Overcoming the Self-Promotion Dilemma: Interpersonal Attraction and Extra Help as a Consequence of Who Sings One's Praises

Abstract

Prior research has shown that positive information presented by a third party shields people from the negative consequences of being perceived as self-promoting. But in many contexts, those third parties are intermediaries with a financial interest in the person being promoted rather than neutral parties. In three experimental studies, the authors demonstrate that even when intermediaries are not neutral, they can be helpful for overcoming the self-promotion dilemma--the need to assert one's competence but not be harmed by the fact that people who self-promote are viewed negatively. The authors find that hiring an agent to sing one's praises results in more favorable perceptions of the client, which contributes, in turn, to a greater willingness to offer that person assistance. It is also shown that even when the intermediary is physically present and seen to be complicit with the client, the positive effects of having someone else speak on one's behalf persist.

Full text

1
Overcoming the Self-Promotion Dilemma:
Interpersonal Attraction and Extra Help as
a Consequence of Who Sings One’s Praises
Jeffrey Pfeffer
Stanford University
Christina T. Fong
University of Washington
Robert B. Cialdini
Arizona State University
Rebecca R. Portnoy
University of Washington
(Wosinska, Dabul, Whetsone-Dion, & Cialdini, 1996).
Because being liked is important to many people in and
of itself, and because likeability is a useful method of
interpersonal influence (e.g., Cialdini, 2001), there
are pressures for people to be modest and not self-
aggrandize in their interactions with others.
On the other hand, there are many situations in
which people need to make claims of competence on
their own behalf or on behalf of their work or output to
prospective employers, clients, or purchasers of their
goods or services. One’s positive characteristics may be
uncertain to a prospective client or employer, and the
very structure of the interaction compels the individual
to make claims of competence or quality. In fact, there
is evidence that not to make positive assertions about
oneself or one’s work can be taken as a negative signal
(e.g., Kenrick, Neuberg, & Cialdini, 2002, p. 129).
One possible way out of this dilemma is to have
another person present the desirable information on
one’s behalf (Jones & Wortman, 1973). If it is self-
promotion and boastfulness that create interpersonal
Authors’ Note: We wish to thank Sanford Devoe, Agnieszka Flizik,
Kevin Steensma, and Lara Tiedens for helpful comments and assis-
tance with this work. Correspondence should be addressed to Dr.
Jeffrey Pfeffer, Graduate School of Business, Stanford University,
Stanford, CA 94305-5015; e-mail: pfeffer_jeffrey@gsb.stanford.edu.
PSPB, Vol. 32 No. 10, October 2006 1-13
DOI: 10.1177/0146167206290337
© 2006 by the Society for Personality and Social Psychology, Inc.
Prior research has shown that positive information presented by
a third party shields people from the negative consequences of
being perceived as self-promoting. But in many contexts, those
third parties are intermediaries with a financial interest in the
person being promoted rather than neutral parties. In three exper-
imental studies, the authors demonstrate that even when inter-
mediaries are not neutral, they can be helpful for overcoming the
self-promotion dilemma—the need to assert one’s competence but
not be harmed by the fact that people who self-promote are viewed
negatively. The authors find that hiring an agent to sing one’s
praises results in more favorable perceptions of the client, which
contributes, in turn, to a greater willingness to offer that person
assistance. It is also shown that even when the intermediary is
physically present and seen to be complicit with the client, the pos-
itive effects of having someone else speak on one’s behalf persist.
Keywords: agents, self-promotion, person perception
The self-promotion dilemma people often face is clear.
On one hand, research shows that self-promotion is per-
ceived negatively by others (e.g., Wortman & Linsenmeier,
1977). People who are modest about their qualities and
performance are better liked than those who are boast-
ful (Forsyth, Berger, & Mitchell, 1981) because modesty
reduces the likelihood of appearing self-aggrandizing
(Baumeister & Ilko, 1995). Moreover, boastful presen-
ters are always less effective, whereas intermediate mod-
esty results in the most positive evaluations of people

