than arguments. A dramadraws the viewer into the action it portrays. Con-versely, an argument holds the viewer at arm's length,rather as a platform speech does (Wells 1988). Whena drama is successful, the audience becomes "lost" inthe story and experiences the concerns and feelings ofthe characters. When an argument is successful, theaudience weighs the evidence and then yields to it.

The Dramatization Scale

We conceive of argument and drama as extremepoints on a continuous scale constructed with plot,character, and narration as attributes that mark tran-sitions along the scale.

At the argument extreme of the scale, there is a nar-rator, but the ad has no plot or character. An argu-ment starts to become a story when plot is introduced.

©JOURNALOFCONSUMER RESEARCH •Vol. 16« December 1989

336 THE JOURNAL OF CONSUMER RESEARCH

Plot comprises a stable state of affairs breached to in-duce a crisis and finally redressed (see Bruner 1986).Although plot usually involves human intentions,that need not be so. Durgee (1988), for example,claims that advertising can impute drama to con-sumer products by portraying a sequence of expec-tancy, tension, and resolution in product use (for ex-ample, the plan, aim, shoot sequence in using a cam-era). It is not difficult to see plot in commercials thatstar products rather than humans. When a detergentfights a stain, or one diaper competes with another toretain water, the events enact a plot. We use the term"demonstration" to refer to a commercial with plotbut no human character.'

The transition from demonstration to story ismarked by the concept of character. Characters areprotagonists who act within the context of a plot, asdistinct from narrators, who address the audience.Character serves to make human values salient(Scholes 1981), so that characters placed in conjunc-tion with products are a resource by which advertisingcan express claims of product value.

So long as an interpreter stands between events andthe audience, we have narration. When the narratoris removed, the story becomes a drama; the distinc-tion is between telling and showing (Booth 196!).Narration, or telling, draws attention to the fact thatevents have been selected from a larger set of pastevents (Scholes 1981) and are being reported to theaudience for a reason. Although narration can under-score an event's meaning, explaining its relevance toa claim, there is a cost to doing so because the appealtends to be processed evaluatively, with conscious-ness of persuasive intent. In a drama, by contrast,events are not so obviously selected and ordered: theyseem to simply unfold (White 1981). Without a narra-tor, advertising has to depend on verisimilitude to es-tablish what the depicted events are worth to itsclaim. If it succeeds, this type of ad builds an em-pathic bond between its audience and the concerns ofits characters. If it fails, it is perceived as contrived orhokey, and empathy gives way to evaluativeness.

In summary, the rationale for the drama scale is thecontrast between argument and drama. Argumenthas three attributes: it is plotless, characterless, andnarrated. Its indicative mood can be quite explicitabout what consumers should believe and why. At theother extreme, drama has plot and character but nonarrator. Its subjunctive mood gives up the ability tomake explicit claims in exchange for the power of em-pathy. In between are mixed forms, such as narrateddrama (for example, in a slice of life with commen-tary) or dramatized argument (for example, when

'This is not to say that demonstration advertising cannot containhuman characters. We are defining demonstration by its necessaryconditions.

protagonists debate the merits of a product). It is alsopossible, as in the California raisins commercial inthe Exhibit, to have character without plot, a tableauin which very little develops. The steps on the dramascale are:

ArgumentDemonstrationStoryDrama

NarratedNarratedNarratedUnnarrated

No characterNo characterCharacterCharacter

No plotPlotPlotPlot

Paths to Persuasion

Persuasion is used here to refer to a change in anaudience's conception of the value of an object.Brand attitude is a more usual measure of persuasion.Value, however, has the advantage over attitude inthat it is potentially a multidimensional construct(see Holbrook and Corfman 1986). This study istherefore able to treat as an empirical question thepossibility that the dimensions of value are affecteddifferentially by argument and drama.

When advertising persuades, the audience can besaid, almost definitionally, to have tested and ac-cepted the truth of a claim of value. Following Bruner(1986), we propose that there are two kinds of truthtest, corresponding to the two types of advertising ap-peal.

Appeals that the audience takes to be open to objec-tive testing suggest that a claim depends on matters offact, so that anyone, given enough data, would reachthe same opinion on its truth. Claims of this kind arethe province of argument. Arguments such as thosefound in comparison advertising lay out a claim'sgrounds and warrants for examination (Deighton1985), risking counterargument in pursuit of rea-soned agreement. Advertising for Oral B tooth-brushes, for example, makes this kind of claim whenit asserts that Oral B brushes are less abrasive thanother brands, then backs the claim by rubbing thefender of a car with two brushes.

