Dynamics of Communicator and Audience Power: The

Dynamics of Communicator and AudiencePower The Persuasiveness of Competenceversus Warmth

DAVID DUBOISDEREK D RUCKERADAM D GALINSKY

The current research offers a new theoretical perspective on the relationship be-tween power and persuasion An agentic-communal model of power is presentedthat proposes power affects both the messages generated by communicators andthe messages that persuade audiences Compared to low-power states high-power states produce a greater emphasis on information that conveys compe-tence As a consequence high-power communicators generate messages withgreater competence information and high-power audiences are persuaded moreby competence information In contrast to high-power states low-power statesproduce a greater emphasis on information that conveys warmth As a result low-power communicators generate messages with greater warmth information andlow-power audiences are persuaded more by warmth information Because ofthese two outcomes a power-matching effect occurs between communicator andaudience power high-power communicators are more effective in persuadinghigh-power audience members whereas low-power communicators are more ef-fective in persuading low-power audience members Four experiments find sup-port for these effects in oral and written contexts with three distinct manipulationsof power Overall these experiments demonstrate that the persuasiveness ofmessages can be affected by the alignment between the psychological sense ofpower of the communicator and the audience

Keywords power persuasion agentic-communal warmth competence

What is the role of psychological power in the persua-sion process Recent research suggests a simple an-

swer those with power are more persuasive than thosewithout power For example interviewees who reflected onpossessing power were more persuasive in a mock interviewcompared to interviewees who reflected on lacking power(Lammers et al 2013) This straightforward answer becomesless convincing when one considers the finding that rela-tively powerless children can deliver certain forms of protec-tive messages more effectively than adults (Pratkanis andGliner 2004) Similarly HIVAIDS-preventive health mes-sages have been found to be effective when conveyed bylow-income individuals as opposed to highly ranked offi-cials (Sachs 2005) So which is it Are the powerful morepersuasive than the powerless or vice versa

The present work tackles this question by offering in-sight into initial conditions under which the powerful

David Dubois (davidduboisinseadedu) is assistant professor of

marketing INSEAD Boulevard de Constance 77305 Fontainebleau

France Derek D Rucker (d-ruckerkelloggnorthwesternedu) is the

Sandy and Morton Goldman Professor of Entrepreneurial Studies in

Marketing Kellogg School of Management Northwestern University

Evanston IL 60208-2001 Adam D Galinsky (adamgalinskycolum-

biaedu) is the Vikram S Pandit Professor of Business Columbia

Business School Columbia University New York NY 10027-6902 This

article is based on the first authorrsquos dissertation The authors thank the edi-

tor associate editor and reviewers for their helpful feedback Support to

the authors from the Kellogg School of Management Northwestern

University INSEAD and HEC Paris is gratefully acknowledged

Eileen Fischer served as editor and Sharon Shavitt served as associate

editor for this article

Advance Access publication February 22 2016

VC The Author 2016 Published by Oxford University Press on behalf of Journal of Consumer Research Inc

All rights reserved For permissions please e-mail journalspermissionsoupcom Vol 43 2016

DOI 101093jcrucw006

68

versus the powerless are more persuasive We proposethat having or lacking power may change the diagnostic-ity of information (Feldman and Lynch 1988) and thusthe extent to which information is used or relied uponWe argue that psychological states of high power leadcommunicators to favor the use of competence-related ar-guments and audiences to rely more heavily on compe-tence information when forming their attitudes Incontrast psychological states of low power lead commu-nicators to favor the use of warmth-related argu-ments and audiences to rely more heavily on warmthinformation in forming their attitudes As a consequenceof these independent effects the success of a persuasivemessage can hinge on the interaction between communi-cator and audience power High-power communicatorsgenerate messages that persuade high-power audienceslow-power communicators generate messages that per-suade low-power audiences

Before developing our hypotheses further we elaborateon the construct of power

POWER AND ITS EFFECTS ONCOMMUNICATORS AND AUDIENCES

Power is defined as asymmetric control over valued re-sources in social relationships (Magee and Galinsky2008 Rucker Galinsky and Dubois 2012) Researchershave studied power as a psychological state or mindsetthat can arise from both structural differences insocioeconomic status (Dubois Rucker and Galinsky2015 Kraus Piff and Keltner 2009 Stephens Markusand Townsend 2007) as well as situational factors such asonersquos social role (eg boss vs employee Anderson andGalinsky 2006) The last decade has witnessed an acceler-ation of research efforts to unpack powerrsquos far-reachingeffects on consumer behavior (Dubois Rucker andGalinsky 2012 Rucker et al 2012) However to datescant evidence exists as to how power systematically af-fects the creation and reception of persuasive messagesThe little work that does exist can be broken into researchfocused on understanding the effects of power on commu-nicators (ie those delivering a message) or on audiences(ie those receiving a message)

With regard to communicators having power in-creases communicatorsrsquo attempts to influence others(Kipnis 1972) and to express onersquos own opinion ingroups (Anderson and Berdahl 2002 Galinsky et al2008) As noted Lammers et al (2013) found that indi-viduals primed with high power were more persuasivethan those primed with low power Furthermore a com-municatorrsquos power can enhance persuasion by serving asa simple heuristic (Petty Cacioppo and Goldman 1981)and thus induce greater compliance in an audience(Cialdini 2009)

With regard to audience power Brinol and colleagues(2007) found that an audiencersquos psychological power af-fected their response to persuasion in two distinct waysFirst increasing an audiencersquos power prior to receiving amessage led to less scrutiny of the message and therefore aweaker discrimination between weak and strong argu-ments Second when the audiencersquos power was increasedafter a message they became more confident in their mes-sage-related thoughts As a consequence powerful individ-uals were more likely to use those thoughts in forming anopinion on the topic which led to more positive attitudesafter strong arguments but more negative attitudes afterweak arguments Although Brinol et al (2007) examinedaudience power they did not manipulate communicatorpower

Beyond these few examples the relationship betweenpower and persuasion remains relatively unexplored Theknowledge accumulated so far suggests a simple main ef-fect high-power communicators are more persuasive thanlow-power communicators In contrast to this idea we in-troduce the notion that power can fundamentally shape thevalue or diagnosticity of information and thus the type ofinformation communicators convey and that audiencesgive weight to Consequently high- or low-power commu-nicators may be more or less persuasive depending onwhether they are communicating to an audience in a stateof high or low power

AN AGENTIC-COMMUNAL MODEL OFPOWER IMPLICATIONS FOR WARMTHVERSUS COMPETENCE INFORMATION

Agency Communion and Power

Bakan (1966) introduced the ideas of agency and com-munion to reflect two fundamental modalities of humanthought and behavior Agency is marked by a focus on theself and produces consequences such as independence andpersonal striving In contrast communion focuses peopleon others and enhances individualsrsquo attention to gettingalong with others and othersrsquo needs Rucker et al (2012also Rucker Dubois and Galinsky 2011 Rucker andGalinsky 2015 2016) proposed an agentic-communalmodel of power by which power affects peoplersquos orienta-tion toward the world in an agentic or communal fashionhigh power fosters an agentic orientation and low powerfosters a communal orientation Because the powerful areless dependent on others they can pursue their own goalsand interests with fewer constraints (ie they are imbuedwith agency) In contrast the dependency of the powerlessrequires them to attend to and incorporate others to achievetheir goals (ie they require communion)

One aspect of agentic and communal orientations is thatthey appear to relate to competence and warmth informa-tion respectively (Asch 1946 Cuddy Fiske and Glick

DUBOIS RUCKER AND GALINSKY 69

2004 Ybarra Chan and Park 2001 also Cuddy Fiskeand Glick 2008 Fiske Cuddy and Glick 2007 Judd et al2005) We propose that an agentic orientation focuses peo-plersquos attention on information that relates to othersrsquo capa-bilities with respect to the skillfulness efficacyintelligence and confidence attached to a target (ie com-petence) Conversely we propose that a communal orienta-tion focuses peoplersquos attention on information associatedwith how good natured trustworthy tolerant friendly andsincere a target is perceived to be (ie warmth) Althoughboth competence and warmth are important dimensions ofdaily interactions our hypothesis suggests that power shiftsthe balance in terms of how concerned people are witheach dimension

The constructs of warmth and competence potentially offerinsights into how power fundamentally affects persuasion Inparticular as described next drawing on the accessibility-diagnosticity perspective (Feldman and Lynch 1988) weposit that both audience and communicator power affect thediagnosticity of warmth and competence information andthus the extent that warmth and competence information isused when generating and assessing message arguments

The Accessibility-Diagnosticity Perspective

In a seminal article Feldman and Lynch (1988) distin-guish between two dimensions of attitude formation the ex-tent to which an individual readily and easily retrieves apiece or type of information (hereafter accessibility) and the

extent to which an individual finds a piece or type of infor-mation relevant for the task at hand (hereafter diagnostic-ity) A core principle of the accessibility-diagnosticitymodel is that information is more likely to be recruited in ajudgment when it is accessible and viewed as diagnostic

Building on the accessibility-diagnosticity perspectivewe propose that high-power states can increase the diag-nosticity and thus the use of competence-related argumentsbecause of the association between high power and agencyIn contrast because of the association between low powerand communion we propose that low-power states can in-crease the diagnosticity and thus the use of warmth-relatedarguments As a result when generating messages high-power communicators are more likely to use competence-related arguments (eg stressing skillfulness and intelli-gence) whereas low-power communicators are more likelyto use warmth-related arguments (eg stressing friendli-ness and trustworthiness) In a similar vein we proposethat when evaluating messages high-power audiences aremore likely to rely on competence-related argumentswhereas low-power communicators are more likely to relyon warmth-related arguments

As a consequence of these conceptual links we predict thatpersuasion can be enhanced when a match exists between com-municator and audience power This outcome is predicated onpower leading to both differential generation and reception ofcompetence versus warmth information (figure 1 shows theconceptual model) Of note in the case of communicatorswhen given the opportunity to generate a message

FIGURE 1

THEORETICAL MODEL OF THE RELATIONSHIP BETWEEN POWER AND PERSUASION

70 JOURNAL OF CONSUMER RESEARCH

spontaneously it is also possible that competence and warmthinformation may be differentially accessible In this researchhowever we focus on testing and providing evidence for thediagnosticity account

FORMAL HYPOTHESES

We introduce formal hypotheses regarding a new role ofpower in the persuasion process First we suggest that therelationship between power and persuasion depends on thepsychological power of the communicator and the audi-ence We propose that communicator and audience powerinteract to affect persuasion with greater persuasion occur-ring whenmdashunbeknownst to either partymdasha match existsbetween the power states of the communicator and the au-dience Formally