perception problems, then having the claims made by
another may have fewer negative consequences.
Consistent with this argument, Inman, McDonald, and
Ruch (2004) found that people who hear descriptions
about a person in the first person (e.g., “I helped him”)
see that individual as both more boastful and less like-
able than when they hear the same description in the
third person (“She helped him”). Their studies suggest
that intermediaries may help their clients by providing
third-person descriptions rather than first-person
descriptions of positive attributes. The Inman et al.
(2004) findings raise an important issue—whether it is
only apparently neutral third parties who can provide
the benefits of signaling competence and avoiding the
self-promotion dilemma. Although it is the case that
sometimes the third parties that are providing the
descriptions of others may be casual friends or acquain-
tances or apparently neutral observers, there are many
situations in which the third parties have an economic
interest in having the individual they are describing be
perceived positively. For instance, literary agents
(Coser, Kadushin, & Powell, 1982) receive a percentage
of what their clients obtain and therefore have an obvi-
ous interest in having their clients and their work be
perceived as positively as possible. Similarly, executive
recruiters sometimes work on a contingent fee basis so
they only make money if they fill a position, which gives
them an incentive to hype their candidate. And even in
noncontingent searches, the fee is typically a percent-
age of the salary offered, which again makes executive
recruiters third parties with a financial interest in how
well their clients do.
Moreover, it is not just the case that people providing
favorable descriptions of others can have an incentive to
do so. In many instances, the intended targets of these
positive statements know about this agency relationship
and the intertwined economic interests between the
client and the intermediary. So the question arises as to
whether third-party claims on behalf of another provide
the benefits of avoiding the self-promotion dilemma or
signaling competence even when the third party has an
economic interest in the transaction.
BACKGROUND AND HYPOTHESES
Much of the existing literature on the use of paid
intermediaries to conduct transactions proceeds largely
from a principal-agent perspective originating in eco-
nomics (e.g., Kurtzberg, Moore, Valley, & Bazerman,
1999; Mnookin & Cohen, 1999), a point of view that pre-
sumes conflicting interests between agents and those
who hire them. Not surprisingly, then, existing empiri-
cal research generally demonstrates the negative effects
of agents. Levitt and Syverson (2004) argued that
because real estate agents receive only a small propor-
tion of the incremental price when a house sells for
more money, the agents have an incentive to sell houses
for less money but more quickly. The authors found that
agents who sold their own houses kept those houses on
the market longer and obtained a higher price than
when they sold the houses of clients. Bazerman, Neale,
Valley, Zajac, and Kim (1992) simulated real estate nego-
tiation and demonstrated that the selling price of a
property was higher when an agent was used but that
the price increase was insufficient to recoup the cost of
the agent and the likelihood of the negotiation reaching
an impasse increased. Yavas, Miceli, and Sirmans (2001)
and Schotter, Snyder, and Zheng (1995) also reported
that using intermediaries increased the likelihood of
impasse and increased the time to reach an agreement
if one were achieved. Kurtzberg et al. (1999), summa-
rizing the evidence on the effects of the use of interme-
diaries, maintained that “the use of intermediaries in
bargaining relationships has been shown to reduce the
likelihood of transaction, increase the sale price, and
increase the time to agreement. . . . Overwhelmingly,
previous research points to the hazards associated with
the use of agents” (p. 286).
The Interpersonal Benefits of Using an Agent
One might wonder if intermediaries have such bad
effects on negotiating outcomes why they are used so
frequently. One answer might be simply custom, habit,
or precedent, but there are discussions in the literature
that hint at some possible benefits that agents can pro-
vide. Rubin and Sander (1988) noted that agents might
have more knowledge of the subject matter and the
negotiating process, thereby providing expertise (see
also Levitt & Syverson, 2004), and also could have
access to networks of influence not available to those
less experienced. Jonas, Schulz-Hardt, and Frey (2005)
suggested that advisors or intermediaries can provide
information to their clients and thereby reduce the
clients’ lack of knowledge.
Using an intermediary also can make a client appear
more competent. There are several mechanisms that can
potentially produce such an effect. If intermediaries
obtain payment only to the extent that their clients are
successful, then the very ability to obtain representation
can be a signal that others see the client as worthy and
skilled. Second, when agents make claims on behalf of
their clients, these claims will be seen as less self-serving
than similar statements made by the client. Third parties
provide the aura of objectivity, and therefore, intermedi-
ary statements of client competence are likely to be more
credible. Although intermediaries can affect perceptions
2 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