Appeals that the audience believes can be verifiedonly subjectively are appeals to feeling. Here the truthtest is personal and discretionary. Imagine that an adclaims that it is exhilarating to drive a Pontiac.Whether the car signifies excitement in general maybe an objective issue, but whether it is true for me isnot a matter on which I need to defer to the evidenceof authority, logic, or anyone else's experience. Ap-peals of this kind are the province of drama.

When a claim rests on subjective grounds, drama'sadvantage over argument is that it does not have toreduce the subjective experience to words and thendepend on the credibility of a narrator to communi-cate it. It can depict the experience directly, with theaim of evoking the feeling itself in its audience.Bruner (1986) observes that the presence of charactermay not convince us of a general truth—it may even

USING DRAMA TO PERSUADE 337

EXHIBIT

EXAMPLES OF COMMERCIAL CLASSIFICATION

Product Ad action sequence Coding

Vision saucepan An aluminum saucepan is placed inside a transparent, glassVision saucepan, which is heated on a hot plate. Thealuminum saucepan becomes red hot and melts. During thisdemonstration, a narrator describes the several merits of theglass saucepan.

C^ifornia raisins Clay models of raisins with human characteristics are showndancing and singing on a teievision set watched by a coupiesnacking on raisins. The raisins' song refers to attractiveattributes of raisins.

Diet Pepsi A young woman calls on a man in the apartment next to hersto borrow a Diet Pepsi. The man. finding he has none, leavesby a back window in search of the product. He enduresseveral adventures before returning with the can, to be toldthat her roommate wants one too.

Coders scored the commercial 0 for character. 4 for plot(referring to the contest between the glass and aluminumsaucepans, which occupied two-thirds of the commercial'sduration), and 0 for narration (reverse-scaled, continuousnarration) for a mean aggregated rating of 1.33 on the 0 to6 scale.

The commercial scored a mean of 5-66 aaoss the threejudges for character, 0 for plot, and 5 for narrationbecause the coders interpreted the singing of the raisinsas character communication rather than narration. Themean of the three components of the rating was 3.56.The commercial scored 6 for both plot and character and amean of 5.33 for narration because it was entirelyunnarrated except for a closing frame with the slogan"Pepsi—Choice of a new generation." The meansummary rating was 5.78.

impede it—but it does vividly instantiate a particularproposition. Drawing on Iser (1978), he states thatplot and character recruit the reader's imagination to"perform" the meaning of the drama. Narration in-terferes with this process. In telling, a narrator doessome of the audience's thinking, explaining theevents and warranting their meaning. Without a nar-rator, the verisimilitude of the events alone, throughtheir ability to build empathy, determines how wellthey back the claim. When verisimilitude is high, theaudience may not, in fact, even notice that a claimis being made. On the positive side, nothing intrudesbetween the audience and the immediacy of the expe-rience shown in the drama. On the negative side,there is no interpreter to underscore the point.

To summarize, we contend that there are two pathsby which advertising can persuade. In one, the adver-tising suggests that a claim is objective, by invokingthe rhetorical form of an argument. In the other, theclaim is framed as subjective, appeals to personal ex-perience, and is not open to objective testing. Dramais a reliable way to invoke this mode of processing.^

Indicators of Persuasion PathsEvidence exists for a variety of mental processes

mediating acceptance of advertising: argument

*We are not being prescriptive here. We do not want to claim thatthere is a "best" path for a particular communication task. Thesame claim can be made by drama or argument, and many factorsinfluence whether one is better than the other, including the formofcompcting claims, the effectiveness of competing claims, and theeffectiveness of the executions. In any event, it is probably not pos-sible to hold content constant while manipulating form, so proposi-tions about the effect of different forms of expression on identicalcontent are somewhat conjectural.

for and against the advocated position, curiositythoughts, and source derogation (Wright 1973, 1980);connections (Krugman 1967); attitude toward the ad(McKenzie, Lutz, and Belch 1986); and variousaffective responses (Batra and Ray 1986; Burke andEdeJl 1989; Holbrook and Batra 1987). Our review ofthe dramatization literature suggests that the media-tors may themselves be conditioned by the form ofthe communication. Differences in mediating re-sponses may mark differences in persuasion paths.

