Cons_Response.pdf

download Cons_Response.pdf

of 14

Transcript of Cons_Response.pdf

  • 7/21/2019 Cons_Response.pdf

    1/14

    CONSUMER RESPONSE TO SEX ROLE PORTR Y LS IN DVERTISEMENTS

    Effects of Incongruity and Prejudices on Emotions and Attitudes

    Ulrich R, Orth and Denisa Holancova

    ABSTRACT: This study investigates how male and female consumers in the Czech R epublic respondcosex role portrayals

    in advertisements for a fictional cell phone service. Special emphasis is placed on integrating consumer prejudices and

    emotions into an incongruity-resolution framework. Prejudices were found to moderate effects ofsexrole incongruity on

    the emotional dimensions approval and disapproval. Surprise, the chird emotional dimension, was directly affected by role

    incongruity. Downstream effects included positive (approval) and negative (disapproval, surprise) influences on attitude

    toward the ad. Extensions into other cultural settings and more in-depth examinations of the elaboration and processing

    paths, including effects on memory, are suggested for future research.

    Congruity theory (Osgood and Tannenbaum 1955) has been

    successfully applied in social psychology for explaining atti-

    tude formation. According to the underlying principle of cog-

    nitive consistency, people value harmony am ong their thou ghts ,

    feelings, and behaviors, and they are motivated to maintain

    uniformity among these elements (Solomon 1996). Advertis-

    ing researchers have applied schema congruity to a variety of

    contexts, including pictorial and verbal ad components (H ous-

    ton, Ch ildets, and Heck ler 1987), brand names (Meyers-Levy,

    Louie, and Curren 1994), and message appeals (Hong and

    Zinkhan 1995). While much theory and research have been

    directed at understanding the role of reason, or systematic

    ptocessing, in the central route to persuasion (Kirmani and

    Shiv 1998; Petty et al. 1988), prior research also provides

    evidence that executional cues in advertising, such as pictures,

    may influence high-involvement consumers' evaluations as

    well, in the form of peripheral-route ptocessing via emotions

    (Ma clnnisand jawors ki 1989)- In addition, congruity research-

    ers suggest th at em otions may arise from consumer ca tegori-

    zation and processing activities (Garbarino and Edell 1997),

    and Alden, Mukherjee, and Hoyer (2000) empirically dem-

    onstrated that reactions to stimulus-schema incongruity in-

    volve surprise as an emotio nal dim ension . The role of emotions

    in shaping attitudes, in turn, is well established (Brown and

    Stayman 1992).

    Within the large number of studies on sex role portrayals,

    researchers have tended to emphasize cognitive influences and

    processes in their examinations ofsex role portrayal effects on

    Ul rich R, O rth (Prof Dr. hahil., Munich University of Technology,

    Germany) is a professor of agribusiness and food marketing, Or-

    egon State University.

    Denisa Holancovais

    Ph.D.

    candidate at Mendel University Brno,

    Czech Republic.

    consumer attitudes (Debevec and Iyer 1988; Leigh, Rethans,

    and Reichenbach Whitney 1987). Sex role portrayal herein

    refers to the characterization of females and males in advertis-

    ing, as well as their sett ing relative to other characters (W ortze

    and Frisbie 1974). No study could be found examining con-

    sum er response to sex role portrayals with in a congruity frame-

    work. For instance, i t is unclear whether advertisements

    featuring sex role portrayals that are congruent w ith consum -

    ers' self-concepts and beliefs are likely to be more effective

    than ones featuring incongruent portrayals. Although it has

    been suggested that sex role portrayals evoke emotions (Batra

    and Ray 1986; Schmitt, Leclerc, and Dub e-Rioux 1988), there

    appears to be no prior research integrating consumer emo-

    tions and their relationship with the cold cognitive pro-

    cessing of sex role portrayals. For example, will sex role

    portrayals that are congruent with consumer schema elicit

    more approving emotions than portrayals thar are perceived

    as incongruent? Does incongruity correlate with viewer feel-

    ings of surprise? And how do the emotions evoked affect con-

    sumer at t i tudes?