H1a High-power communicators will be more effective at

persuading high-power audiences than low-power

audiences

H1b Low-power communicators will be more effective at

persuading low-power audiences than high-power

audiences

Next we propose that this matching effect occurs be-cause power shifts both the communicatorrsquos use and theaudiencersquos reliance on competence versus warmth-relatedarguments More formally power shapes the argumentscommunicators generate as follows

H2a High-power communicators compared to low-power

communicators are more inclined to generate and select ar-

guments related to competence

H2b Low-power communicators compared to high-power


guments related to warmth

Turning to audiences we propose that power shapes thearguments that audiences rely on to form their attitudes inthe following manner

H3a High-power audiences compared to low-power audi-

ences are more inclined to rely on competence-related argu-

ments to form their attitudes

H3b Low-power audiences compared to high-power audi-

ences are more inclined to rely on warmth-related argu-


OVERVIEW

Four experiments test our hypotheses Experiment 1 inves-tigates whether matching communicatorsrsquo and audiencesrsquopsychological state of power enhances persuasion in an oralcontext Experiments 2 and 3 examine communicator and au-dience power in written contexts and test whether this effectcan be accounted for by a differential use of warmth or

competence at the communication stage and a potential dif-ferential in reliance on warmth or competence in persuasivemessages at the reception stage Finally experiment 4 demon-strates that differences in power shift audiencersquos reliance onwarmth versus competence information when assessing per-suasive messages

We report all manipulations and any data exclusions inour experiments Sample sizes were based on subject avail-ability as well as unrelated research projects run in con-junction with these experiments No additional data wereadded after analyses In some experiments we collected ad-ditional measures after the key hypothesis-related measuresfor exploratory purposes A discussion of these measures isavailable from the authors upon request

EXPERIMENT 1 MATCHINGCOMMUNICATOR AND AUDIENCE

POWER

Experiment 1 manipulated both the psychological stateof power of the communicator and that of the audience inan oral transmission context Of central interest was the au-diencersquos responsiveness to the message We placed partici-pants into dyads and instructed one participant to persuadethe other to use a gym facility Although power does nottypically affect mood (Brinol et al 2007 Rucker et al2011 Smith and Galinsky 2010 Smith and Trope 2006)we measured and controlled for differences in mood givenprior research documents that mood can affect the mes-sages communicators generate (Bohner and Schwarz 1993Forgas 2006) and whether audiences are persuaded(Labroo and Rucker 2010 Schwarz Bless and Bohner1991)

Participants and Design

A total of 120 business undergraduates (72 women) at aFrench university participated as part of a classroom aca-demic exercise on persuasion Participants were randomlyassigned to a 2 (communicator power high vs low) 2(audience power high vs low) between-participants de-sign Participants were assigned to the role of communica-tor or audience member (ie recipient) in one of fourdyads high-power communicator and high-power audi-ence high-power communicator and low-power audiencelow-power communicator and low-power audience low-power communicator and high-power audience Thus eachaudience member was yoked to a single communicator

Procedure

All written materials were presented in French All partici-pants first completed a one-page questionnaire presented as acognitive warm-up which included a scrambled sentencestask (Smith and Trope 2006) In reality this served as the


power manipulation Next participants were randomly as-signed to the role of communicator or audience member aspart of a persuasion exercise conducted over two consecutivesessions Participants assigned to the role of communicatorwere told they would be paired with another participant andthat their task was to persuade this participant to use the newon-campus gym facilities Communicators were subsequentlygiven one minute to think individually about what theywanted to say Participants assigned to the audience conditionwere informed they would be paired with another participantand would listen to a short speech from that participant

Next participants sat down in dyads and the persuasionexercise began Communicators were given one and a halfminutes to convince the audience member to start using anew on-campus gym facility A timer signaled the begin-ning and the end of the task During the persuasion taskneither communicators nor audiences were aware of thepower condition they were assigned to nor were they in-formed that the warm-up task was meant to induce powerAfter the persuasive task ended both communicators andaudience members completed a final questionnaire For au-dience members the questionnaire asked them to reporthow persuasive the communicator was as well as their like-lihood of using the new on-campus gym facilities For bothcommunicators and audience members the questionnairecontained manipulation checks

Independent Variables

Power Both communicators and audience memberscompleted a scrambled sentences priming task (Smith andTrope 2006) containing 16 sets of words For each set par-ticipants were presented with five words and instructed tomake a sentence using four of the five words listed In thehigh-power condition 8 of the 16 sets included a word re-lated to having power (ie authority captain commandscontrols dominates executive influenced privileged) Inthe low-power condition 8 of the 16 sets included a wordrelated to lacking power (ie complied janitor obey pas-sive servant submits subordinate yield)

CommunicatorAudience Role Participants either heldthe role of communicator (ie tried to convince the audi-ence member to start using the new on-campus gym facili-ties) or audience member (ie listened and assessed howconvinced they were by the communicatorrsquos speech)

Dependent Variables

Communicator Persuasiveness Audience members re-ported the extent to which they found the communicator tobe convincing and persuasive (anchored at 1frac14Extremelyunconvincing and 7frac14Extremely convincing and 1frac14Extremely unpersuasive and 7frac14Extremely persuasive)These two items were averaged to form a persuasion index(rfrac14 94)

Likelihood of Using the New Gym Facilities Audiencemembers reported how likely they would be to start usingthe new gym facilities on a 7-point scale anchored at1frac14Very unlikely and 7frac14Very likely

Manipulation Checks At the end of the session partic-ipants answered three questions assessing the extent towhich they felt powerful during the persuasion task on7-point scales anchored at powerless-powerful withoutcontrol-in control weak-strong Because these items werehighly correlated (afrac14 92) they were aggregated into asingle power index In addition mood was assessed with a7-point scale anchored at 1frac14Sad and 7frac14Happy

Results and Discussion

Manipulation Checks A two-way analysis of variance(ANOVA) revealed that participants reported feeling sig-nificantly less powerful in the low-power condition(Mfrac14 359 SDfrac14 101) compared to the high-power condi-tion (Mfrac14 434 SDfrac14 141 F (1 116)frac14 1083 pfrac14 001gp

2frac14 08) Importantly no main effect of role (Flt 1) orinteraction between power and role was present (Flt 1)suggesting that participantsrsquo experienced power was not af-fected by whether they were assigned to a role of commu-nicator or audience member As an additional check ofwhether our manipulation produced dyads of similar anddissimilar psychological power we computed the differ-ence score in power between each communicator and eachaudience member within matching conditions (low-powercommunicator and low-power audience member high-power communicator and high-power audience member)and mismatching conditions (low-power communicatorand high-power audience member low-power communica-tor and high-power audience member) A one-wayANOVA on this difference score revealed that the gap infeelings of power between communicators and audiencemembers was lower in the matching conditions (Mfrac14 64SDfrac14 63) than in the mismatching conditions (Mfrac14 111SDfrac14 82) F(1 58)frac14 608 pfrac14 02 gp

2frac14 09) further sug-gesting that our power manipulation produced the intendedeffect within dyads Finally a two-way ANOVA on themood measure found no main of power role or their inter-action on participantsrsquo mood (Flt 1)

Communicator Persuasiveness A two-way ANOVAon the audiencersquos ratings of communicator persuasivenessrevealed a significant communicator power audiencepower interaction F (1 56)frac14 1581 plt 001 gp

2frac14 22No additional effects were found for communicator poweror audience power (Flt 1) High-power communicatorswere more persuasive when addressing a high-power audi-ence member (Mfrac14 503 SDfrac14 128) than a low-power au-dience member (Mfrac14 393 SDfrac14 108) F(1 56)frac14 560pfrac14 02 dfrac14 92) In contrast low-power communicatorswere more persuasive when addressing a low-power


audience member (Mfrac14 520 SDfrac14 159) compared to ahigh-power audience member (Mfrac14 373 SDfrac14 94) F(156)frac14 1032 pfrac14 002 dfrac14 112)

Likelihood of Using the New Gym Facilities A two-way ANOVA on audiencesrsquo likelihood of going to the gymrevealed a significant communicator power audiencepower interaction F (1 56)frac14 1084 pfrac14 002 gp

2frac14 16No other effects emerged (Flt 1) High-power communica-tors were more effective in convincing a high-power audi-ence to use the new gym facilities (Mfrac14 500 SDfrac14 146)than a low-power audience (Mfrac14 380 SDfrac14 126) F(156)frac14 486 pfrac14 031 dfrac14 87) In contrast low-power com-municators were more effective in convincing a low-poweraudience to use the new gym facilities (Mfrac14 526SDfrac14 183) compared to a high-power audience (Mfrac14 393SDfrac14 133) F(1 56)frac14 601 pfrac14 01 dfrac14 83)

Experiment 1 offers evidence that persuasion can be ajoint function of the psychological state of power of boththe communicator and audience in an oral contextSpecifically matching communicator and audience powerled to more persuasion (support of hypothesis 1a and hy-pothesis 1b) Where past findings suggest that high-powercommunicators are more persuasive than low-power com-municators (Kipnis 1972 Lammers et al 2013) the currentexperiment demonstrates that both low- and high-powercommunicators can be more persuasive based on an audi-encersquos power mindset One limitation however is that wedid not measure whether this matching effect is a result ofa differential use of warmth versus competence informa-tion We address this concern in experiment 2

EXPERIMENT 2 THE MEDIATING ROLEOF WARMTH AND COMPETENCE

Experiment 2 aimed to replicate the findings of experi-ment 1 in a written context and to provide evidence for theunderlying process Based on our agentic-communal ac-count high- and low-power states affect the use and relianceon information related to competence and warmth Becausehigh-power communicators view competence as more diag-nostic they are more likely to use competence in the genera-tion of their arguments because high-power audiences viewcompetence as more diagnostic they are more likely to relyon it when forming their attitudes and thus more persuadedby messages from high-power communicators Converselybecause low-power communicators view warmth as morediagnostic they are more likely to use warmth in the genera-tion of their arguments because low-power audiences viewwarmth as more diagnostic they are more likely to rely onwarmth information when forming their attitudes and thusthey are in turn more persuaded by messages from low-power communicators

To test this perspective we assigned independent codersto measure the warmth and competence of the messages

generated by communicators to see if (1) communicatorsgenerated messages that differed along these dimensionsand (2) these measures explained the differential persua-sion among audiences We also added baseline conditionsfor purposes of establishing the locus of the effect Finallywe examined message length to test whether the effect ofcommunicator power on persuasion stems from changes inmessage format as opposed to message content


