Consumer Disbelief and Attitudes: An Implicit Cognition

Download Report

Transcript Consumer Disbelief and Attitudes: An Implicit Cognition

Consumer Response to False Information:
Is Believability Necessary for Persuasion?
•Claudiu V. Dimofte
•Georgetown University
•Richard F. Yalch
•University of Washington
•May 7, 2005
•Vancouver, BC
Outline
• Crisis Management
• Damage Control (Study 5)
• False Marketplace Information
• Negative Rumors about Company/Brand
• Overselling of Product/Brand Features
• The Implicit Account
• Rumors
• Information Processing (Studies 1, 2, 3)
• Infomercials
• Curious Disbelief (sic) (Study 4)
• Discussion/Conclusions
PRODUCT
INFORMATION
FALSE
TRUE
+
-
+
-
ADS/WOM
CRISES
DECEPTIVE
SALES
RUMORS
PRODUCT
INFORMATION
FALSE
TRUE
+
ADS/WOM
CRISES
+
-
DECEPTIVE
SALES
RUMORS
Coping with Crises
• Examples
• Celebrities
•
•
•
•
•
•
Bill O’Reilly – sexual harassment
Michael Jackson – being weird
Pat O’Brien – alcoholic leaving obscene messages
Martha Stewart – prison time for lying
Janet Jackson – Superbowl
Paul Abdul – misbehavior with contestant
Damage Control
1. Do Nothing – Bill O’Reilly
2. Refutation – Michael Jackson
3. Apologize and Go On – Pat O’Brien
4. Retrieval – Martha Stewart “Apprentice spinoff ”
5. Storage – Janet Jackson “Equipment malfunction”
6. Counterattack – Paula Abdul
Study 5: Damage Control
• Goals:
• Look at all damage control strategies in one experiment
• Participants:
• 133 undergrads from introductory GT Marketing class
• Method:
• Participants are exposed to news coverage on Bank of
America losing customer data
• Participants take IAT and provide explicit truth ratings
Study 5 - Stimuli
•
•
Explicit Procedure Stimuli:
•
Participants read web news story on Bank of America’s
losing data
•
After short delay, they are given 6-item questionnaire about
problem gravity, responsibility, BA evaluation, BA data safety,
importance of apology, safety of BA vs. WF
•
Conditions: apology, do nothing, deny, counterattack, storage,
retrieval
Implicit Procedure Stimuli:
•
IAT Bank of America/Wells Fargo, safe/unsafe
CRISIS
APOLOGY
REFUTE
STORAGE
ATTACK
RETRIEVAL
ATTACK
Study 5 - Gravity
"Not Serious at All"
3
2.5
2
1.5
1
0.5
0
Do Nothing
Apologize
Refute
Storage
Retrieval
Counterattack
Study 5 – BofA’s Responsibility
"Not their fault"
3
2.5
2
1.5
1
Do Nothing
Apologize
Refute
Storage
Retrieval
Counterattack
Study 5 – Data Safe & vs Wells Fargo?
"Exremely Safe"
4.5
4
3.5
3
2.5
2
1.5
1
Do Nothing
Apologize
Refute
Storage
Retrieval
Counterattack
Study 5 – Overall Evaluation of BofA
"Extremely Favorable"
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
Do Nothing
Apologize
Refute
Storage
Retrieval
Counterattack
Study 5 – Implicit Associations
BofA
- Safe
0.04
0.02
0
Do Nothing
Apologize
Refute
Storage
Retrieval
Counterattack
-0.02
-0.04
-0.06
-0.08
BofA
- Unsafe
Study 5 – Conclusions
• Explicit Results:
• Apologize: Admitting fault was generally worst strategy.
• Ignore: left doubt and did not help. Second worst strategy.
• Refute: slight help with most issues.
• Storage: lowered perceived severity & BofA concerns. Best strategy.
• Retrieval: only lowered severity but aided overall evaluation.
• Counterattack: Minimized severity but kept low safety ratings. Similar but
not as good as retrieval for overall evaluations.