It is reasonable to expect expression of belief to me-diate both persuasion paths. A partisan communica-tion must logically be believable to be persuasive. Butour review suggests two different ways to be believ-able. The first, the appeal to objectivity through argu-ment, invites counterargument and overt expressionsof belief. Wells (1988, p. 15) claims that lectures are"ideas that other people are trying to impose on me"and that defenses are erected against them. Althoughan argument may generate counterargument, it mustalso evoke positive beliefs if it is to be persuasive. Inthe appeal to subjective truth through drama, coun-terargument is less likely to occur. However, dramamust evoke expressions of feeling and meet the test ofverisimilitude or plausibility of the depicted events.Belief is the product rather than an indicator of theprocess of persuasion here.

The Figure illustrates these divergent paths to per-suasion. Form, an indicator of the degree of advertis-ing dramatization, determines whether belief is built(I) by a process involving expression of feeling andverisimilitude or (2) by a path through counterargu-ment on a direct route to belief. Belief in turn affectsjudgments of value. Subjects' prior opinions of valuealso influence belief and posterior value. Our hypoth-eses refer to these paths in the Figure.

338 THE JOURNAL OF CONSUMER RESEARCH

FIGURE

PATHS TO PERSUASION

Pre-Walue

HI: The more dramatized a commercial, the lessit will (directly) elicit counterargument andexpression of belief, and the more it willelicit expression of feeling and verisimili-tude.

H2: Counterargument impedes the expressionof belief and the communication of value.

H3: Expression of feeling facilitates the expres-sion of belief and communication of value.

H4: Verisimilitude facilitates the expression ofbelief and communication of value, butmore for drama than for argument.

THE STUDYTo test the hypotheses, we measured value, adver-

tising form, counterargument, expression of belief,expression of feeling, and verisimilitude. The value ofa product or service was measured through ratings of10 descriptive qualities on a six-point scale rangingfrom "does not describe (the product or service in thead) at all" to "describes extremely well." The 10 ad-jectives or descriptive phrases used were: important,relevant, useful, helps get things done, necessary, en-joyable, reflects my personality, gives me pleasure,exciting, and desirable. These items were chosen tospan a range of value dimensions, including func-tional, symbolic, and experiential (Park, Jaworski,and Maclnnis 1986), and a factor analysis of re-sponses to the scale suggested that subjects discrimi-nated between two kinds of value, one having to dowith usefulness and the other with enjoyment. Pathcoefficients in the model to be presented later did not,

however, differ materially when estimated separatelyfor each dimension. In the interest of simplicity,therefore, we present value as the sum of these itemsin this article, while recognizing that it is a multidi-mensional construct.

Advertising form was rated by three trained coderswho viewed each member of a set of television com-mercials several times. The commercials were dividedinto units of five seconds duration, and the followingquestions were answered yes (coded I) or no (coded0) for each five-second unit:

Narration: Was the unit unnarrated? Narration refersto speech or writing (including pack shot) directed tothe audience.

Character: In the unit, were one or more protagonistsshown or heard acting as if they were unaware of theexistence of the camera?

Plot: In the unit, did you see or hear the working out ofastory? A story is a fictional or true account of how theexpectations or wishes (of a person) or the inclinationsor tendencies (of a person or product) are first opposed,frustrated, or are otherwise in doubt, then in some wayprevail, succeed, or are redressed.

The three judges' binary judgments for a commercialwere summed and divided by the duration of the com-mercial in seconds. A linear transformation placedthis score on a five-point scale ranging from pure ar-gument to pure drama.

The four indicators of persuasion path were mea-sured in the tradition of the reaction profile (Wells,Leavitt, and McConville 1971) and viewer responseprofile (Schlinger 1979) as consumer judgments ofthe experience of viewing a commercial. They are notintended as measures of attitude toward the ad, butas self-reports of responses that occurred during theprocessing of the commercial. They were rated on six-point scales as follows:

Counterargument: A two-item scale asked subjects torate their inclination to argue back to the commercialand asked if they thought of reasons not to use the prod-uct while viewing the commercial.

Expression of belief: A five-item scale measuredwhether the ad was believable, was personally impor-tant, had shown the product had advantages and whata really good product of this type should do. Subjectswere also asked whether the ad convinced them that theproduct was one they needed or could use.

Expression of feeling: Five items were constructed totap the extent to which the commercial aroused appre-ciation of its quality and cleverness, as well as feelingsof positive affect, comfort, enjoyment, entertainment,and excitement.