A total of 360 students at Northwestern University (173men) were assigned to a 3 (communicator power baselinelow high) 3 (audience power baseline low high) be-tween-participants design Each audience member wasyoked to a single communicator and only saw that one mes-sage yielding 180 observations on the key measure of audi-ence attitudes

Procedure

We manipulated power using an episodic recall taskParticipants were subsequently assigned to either the roleof communicator or audience member

In the communicator sessions participants were in-structed to write a persuasive message promoting their uni-versity to prospective students Specifically participantswere told

Imagine that you have been chosen by the deanrsquos office to

promote your university at several top high-schools in the

country We would like you to write a persuasive speech

promoting your university as if you were to try to convince

an audience to apply to your university

In the audience sessions participants were randomlypresented with a message generated by a previous partici-pant in the high-power low-power or baseline conditionNeither audiences nor communicators were aware of thepower condition of the other Finally we asked audiencesto provide their attitude toward the target stimuli


Power Power was manipulated by having participantswrite about a time they had or lacked power (GalinskyGruenfeld and Magee 2003) Participants in the baselinecondition were asked to recall the last time they went tothe grocery store (Rucker and Galinsky 2008)

Dependent Variables

Attitudes Toward the University Participants in the au-dience condition responded to three 7-point scales to assesstheir attitudes toward the university The scales were an-chored at the end points bad-good unfavorable-favorablenegative-positive (Petrocelli Tormala and Rucker 2007)


and combined to form a single measure of attitudes(afrac14 94)

Message Content We asked two coders blind to thehypotheses to code the messages generated by the commu-nicators on two dimensions aggregated into two indicescompetence (four items capable skillful intelligent confi-dent afrac14 88) and warmth (five items good natured trust-worthy tolerant friendly sincere afrac14 89 Cuddy et al2008) conveyed by the message All items were assessedon 7-point scales and discrepancies between coders wereaddressed through discussion (table 1 shows intercoder re-liability by dimension) Finally we accounted for messagelength via counting the number of words

Manipulation Checks At the end of the experimentwe included questions to assess the extent to which the es-say made participants feel powerful on 7-point scales an-chored at powerless-powerful without control-in controlweak-strong We aggregated these measures into a singlepower index (afrac14 90)


Manipulation Checks A two-way ANOVA (power androle) on the power index revealed only a main effect ofpower Participants felt significantly less powerful in thelow-power condition (Mfrac14 281 SDfrac14 110) compared tothe baseline condition (Mfrac14 347 SDfrac14 120) and high-power condition (Mfrac14 449 SDfrac14 152 F (2 354)frac14 5171plt 001 gp

2frac14 22) suggesting the power manipulationwas successful Neither the main effect of role F (1354)frac14 125 pfrac14 26 gp

2frac14 004 nor the interaction be-tween power and role was significant F (2 354)frac14 119pfrac14 30 gp

2frac14 007 Post hoc tests revealed that participantsfelt significantly more powerful in the high-power condi-tion compared to both the baseline condition F (1357)frac14 3749 plt 001 dfrac14 74 and the low-power condi-tion F (1 357)frac14 10188 plt 001 dfrac14 126 In additionlow-power participants felt significantly less powerful thanbaseline participants F (1 357)frac14 1576 plt 001 dfrac14 57

As in experiment 1 playing the role of communicator oraudience member did not affect participantsrsquo sense ofpower

Audience Attitudes We submitted participantsrsquo attitudeindex to a 3 (communicator power) 3 (audience power)ANOVA

A marginal main effect of communicator poweremerged F (2 171)frac14 295 pfrac14 055 gp

2frac14 03 High-power communicators (Mfrac14 444 SDfrac14 139) were morepersuasive than baseline communicators (Mfrac14 388SDfrac14 116) F (1 171)frac14 588 pfrac14 01 dfrac14 43) but did notdiffer from low-power communicators (Mfrac14 418SDfrac14 135) F (1 171)frac14 127 pfrac14 26 dfrac14 18 Low-powercommunicators did not significantly differ from baselinecommunicators F (1 171)frac14 168 pfrac14 19 dfrac14 23 No ef-fect of audience power on audience attitudes emerged(Flt 1) (figure 2)

Of central importance a significant communicatorpower audience power interaction emerged F (4 171)frac14428 pfrac14 003 gp

2frac14 09 which we decomposed by audi-ence type First among high-power audiences a main ef-fect of communicator power emerged F (2 57)frac14 588pfrac14 005 gp

2frac14 17 high-power communicators persuadedhigh-power audiences more effectively (Mfrac14 505SDfrac14 162) than both baseline communicators (Mfrac14 383SDfrac14 121 F (1 57)frac14 834 pfrac14 005 dfrac14 85) and low-power communicators (Mfrac14 376 SDfrac14 110 F (1 57)frac14928 pfrac14 003 dfrac14 93) High-power communicators didnot differ statistically in persuading low-power and base-line communicators pfrac14 87 dfrac14 06 Among low-poweraudiences a main effect of communicator power occurredF (2 57)frac14 437 pfrac14 01 gp

2frac14 13 low-power communi-cators produced greater persuasion (Mfrac14 490 SDfrac14 150)than both baseline communicators (Mfrac14 373 SDfrac14 111F (1 57)frac14 811 pfrac14 006 dfrac14 88) and high-power com-municators (Mfrac14 403 SDfrac14 121 F (1 57)frac14 447pfrac14 039 dfrac14 65) while no difference was present betweenhigh-power and baseline audiences (pfrac14 46) Among base-line audiences communicator power did not affect atti-tudes (Flt 1)

Overall high-power communicators persuaded high-power audiences more effectively than low-poweraudiences (hypothesis 1a) In contrast low-power commu-nicators persuaded low-power audiences more effectivelythan high-power audiences (hypothesis 1b)

Message Competence Judges rated high-power com-municators as delivering messages conveying greater com-petence (Mfrac14 471 SDfrac14 156) than both baseline(Mfrac14 391 SDfrac14 134 F (1 177)frac14 1023 pfrac14 002dfrac14 55) and low-power communicators (Mfrac14 374SDfrac14 112 F (1 177)frac14 1512 plt 001 dfrac14 71) Messageratings between the baseline and low-power communica-tors did not differ pfrac14 69 These findings supporthypothesis 2a

TABLE 1

INTERCODER RELIABILITY BY DIMENSION EXPERIMENT 2

DimensionPearson r (dimensionsassessing message)

Capable 901Skillful 866Intelligent 846Confident 916Good natured 904Trustworthy 848Tolerant 861Friendly 851Sincere 871


Message Warmth Judges rated low-power communi-cators as delivering messages conveying more warmth(Mfrac14 473 SDfrac14 136) than both baseline (Mfrac14 352SDfrac14 102 F (1 177)frac14 3459 plt 001 dfrac14 100) andhigh-power communicators (Mfrac14 383 SDfrac14 98 F (1177)frac14 1921 plt 001 dfrac14 76) Message ratings betweenbaseline and low-power communicators did not differ(pfrac14 13) Together these findings support hypothesis 2b

Length Number of words did not vary as a function ofcommunicator power (Flt 1)

Mediation Analyses We tested whether the effect ofcommunicator power on persuasion was mediated by mes-sage competence andor warmth and whether these poten-tial mediating paths were moderated by audience power(Hayes 2013 model 14 figure 3) Given our theoreticalmodel and specific hypotheses we focused on the high-and low-power conditions We coded high power as 1 andlow power as 1

First replicating the ANOVA results we found thatcommunicator power predicted message competence(bfrac14 52 t(78)frac14 342 pfrac14 001) and message warmth(bfrac1444 t(78)frac14 319 pfrac14 002) Next a comprehensiveregression predicting persuasion from communicatorpower audience power the two mediators (message com-petence and message warmth) and their interaction withaudience power revealed a main effect of message compe-tence (bfrac14 52 t(73)frac14 812 plt 001) and message warmth(bfrac14 39 t(73)frac14 541 plt 001) Of central importanceboth the audience powermessage competence (bfrac14 37t(73)frac14 735 plt 001) and audience powermessagewarmth (bfrac1450 t(73)frac14 622 plt 001) interactionswere significant suggesting the presence of moderatedmediation Supporting this proposition the index of

moderated mediation was significant for both warmth(95 [CI] 17ndash76) and competence (95 CI 15ndash65 fig-ure 3 shows complete path coefficients) confidence inter-vals No other significant effect or interaction emerged(pgt 16) Together these results indicate that high-powercommunicators compared to low-power communicatorsused more competence-related arguments and that mes-sages with competent arguments were more persuasiveamong high-power audiences than among low-power audi-ences In contrast low-power communicators compared tohigh-power communicators used more warmth-related ar-guments and messages with warm arguments were morepersuasive among low-power audiences than among high-power audiences

FIGURE 2

COMMUNICATORSrsquo PERSUASIVENESS AS A FUNCTION OF AUDIENCE POWER AND COMMUNICATOR POWER EXPERIMENT 2

300

350

400

450

500

550

600

Baseline High Low

Audience Power

High Power Communicator Baseline Communicator Low Power Communicator

FIGURE 3

MEDIATION THROUGH WARMTH AND COMPETENCEJUDGMENTS EXPERIMENT 2

Communicator Power (-1 = Low 1 = High)

Attitudes

52

-44

37

-01NS

= p lt 05 = p lt 01

Message Warmth

Message Competence

Audience Power (-1 = Low 1 = High)


-50 39

CI = [15 65]

CI = [17 76]

52


Overall this study replicated the power-matching effectobtained in experiment 1 in a written context We also foundevidence consistent with our proposed account of the processat the communication stage having power led communicatorsto emphasize greater competence whereas lacking power ledcommunicators to emphasize greater warmth At the recep-tion stage audiences in a state of high power were more per-suaded by competence-based messages than warmth-basedmessages In contrast audiences in a state of low power weremore persuaded by warmth-based messages than compe-tence-based messages Together these accounts are consistentwith the general idea that both communicators and audiencesplace differential diagnosticity on warmthcompetence as afunction of power

EXPERIMENT 3 DIFFERENTIALDIAGNOSTICITY OF WARMTH AND

COMPETENCE

In prior experiments communicators generated their ownmessages which introduces two limitations First it is possi-ble that the messages generated varied in content beyondwarmth and competence Second it is possible that beyondviewing competence and warmth as more diagnostic commu-nicators drew on this information because it was more acces-sible Experiment 3 aimed to address these issues by havingcommunicators construct messages from a pool of argumentspreselected to vary in warmth and competence but not otherdimensions such as valence or abstractness