Study 5 – Conclusions
• Implicit Results:
– Bank of America is strongly automatically associated with ‘unsafe’
in all cases except:
 apology (only time better thanWells Fargo)
 storage (equal to Wells Fargo)
• Overall Insights:
– Storage works at both levels, by turning – into +
– Apology
– – for explicits but + for implicits
– Explicit judgments, think about why they apologized
– Implicit judgments, think about BofA not the apology or safety
PRODUCT
INFORMATION
FALSE
TRUE
+
-
+
-
ADS/WOM
CRISES
DECEPTIVE
SALES
RUMORS
False Negative
Information
• Rumors about Company/Brand
• Rumors: specific propositions or beliefs passed along from person to
person without any secure standards of evidence
(Allport and Postman 1947)
• Originate in unconscious desires that are transformed to become
conscious (Rossignol 1973)
• “[…] the rumor could exist at various levels of consciousness and
could lead one to get a pizza or a taco without being aware of why
one did so” (Koenig 1985)
Unfounded Rumor Examples
• Corona
• Mountain Dew
• Procter & Gamble
Failing to Combat Rumors can
have Severe Consequences
Rumor Quelling Strategies
• Tybout et al. (1981): refutation does not work, use storage or
retrieval strategies
• Iyer and Debevec (1991): rumors are less credible when propagated
by someone that can gain out of its dissemination
• Koller (1992): the best way to fight rumors is to explain their rumor
and lack of veracity via positive advertising
• Kamins et al. (1997): rumors are generally more easily spread when
they are personally relevant and favorable
• Bordia et. al (2000): best way to fight rumors is via honest denial
Information Processing
Insights
•
Tybout et al. (1981)
•
Directly refuting rumors is the least effective way to deal with
them; offered two alternative strategies:
1.
STORAGE STRATEGY: expose consumers to a secondary
stimulus during encoding of rumor information, making the
brand more likely to be associated with that stimulus rather than
the rumor
2.
RETRIEVAL STRATEGY: expose consumers to a secondary
stimulus during retrieval of brand information, thus lessening the
chance of the joint retrieval of brand and rumor
Information Processing Insights (cont’d)
•
Tybout et al. (1981) design: employed rumor of
McDonald’s’ use of red worm meat in their burgers
1. STORAGE STRATEGY: during encoding of worm
rumor information, consumers were exposed to a
secondary stimulus (confederate claiming a famous,
pricey local French restaurant uses tasty worm
sauces)
GOAL: make worm meat more desirable or more
associated with French restaurants than McDonalds.
Information Processing Insights (cont’d)
2. RETRIEVAL STRATEGY: during retrieval of
McDonald’s information, consumers were exposed to a
secondary stimulus (questionnaire about the McDonald’s
location they frequent the most)
GOAL: lessen likelihood of joint retrieval of
McDonald’s and worm meat rumor
3. REFUTATION STRATEGY: after exposure to worm
rumor, consumers were exposed to McDonald’s’ claim
that red worm meat is too expensive to use.
GOAL: alter believability of the worm rumor
Tybout (1981) results
12
10
8
6
4
2
0
No Rumor
Rumor
Refutation
Storage
Retrieval
Information Processing Insights (cont’d)
•
Problem: no clear understanding of what the processing mechanisms
behind these effects are
•
STORAGE: did it disrupt the McDonald’s – worms association or did it
make eating worms more positive by relating it to French food?
•
RETRIEVAL: did it block the activation of the McDonald’s – worm
association or inhibited its retrieval in relation to other McDonald’s
thoughts?
•
REFUTATION: after study, subjects in all conditions showed strong
disbelief in the rumor’s veracity so why were they still affected by it?
Explicit vs. Implicit Processing
"Are you aware of the unconscious hostility you are
exhibiting to us right now?“
Doctor X, Psychiatrist
”How could I be aware if it's unconscious?”
Leonard, Mental Hospital patient
in the movie The Awakening
Explicit vs. Implicit Processing
• Explicit (implicit) processing: those aspects of cognition which
are (un)available to the individual's conscious awareness.
• The methods used to test for each differ:
• explicit processing is typically examined directly, by asking
individuals to evaluate their own thought processes;
• implicit processing is typically examined indirectly by
evaluating performance on tests that depend on thought
processes that are not subject to introspection.
Explicit vs. Implicit Processing
• Fazio et al. (1986): attitudes characterized by a strong
association between the attitude object (AO) and its
evaluation are capable of being activated from memory
automatically upon mere presentation of AO.