    One problem in interpreting the results of studies within

    both streams of research on incongruity and sex tole portray-

    als is that the conceptual bases of cognition and affect are not

    clearly identified. Although cognition has been clearly dis-

    tinguished from affect, researchers have been less than spe-

    cific in defining affect and distinguishing it from the closely

    related, although not identical, concepts of attitu de and em o-

    tion. Eagly and Chaiken (1993) define affect as experientia

    The authors thank BorisW Becker, Oregon State University, for his

    comments on previous draft.s of the manuscript. They also thank Ronald

    Faber and three dnonym ousJourt/al vj AJvertising reviewers for their

    insightful comments and guidance, which helped improve the

    presentation of this research. Financial support from Eurotel and Metra

    for parts of the research is gratefully acknow ledged.

    ourn lof Adi ertiiing, vol 32. no, 4 (Winter 2 003-4). pp. 77-

    2004 A merican Academy ot Atlverrising. All rights reserve

    ISSN OOyi-3367 / 3004 $9-50 - 0.00

  • 7/21/2019 Cons_Response.pdf

    2/14

    78

    The Journal of dvertising

    feelings directed toward attitude objects. Yet emotions are

    un iqu e vis-a-vis affect in th at they are valenced, and can the re-

    fore be classified as positive or negative (Vanden Abeele and

    MacLachlan 1994); they are spontaneous, temporary states

    (Murry, Lastovicka, and Singh 1992); and they can be classi-

    fied by individuals according to a number of directions or

    orientation s, such as happy, sad, angry, or mad (Oliver 1992).

    In contrast, attitudes have been defmed as relatively stable

    Opinions containing a cognitive element and an emotional

    element (Wade and Tavris 1996). More important, attitudes

    include a cognitive and behavioral compo nent not present in

    the conceptual definition of emotion (Fishbein and Ajzen

    1975).

    In a related stream of

    research.

    Haddock and Zanna (199.^)

    applied this concept of attitude to the domain of intergroup

    attitudes and found that evaluations of advertising stimuli

    could be biased by neglecting differences in consumer preju-

    dices. Lysonski and PoUay (1990) report that both female and

    male sex role stereotyping occuts, and may lead to the devel-

    opment of prejudices. And consumer prejudices, in turn, have

    been found to affect consumer evaluation of advertisements,

    particularly emo tions (Bhat, Leigh, and W ardlow 199H). Given

    further evidence that incongruity does not perfectly correlate

    with consumer emotional response butratherdepends on con-

    sumer predisposit ions (Meyer, Rudolph, and Schutzwohl

    1991), it seems reasonable to assume that sex role portrayals

    in advertisements may elicit different responses depending

    not only on stimulus-schema incongruity, but also on con-

    sumer prejudice.

    Building on these studies, this paper extends tesearch on

    consumer response to sex role portrayals in advertising in at

    least three ways: First, it demonstrates how congruity theory

    is useful theoretically for explaining both women's and men's

    response to sex role portrayals in advertising . Second, it in te-

    grates consumer emotional responses, and hence develops the

    linkages between cognition and emotions as key variables in

    an incongruity-resolution framework. Third, it incorporates

    consumer prejudices as an individual difference variable and

    moderator of incongruity effects. The choice of the country

    sett ingthe Czech Republicwas primarily motivated by

    the researchers' academic and national affiliation.

    CONCE PT U L FR M E WORK

    Perceived Sex Role Incongruity

    To examine individual responses of both females and males to

    sex role portrayals in advertising, this study ad opts a cong ru-

    ity perspective. Congruity theory predicts both the degree

    and direction of attitude change, postulating that this change

    occurs in the ditection of increased congruity within the

    subject's cognitive schema (Meyers-Levy, Louie, and Curren

    1994). Results of congruity research demonstrate that whe

    information

    is

    somehow incongruent w ith prior expectations

    effects occur on consumer attitudes (Alden, Mukherjee, and

    Hoyer 2 00 0; Meyers-Levy and Tybout 1989), and also on the

    processing of information (Kirmani and Shiv 1998; Stayman,

    Alde n, and Smith 1992). Researchers used a variety of term i-

    nologies interchangeably, such as congruent/incongruent, ex-

    pected/ unexpected, and consistent/inconsistent (Heckler and

    Childers 1992).

    In the current con text of magazine advertising, (in)cong ruity,

    as the term is used here, refers to the extent to which (no)

    structural correspondence is achieved between the entire con-

    figuration of sex role portrayal and the configuration speci-

    fied by a person's schema or beliefs. Adop ting an incong ruity

    perspective entails recognizing that individuals' understand-

    ing of nonverbal cues varies (Burgoon 1994), and consumers

    may thus perceive pictorial role portrayals differently. More

    specifically, one part of an audience may perceive portrayals

    of a female boss and a male secretary as congruent with theit

    schema and beliefs, whereas another part of the audience m ay

    perceive this portrayal as incongruent. According to congru-

    ity theory, sex role portrayals that are congruent with con-

    sumer self-schema and beliefs are then likely to generate more

    favorable attitudes toward the brand than portrayals that are

    incongruent with consumer schema.