We expected high-power communicators to select agreater number of competence-related arguments but low-power communicators to select a greater number ofwarmth-related arguments In turn we expected a high-power audience to be more persuaded by messages withcompetence-related arguments but a low-power audienceto be more persuaded by messages with warmth-related ar-guments Having communicators choose arguments helpedus to better isolate the importance of message content re-lated to warmth and competence In addition holding theaccessibility constant (ie everyone read and selected asubset of arguments from the same pool) allowed us to bet-ter isolate diagnosticity on the part of communicators

EXPERIMENT 3 PRETEST

Sixty-two participants (35 women) from the Paris metro-politan area took part in a laboratory session Participantswere recruited through the use of flyers and online advertisingand were compensated E12 for an hour As part of a largerexperimental session participants were exposed to 18 argu-ments about a restaurant (appendix 1) Nine of the argumentswere designed to consist of information primarily associatedwith warmth whereas nine of the arguments were designedto consistent of information primarily associated with

competence The order of the arguments was counterbal-anced Warm arguments emphasized communal aspects ofthe restaurant and the disposition of the staff (eg ldquoThechefrsquos friendly reputation makes this restaurant very invitingrdquoldquoYou will find very amiable sommeliers able to assist you inyour choice of winerdquo) In contrast competent arguments fo-cused more on skills and abilities associated with the restau-rant (eg ldquoThe chef has built a reputation based on his skillsand competencerdquo ldquoYou will find capable sommeliers able toassist you in your choice of winerdquo the appendix provides thefull list of arguments) Note that arguments as with all mate-rials in this experiment were presented originally in French

For each argument participants were presented with a se-ries of adjectives and asked on a 7-point scale whether theyendorsed the adjective as associated with the statement (ieldquoTo what extent does this statement convey ___rdquo on a scalefrom 1frac14Not at all to 7frac14Extremely) Participants were pre-sented with two items to assess perceived warmth (warmthtrustworthiness) and two items to assess perceived compe-tence (competence skillfulness) In addition we included twoitems to assess action orientation (action passivity the sec-ond item was reverse-coded prior to data analysis) and twoitems to assess abstraction (abstraction concreteness the sec-ond item was reverse-coded prior to data analysis) Last butnot least we assessed the persuasiveness of each statementthrough two items (how persuasiveconvincing is this argu-ment to you) The items designed to assess warmth (rfrac14 81plt 001) competence (rfrac14 83 plt 001) action orientation(rfrac14 73 plt 001) abstraction (rfrac14 82 plt 001) and persua-sion (rfrac14 84 plt 001) were correlated and thus we averagedthem to construct five independent indices

Next we performed a series of repeated-measuresANOVAs to examine participantsrsquo perceptions of the argu-mentsrsquo warmth and competence with two factors argumenttype (ie warmth vs competence) and argument dimension(ie the repeated measure assessing the target dimension ofwarmth vs competence) These analyses revealed that as awhole the warm and competent arguments differed inwarmth and competence but did not differ in abstraction ac-tion orientation abstraction or persuasiveness (table 2)

MAIN EXPERIMENT


A total of 160 participants (96 women) from the Parismetropolitan area were randomly assigned to a 2 (communi-cator power high vs low) 2 (audience power high vslow) between-participants design as part of laboratory ses-sions They were recruited through the use of flyers and on-line advertising and were compensated E12 for an hour Asin prior experiments we used a yoked design where low- orhigh-power audiences were given a message for a restaurantcomposed by a low- or high-power communicatorHowever unlike prior experiments instead of generating


their own arguments communicators selected argumentsfrom the list developed in our pretest

Procedure

We conducted two types of lab sessions one in whichparticipants were assigned to the role of a communicatorand another in which participants were assigned to the roleof an audience member Regardless of the type of sessionparticipantsrsquo power was manipulated first by assigning themto a role of boss or employee (Anderson and Berdahl 2002Galinsky et al 2003) Participants were told they would ei-ther serve as a boss in charge of employees (high power) oras an employee who would follow the orders of a boss (lowpower) in a task that would occur later in the experimentalsession All written materials were presented in French

Participants in the communicator condition were instructedto design a persuasive message by picking out 6 argumentsfrom the list of the 18 arguments from our pretestParticipants in this condition were told that their goal was tocompose a message for a restaurant by selecting the most per-suasive arguments from the list Participants assigned to theaudience condition were randomly presented with a messagefrom a low- or high-power communicator in a subsequent labsession As in prior experiments audiences were unaware ofcommunicatorsrsquo power and vice versa Participants in the au-dience condition then provided their attitude toward the res-taurant Finally at the end of the experimental session allparticipants were thanked and debriefed All materials werepresented in French


Communicator and Audience Power The power manip-ulation for both communicators and audiences was identicaland consisted of assigning participants to a role of boss or em-ployee for a subsequent task Participants first completed aleadership questionnaire and were told that they would be

assigned to a role as part of a group task on the basis of theiranswers to the questionnaire as well as the experimenterrsquos ob-servation of their nonverbal behaviors Participants in thehigh-power condition were given a written description of theirrole that read as follows (translated from French)

As a boss you are in charge of directing your subordinates

in creating different products and managing work teams

You decide how to structure the process of creating products

and the standards by which the work done by your em-

ployees is to be evaluated As the boss you have complete

control over the instructions you give your employees In

addition you also evaluate the employees in a private ques-

tionnairemdashthat is the employees never see your evaluation

The employees have no opportunity to evaluate you

In contrast participants in the low-power condition read(translated from French)

As an employee you are responsible for carrying out the or-

ders of the boss in creating different products The boss de-

cides how to structure the process of creating these products

and the standards by which your work is to be evaluated As

the employee you must follow the instructions of the boss In

addition you are evaluated by the boss and this evaluation

will be private that is you will not see your bossrsquos evaluation

of you This evaluation will help determine the bonus reward

you get You have no opportunity to evaluate your boss

Manipulation Checks At the end of the experimentwe assessed the extent to which the participantsrsquo assignedrole made them feel powerful on 7-point scales anchored atpowerless-powerful without control-in control weak-strong afrac14 91)

Dependent Variables

Communicatorrsquos Choice and Rank of Statements Wedummy-coded the arguments used (Competentfrac14 1Warmfrac14 0) and then summed across all six arguments

TABLE 2

RESULTS OF REPEATED-MEASURES ANOVAS ON EACH DIMENSION OF THE PRETEST WITH TWO FACTORS ARGUMENT TYPE(IE WARM VS COMPETENT) AND ARGUMENT DIMENSION (IE THE REPEATED MEASURE ASSESSING THE TARGET

DIMENSION) PRETEST EXPERIMENT 3

Dimension Warm arguments (9 items) Competent arguments (9 items) Test

Warmth Mfrac14477 Mfrac14295 F (1 61)frac1442775 plt 001SDfrac14144 SDfrac14132 gp

2frac14 87Competence Mfrac14287 Mfrac14470 F (1 61)frac1440921 plt 001

SDfrac14134 SDfrac14143 gp2frac14 87

Action orientation Mfrac14394 Mfrac14384 F (1 61)frac141119 pfrac14 29SDfrac14176 SDfrac14174 gp

2frac14 02Abstraction Mfrac14402 Mfrac14415 F (1 61)frac14180 pfrac14 19


Persuasiveness Mfrac14418 Mfrac14417 F (1 61)frac14 010 pfrac14 92SDfrac14151 SDfrac14161 gp

2frac14 00

NOTEmdash None of the analyses revealed a significant effect of argument dimension or a significant argument dimension argument type interaction the F tests

reported are all main effects of argument type


chosen This approach yielded a measure of the number ofcompetent arguments used by participants and thus the over-all degree of competence (vs warmth) emphasized in themessage This measure could range from 0 (ie no compe-tent arguments and all warm arguments) to 6 (ie all com-petent arguments and no warm arguments) In addition weexamined the order in which communicators placed theirstatements within their message Research on primacy ef-fects in persuasion suggests that communicators mightsometimes place statements they value earlier in their mes-sage especially in a scenario where they are encouraged toinvest resources into constructing the message (Haugtvedtand Wegener 1994) such as ours To confirm this intuitionwe asked 25 individuals from the same population to imag-ine they had to craft a persuasive message composed of sixarguments one of which was stronger than all others Theywere then asked to report where they would place this argu-ment 68 chose to rank it first (17 of 25) 4 second (1 of25) 12 third (3 of 25) 4 fourth (1 of 25) 0 fifth (0 of25) and 12 sixth (3 of 25) v2(5)frac14 4196 plt 001

Audience Attitudes Participants assigned to the audi-ence condition were asked to report their attitudes towardthe message topic using the same three items as in experi-ment 2 bad-good unfavorable-favorable negative-posi-tive afrac14 94)


Manipulation Checks A two-way ANOVA on thepower index revealed participants felt significantly lesspowerful in the low-power condition (Mfrac14 353SDfrac14 131) than in the high-power condition (Mfrac14 466SDfrac14 141 F (1 156)frac14 2733 plt 001 gp

2frac14 14) No ef-fect of communicator versus audience role or apower role interaction was present (Flt 1) suggestingthat participantsrsquo power did not depend on whether theywere assigned to the communicator or audience condition

Communicatorrsquos Choice and Rank of Statements Aone-way ANOVA on the number of competent argumentsused revealed a main effect of power F (1 78)frac14 3878plt 01 gp

2frac14 33 high-power communicators selectedmore competent arguments (Mfrac14 357 SDfrac14 117) thanlow-power communicators (Mfrac14 200 SDfrac14 138) As agroup high-power communicators used 143 competent ar-guments and 97 warm arguments This pattern was re-versed among low-power communicators who used 80competent arguments and 160 warm argumentsv2(1)frac14 3219 plt 001 In fact compared to a possible bal-anced distribution of 120 warm statements and 120 compe-tent statements a state of high power significantlyincreased communicatorsrsquo likelihood to use competentstatements v2(1)frac14 407 pfrac14 03 whereas a state of lowpower significantly increased communicatorsrsquo likelihoodto use warm statements v2(1)frac14 1303 pfrac14 01 Overall

high-power (low-power) communicators used a signifi-cantly greater number of competent (warm) statementsthan predicted by chance

In addition a one-way ANOVA on the average rankingof competent statements where lower numbers reveal ear-lier placement yielded a main effect of communicatorpower F (1 74)frac14 748 pfrac14 008 gp