• Devine (1989): dissociation of automatic and controlled
processes involved in prejudice
• social stereotype is automatically activated in the presence of a
member of stereotyped group and that low-prejudice
responses require controlled inhibition thereof
Explicit vs. Implicit Processing
• Bargh et al. (1996): stereotypes become active automatically
in the presence of relevant behavior or stereotyped-group
features
• participants for whom an elderly stereotype was primed walked
more slowly down the hallway when leaving experiment
• Greenwald et al. (1998): measured individual differences in
implicit cognition with implicit association test (IAT)
• Bottom line: automatic cognition occurs and is driven by associations
that we generally cannot control.
Rumor Processing: The Implicit Account
• False information persuades via an unconscious route by
building automatic associations between brands and the
information cues in the message.
• While explicitly rejecting the veracity of outlandish brand
rumors, consumers lack control over the implicit
associations occurring at exposure and being practiced
during subsequent evaluations.
Study 1 - Storage
•
•
Goal:
•
Replicate Tybout et al. (1981) storage strategy results
•
Disentangle processing mechanisms behind strategy
Participants:
•
•
229 undergrads from introductory Marketing class
Method:
•
Participants are exposed to rumor and secondary stimulus
•
Filler task for 5 minutes
•
Participants take IAT, then provide explicit brand ratings
Study 1 - Stimuli
•
•
Explicit Procedure Stimuli:
•
Participants read series of supposed New York Times
news stories on politics, business (McDonald’s worm
rumor troubles in Asia), sports, and leisure.
•
Half read about gourmet worm food from Asia
Half read about house decorating.
Implicit Measures:
•
IAT McDonald’s/Burger-King, food-related/worm-related
•
IAT food-related/worm-related, pleasant/unpleasant
Study 1 – Stimuli (cont’d)
press “d” for
press “k” for
Worm-related
Food-related
OR
OR
McDonald’s
Burger King
Study 1 – Stimuli (cont’d)
press “d” for
press “k” for
Worm-related
Food-related
OR
OR
McDonald’s
Burger King
larva
Study 1 – Stimuli (cont’d)
press “d” for
press “k” for
Worm-related
Food-related
OR
OR
McDonald’s
Burger King
beef
Study 1 – Stimuli (cont’d)
press “d” for
press “k” for
Worm-related
Food-related
OR
OR
McDonald’s
Burger King
Study 1 – Stimuli (cont’d)
press “d” for
press “k” for
Worm-related
Food-related
OR
OR
Pleasant
Unpleasant
beef
Study 1 – Stimuli (cont’d)
press “d” for
press “k” for
Worm-related
Food-related
OR
OR
Pleasant
Unpleasant
rainbow
Study 1 – Stimuli (cont’d)
press “d” for
press “k” for
Worm-related
Food-related
OR
OR
Pleasant
Unpleasant
hurricane
Study 1 – Explicit Results
Brand Rating
80
60
40
20
0
Control
Storage
• Tybout et al. (1981) results replicate: consumers in the storage
condition have higher evaluations of McDonald’s than those in the
rumor-only condition: t(228) = -2.20, p < .03
Study 1 – IAT Results
McDonald’s Worms
Worms Unpleasant
0.5
D-measure
0.4
0.3
0.2
0.1
0
McDonald’s Food
McDonald's-Worm s IAT
Control
Worm s-Affect IAT
Worms Pleasant
Storage
• McDonald’s is equally (and weakly) associated with worms in both conditions:
t(117) < 1, ns
• Worms are associated with unpleasant in both conditions, but less so in storage:
t(110) = 2.34, p < .03
Study 1 – Conclusion & Unresolved Issue
•
Conclusion: the storage strategy works by minimizing the
negative feelings associated with the rumor via the positive
secondary stimulus
•
Problem: the McDonald’s – Worms IAT showed little
association, even in the control condition
•
Stronger rumor induction may be needed (vivid images or
repetition).
Study 2 – Retrieval vs Refutation
•
Goal: compare two strategies in terms of implementation delay
following rumor exposure.