Verisimilitude: A six-item scale measured whether thesubject felt drawn into the commercial, whether the ac-tions depicted seemed authentic, whether the commer-

USING DRAMA TO PERSUADE 339

cial had portrayed feelings the subject could relate toand had made the subject want to join in the action.

Forty television commercials were selected for thestudy. The selectors were not apprised of the argu-ment/drama distinction or of the hypotheses of thisstudy, but were attempting to construct a broad cross-section of prime time television advertising by na-tional advertisers. Twenty-five different brands ofconsumer products and services, including food, bev-erages, clothing, household supplies, a magazine,transportation services, and financial services, wererepresented. Fifteen of the brands were representedby two commercials each. Three commercials were60 seconds long, and the remainder were 30 secondslong.

Subjects were recruited in malls in 10 citiesthroughout the United States and were qualified asusers of one of the 25 products in the study. On a com-puter-administered questionnaire, each subject ratedhis or her opinion of and experience with five brandsin that product category and rated the target brand onthe 10 items of the value scale. The subject then sawa television commercial for that brand. The subjectrated the commercial on each of the items on the pro-file of responses, rerated the brand on the 10 items ofthe value scale, and supplied demographic informa-tion. The interview lasted about 25 minutes and in-volved only one commercial. Subjects were paid fortheir time, and each commercial was rated by be-tween 29 and 31 subjects. In all, 1,215 people wereinterviewed.

RESULTS

The Exhibit gives examples of how the commer-cials were coded. Agreement among the three coderswas high, with the lowest pairwise correlation at 0.97.The median on the form scale was 3.22, with a rangefrom 1 to 4.85; thus, some commercials were rated aspure argument, but none as pure drama. (This is to beexpected, as it is common to find some device, suchas a final frame with a pack shot or a tag line, thatbreaks the drama mode.)

Reliability of the lO-item value scale was assessedseparately for the 1,215 responses collected before(pre-value) and after (post-value) subjects saw a com-mercial. The results indicated high reliability andunidimensionality, and each of the four path indica-tors achieved adequate reliability.

Measures

Pre-valuePost-valueCounterargumentCommunication of feeling

Cronhach 'salpha

.91

.93

.53

.88

Communication of belief .85Verisimilitude .84

The counterargument scale is low, but satisfactory fora two-item scale. Neither pre-value nor post-valuewas correlated with form, i.e., drama versus argu-ment. We shall first report mean scores, mainly forthe sake of exposition, and then a path analytic testof the hypotheses.

Descriptive Results

Our discussion predicted that a persuasive dramafollows a particular path to persuasion. It does notarouse counterargument but generates expressions offeeling and convinces the audience of its verisimili-tude. By implication, an unpersuasive drama fails tofollow this path in some or all of its particulars. A per-suasive argument, on the other hand, may producemore counterargument than a drama, but less than anunpersuasive argument. It does not depend on feelingor verisimilitude to persuade, but does evoke expres-sions of belief.

Table 1 summarizes what our data have to sayabout these predictions. The respondents are parti-tioned into persuaded (« = 616) and unpersuaded {n= 599), defining persuasion as an increase in the valuescore for the advertised product after exposure to thead. Respondents are then split according to whetherthe ad they saw fell into the lower (argument) or upper(drama) half of the form scale. For each of these fourcells, means for the four path indicators are reported.

The median splits reveal that the profile of a persua-sive argument commercial differs significantly fromthe profile of a persuasive drama commercial, sup-porting the prediction that good drama induces morefeeling, more verisimilitude, and less counterargu-ment (the latter at alpha - 0.07) than good argument.The implication that good drama is persuasive be-cause it elicits less counterargument and more feelingand verisimilitude than poor drama is also supported.Although the difference between good and poor argu-ment with respect to counterargument is not signifi-cant, good argument elicits more belief than unsuc-cessful argument, and none of the predictions is con-tradicted. It is worth noting that there was nodifference in the mean beliefs of good arguments andgood dramas, consistent with the idea that one formis not consistently more persuasive than the other.