2frac14 09 High-powercommunicators placed competent arguments earlier(Mfrac14 318 SDfrac14 69) than low-power communicators did(Mfrac14 374 SDfrac14 108) In contrast a one-way ANOVA onthe average ranking of warm statements revealed a ten-dency albeit nonsignificant (F (1 74)frac14 227 pfrac14 12gp

2frac14 03 to place warm arguments earlier among low-power communicators (Mfrac14 344 SDfrac14 68) than amonghigh-power communicators (Mfrac14 374 SDfrac14 104)

Audience Attitudes A two-way ANOVA on the audi-encersquos attitudinal index revealed a significant interactionF (1 76)frac14 1240 pfrac14 001 gp

2frac14 14 High-power commu-nicators were more effective at persuading high-power au-diences (Mfrac14 448 SDfrac14 149) compared to low-poweraudiences (Mfrac14 356 SDfrac14 131) F (1 76)frac14 469pfrac14 034 gp

2frac14 06 In contrast and again replicating theprior experiments low-power communicators persuadedlow-power audiences more effectively (Mfrac14 486SDfrac14 178) than high-power audiences (Mfrac14 348SDfrac14 119) F (1 76)frac14 792 pfrac14 006 gp

2frac14 09)

Mediation To test the role of communicatorsrsquo choiceof warm and competent arguments in audiencersquo attitudestoward the restaurant we used a moderated mediation anal-ysis (Hayes 2013 model 14) Low-power communicatorswere coded as 1 and high-power communicators werecoded as 1 Our mediator consisted of the number of com-petent arguments in each message Specifically this mea-sure could range from 0 (all warm arguments and nocompetent arguments) to 6 (all competent arguments andno warm arguments) and reflected the extent to whichcommunicators used competent versus warm arguments(ie number of competent argumentsfrac14 6-number of warmarguments)

We found that communicator power predicted the num-ber of competent arguments such that high-communicatorpower was associated with the use of more competent ar-guments in a message (bfrac14 79 t(78)frac14 623 plt 001)Next a simultaneous regression predicting persuasion fromcommunicator power audience power and the mediator(ie number of competent arguments) found a significantaudience power number of competent arguments interac-tion (bfrac14 81 t(75)frac14 864 plt 001) Moreover as an in-dex of moderated mediation we computed whether thenumber of competent arguments mediated persuasion ateach level of audience power This analysis revealed thatthe number of competent arguments successfully explainedthe difference in persuasion among both low-power (95CI 108 to 48) and high-power audiences (95 CI


33ndash83) The positive and negative CIs further reflect thenature of these effects low-power audiences were morepersuaded as the number of arguments slanted towardwarmth high-power audiences were more persuaded as thenumber of arguments slanted toward competence

Experiment 3 provides additional evidence for our hy-pothesis that differences in message warmth and compe-tence underlie the power-matching effect Unlike priorexperiments where communicators generated argumentsthat might have varied on additional dimensions we findpower caused communicators to select differentially amongarguments designed primarily to vary in competence andwarmth High-power communicators picked a greater num-ber of competent arguments when crafting their messagebut low-power communicators picked a greater number ofwarm arguments (in support of hypotheses 1a and 1b) Thissuggests a difference in potential diagnosticity because weheld accessibility constant by giving everyone access to thesame arguments Furthermore audiences appeared to relydifferentially on the selection of warm versus competent ar-guments (in support of hypotheses 2a and 2b) because theamount of persuasion that occurred was linked to the type ofargument present

EXPERIMENT 4 AUDIENCE POWER ANDMESSAGE DIAGNOSTICITY

Experiment 4 tested whether power affects perceptions ofthe diagnosticity of warmth versus competence informationBecause our emphasis was on understanding why audienceswere persuaded by competent versus warm arguments wedid not manipulate communicator power but manipulatedmessage type directly That is we assigned all participants tobe an audience member and assessed the perceived diagnos-ticity of messages that emphasized either warmth or compe-tence In addition the experiment aimed to examine whetherour findings might be explained by the audiencersquos feelings ofsimilarity with the communicator For example perhaps audi-ences make inferences about the source of the message basedon seeing information that they themselves value and thuspersuasion is more a result of inferences about the communi-cator than properties of the message itself To test this alterna-tive perspective we included a measure of similarity


A total of 101 participants (55 women) from the Parisarea were randomly assigned to a 2 (audience power highvs low) 2 (message type competent vs warm) between-participants design in a laboratory session

Procedure

All written materials were presented in French We firstmanipulated power among participants and presented them

with either warm or competent messages about a restaurantusing arguments adapted from the pretest from experiment3 Next in addition to measuring participantsrsquo attitudes wemeasured how diagnostic they viewed each argument givenWe hypothesized that low-power individuals would ratewarmth-related messages as more diagnostic than compe-tence-related messages and high-power individuals wouldrate competence-related messages as more diagnostic thanwarmth-related messages


Audience Power Low- and high-power states weremanipulated through the episodic recall task described inexperiment 2 (Galinsky et al 2003)

Message Content As part of a consumer survey par-ticipants were presented with one of two messages for arestaurant Both messages contained competent andwarm arguments as we aimed to assess whether powershifted participantsrsquo diagnosticity of each type of argu-ment As a consequence we refer to our conditions asldquocompetence-skewedrdquo and ldquowarmth-skewedrdquo respec-tively The competence-skewed message contained twocompetent arguments and one warm argument (translatedfrom French)

The chef has built a reputation based on his skills and com-

petence The intelligent design of the place contributes to

deliver a highly competent service by limiting both the

kitchen and the wait staffrsquos walking The wait staffrsquos

friendly manners create a very inviting atmosphere in the

restaurant

In contrast the warmth-skewed message contained twowarm arguments and one competent argument (translatedfrom French)

The chefrsquos friendly reputation makes this restaurant very in-

viting The dining roomrsquos interior design provides a very

cozy feel to the place that makes customers feel at home

The wait staffrsquos skills and training make them experts at

their jobs

Manipulation Checks Finally we assessed the extentto which participants viewed the message as conveyingwarmth versus competence on two 7-point scales an-chored at ldquonot at allrdquondashldquovery muchrdquo In addition partici-pants answered three questions assessing the extent towhich they felt powerful during the persuasion task on7-point scales (powerless-powerful without control-incontrol weak-strong) aggregated into a single power in-dex (afrac14 89)

Dependent Variables

Attitudes Toward the Restaurant Participantsrsquo atti-tudes toward the restaurant were assessed through three


7-point scales assessing participantsrsquo opinion anchored atbad-good unfavorable-favorable negative-positive (a frac14 92)

Diagnosticity Following past work we assessed diag-nosticity for each argument by asking participants aboutthe extent to which the information was relevant and im-portant We used two 7-point scales anchored at not at allndashvery much (Aaker 2000 Aaker and Sengupta 2000)Participantsrsquo responses to the two items were strongly cor-related (rfrac14 84 for each argument or higher) Because eachmessage contained both competent and warm argumentsthis allowed us to create measures of individual-item diag-nosticity within participants for both competent and warmarguments Specifically for each argument we first aver-aged across our two measures of diagnosticity (ie rele-vance and importance) Subsequently we averaged acrosssimilar arguments to form a composite score of overalldiagnosticity For example if a participant was exposed totwo competent arguments and one warm argument wewould first compute diagnosticity for each separate item(ie mean of relevance and importance) and then averageacross the two means of the competence item to achieveour composite measure of diagnosticity for competenceand take the single warmth item composite to achieve ourcomposite measure of diagnosticity for warmth

Similarity We asked participants the extent to whichthey felt the message source was similar to them using a7-point scale (not at allndashvery much)


Manipulation Checks A two-way ANOVA on thepower index revealed participants reported feeling signifi-cantly more powerful in the high-power condition(Mfrac14 418 SDfrac14 153) compared to the low-power condi-tion (Mfrac14 315 SDfrac14 119 F (1 97)frac14 1419 pfrac14 001gp

2frac14 12) No other effects emerged (pgt 49) Second atwo-way ANOVA on the item measuring warmth revealedparticipants viewed messages as conveying more warmthin the warmth-skewed message condition (Mfrac14 446SDfrac14 158) than in the competence-skewed condition(Mfrac14 337 SDfrac14 121 F (1 97)frac14 1522 plt 001gp

2frac14 13) No other significant effect or interactionemerged on this item (pgt 27) Third a two-way ANOVAon the item measuring competence revealed participantsviewed messages as conveying more competence in thecompetence-skewed condition (Mfrac14 429 SDfrac14 185) thanin the warmth-skewed condition (Mfrac14 322 SDfrac14 109 F(1 97)frac14 1227 pfrac14 001 gp

2frac14 11) No other significanteffects emerged on this item (pgt 31)

Attitudes Toward the Restaurant A two-way ANOVAon participantsrsquo attitude score revealed a significant audi-ence power message type interaction F (1 97)frac14 1695plt 001 gp

2frac14 15 No main effect of power or messagetype was present (pgt 36) Low-power participants liked

the restaurant more following the warmth-skewed message(Mfrac14 420 SDfrac14 126) compared to the competence-skewed message (Mfrac14 318 SDfrac14 109) F (1 97)frac14 798pfrac14 006 gp

2frac14 07) In contrast high-power participantsliked the restaurant more following the competence-skewed message (Mfrac14 437 SDfrac14 126) compared to thewarmth-skewed message (Mfrac14 344 SDfrac14 114) F (197)frac14 864 pfrac14 004 gp

2frac14 08)

Similarity A two-way ANOVA on similarity revealedno effect of power message type or a power messagetype interaction (F lt 1)

Diagnosticity A repeated ANOVA with power andmessage content as between-subject factors and argumenttype (warmth vs competence) as a repeated factor revealedonly a significant power argument type interaction F (197)frac14 3403 plt 001 gp

2frac14 26 No three-way interactionemerged with message content Post hoc tests revealed thenature of this interaction Low-power audiences ratedwarm arguments as more diagnostic (Mfrac14 370 SDfrac14 143)than high-power audiences (Mfrac14 251 SDfrac14 89) F (197)frac14 2443 plt 001 gp

2frac14 20) but high-power audiencesrated competent arguments as more diagnostic (Mfrac14 356SDfrac14 151) than low-power audiences (Mfrac14 265SDfrac14 80) F (1 97)frac14 1455 plt 001 gp