    Schema Incongruity and Consumer Emotions

    Researchers have argued that focusing on cognitive ptocesses

    and failing to understand the role of emotions impedes un-

    derstanding of various consumer behaviors (Allen, Machleit

    and Kleine 1992). Past research in social psychology (see, e.g.,

    Fiske 1982) and advertising (see, e.g., Alden, Muk herjee, and

    Hoyet 2000) indicates that congruity theory may be a par-

    ticularly useful framework for understanding the process of

    the generation of emotions. Fiske (1982) suggested that if an

    item is cong ruent with an existing schem a, it will receive the

    feelings linked to that schema.

    More recent research provides evidence tha t emo tions may

    arise not only as a result of che evaluation ot information, but

    also from consumer categorization and processing activities

    (Garbarino and Edell 1997). Whether the stimulus is evalu

    ated more positively or more negatively depends on how m uch

    effort is involved in resolving the incongruity. Alternatives

    that

    require m ore effortful processin g are less preferred than

    alternativ es tha t teq uire less effortful processing because of

    the ptocess-induced negative feelings generated (Garbarino

    and Edell 1997).

    Cohen and Basu (1987, p. 470) add another perspective:

    Since catego tization p roduces a reduction in u ncertainty

    positive affect may result from a successful fit and negative

    affect from an inab ility to catego rize an item. As a resu lt

  • 7/21/2019 Cons_Response.pdf

    3/14

    Winter

    2003^ 7

    researchers have begun building a theory that incorporates

    both the cognit ive and the emotional components that help

    explain how in cong ruity of advertising cues affects attit ud e

    toward the ad and brand attitude (Alden, Mukherjee, and

    Hoy er 20 00 ). Exam ining effects of different levels of perceived

    humorousness in advertising, the authors reported that ad

    content incongruity generates surprise. Although this find-

    ing confirms the existence of

    l ink between incongruity and

    consumer emotion, the study focuses on a single emotion and

    does not examine the linkages with a broader range of emo-

    tions tor an important social issue.

    Generally, the role of emotions as affective responses to

    advertisements, as well as their impact on attitudes, are well

    established {Brown and Stayman 1992). Emotions have been

    found to influence the evaluation of both attitude toward an

    ad and toward rhe adverrised brand (Edell and Burke 1987;

    Holbroo k and Batra 19H7). Con sidering further evidence thar

    emo tions are evoked by means of execution al cues in ads (Barra

    and Ray 1986 ), such as role portrayals, it is surprising that no

    study could be found ex amining con sumer emotional response

    to sex role portrayals in advertising. Instead, prior research on

    the processing of ads featuring sex role portrayals focused on

    how content and imagery affect cognitive responses and atti-

    tude toward the ad, which in turn affect attitude toward the

    brand and purchase intention (Jaffe 1994; Whipple and

    Courtney 1985).

    The studies cited reflect a diversity of theoretical domains

    and research contexts. Com mon through out them , thoug h, is

    a fundamental result : st imulus-schema incongruity appears

    to be a particularly useful framework for und erstan ding the

    linkages between cold cogn itive processing and emo tions.

    This raises the question of how to measure emo tional response

    to advertisements featuring sex role portrayals.

    Emotional responses to advertising are complex and mul-

    tidimensional in their scope. Typologies have been created

    including numerous feelings, such as Holbrook and Batra's

    (1987) Standard Emotional Profile. A variety of empirical stud-

    ies, however, indicates that those batteries of emotions could

    be pared down to two or three core emotional dimensions.

    Eot example, in their study on the effect of consumer preju-

    dices on ad processing, Bhat, Leigh, and Wardlow (1998) re-

    duced Holbrook and Batra's 29-item emotional battery by

    means of principal component analysis to three general emo-

    tional dimensions: approval, disapproval, and surprise.

    Given rhe findings of the literature review, it is assumed

    that sex role portrayals that are incongruent with consumer

    self-schema are likely to arouse strong emotional responses.

    Assuming several dimensions of emotional response, such as

    approval, disapproval, and surprise, leads to the hypothesis

    that the presence of sex role portrayals in ads that are incon-

    gruent with consumer self-schema will result in less approv-

    ing and more disapproving emotional response. Building on

    the results of Alden , Mukherjee, and Hoyer's 20 00 study, it is

    further hypothesized that sex role portrayals that are incon-

    gruent with consumer self-concept and beliefs will generate

    surprise.