Tests of the HypothesesThe path model illustrated by the Figure allows the

pre-value score to be correlated with the other exoge-nous variable, the form of the ad. The linear systemof equations implied by the model was estimated withEQS (Bentler 1985). EQS is an approach to linear

340 THE JOURNAL OF CONSUMER RESEARCH

Ad form andeffectiveness

Persuasive argumentPersuasive drama

Unpersuasive argumentUnpersuasive drama

TABLE 1

MEAN SCORES FOR PATH INDICATORS

Counterargument

2.51 ' "2.35'

2.68''2.58"

Expressionof belief

4.16"4.08'

3.70''3.51"=

Expressionof feeling

4.18'4.38"

3.66*=3.95"

• • '

Verisimilitude

3.68'3.94"

3.34^3.43=

314302

291308

"^ Two means in the same column are different (at alpha = 0.05) only when they do not share a common superscnpt.NOTE: Argument and drama are defined by a median split on the form variable. Persuasive cases are those for which a respondent's rating of value increased afterexposure lo the commercial; unpersuasive cases are ihose (or which it declined.

structural equations modeling, similar to LISREL(Joreskog and Sorbom 1983) but more suitable formodels that do not contain latent variables. Maxi-mum likelihood estimation, based on multivariatenormal distribution theory, was used.

The model fits the data well. The standardized re-sidual covariance matrix values are small (averageoff-diagonal absolute value = 0.10) and evenlydistributed among the variables (maximum value= 0.45). The fit can be compared to the fit of a modelthat assumes complete independence among the vari-ables, by the Bentler-Bonett normed fit index. The in-dex is given by 1 - Q/Qo, defining

where 5" is the vector of observed variances and co-variances, ffisa model for the data, a function of morebasic parameters 0 , which are estimated so as to min-imize Q, and W'ls a weight matrix to accommodatethe data distribution assumptions. QQ is the samefunction as Q, evaluated under the independencemodel. The Bentler-Bonett normed index is 0.93.

Table 2 gives the unstandardized values of the pathcoefficients. The coefficients on the paths involved inthe hypotheses all pass a univariate large-sample nor-mal z test of difference from zero at the 0.05 level ofconfidence.

Effects of Form on Path Indicators. Hypothesis Irefers to these effects and is supported. As the rangeof the form variable is from 1 (pure argument) to 5(drama), a positive coefficient indicates that the re-sponse is more likely to be generated by an ad in thedrama form. Thus, argument advertising was associ-ated more with counterargument and belief (by thedirect path) than drama advertising, which was asso-ciated more with feeling and verisimilitude.

Effects of Path Indicators on Belief and Value.Hypotheses 2-4 refer to these effects. As hypothe-sized, counterargument is negatively associated withbelief, while feeling and verisimilitude are positively

associated. Belief is associated with value. To test thesecond part of Hypothesis 4, that verisimilitude in-fluences the communication of value more by dramasthan by arguments, we reestimated the model on thetwo subsets of the data (dramas and arguments) cre-ated by a median split on the form variable. The re-sults are in Table 2? For the hypothesis to hold, theverisimilitude-belief path coefficient in the argumentmodel (0.33) must be significantly less than in thedrama model (0.45). To test if that was so, the dramamode! was reestimated with the coefficient con-strained to 0.33. The model chi-square statistic in-creased by 10.3 foronedegreeof freedom gain, so thehypothesis of no difference was rejected at the 0.01significance level.

Although no other contingent effect of dramatiza-tion on path coefficients had been hypothesized, weused the same test to see whether the other three pathsentering belief were significantly different in the twosubmodels. The direct form-belief path was signifi-cantly larger in the argument model. The paths fromcounterargument and feeling to belief were not sig-nificantly different between the submodels.

Value, as we measure it, is potentially multidimen-sional. A factor analysis of our data suggests that sub-jects discriminated usefulness from enjoyment in rat-ing value. When we estimated submodels for thesetwo components of value separately, however, wefound no material differences in the pattern of co-efficients, whether for arguments or dramas or for allcommercials together. We do not rule out the possi-bility that argument might be more effective in estab-lishing usefulness and drama in establishing enjoy-ment value, but the question could not be resolved byour data because the measures of each value compo-nent were correlated.

^For the form effects (the first four kinds of effects listed in Table3), we attach no significance to differences between coefficients esti-mated on the argument and drama subsets of the data. The mediansplit on the form variable attenuates its range and makes cotnpari-son of these path coefficients unwise.