2frac14 13)These effects add support to our account that audience

power induces differences in diagnosticity of warmth ver-sus competence information and ultimately affects whattypes of messages people are drawn toward Of note wealso examined whether individual-item measures of diag-nosticity mediated audience attitudes in this experimentAs shown in the previous analyses power affects how di-agnostic the arguments were Furthermore diagnosticitywas correlated with attitudes (rfrac14 217 pfrac14 029)However mediational analyses did not find evidence of asignificant indirect effect We suspect this lack of statisti-cal mediation may have occurred because our items weredesigned to assess the individual diagnosticity of each itemas opposed to participantsrsquo gestalt diagnosticity of the mes-sage It may very well be that participantsrsquo overall weight-ing is based on a gestalt diagnosticitymdashwhich our items donot capturemdashas opposed to individual-item diagnosticity

Mediation Through Power Given that our diagnostic-ity items were not sensitive enough to capture mediationone might raise the concern that the findings of this experi-ment operate through a mechanism other than audiencepower To address this potential concern we testedwhether the effect of audience power on attitudes towardthe restaurant was mediated by participantsrsquo feelings ofpower (ie our manipulation check) and whether these ef-fects were moderated by whether the message content em-phasized warmth versus competence (Hayes 2013 model14) We used the linear code low powerfrac141 and highpowerfrac14 1 In addition the warmth-skewed message


condition was coded as 1 and the competence-skewedmessage condition as 1

First replicating the ANOVA results we found that au-dience power predicted attitudes toward the restaurant(bfrac14 52 t(99)frac14 379 plt 001) Next a comprehensive re-gression predicting attitudes toward the restaurant from au-dience power participantsrsquo feelings of power and messagecontent (warmth vs competence) as a moderating variablerevealed a main effect of message content (bfrac1476t(96)frac14 222 pfrac14 03) and a significant message con-tent audience power interaction (bfrac14 20 t(96)frac14 230pfrac14 03) Neither participantsrsquo feelings of power (bfrac14 01t(96)frac14 11 pfrac14 91) nor audience power (bfrac14 09t(96)frac14 69 pfrac14 48) were significant factors Crucially theindex of moderated mediation around participantsrsquo feelingsof power did not contain zero (95 CI 02ndash51 Hayes2013)

GENERAL DISCUSSION

Contributions to the Power Persuasion andSocial Judgment Literatures

We believe this research makes several contributionsFirst for the literature on power this work represents aneffort to study communicator power above and beyondsource effects In the study of source effects the audienceis aware of the communicatorrsquos power and it is this ex-plicit awareness that appears to increase the audiencersquospersuasion We demonstrate that psychological states ofpower independent of the awareness of the power of thesource shape the message a communicator self-generatesor selects Moreover we show the efficacy of such argu-ments can depend on the audiencersquos power

Conceptually this work also extends the literature onpower and persuasion Lammers et al (2013) manipulatedintervieweesrsquo power and found that in mock interviewseven when unaware of intervieweesrsquo power interviewerspreferred high-power interviewees over low-power inter-viewees Our work complements this finding and suggestsone potential reason for this relationship interviewersmight have preferred high-power interviewees becausethey themselves felt high power (ie they had power overthe intervieweersquos outcome) and thus focused on high-power intervieweesrsquo competence However the currentfindings also qualify prior work by suggesting that high-power communicators might not always be more persua-sive such as when audiences are psychologically in a stateof powerlessness

In addition this research reveals the complex and dy-namic nature of psychological power Whereas much pastresearch on psychological states of power has examinedthe effects of power on a single individual this work high-lights the importance of considering how psychologicalstates of power can influence social dynamics between

individuals We have studied the implications of these ef-fects for persuasion but power researchers could examinethe interplay of different power states across a host ofdomains (Ronay et al 2012)

Finally for the literature on social judgments this workoffers a new psychological antecedent to peoplersquos focus onwarmth and competence lacking or having power respec-tively This is consistent with past findings that tradition-ally deprived groups such as the elderly or mentallydisabled are perceived as more warm than competentwhereas the opposite holds for favored groups such aswealthy individuals (Cuddy et al 2008) However novel tothe present work is the finding that momentary psychologi-cal states such as power can make people more or lesssensitive to warmth and competence

Limitations Future Research and PracticalImplications

One direction for future research is to better understandthe extent to which if at all different cultures moderateour findings According to the agentic-communal perspec-tive of power (Rucker et al 2012) a sense of power en-hances agency and can reinforce an individualrsquos chronicorientation In individualistic contexts a sense of power in-creases onersquos self-focus and action orientation In collectiv-ist contexts however increases in power might foster asense of responsibility and benevolence (Chen et al 2001Torelli and Shavitt 2011) In light of research suggestingthat competence is associated with self-profitable motiveswhereas warmth is associated with other-profitable motives(Peeters 2002) we hypothesize that cultural orientationcould moderate if not reverse the present effects That iswithin a collectivist culture a sense of power might in-crease the use and reliance on warmth information notcompetence information

A limitation of the present research is that we did nottest the role of power in changing the accessibility ofwarmth and competence information but instead focusedon diagnosticity and the extent to which individuals placegreater weight on warm and competent arguments Basedon the accessibility-diagnosticity framework (Feldman andLynch 1988) power might also affect accessibility in addi-tion to diagnosticity As such factors hindering informa-tion accessibility may moderate the findings we observedThis question represents a potential boundary condition forfuture exploration

We did not test how making the communicatorrsquos andorthe audiencersquos power salient alters persuasion When acommunicator is known to have high power audiencesmay expect and desire competence in their statements re-gardless of their own state of power That is knowledgeabout the communicatorrsquos power could create expectancies(ie expressions of competence for high power expres-sions of warmth for low power Rucker Hu and Galinsky


2014) which if not met would produce a negative re-sponse Alternatively when the sourcersquos power is knownit may simply serve as a heuristic cue to persuasion Whenhierarchies are known and established the mere associa-tion with the role may dictate persuasion regardless of themessage content (ie employees have to listen to bosses)Such outcomes may be particularly likely in situationswhere individuals tune out from the message argumentsand simply focus on source power as a heuristic (Pettyet al 1981)

Conversely when the audiencersquos power is known com-municators might tune their message according to thepower of the audience One possibility is that low-powerindividuals who possess greater perspective-taking abilities(Galinsky et al 2006) will be better able to strategicallyshift message content toward greater competence Anotherpossibility is that high-power individuals because they seeothers through an instrumental lens might become morestrategic and tune their message according to the powerlevel of their audience (Gruenfeld et al 2008) Of criticalimportance in settings where the hierarchy is known andexplicit to individuals factors other than the warmth andcompetence expressed in the messages may underliepersuasion

Prior work in persuasion has made a distinction betweenthe use of strong and weak arguments (eg Petty andCacioppo 1984) In the present research we either encour-aged participants to generate compelling arguments (ieexperiments 1 and 2) or used relatively strong arguments(ie experiments 3 and 4) If weak (ie specious) argu-ments are used one possibility is that audiences will notview them as diagnostic regardless of whether the argu-ments match or mismatch the audiencersquos focus on compe-tence or warmth This would represent an importantboundary condition of the present effects Of additionalnote the present experiments focused on relatively high-ef-fort situations where both communicators and audienceswere encouraged to process the information carefully It ispossible that under circumstances where elaboration oramount of processing is not constrained to be high the in-troduction of weak arguments may backfire when theymatch the type of information viewed as diagnostic Forexample under moderate elaboration conditions matchingmight encourage greater processing of information (egWheeler Petty and Bizer 2005) To the extent that match-ing produces greater information scrutiny it might enhancepersuasion when arguments are strong but reduce persua-sion when they are weak In some circumstances the useof weak arguments may attenuate or reverse the matchingeffects observed in the present article

Finally future research could explore when matchingversus power complementarity (ie mismatching) produ-ces greater persuasion Existing work suggests that hierar-chical differentiation can sometimes produce positiveconsequences on performance At the perceptual level

participants like others more when an individual adopts apower posture opposite to their own and individuals alsotend to adopt an opposite stance of a high- or low-powerother (Tiedens and Fragale 2003) Furthermore grouptasks can produce better outcomes when there is a clearhierarchy as opposed to equality (Ronay et al 2012)Such dominance complementarity is argued to be attrac-tive because it facilitates interactions and performance(Maissen 2009) in dyadic task-oriented relationships(Tiedens Unzueta and Young 2007) The present re-search suggests an intriguing possibility Due to the dif-ferential diagnosticity of warmth and competencemismatched dyads might become more dysfunctionalover time For example when working through a projectit is possible that those with power increasingly empha-size competence and those with low power increasinglyemphasize warmth which would interfere with task per-formance Future research should investigate dominancecomplementarity to better understand when positive ver-sus negative effects of complementarity are present Forexample perhaps in conversational dialogues where bothpositions of power are known to one another dominancecomplementarity may work better than matching butcomplementary may be less effective when power isunknown

CONCLUSION

This research represents the first systematic effort toinvestigate the interplay between the psychological powerof the communicator and of the audience on persuasionWe demonstrate that a kind of ldquopsychological attune-mentrdquo can occur between communicators and audiencesthrough powermdashpeople are more persuasive to individ-uals in a similar power mindset despite neither party be-ing aware of the power level of the other party Thesefindings not only shed light on the effect of power on afundamental consumer activitymdashpersuasionmdashbut also il-lustrate the role of warmth and competence in persuasivecontexts

DATA COLLECTION INFORMATION

The first author supervised data collection for all fourexperiments The first experiment was conducted atHEC Paris during the spring of 2012 The second experi-ment was conducted at Northwestern during the springof 2010 The third experiment was conducted at theINSEAD-Sorbonne Behavioural Lab during the springof 2012 The fourth experiment was conducted at theINSEAD-Sorbonne Behavioural Lab during the fall of2015 The first author was responsible for data analysisand storage with guidance from the second and thirdauthors


APPENDIX

APPENDIX 1 LIST OF COMPETENT ANDWARM ARGUMENTS EXPERIMENT 3

(TRANSLATED FROM FRENCH)

List of warm arguments

1 The chefrsquos friendly reputation makes this restau-rant very inviting

2 The dining roomrsquos interior design provides a verycozy feel to the place that makes customers feel athome

3 The wait staffrsquos friendly manners create a very in-viting atmosphere in the restaurant

4 The chef regularly comes out to the dining room togreet and thank the customers

5 You will find very amiable sommeliers able to as-sist you in your choice of wine

6 The wallsrsquo covering with warm colors gives a veryfriendly feel to the restaurant

7 The inviting smells coming out of the kitchen willcertainly remind you of your own home

8 You will find polite waiters willing to make youfeel at ease in this restaurant

9 The waiters are very trustworthy in providing justthe right meal

List of competent arguments

10 The chefrsquos cuisine skills have been recognized andsingled out in a specialized food magazine

11 The waiters follow a specific training that makesthem able to give customers full information aboutthe food served as well as where it comes from