    HI: The incongruity of an ac/pertisernent s sex ro/e portrayal

    will affect the strength and type of

    consumer

    emotional

    response.

    (In)congruent sex role portrayals will result in ap proving

    (disapprm ing) responses. Higher levelsof incongruity will result

    in stronger surprise.

    T h e M o de r a t ing Role o f Con sum e r P r e jud i c e s

    As mentioned before, stimulusschema incongruity is a use-

    ful framework for examing consumer response to sex role por-

    trayals in advertising. However, incongruity is unlikely to

    perfectly correlate with consumer emotional response. A strong

    correlation may occur, for exam ple, when the em otion evoked

    is surprise {Meyer, Rudolph, and Schutzwohl 1991). In such

    cases, consumers may have no well-formed a priori attitudes

    toward the sex role portrayal issue due to lack of awareness

    Alternatively, an advertisement might elicit strong emotiona

    reactions of approval or disapproval when co nsumers are m ore

    aware and prejudiced toward the sex role portrayal issue. Eor

    example, consumers with a strong negative attitude toward

    stereotyped role portrayals in advertising may exhibit strong

    emo tions of disapproval when show n a portrayal incong ruenr

    with their self-schema and beliefs. Thus, an advertisemen

    might generate different emotions depending on consumers

    prior exposure and prejudices. The effect of incongruity is

    likely to be magnified as prejudices increase.

    Strong support for this proposition is provided by past re-

    search indicating that neglecting differences in consumer

    prejudices could bias consumer evaluarions (Haddock and

    Zan na 19 93 ; Lysonski and PoUay 199{)) and th at a priori a tti -

    tudes correlate with diminished effectiveness of a particular

    ad {Bhat, Leigh, and W ardlow 1998). A series of studies ex

    amined consumer general perception of sex role portrayals in

    advertising and reported stereotypes being reflected. Women

    were found to be primarily depicted in conformity with tra-

    ditional gender roles, relationships, and situations {Ferguson

    Kreshel, and Tinkham 1990; Wiles, Wiles, and Tjernlund

    1995).

    Similarly, stereotypes were reported for the portrayal

    of men. Most male models were shown in conformity with

    traditional gender roles {e.g., as businessmen, sportsmen, au-

    thority and father f igures) (Kolber and Albanese 1996

    Wolheter and Lammers 1980).

    In their landmark study, Lundstrom and Sciglimpaglia

    {1977) introduced a scale for measuring respondents' genera

    attitudes toward sex role portrayals in advertising. The in-

    strum ent was employed in subsequen t studies and generally

    was found to be useful for assessing overall issue awareness

  • 7/21/2019 Cons_Response.pdf

    4/14

    80

    The Journal nf Advertising

    (Ford, LaTour, and Lundstrom 199 1; Lundstrom, W hite , and

    Chopoorian 1999)- Lundstrom and Sciglimpaglia (1977) fur-

    ther reported thar some groups of female consumers, usually

    defined by demographic variables (i.e., income, education),

    were more aware of stereotyped role portrayals in advertising

    than others and that corresponding prejudices influenced

    women s attitude s toward firms and produ cts. Their findings

    were complemented by more recent research demonstrating

    that females who were more prejudiced toward stereotyped

    sex role portrayals in advertising were more likely to exhibit

    negative attitudes toward firms or products associated with

    stereotyped role portrayals (Lysonski and Pollay 1990). At

    this point, it is importatit to note that whereas this stream of

    research examined primarily female consumer perceptions and

    attitudes at an aggregated level, the Lundstrom and Sciglim-

    paglia (1977) scale is getierally considered a useful tool for

    assessing consumer prejudices on the issue of sex role portray-

    als in advertising.

    Given the results of these studies, it is expected that con-

    sumers vary considerably in their self-reported prejudices

    toward sex role portrayals because of differences due to ex-

    perience, family values, or education. It also seems reason-

    a b l e to e xpe c t c onsum e r p r e jud i c e s , as a n i nd iv idu a l

    difference variable, to interact with sex role incongruity to

    strongly inf luence emotio nal responses, with subseq uent

    effects on consumer attitudes. In other words, it is proposed

    that consumers emotional and att i tudinal responses to an

    ad incorporating sex role imagery will be moderated by their

    prejudices:

    H2:

    Effects of sex role portrayal incongruity on consume r

    emotionswill emoderated hy an individual s prejudtce towa rd

    sexroleportrayals. The more (less) prejudiced an individual

    is, the stronger (weaker) his or her emotional response to an

    incongruent sexroleportrayal will be.