USING DRAMA TO PERSUADE 341

TABLE 2

COEFFICIENTS ON CONSTRUCT EOUATIONSFOR MODEL IN FIGURE

Effects

From - * To

Form -* counterargumentForm -* belief (direct)Form -*• feelingForm -- verisimilitudeCounterargument — beliefFeeling -* beliefVerisimilitude -* beliefPre-value - • beliefBelief -» post-valuePre-value - * post-value

Bentler Bonett normed fit

1

Aggregate

model(n= 1.215)

-0.158--0.743*

0.653*0.599*

-0.173'0.247-0.399*0.402*0 288*0.520*

0.929

Path coefficients

Model fit toarguments

only(n = 605)

-0.160-0.710-0.988'1.445*

-0.212*0.281-0.345-0.435*0.213*0.482*

0.917

Model fit todramas

only(n = 615)

-0.273'-0.2030.4750.070

-0.134*0.221-0.450*0.386*0.208-0.564'

0.930

NOTE: • indicales that coefficiems are significant at alpha = 0.05.

CONCLUSIONS AND DISCUSSIONThis study suggests that the degree of dramatiza-

tion in television commercials influences how this ad-vertising will be processed. An argument's appeal isprocessed evaluatively, with opponent processes ofcounterargument and expression of belief determin-ing the degree of persuasion. A drama's appeal is pro-cessed empathically. Viewers are less disposed to ar-gue and believe the appeal to the extent that they ac-cept the commercial's verisimilitude and respond toit emotionally.

Conceptual DiscussionThere are many ways to think about persuasion,

but the main body of advertising research uses few ofthem. One largely unmined tradition deals with therhetoric of fiction (Booth 1961). Despite the absenceof propositional claims, reasons, evidence, and othertrappings of argument, fictional literature is persua-sive. One product of literary criticism is a catalog ofthe tactics of organization and structure that story-tellers use to impose their invented worlds on audi-ences. These insights bear on persuasion by drama-tized advertising.

The most provocative difference between the meth-ods of inquiry in literature and advertising is that lit-erary criticism tends to ignore the psychology of read-ers almost entirely in favor of attributes of the text,while advertising research does the opposite. The po-tential for synthesis of the two traditions is consider-able. In this article, we have borrowed an assumptionof some literary analysts, that an audience's response

to a presentation is shaped by its form,** and tested itby looking for indicators of cognitive process.

Other disciplines besides literature have had to dealwith whether, why, and how stories persuade. Thehistorian Hexter (1971, p. 47) attempts to account forthe necessity of storytelling in historical explanationwith an example from baseball. The New York Yan-kees won the American League in 1939 for reasonsthat can be summarized in statistical tables. The NewYork Giants won the National League in 1951 for rea-sons that are not easily understandable exceptthrough stories. The argument of the first method ofexplaining is not superior to the narrative of the sec-ond: both may convey "good" reasons. In fact, Gold-berg (1982), noting the popularity of narrative overargument in theological writing, claims narrative islogically prior to argument in human understanding.Propositions, he contends, gain their sense and mean-ing from narrative and are, in fact, abstractions fromit. In his analysis of the rhetorical character of eco-nomic inquiry, McCloskey (1985, p. 78) makes a sim-ilar claim: "The word 'story' has in fact come to havea technical meaning in mathematical economics,though usually spoken in seminars rather than writ-ten in papers. It means an extended example of theeconomic reasoning underlying the mathematics, of-ten a simplified version of the situation in the realworld that the mathematics is meant to characterize."One reason stories are persuasive may be that muchknowledge is contained and transmitted in stories.

Technical Discussion

The findings of this study can be discussed in termsof (1) dramatization's effect on indicators of persua-sion process and (2) the effect of the indicators on per-suasion. The first set of findings shows that advertis-ing form can be a significant determinant of interme-diate responses to advertising. This is the moreinteresting result because it establishes that there arematerial implications to the argument-drama distinc-tion. Studies that deal with cognitive and affective re-sponses generated by a number of different ads woulddo well to anticipate this source of variance and theproblem it poses for aggregation across commercials.

The second set of findings, that the path indicatorsinfluence persuasion, are largely consistent with ear-lier work, although the role of verisimilitude has notpreviously been measured in advertising research.Our conclusion that counterargument inhibits per-suasion, while expressions of belief facilitate it, addslittle to earlier studies. Similarly, the finding that feel-ings contribute uniquely to persuasion is not surpris-

*0r, in the pithy terms of Kenneth Burke (1950, p. 58), "A yield-ing to the form prepares for assent to the matter identified with it."

342 THE JOURNAL OF CONSUMER RESEARCH

ing{Edell and Burke 1987; Holbrook and Batra 1987)and might even be open to the criticism that a com-posite feeling measure can mask countervailing in-fluences of separate feelings (Burke and Edell 1988).The fact that our measures of advertising response be-have in a manner consistent with different measuresof the same concepts in other studies does, however,offer some evidence of their validity.