12 You will find capable sommeliers able to assist youin your choice of wine

13 The chefrsquos expertise ranges from French toJapanese and to Brazilian culinary techniques

14 The chef has built a reputation based on his skillsand competence

15 The wait staffrsquos skills and training make them ex-perts at their jobs

16 The intelligent design of the place contributes todeliver a highly competent service by limiting boththe kitchen and the wait staffrsquos walking

17 The inviting smells coming from the kitchen testifyto the chefrsquos extended skills

18 The capabilities of this restaurant from waiters tocooks and interior design will meet your expectations

REFERENCES

Aaker Jennifer (2000) ldquoAccessibility or DiagnosticityDisentangling the Influence of Culture on PersuasionProcesses and Attitudesrdquo Journal of Consumer Research 26(March) 340ndash57

Aaker Jennifer and Jaideep Sengupta (2000) ldquoAveraging versusAttenuation The Role of Culture in the Resolution of

Information Incongruityrdquo Journal of Consumer Psychology9 (2) 67ndash82

Anderson Cameron and Jennifer L Berdahl (2002) ldquoTheExperience of Power Examining the Effects of Power onApproach and Inhibition Tendenciesrdquo Journal of Personalityand Social Psychology 83 (6) 1362ndash77

Anderson Cameron and Adam D Galinsky (2006) ldquoPowerOptimism and Risk-Takingrdquo European Journal of SocialPsychology 36 (4) 511ndash36

Asch Solomon E (1946) ldquoForming Impressions of PersonalityrdquoJournal of Abnormal and Social Psychology 41 (3)258ndash90

Bakan David (1966) The Duality of Human Existence Isolationand Communion in Western Man Chicago Rand McNally

Bohner Gerd and Norbert Schwarz (1993) ldquoMood StatesInfluence the Production of Persuasive ArgumentsrdquoCommunication Research 20 (5) 696ndash722

Brinol Pablo Richard E Petty Carmen Valle Derek D Ruckerand Alberto Becerra (2007) ldquoThe Effects of MessageRecipientsrsquo Power Before and After Persuasion A Self-Validation Analysisrdquo Journal of Personality and SocialPsychology 93 (6) 1040ndash53

Chen Serena Annette Y Lee-Chai and John A Bargh (2001)ldquoRelationship Orientation as a Moderator of the Effects ofSocial Powerrdquo Journal of Personality and SocialPsychology 80 173ndash87

Cialdini Robert B (2009) Influence Science and PracticeBoston Pearson Education

Cuddy Amy J C Susan T Fiske and Peter Glick (2004) ldquoWhenProfessionals Become Mothers Warmth Doesnrsquot Cut theIcerdquo Journal of Social Issues 60 (4) 701ndash18

mdashmdashmdash (2008) ldquoWarmth and Competence as UniversalDimensions of Social Perception The StereotypeContent Model and the BIAS Maprdquo in Advances inExperimental Social Psychology New York AcademicPress 61ndash149

Dubois David Derek D Rucker and Adam D Galinsky (2012)ldquoSuper Size Me Product Size as a Signal of Statusrdquo Journalof Consumer Research 38 (6) 1047ndash62

mdashmdashmdash (2015) ldquoSocial Class Power and Selfishness When andWhy Upper and Lower Class Individuals BehaveUnethicallyrdquo Journal of Personality and Social Psychology108 (3) 436ndash49

Feldman Jack and John G Lynch Jr (1988) ldquoSelf-GeneratedValidity and Other Effects of Measurement on BeliefAttitude Intention and Behaviorrdquo Journal of AppliedPsychology 73 421ndash35

Fiske Susan T Amy J C Cuddy and Peter Glick (2007)ldquoUniversal Dimensions of Social Cognition Warmth andCompetencerdquo Trends in Cognitive Sciences 11 (2) 77ndash83

Forgas Joseph P (2006) Affect Cognition and Social BehaviourNew York Psychology Press

Galinsky Adam D Deborah H Gruenfeld and Joe C Magee(2003) ldquoFrom Power to Actionrdquo Journal of Personality andSocial Psychology 85 (3) 453ndash66

Galinsky Adam D Joe C Magee Deborah H GruenfeldJennifer A Whitson and Katie A Liljenquist (2008) ldquoPowerReduces the Press of the Situation Implications forCreativity Conformity and Dissonancerdquo Journal ofPersonality and Social Psychology 95 (6) 1450ndash66

Galinsky Adam D Joe C Magee M Ena Inesi and Deborah HGruenfeld (2006) ldquoPower and Perspectives Not TakenrdquoPsychological Science 17 (12) 1068ndash74


Galinsky Adam D Derek D Rucker and Joe C Magee(2013) ldquoPower Past Findings Present Considerationsand Future Directionsrdquo in APA Handbook of Personalityand Social Psychology vol 1 Attitudes and SocialCognition Washington DC American PsychologicalAssociation

Gruenfeld Deborah H M E Inesi Joe C Magee and Adam DGalinsky (2008) ldquoPower and the Objectification of SocialTargetsrdquo Journal of Personality and Social Psychology 95(1) 111ndash27

Haugtvedt Curtis P and Duane T Wegener (1994) ldquoMessageOrder Effects in Persuasion An Attitude StrengthPerspectiverdquo Journal of Consumer Research 21 205ndash18

Hayes Andrew F (2013) Introduction to Mediation Moderationand Conditional Process Analysis New York GuilfordPress

Judd Charles M Laurie James-Hawkins Vincent Yzerbytand Yoshihisa Kashima (2005) ldquoFundamentalDimensions of Social Judgment Understanding theRelations Between Judgments of Competence andWarmthrdquo Journal of Personality and Social Psychology89 (6) 899ndash913

Keltner Dacher Deborah H Gruenfeld and Cameron Anderson(2003) ldquoPower Approach and Inhibitionrdquo PsychologicalReview 110 (2) 265ndash84

Kipnis David (1972) ldquoDoes Power Corruptrdquo Journal ofPersonality and Social Psychology 24 (1) 33ndash41

Kraus Michael W Paul K Piff and Dacher Keltner (2009)ldquoSocial Class Sense of Control and Social ExplanationrdquoJournal of Personality and Social Psychology 97 (6)992ndash1004

Labroo Aparna A and Derek D Rucker (2010) ldquoTheOrientation-Matching Hypothesis An Emotion-SpecificityApproach to Affect Regulationrdquo Journal of MarketingResearch 47 (5) 955ndash66

Lammers Joris David Dubois Derek D Rucker and Adam DGalinsky (2013) ldquoPower Gets the Job Priming PowerImproves Interview Outcomesrdquo Journal of ExperimentalSocial Psychology 49 (4) 776ndash79

Magee Joe C and Adam D Galinsky (2008) Social HierarchyThe Self-Reinforcing Nature of Power and Status ed J PWalsh and A P Brief London Taylor amp Francis

Maissen Lily Benjamin (2009) ldquoSuperior and Subordinate StyleFit Dominance Complementarity as a Determinant ofSubordinate Successrdquo unpublished dissertation ColumbiaUniversity New York NY

Peeters Guido (2002) ldquoFrom Good and Bad to Can and MustSubjective Necessity of Acts Associated with Positively andNegatively Valued Stimulirdquo European Journal of SocialPsychology 32 (1) 125ndash36

Petrocelli John V Zakary L Tormala and Derek D Rucker(2007) ldquoUnpacking Attitude Certainty Attitude Clarity andAttitude Correctnessrdquo Journal of Personality and SocialPsychology 92 (1) 30ndash41

Petty Richard E and John T Cacioppo (1984) ldquoThe effects ofinvolvement on response to argument quantity and qualityCentral and peripheral routes to persuasionrdquo Journal ofPersonality and Social Psychology 46 69ndash81

Petty Richard E John T Cacioppo and Rachel Goldman (1981)ldquoPersonal Involvement as a Determinant of Argument-BasedPersuasionrdquo Journal of Personality and Social Psychology41 (5) 847ndash55

Pratkanis Anthony R and Melissa D Gliner (2004) ldquoAnd When Shalla Little Child Lead Them Evidence for an Altercasting Theoryof Source Credibilityrdquo Current Psychology 23 (4) 279ndash304

Ronay Richard Katharine Greenaway Eric M Anicich andAdam D Galinsky (2012) ldquoThe Path to Glory Is Paved withHierarchy When Hierarchical Differentiation IncreasesGroup Effectivenessrdquo Psychological Science 23 669ndash77

Rucker Derek D David Dubois and Adam D Galinsky (2011)ldquoGenerous Paupers and Stingy Princes Power DrivesConsumer Spending on Self versus Othersrdquo Journal ofConsumer Research 37 (6) 1015ndash29

Rucker Derek D and Adam D Galinsky (2008) ldquoDesire toAcquire Powerlessness and Compensatory ConsumptionrdquoJournal of Consumer Research 35 (2) 257ndash67

mdashmdashmdash (2015) ldquoPower and Consumer Behaviorrdquo in CambridgeHandbook of Consumer Psychology ed Michael I NortonDerek D Rucker and Cait Lamberton Cambridge MACambridge Press

mdashmdashmdash (2016) ldquoAn Agentic-Communal Model of PowerImplications for Consumer Behaviorrdquo Current Opinions inPsychology 10 1ndash5

Rucker Derek D Adam D Galinsky and David Dubois (2012)ldquoPower and Consumer Behavior How Power Shapes Whoand What Consumers Valuerdquo Journal of ConsumerPsychology 22 (3) 352ndash68

Rucker Derek D Miao Hu and Adam D Galinsky (2014) ldquoTheExperience versus Expectations of Power A Recipe forAltering Powerrsquos Effects on Behaviorrdquo Journal of ConsumerResearch 41 381ndash96

Sachs Jeffrey (2005) The End of Poverty Economic Possibilitiesfor Our Time New York Penguin Press

Schwarz Norbert Herbert Bless and Gerd Bohner (1991) ldquoMoodand Persuasion Affective States Influence the Processing ofPersuasive Communicationsrdquo Advances in ExperimentalSocial Psychology 24 161ndash99

Slabu Letitia Ana Guinote and David Wilkinson (2013) ldquoHowQuickly Can You Detect It Power Facilitates AttentionalOrientingrdquo Social Psychology 44 (1) 37ndash41

Smith Pamela K and Adam D Galinsky (2010) ldquoThe noncon-scious nature of power Cues and consequencesrdquo 4 918ndash38

Smith Pamela K and Yaacov Trope (2006) ldquoYou Focus on theForest When Yoursquore in Charge of the Trees Power Primingand Abstract Information Processingrdquo Journal of Personalityand Social Psychology 90 (4) 578ndash96