    Given the range of prejudices among consumers and in the

    light of earlier findings of direct effect ofsex role portrayals

    on brand att i tude (Leigh, Rethans, and Reichenbach W hitney

    1987),

    it is further hypothesized that sex role incongruity, as

    moderated by consumer prejudices, influences brand a ttitu de.

    Consid ering the f indings of previous mediational m odels

    (Burns, Biswas, and Babin 1993), emotional response and at-

    titude toward the ad are expected to mediate the effect of sex

    role incongruity on brand att i tude. Given the f indings of

    Holb rook and B atra s (1987 ) research on th e role of emo tions

    in shaping attitudes, it is hypothesized that role incongruity

    influences attitude toward the ad, with the effect moderated

    by consumer prejudices. Moreover, the effect of role incon-

    gruity on attitude toward the ad is expected to be attenuated

    when em otional response is included in the m odel as an addi-

    tional predictor (Batra and Ray 1986; Holbrook and Batra

    1987):

    H3: Effects of the incongruity of an advertisement s sex role

    portrayal on consume r emotional

    response

    are moderated by

    consum ers prejudices and, in turn, m ediateeffects on attitude

    toward the advertisement.

    Figure I summarizes the postulated relationships.

    M E T H O D

    Subje c t s a nd D e s ign

    The data was collected at two mall and two city center loca-

    tions in Brno, Czech Republic. To obtain a balanced gender

    quota, the selection process combined quota sampling with a

    systematic approach in which every third person who passed

    the interviewer was asked to participate. Three hundred and

    twenty adults (1 61 females and 159 m ales) were selected ran-

    domly within a narrow age bracket (18 to 35 years) to ex-

    clude as many potential ly distor t ing effects of personal

    variables as possible. Respondent profiles at the sampling lo-

    cations showed no significant differences.

    Consumers were assigned randomly to one of eight stimu-

    lus schemes. Each individual was given an envelope contain-

    ing the relevant experimental ad treatm ent and a questionnaire

    measuring emotional responses, at t i tudes, prejudices, and

    demographic characteristics.

    Translation and back-translation techniques were used in

    the development of the instrument to ensure cultural equiva-

    lence. The questionnaire was written in English, translated

    into Czech by a bilingual graduate student, back-translated

    into English by a bilingual faculty member, and then com-

    pared with the original questionnaire to check for congruence.

    Some rewording of questions was necessary to accommodate

    phraseology commonly used in Czech, and after minor revi-

    sion, the questionnaire was printed in Czech.

    S t im ul i

    To classify and choose the specific ads used in the experiment,

    established procedures were followed (Bhat, Leigh, and

    Wardlow

    1998;

    Jaffe 199 4). A cell phon e service was selected

    as an item popular with the target group; it appealed to both

    female and male consumers and there was frequent use of sex

    role portrayals in recent advertisements. The cell phone ser-

    vice was given a fictional brand name to remove influences of

    prior brand evaluation and knowledge.

    A focus group of ad practitioners in the Czech Republic

    was asked to suggest types of images that could plausibly be

    used in advertising to portray sex roles. Based on the focus

    group recommendations, a portfolio of storyboards (settings

    and role portrayals) was developed. Colored magazine-style

    advertisements were created as the experimental stimuli be-

  • 7/21/2019 Cons_Response.pdf

    5/14

    Winter 2005^ 8

    FIGURE 1

    Study Variables and Expected Linkages

    Role In

    (

    icongruity

    Rl

    HI

    H3

    Attitude

    Toward the Ad

    ^aa

    Consumer

    Prejudices

    iPREJ

    H2

    w

    ^

    Emotions

    Disappn

    tval

    Approval

    Surprise

    H 3

    Attitude

    Toward the Brand

    cause magazine ads were commonly used to promote the se-

    lected service.

    Past research (Linville and Carlston 1994) suggests that a

    person's self-concept is accessed through the social situation,

    which is defmed as the (1) physical aspects of the situation

    (e.g., places) and (2) social surroutidings (e.g., other people

    present in the situation, their traits, and their roles). In addi-

    tion, sex role portrayal researchers frequently employed the

    setting as a nonverbal cue for conveying sex roles (Belkaoui

    and Belkaoui 1976; Wagner and Banos 1973). This praxis

    acknowledges tha t nonverbal cues are fundamental in under-

    standing interpersonal interactions and are frequently being

    used in the information processing of social interaction be-

    tween men and women (Burgoon 1994). Accordingly, role

    portrayal was manipulated in this study by modifying the

    characterization and setting of the models relative to each

    other.