•
Method:
•
113 participants exposed to rumor
•
Followed by secondary stimulus
• Retrieval – questions unrelated to rumor
• Refutation – statement why the rumor is false
•
Filler task for 5 or 30 minutes
•
Participants take IAT, then provide explicit brand ratings
Study 2 - Stimuli
•
•
Explicit Procedure Stimuli:
•
Participants read series of supposed New York Times
news stories on politics, business (McDonald’s worm
rumor troubles in Asia), sports, and leisure.
•
After short/long delay, they are given 4-item
questionnaire about the McDonald’s/Circuit City they
frequent, or read McDonald’s press release refuting
rumor.
Implicit Measures:
•
IAT McDonald’s/Burger-King, food-related/wormrelated
Study 2 – Explicit Results
Brand Rating
80
70
*
60
50
*
40
Control
Retrieval
Short Delay
Refutation
Long Delay
* p < .05
• Tybout et al. (1981) results do not replicate:
– interaction Condition x Delay: F(2, 96) = 3.14, p < .05
• Retrieval strategy does not work, fast refutation does
Study 2 – IAT Results
McDonald’s Food
0.03
*
D-measure
0
-0.03
-0.06
-0.09
*
McDonald’s Worms
-0.12
Control
Retrieval
Short Delay
Refutation
* p < .05
Long Delay
• Main effect of Delay: F(2, 104) = 3.34, p = .07
• Retrieval does work implicitly if positive thoughts elicited fast enough
• Rumors are difficult to fight after delay
Study 3
• Automatic associations quickly established with
initial exposures and explicit and implicit – but is this
limited to novel stimuli?
• If so, we are truly not as susceptible to undue influence
• If not, can anything be done to prevent undue
associations from emerging?
• Test: employ strong pre-existing association, try to
change it[s strength] via rumor
Study 3
• Goals:
•
Look at strong, universally prevalent association:
• Females and caring
•
Participants:
•
•
133 undergrads from introductory GT Marketing class
Method:
•
Participants are exposed to male pregnancy rumor or
message on home decoration
•
Participants take IAT and provide explicit truth ratings
Study 3 - Stimuli
•
•
Explicit Procedure Stimuli:
•
Participants read supposed CNN story on Asian
hospital and its male pregnancy program
(house decoration in control)
•
After short delay, they are given 4-item questionnaire
about the likelihood of this story’s being true, etc.
•
All asked about possibility of male pregnancy
Implicit Procedure Stimuli:
•
IAT female/male names, caring/uncaring
Study 1 – Stimuli (cont’d)
press “d” for
press “k” for
Male
Female
OR
OR
caring
uncaring
John
Study 1 – Stimuli (cont’d)
press “d” for
press “k” for
Male
Female
OR
OR
caring
uncaring
nurturing
Study 3 - Results
•
Explicit Results:
•
Rumor elicits more curiosity:
t(132) = 130.34, p < .001
•
Rumor is not credible:
t(132) = 125.13, p < .001
•
Male pregnancy is possible:
t(132) = 131.20, p < .001
•
Implicit Results:
•
Both conditions associate female names more with ‘caring’
•
Slightly less so in rumor condition, ns
Studies 1 – 3 Conclusions
• Rumors are difficult to manage because explicit beliefs are
often not a consideration
• Worm rumor not believed but  McDonalds attitudes
• Male pregnancy believed but  males no more caring
• Believability does not appear to be a major consideration in
using information to make judgments.
Studies 1 – 3 Explanations
• Cartesian view on comprehension and acceptance of assertions:
messages are at once comprehended and rejected if assessed to
be unbelievable
• Spinozan view: all information is first accepted during
comprehension, and only subsequently rejected upon
believability assessment (Gilbert et al. 1990).
• Support for the latter: all information is automatically awarded
bona fide status upon encounter, unless strong prior associations
of different valence exist.
False Positive Information
• Overselling of Product/Brand Features
• “Too-good-to-be-true” commercial/infomercial claims
• Pragmatic Implications
• Consumers drawing unwarranted inferences (Kardes 1988)
due to processing or production deficits (Gaeth and Heath
1987)
• Richards (1990) lists 14 types of deceptive advertising
(preemptive – Wonder Bread builds Strong Bodies 12 ways)
False Positive Information
• The “Curious
Disbelief ”
Phenomenon
• “Hard to Believe”
Advertising Claims
may motivate trial
more than believable
claims
Maloney (1962)
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%
Believers
Curious
Nonbelievers
Disbelievers
Willingness to Try Product
Infomercial Persuasion:
The Implicit Account
• False information persuades via an unconscious route by
building automatic associations between brands and the
novel information cues in the message.