Stephens Nicole M Hazel R Markus and Sarah S M Townsend(2007) ldquoChoice as an Act of Meaning The Case of SocialClassrdquo Journal of Personality and Social Psychology 93 (5)814ndash30

Tiedens Larissa Z and Alison R Fragale (2003) ldquoPower MovesComplementarity in Dominant and Submissive NonverbalBehaviorrdquo Journal of Personality and Social Psychology 84(3) 558ndash68

Tiedens Larissa Z Miguel M Unzueta and Maia J Young (2007)ldquoAn Unconscious Desire for Hierarchy The MotivatedPerception of Dominance Complementarity in TaskPartnersrdquo Journal of Personality and Social Psychology 93(3) 402ndash14

Torelli Carlos J and Sharon Shavitt (2011) ldquoThe Impact ofPower on Information Processing Depends on CulturalOrientationrdquo Journal of Experimental Social Psychology47 959ndash67

Wheeler S Christian Richard E Petty and Georges YBizer (2005) ldquoSelf-schema matching and attitude


change Situational and dispositional determinants ofmessage elaborationrdquo Journal of Consumer Research31 787ndash97

Wilson Elizabeth J and Daniel L Sherrell (1993) ldquoSourceEffects in Communication and Persuasion Research A Meta-

Analysis of Effect Sizerdquo Journal of the Academy ofMarketing Science 21 (2) 101ndash12

Ybarra Oscar Emily Chan and Denise Park (2001) ldquoYoung andOld Adultsrsquo Concerns About Morality and CompetencerdquoMotivation and Emotion 25 (2) 85ndash100


Copyright of Journal of Consumer Research is the property of Oxford University Press USAand its content may not be copied or emailed to multiple sites or posted to a listserv withoutthe copyright holders express written permission However users may print download oremail articles for individual use

versus the powerless are more persuasive We proposethat having or lacking power may change the diagnostic-ity of information (Feldman and Lynch 1988) and thusthe extent to which information is used or relied uponWe argue that psychological states of high power leadcommunicators to favor the use of competence-related ar-guments and audiences to rely more heavily on compe-tence information when forming their attitudes Incontrast psychological states of low power lead commu-nicators to favor the use of warmth-related argu-ments and audiences to rely more heavily on warmthinformation in forming their attitudes As a consequenceof these independent effects the success of a persuasivemessage can hinge on the interaction between communi-cator and audience power High-power communicatorsgenerate messages that persuade high-power audienceslow-power communicators generate messages that per-suade low-power audiences

Before developing our hypotheses further we elaborateon the construct of power

POWER AND ITS EFFECTS ONCOMMUNICATORS AND AUDIENCES

Power is defined as asymmetric control over valued re-sources in social relationships (Magee and Galinsky2008 Rucker Galinsky and Dubois 2012) Researchershave studied power as a psychological state or mindsetthat can arise from both structural differences insocioeconomic status (Dubois Rucker and Galinsky2015 Kraus Piff and Keltner 2009 Stephens Markusand Townsend 2007) as well as situational factors such asonersquos social role (eg boss vs employee Anderson andGalinsky 2006) The last decade has witnessed an acceler-ation of research efforts to unpack powerrsquos far-reachingeffects on consumer behavior (Dubois Rucker andGalinsky 2012 Rucker et al 2012) However to datescant evidence exists as to how power systematically af-fects the creation and reception of persuasive messagesThe little work that does exist can be broken into researchfocused on understanding the effects of power on commu-nicators (ie those delivering a message) or on audiences(ie those receiving a message)

With regard to communicators having power in-creases communicatorsrsquo attempts to influence others(Kipnis 1972) and to express onersquos own opinion ingroups (Anderson and Berdahl 2002 Galinsky et al2008) As noted Lammers et al (2013) found that indi-viduals primed with high power were more persuasivethan those primed with low power Furthermore a com-municatorrsquos power can enhance persuasion by serving asa simple heuristic (Petty Cacioppo and Goldman 1981)and thus induce greater compliance in an audience(Cialdini 2009)

With regard to audience power Brinol and colleagues(2007) found that an audiencersquos psychological power af-fected their response to persuasion in two distinct waysFirst increasing an audiencersquos power prior to receiving amessage led to less scrutiny of the message and therefore aweaker discrimination between weak and strong argu-ments Second when the audiencersquos power was increasedafter a message they became more confident in their mes-sage-related thoughts As a consequence powerful individ-uals were more likely to use those thoughts in forming anopinion on the topic which led to more positive attitudesafter strong arguments but more negative attitudes afterweak arguments Although Brinol et al (2007) examinedaudience power they did not manipulate communicatorpower

Beyond these few examples the relationship betweenpower and persuasion remains relatively unexplored Theknowledge accumulated so far suggests a simple main ef-fect high-power communicators are more persuasive thanlow-power communicators In contrast to this idea we in-troduce the notion that power can fundamentally shape thevalue or diagnosticity of information and thus the type ofinformation communicators convey and that audiencesgive weight to Consequently high- or low-power commu-nicators may be more or less persuasive depending onwhether they are communicating to an audience in a stateof high or low power

AN AGENTIC-COMMUNAL MODEL OFPOWER IMPLICATIONS FOR WARMTHVERSUS COMPETENCE INFORMATION

Agency Communion and Power

Bakan (1966) introduced the ideas of agency and com-munion to reflect two fundamental modalities of humanthought and behavior Agency is marked by a focus on theself and produces consequences such as independence andpersonal striving In contrast communion focuses peopleon others and enhances individualsrsquo attention to gettingalong with others and othersrsquo needs Rucker et al (2012also Rucker Dubois and Galinsky 2011 Rucker andGalinsky 2015 2016) proposed an agentic-communalmodel of power by which power affects peoplersquos orienta-tion toward the world in an agentic or communal fashionhigh power fosters an agentic orientation and low powerfosters a communal orientation Because the powerful areless dependent on others they can pursue their own goalsand interests with fewer constraints (ie they are imbuedwith agency) In contrast the dependency of the powerlessrequires them to attend to and incorporate others to achievetheir goals (ie they require communion)

One aspect of agentic and communal orientations is thatthey appear to relate to competence and warmth informa-tion respectively (Asch 1946 Cuddy Fiske and Glick









FIGURE 1




FORMAL HYPOTHESES




audiences



audiences















OVERVIEW





POWER




Procedure








Dependent Variables





















Procedure











Dependent Variables





















TABLE 1










FIGURE 2


300

350

400

450

500

550

600

Baseline High Low

Audience Power


FIGURE 3



Attitudes

52

-44

37

-01NS

= p lt 05 = p lt 01

Message Warmth

Message Competence



-50 39

CI = [15 65]

CI = [17 76]

52




COMPETENCE








MAIN EXPERIMENT





Procedure


























Dependent Variables


TABLE 2











2frac14 00


















2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES







































































FIGURE 1




FORMAL HYPOTHESES




audiences



audiences















OVERVIEW





POWER




Procedure








Dependent Variables





















Procedure











Dependent Variables





















TABLE 1










FIGURE 2


300

350

400

450

500

550

600

Baseline High Low

Audience Power


FIGURE 3



Attitudes

52

-44

37

-01NS

= p lt 05 = p lt 01

Message Warmth

Message Competence



-50 39

CI = [15 65]

CI = [17 76]

52




COMPETENCE








MAIN EXPERIMENT





Procedure


























Dependent Variables


TABLE 2











2frac14 00


















2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES

































































FORMAL HYPOTHESES




audiences



audiences















OVERVIEW





POWER




Procedure








Dependent Variables





















Procedure











Dependent Variables





















TABLE 1










FIGURE 2


300

350

400

450

500

550

600

Baseline High Low

Audience Power


FIGURE 3



Attitudes

52

-44

37

-01NS

= p lt 05 = p lt 01

Message Warmth

Message Competence



-50 39

CI = [15 65]

CI = [17 76]

52




COMPETENCE








MAIN EXPERIMENT





Procedure


























Dependent Variables


TABLE 2











2frac14 00


















2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES





































































Dependent Variables





















Procedure











Dependent Variables





















TABLE 1










FIGURE 2


300

350

400

450

500

550

600

Baseline High Low

Audience Power


FIGURE 3



Attitudes

52

-44

37

-01NS

= p lt 05 = p lt 01

Message Warmth

Message Competence



-50 39

CI = [15 65]

CI = [17 76]

52




COMPETENCE








MAIN EXPERIMENT





Procedure


























Dependent Variables


TABLE 2











2frac14 00


















2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES










































































Procedure











Dependent Variables





















TABLE 1










FIGURE 2


300

350

400

450

500

550

600

Baseline High Low

Audience Power


FIGURE 3



Attitudes

52

-44

37

-01NS

= p lt 05 = p lt 01

Message Warmth

Message Competence



-50 39

CI = [15 65]

CI = [17 76]

52




COMPETENCE








MAIN EXPERIMENT





Procedure


























Dependent Variables


TABLE 2











2frac14 00


















2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES


















































































TABLE 1










FIGURE 2


300

350

400

450

500

550

600

Baseline High Low

Audience Power


FIGURE 3



Attitudes

52

-44

37

-01NS

= p lt 05 = p lt 01

Message Warmth

Message Competence



-50 39

CI = [15 65]

CI = [17 76]

52




COMPETENCE








MAIN EXPERIMENT





Procedure


























Dependent Variables


TABLE 2











2frac14 00


















2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES





































































FIGURE 2


300

350

400

450

500

550

600

Baseline High Low

Audience Power


FIGURE 3



Attitudes

52

-44

37

-01NS

= p lt 05 = p lt 01

Message Warmth

Message Competence



-50 39

CI = [15 65]

CI = [17 76]

52




COMPETENCE








MAIN EXPERIMENT





Procedure


























Dependent Variables


TABLE 2











2frac14 00


















2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES


































































COMPETENCE








MAIN EXPERIMENT





Procedure


























Dependent Variables


TABLE 2











2frac14 00


















2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES

































































Procedure


























Dependent Variables


TABLE 2











2frac14 00


















2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES














































































2frac14 09)










Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES






































































Procedure











restaurant






their jobs


Dependent Variables















2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES












































































2frac14 08)











GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES


































































GENERAL DISCUSSION

















CONCLUSION





APPENDIX























REFERENCES





































































CONCLUSION





APPENDIX























REFERENCES
































































APPENDIX























REFERENCES
















































































































Documents

Dynamics of Communicator and Audience Power: The