    Following the guidelines on selecting pictorial ad elem ents

    (Phillips 1997), a total of six storyboards were created, three

    in an occupational settin g and three in a nonoccupational set-

    ting, with four different sex role schemes each. A panel of

    local independent judges (marketing faculty, ad profession-

    als,

    target audience members) then rated the pretest ads on

    their ability to capture a range of responses to sex role por-

    trayals. Eight advertisements within the two storyboards that

    generated a wide range of responses were used as the ad stim uli

    in this study (see Appendix). In the occupational setting, the

    two models were placed in an office with the boss sitti ng to

    the left at his or her desk while talking on a cell phone an

    the secretary serving coffee from th e right. In the nonoccu

    pational setting, the two models were in a house that wa

    being renovated. W hile the leader was talking on the ce

    pho ne, the follower was sweep ing the floor. Th e ad cop

    size,

    and layout were made identical to control for potenti

    confounds. Text was limited to the brand's logo and a singl

    catchphrase to enhance realism and minimize text as a con

    found. For the occup ational settin g the catchph rase was Yo

    wan t reliable employees- You definitely w ant our service

    and for the nonoccupa tional setti ng it was Your partn er ma

    dump youOur signal will stay with you.

    Measu r e s

    Attitudes

    The attitudes examined were consumer prejudices, attitud

    toward the ad, and brand attitude. The survey instrument fo

    consumer prejudice was identical to the one used in th

    Lundstrom and Sciglimpaglia (1977) study and follow-u

    research (Ford, LaTour, and Honeycutt 1997; Lundstrom

    White, and Chopoorian 1999; Lysonski and Pollay 1990

    Response to ten attitude statements pertaining to the gener

    perception of advertising role portrayals was measured usin

    a seven-point Likert-style scale tanging from 1 = strong

    disagree to 7 ^ strongly agree. To avoid response set pro

    lems, some items were reversed in the actual questionnair

  • 7/21/2019 Cons_Response.pdf

    6/14

    82 The Journal of Advertising

    TABLE I

    CFA Results for the Emotions Scale: Revised Mode

    Exogenous construct

    Variable

    Standard factorial loading

    t values

    Disapproval

    Approval

    Surprise

    Correlations among latent

    cons tructs t values)

    isapproval

    Approval

    urprise

    Note CFA = confirmatory t^accor analysis.

    Angry

    Contemptuous

    Disgusted

    Distrustful

    Irri tated

    Revolted

    Scornful

    Skeptical

    Uneasy

    Wor r i ed

    Curious

    Envious

    Excited

    Interested

    Involved

    Stimulated

    Wishfu l

    Surprised

    Astonished

    isapproval

    1.000

    -.661 -5.94)

    .233 2.43)

    .66

    .75

    .65

    .60

    .66

    .60

    .79

    .66

    .79

    .63

    .64

    .70

    .66

    .68

    .59

    .73

    .75

    .75

    .64

    Approval

    1.000

    - .129 -1.83)

    8.90

    10.43

    9.34

    9.44

    9.86

    6.43

    12.69

    8.46

    11.43

    8.45

    9.54

    10.64

    8.92

    9.14

    6.24

    12.64

    10.05

    9.91

    9.69

    Surprise

    LOOO

    In this study, a prejudice toward sex role portrayal would be

    indicated through consistently low values. Six corresponding

    items that exceeded the threshold factor loading of .5 were

    assigned to a single common factor by factor anaiysis

    {KMO

    [Kaiser-Meyer-Olkin} = .881, explained percentage of vari-

    ance ^ .597). Reliability of the scale of selected items was

    sufficient (Cronbach s a

    =

    .791 ). A single measure for con-

    sumer prejudice was calculated for each respondent as the av-

    erage score on six items of the scale.

    The measures for attitude toward the ad and brand atti-

    tude were two three-item, seven-point semantic differential

    scales adapted from previous operationalizations (MacKenzie

    and Lutz 1989). Corresponding items were assigned to com-

    mon factors by a confirmatory factor analysis (Hair eta l. 1998).

    Factor loadings and / values for attitude toward the ad were

    both significant and positive (bad/good

    =

    .7 51 , / ^ 12..37,

    [un}favorable - .65 7, r - 9 .19 , tdisjlik ed - . 6 5 5 , / - 10.18),

    as were the parameters for brand attitude (bad/good ^ .702,

    / = 12.17,[un}favorable = .637,?