This second set of findings contains some interest-ing contingencies. Verisimilitude affects belief morewhen advertising is dramatized, and advertising elic-its belief directly only when it is in argument form.But neither counterargument nor expression of feel-ing seems to be contingent on advertising form. Thatis to say, counterargument, once elicited, is no less anobstacle to persuasion by drama than by argument,even though drama is less likely to elicit it. Were itnot for argument's direct path tobelief, therefore, thisstudy might seem to say that drama was always morepersuasive than argument. As it happens, we can saythat no one form of advertising consistently beat theother in our sample. And we cannot say from thisstudy if there are contingencies under which one formwould always dominate the other. In fact, given theinventiveness of advertising writers, we see little pointin looking for such contingencies. Once a form hasbeen selected, however, the study does suggest thatdifferent diagnostic criteria apply. The search for in-dicators of the quality of the argument (Areni andLutz 1987) and the quality of the drama (perhaps in-dicated by verisimilitude) appear to be separable ar-eas of inquiry.

This research is open to the criticism that it at-tempts to reach causal conclusions from a correla-tional design. Our response is that in the study of ad-vertising form effects, experimentation is not the wayto go. It is probably impossible to diserttangle adver-tising form and content experimentally. Even if itwere possible, it would be impractical to try to repli-cate the production values of real advertising withlaboratory stimuli that vary according to an experi-mental design. For example, it takes considerableskill to tell a story with verisimilitude. Any attemptto manipulate this factor with stimuli of lower qualitythan television advertising would be of very dubiousinternal validity. We therefore approached the issuewith a large sample correlational study of real adver-tising, contending that the benefits of that approachoutweigh the costs.

Our stimuli were ail good advertising, by the cri-teria that firms were willing to invest to air them. Welost some power to test the theory by using a samplewith less variance than we might have had with inclu-sion of some bad advertising. We are, however, ableto say something about the existence of form effectson persuasion and can report findings about the sizeof these effects among real consumers.

The use ofa pre-post design raises questions of ex-ternal validity. Measuring the change in judgment ofvalue following forced exposure to a commercial doesnot capture longer term in-market effects. Neverthe-less, it is consistent with our limited aim of showingthat the immediate persuasive impact of the pool of40 commercials was mediated by indicators of pro-cessing mode and dramatization. The between-sub-jects pattern of responses is too complex to be attrib-uted to demand or to a tendency to anchor on the pre-measure of value.

Although this study is suggestive, research isneeded to investigate more directly the cognitive pro-cessing that is hypothesized to operate in response toarguments and dramas. In particular, alternativemeasures of persuasion path might be explored. Be-havioral measures of persuasion might be found tosupplement the self-reports. Further research mightalso consider whether forced exposure, as in ourstudy, gives drama an advantage that it would not re-ceive in more natural exposure settings. It may be thatsome television viewers do not give a commercial theserial attention needed to follow a drama that unfoldsover the duration of the ad. They may sample frag-ments of the execution long enough to encode a claimor a sentiment, but not long enough to encode the plotof a story. Our forced exposure design encouragedsubjects to get involved in the plot of drama ads, per-haps more than they would naturally do. This conjec-ture is worth testing.

[Received November 1988. Revised July 1989.]

REFERENCESAaker, David A. and John G. Myers (1987), Advertising

Management. Englewood Cliffs, NJ: Prentice-Hall.Areni, Charles S. and Richard J. Lutz (1987), "The Role of

Argument Quality in the Elaboration LikelihoodModel," in Advances in Consumer Research, Vol. 15,ed. Michael J. Houston, Provo, UT: Association forConsumer Research, 197-203.

Batra, Rajeev and Michael L. Ray (1986), "Affective Re-sponses Mediating Acceptance of Advertising," Jour-nal of Consumer Research, 13 (September), 234-249.

Bennett, W. Lance and Martha S. Feldman (1981), Recon-structing Reality in the Courtroom. New Brunswick,NJ: Rutgers University Press.

Bentler, Peter M. (1985), Theory and Implementation ofEQS: A Structural Equations Program. Los Angeles:BMDP Statistical Software, Inc.

Booth, Wayne C. (1961). The Rhetoric of Fiction. Chicago:University of Cfiicago Press.