• While explicitly rejecting the veracity of outlandish
infomercial claims, consumers lack control over the
implicit associations occurring at exposure and being
practiced during subsequent evaluations.
Study 4: Power of Infomercials
•
•
Goals:
•
Replicate Maloney (1962) explicit results
•
Disentangle processing mechanisms behind curious
disbelief
Method:
•
62 participants are exposed to Ultimate Chopper
infomercial: alone, preceded or followed by discounting
cue (video or text)
•
Explicit dependent measures collected
•
Participants take two IATs: interesting/believable
Study 4 - Stimuli
•
•
Explicit Procedure Stimuli:
•
Participants watch about 90 seconds-worth of edited
Ultimate Chopper infomercial or read equivalent text
•
Discounting cue: about 90 seconds-worth of edited
20/20 footage on infomercial scams
Implicit Procedure Stimuli:
•
IAT Ultimate Chopper/Henckels knives,
boring/interesting
•
IAT Ultimate Chopper/Henckels,
believable/unbelievable
Infomercial
Infomercial - Expose
Study 4 – Explicit Results
Rating
10
8
a,b
a
6
b
4
2
0
Infomercial
Cue+Infomercial
Infomercial+Cue
Text
• There is something special about (video) infomercials:
Explicit Rating
• main effect: F(3, 61) = 2.40, p = .09; a,b: p < .03
Study 4 – Explicit Results
Believability
2
1.5
1
a,b
0.5
0
a
-0.5
b
-1
Infomercial
Cue+Infomercial
Infomercial+Cue
Text
• Infomercials are generally not credible: all conditions show
disbelief: text info best (slightly above .5 – range: -3 to 3)
F(3, 61) = 2.3, p = .08; a,b: p < .05
• Check via suspicion
Study 4 – Explicit Results
Suspicion
2.5
a,c
2
b,d
1.5
1
a,b
c,d
0.5
0
Infomercial
Cue+Infomercial
Infomercial+Cue
Text
• Supports believability results and shows that cue instills suspicion:
F(3, 64) = 4.6, p < .006; a,b: p < .05; c,d: p < .005
• Again, we have explicit disbelief
Study 4 – Explicit Results
Curiosity
1
0.75
0.5
0.25
0
Infomercial
Cue+Infomercial
Infomercial+Cue
Text
• Not a lot of explicit curiosity elicited: irrelevant category?
F(3, 61) = .2, ns
Study 4 – Explicit Results
Makes Peanut
Butter
1
a,b
c,d
0.8
0.6
a,c
b,d
0.4
0.2
0
Infomercial
Cue+Infomercial Infomercial+Cue
Text
• Similar shape/results for several claims (e.g., crushes ice, will never dull)
• Peanut Butter main effect: F(3, 63) = 5.1, p < .003
a,c: p < .03; b,d: p < .003
• Nonbelief drives evaluations
Study 4 – IAT Results
D-measure
ChopperInteresting 0.5
ChopperBelievable
0.25
0
-0.25
Chopper- -0.5
Boring
Chopper-Interesting IAT
Infomercial
Cue+Infomercial
Chopper-Believable IAT
Infomercial+Cue
Text
ChopperUnbelievable
Study 4 – IAT Results
• Interesting IAT: F(3, 58) = 3.3, p < .03 (we have some
implicit curiosity for all infomercial conditions, text quite
boring: an issue of delivery?)
• Believable IAT: F(3, 58) = 1.9, p = .14 (no implicit disbelief,
infomercial seems to build instant automatic believability:
an issue of testimonials?)
• It appears the “curious nonbelief” phenomenon translates
into “curious implicit belief”
Support for The Implicit Account
• False information may persuade via an automatic route by
building automatic associations between brands and the novel
information cues in the message.
• While explicitly rejecting the veracity of outlandish infomercial
claims or brand rumors, consumers lack control over the implicit
associations occurring at exposure and being practiced during
subsequent evaluations.
• Is believability really important? The response is automatic…