    =

    10.44, [dislliked

    =

    .660,

    / ^ 11.57). Both constructs were highly correlated (/? = 720,

    t

    = 14.17), indicating an

    A^ ^

    effect consistent with prior re-

    search (Brown and Stayman 1992). With fit statistics indi-

    cating an acceptable fit of the structural equation model to

    the data (x [7] - 13.15,

    p

    - .110,

    GFI

    [goodness-of-fit in-

    dex] = .994,

    AGFI

    [adjusted goodness-of-fit index} = .9 68 ,

    NFI

    tnormed fit index] = .970,

    RMSR

    [root mean square re-

    sidual] ^ .081), the mean scores of the ratings were used as

    aggregated measures for attitude toward the ad and brand

    attitude in the further analysis.

    Emotions

    The disapproval/approval/surprise scale (Bhat, Leigh, and

    Wardlow 1998) was employed for measuring consumer emo-

    tional response. To ensure construct equivalence in the se-

    lected cultural con text, the validity of the m easurement model

    was examined through confirmatory factor analysis. Since the

    initial model fit statistics indicated that the model proposed

    by Bhat, Leigh, and Wardlow (1998) could be improved, the

    loadings (consistency within and across constructs) and cor-

    relations were examined to determine which exogenous vari-

    ables should be dropped (Hair et al. 1998). In the revised

    model, the variables happy and loving were removed, with

    19 remaining variables loading on three constructs (Table 1).

  • 7/21/2019 Cons_Response.pdf

    7/14

    8

    FIGURE 2

    Alternative Models

    Model 1

    Incongruity

    Emotions

    Model 2

    Consumer

    Prejudices

    Incongruity

    Emotions

    Model 3

    Incongruity

    Consumer

    Prejudices

    Emotions

    The fit statistics for the revised model were considered satis-

    factory: X'(149) -

    215,p

    .057,GFI

    .9\d,AGFl

    .898,

    NFl =. 8 9 1 , RMSR =.07 3. Accordingly, the revised model

    was accepted for use of the constructs disapproval, approval,

    and surprise in the subsequent analysis.

    RE SU L T S

    M a n i p u l a t i o n C h e c k

    To check the role incongruity manipulation, respondents were

    asked the extent to which they either agreed or disagreed w ith

    the following statem ent: The sex roles portrayed in the ad-

    vertisement are congruent with what I believe are appropriate

    roles for men and wo men in society (scale from 1

    strongly

    disagree to 7 - strongly agree). To test the effectiveness of the

    man ipula tion, d ata was analyzed by using an analysis of vari-

    ance (ANOVA) model in which incongruity was the depen-

    dent variable and the setting and scheme treatments were the

    independe nt variables. Results indicate that incongruity scores

    were significantly affected by the selected sex role portrayals

    F ^ 14.843;/J .001; X^= .275). Across genders, consumers

    exposed to ads portraying women in roles superior to men

    (stimuli 3 and 5) reported h igher inc ongru ity, whereas con-

    sumers exposed to ads featuring men in superior roles (stimuli

    4 and 8) reported lower incongruity.

    Test of H I an d H2 : Ef fec ts of Role Inco ngr ui ty

    o n E m o t i o n s

    Before examining effects of role incongruity on respondent

    attitudes, it first appeared to be necessary to determine ho

    consumer prejudice, a presumably moderating variable, i

    teracts with sex role incongruity in influencing emotion

    response. An adequate model for the interaction of the var

    ables had to be identified. Th ree alternative m odels were eval

    ated (see Figure 2): Model (1): Role incongruity is the on

    variable affecting consumer emotional response; Model (2

    Both role incongruity and consumer prejudices jointly affe

    emotions; and Model (3): Consumer prejudices moderate t

    effect of role incongruity on emotions.

    To identify the most likely constellation, three regressi

    analyses were cond ucted. Since the m odels hold different d

    grees of freedom, the most appropriate model could be ide

    tified considering (high) R^^^and (low) Akaike Informati

    Criterion (AIC; Akaike 1987). Approval, disapproval, a

    surpr ise were regressed separa te ly aga ins t incongrui

    incongruity + prejudice, and incongruity + incongruity pre

    dice (see Table 2).