(1974), Modern Dogma and the Rhetoric of Assent,Chicago: University of Chicago Press.

Bruner, Jerome (1986), Actual Minds, Possible Worlds,Cambridge, MA: Harvard University Press.

Burke, Kenneth (1950), A Rhetoric of Motives, EnglewoodCliffs, NJ: Prentice-Hall.

USING DRAMA TO PERSUADE 343

Burke, Marian C. and Julie A. Edell (1989), "The Impactof Feelings on Ad-Based Affect and Cognition," Jour-nal of Marketing Research, 26 (February), 69-83.

O\^\mzn,%eymo\ii (X'^l^), Story and Discourse: NarrativeStructure in Fiction and Film, Ithaca, NY: CornellUniversity Press.

Deighton, John (1985), "Rhetorical Strategies in Advertis-ing," \xi Advances in Consumer Research, Vol. 12, eds.Morris Holbrook and Elizabeth Hirschman, Ann Ar-bor, MI: Association for Consumer Research, 432-436.

Durgee, Jeffrey F.( 1988), "Product Dizm^i" Journal of Ad-vertising Research, 28 (February/March), 42-49.

Edell, Julie A. and Marian C. Burke (1987), "The Power ofFeelings in Understanding Advertising Effects," Jour-nal of Consumer Research, 14 (December), 421-433.

Fisher, Walter R. (1984), "Narration as a Human Commu-nication Paradigm: The Case of Public Moral Argu-ment," Communication Monographs. 51 (I).

Goldberg, Michael (1982) Theology and Narrative, Nash-ville, TN: Abington.

Hexter, J.H. (1971), The History Primer, New York: BasicBooks.

Holbrook, Morris B. and Rajeev Batra (1987), "Assessingthe Role of Emotions as Mediators of Consumer Re-sponse to Advertising," Journal of Consumer Re-search, 14 (December), 404-420.

and Kim P. Corfman (1986), "Quality and Value inthe Consumption Experience: Phaedrus Rides Again,"in Perceived Quality, eds. Jacob Jacoby and Jerry C.Olson, Lexington, MA: Lexington Books.

Iser, Wolfgang (1978), The Art of Reading. Baltimore, MD:Johns Hopkins University Press.

Joreskog, Karl G. and Dag Sorbom (1983), LISREL V Us-ers' Guide, Chicago: International Educational Ser-vices.

Krugman, Herbert (1967), "The Measurement of Advertis-

ing Involvement," Public Opinion Quarterly, 30, 583-596.

McCloskey, Donald N. (1985), The Rhetoric of Economics,Madison: University of Wisconsin Press.

MacKenzie, Scott B., Richard J. Lutz, and George E. Belch(1986), "The Role of Altitude Toward the Ad as a Me-diator of Advertising Effectiveness: A Test of Compet-ing Explanations," Journal of Marketing Research, 23(May), 119-129.

Park, C. Whan, Bernard J. Jaworski, and Deborah J. Mac-lnnis (1986), "Strategic Brand Concept-Image lAzn-agcmenl," Journal of Marketing. 50(4), 135-145.

Ray, Michael (1982), Advertising and CommunicationsManagement. Englewood Cliffs, NJ: Prentice-Hall.

Rothschild, Michael L. (1987), Marketing Communica-tions. Lexington, MA: D.C. Heath.

Schlinger, Mary Jane (1979), "A Profile of Responses toCommercials," Journal of Advertising Research, 19(2), 37-46.

Scholes, Robert (1981), "Language, Narrative and Anti-Narrative," in On Narrative, ed. W.J.T. Mitchell, Chi-cago: University of Chicago Press.

Wells, William D. (1988), "Lectures and Dramas," in Cog-nitive and Affective Responses to Advertising, eds. PatCafferata and Alice Tybout, Lexington, MA: D.C.Heath.

, Clark Leavitt, and Maureen McConville {1971), "AReaction Profile for Commercials," Journal of Adver-tising Research. 11 (December), 11-17.

White, Hayden (1981), "The Value of Narrativity in theRepresentation of Reality," in On Narrative, ed.W.J.T. Mitchell, Chicago: University of Chicago Press.

Wright, Peter {1973), "The Cognitive Processes Mediatingthe Acceptance of Advertising," Journal of MarketingResearch, 10 (February), 53-67.

(1980), "Message-Evoked Thoughts: Persuasion Re-search Using Thought Verbalizations,"sumer Research, 7 (September), 151-175.