    Significant effects were found for all emotional dimension

    with relatively high values for the explained variance. Co

    sidering

    R\^

    and AIC, Model 3, representing a moderati

    role of consumer prejudices, proves to be the superior mod

  • 7/21/2019 Cons_Response.pdf

    8/14

    84 The Journal of Advertising

    TABLE 2

    Results of the Regression Analyses on Effects of Role Incong ruity ft/)

    and Consumer Prejudices

    PREJ)

    on Emo tional Dimensions

    Independent variable

    R/

    Model I)

    R/ + PRE/

    Model 2)

    R/ + fi/ PREy

    Model 3)

    Dependent variable

    Parameters

    2.435

    .045

    ,079

    N0.43

    2.842

    - .028

    -.029

    168.16

    1 428

    - . 1 1 3

    .267

    317.87

    1.282***

    .185***

    253***

    .137

    102.15

    1 596

    - . 1 9 4

    - .3 4 0

    .179

    162.15

    1 953

    - . 0 1 4

    -. 116

    ,084

    389.22

    1 624

    .188

    .306

    .288

    82.55

    1 588

    - . 1 9 4

    - . 4 6 1

    ,317

    136.39

    1.291***

    - . 2 6 2

    - ,1 9 4

    ,108

    387.37

    Approval

    AIC

    Disapproval

    AiC

    Surprise

    AIC

    p< .05 .

    *** / - < .01 .

    for effects on disapproval and approval (comparatively small-

    est AIC and highest R\J. For surpris e, a different m echa-

    nism was identified: According to the quality parameters, role

    inco ngru ity has a direct effect on surprise. These findings su p-

    port H1 and H2, and are consistent with prior research (Alden,

    Mukherjee, and Hoyer 2000; Garbarino and Edell 1997).

    Test of

    H 3 :

    Effec ts of Role Incongrui ty and

    E m ot ions on A t t i t ude T ow a r d the A d

    Direct and mediating effects of role incongruity were evalu-

    ated by perfotming additional regression analyses. This proce-

    dure follows Baron and Kenny (1986, p. 1177), who state,

    There is no need for hierarchical or stepwise regression or the

    com putatio n of any partial or semipartial correlations. Sub-

    jects of the evaluation were as follows: the effect of the inde-

    pendent variable (incongruity) on the dependent variable

    A

    (Model 1), the effect of the mediator variable (emotions) on the

    dependent variable (Model 2), and effects of the independent

    variable and the med iator on the depen dent v ariable (Model 3).

    A significant, but weak, direct effect of role incongruity on

    attitu de toward the ad was found (Model 1). In addition , the

    emotional dimensions directly infiuenced consumer attitude

    toward the ad, strongly in the case of disapproval and ap-

    proval (Model 2). More detailed role incongruity and approval

    affected attitude toward the ad positively, whereas disapproval

    and surprise yielded negative effects. Again, these fmdings

    are in line with previous research (Alden, Mukherjee, and

    Hoyer 2000).

    Mediating effects of emotions on A^^ could be proved by

    demon strating th at ( l) incongruity, moderated by prejudices,

    affects emo tions ; (2) the emo tional dim ensio ns affect a ttit ud e

    toward the ad; and (3) the effect of incongruity on attitude

    toward the ad is attenuated when att itud e is regressed on bo th

    emotions and incongruity (Baron and Kenny 1986). The first

    two conditions have already been established in the previous

    section. In addition. Table 3 shows effects of incongruity on

    attitude toward the ad with four independent variables (Model

    3) and in combination with one emotional dimension at a

    time (Models 3 a, b, c). The correspon ding parameters indi-

    cate that the effect of incongruity on attitude toward the ad is

    reduced when attitude is regressed on the emotional dimen-

    sions and incongruity simultaneously, hence satisfying the

    third condition. Since the infiuence of incongruity and sur-

    prise on attitu de toward the ad appears to be relatively sm all,

    two additional regression analyses were conducted to exam-

    ine the predictive power of these variables m ore closely (Models

    4 and 5). The results indicate that drop ping both variables

    from the set of predictor variables in each case leads to a de-

    crease in

    R~^^ .

    Hence, Model 3, which includes incongruity

  • 7/21/2019 Cons_Response.pdf

    9/14

    int r

    2005^

    85

    TABLE

    3

    Evaluation of Dire ct and Mediating Effects of Role Incongruity (R/) and Emotions on Att i tu de To wa rd the A d

    Model

    Parameter

    Approval Disapproval

    Surprise

    lla:

    A^^= f(approval)

    lib:

    A^ = f(disapp roval)

    lie:

    A^ = / (surp rise)

    III:A^^

    =

    f{RI approval,

    disapproval,

    surprise)

    Ilia:

    A

    =

    f{Rl

    approval )

    Illb:

    A = f (R/, disap prova l)

    Illc: A = f(R/, surprise)

    IV:

    A =

    f(approva l, disapproval, surprise)

    = f(R/, app roval, disapproval)

    * / - < . 1 0 .

    **/ j < .05 .

    * * * / >