of client competence and these effects can be beneficial,
we were primarily interested in the interpersonal affect
and liking effects of using an intermediary.
Intermediaries may potentially offer more than just
knowledge, contacts, or credible claims of competence.
They can affect the interpersonal climate of the transac-
tion and buffer the parties from the negative conse-
quences of tense and difficult interactions. Rubin and
Sander (1988) argued that intermediaries might offer
the advantage of emotional detachment and could be
used tactically, such as playing the role of “good cop” or
“bad cop.” Others also have theorized that agents may be
useful in decreasing the negative affect associated with
negotiations, therefore creating a more positive atmos-
phere between parties (Coser et al., 1982; Kurtzberg &
Medvec, 1999).
In this article, we extend the focus on interpersonal
benefits of employing an agent by examining the possi-
bility that intermediaries are able to resolve the self-
promotion dilemma. Agents can speak of their clients’
positive attributes on behalf of the clients, thereby pro-
viding the positive information without having the
clients self-promote and incur the resulting interper-
sonal costs. For intermediaries to accomplish this task
in situations where others know that they are acting on
behalf of their clients, it would have to be the case that
even though observers know of the economic relation-
ship between the parties, they are unwilling or unable
to correct completely for this situational reality and its
implications for intermediary behavior.
In fact, there is a long research tradition in social
psychology going back to the classic Jones and Harris
(1967) study that shows that even when people are
given relatively heavy-handed reminders of the effect of
situational factors, they may correct for this informa-
tion but not enough to completely remove the effect on
their judgments. Consequently, we argue that Inman
et al.’s (2004) findings about the benefits of third-party
promotion may hold even when the third party pre-
senting the favorable information has a relational and
economic stake in the information being presented.
Therefore, it is likely that people who have favorable
information presented on their behalf by intermedi-
aries will appear more likeable than those who present
the information for themselves.
The large literature on ingratiation (e.g., Giacalone &
Rosenfeld, 1989; Jones & Wortman, 1973) shows that
likeability and positive person perception confer many
advantages, including benefits such as being hired and
earning a higher salary (Higgins, Judge, & Ferris, 2003).
But, in many situations, there are two outcomes to an
interaction such as a negotiation. First, there is the actual
deal struck, such as the price obtained for a house, the
advance for a book, or the salary for an employee.
Although liking may have an effect on these outcomes,
such effects may be constrained because conscientious
negotiators are expected to suppress the influence of
interpersonal attraction during the negotiating process
(e.g., Shell, 1999, pp. 66-67). Furthermore, as Higgins
and his colleagues (2003) point out, direct measures of
success, such as salaries and promotions, are often
affected by a number of factors that are outside the
realm of interpersonal interaction, so impression man-
agement and influence tactics may not have strong
effects in these domains.
Our research focuses, therefore, on a second out-
come that occurs in most transactions: the sentiments
that result from the transaction process. The social
dynamics between the two parties can influence the
purchaser’s or employer’s willingness to do more than
is required down the road. In general, people can and
do perform favors for others above and beyond what is
specified by the explicit terms of a transaction. A pub-
lisher, for instance, has some discretion on how much
to emphasize and promote a book apart from what is
formally stipulated in the contract and an employer can
make an employee’s life more or less difficult on the
job. This willingness to go above and beyond the formal
requirements is likely to be related to the interpersonal
relationship created in the process of reaching an
agreement. Therefore, we argue that people repre-
sented by intermediaries will obtain more extra benefits
and help than those who represent themselves because
people who are represented by paid agents are seen as
being more likeable.
In the following three studies, we test these hypothe-
ses in a number of ways across two negotiation settings.
Study 1 examines the effects of using an agent in get-
ting a job, whereas Studies 2 and 3 replicate our effects
in a negotiation setting in which an author is seeking an
advance from a publisher. We also explore other poten-
tial mechanisms that may explain the benefits of using
an agent. In Study 1, we examine the separate effects of
an agent on perceptions of competency and explore
differential pathways to voluntary help and financial
benefits. In Studies 2 and 3, we examine two alternative
explanations for the benefits of using an agent: social
proof (Study 2) and the physical absence of the client
(Study 3).
STUDY 1
To test our hypotheses, we chose a situation where
the self-promotion dilemma is quite salient and where
the interpersonal or emotional climate of the interac-
tion is crucial: the job interview. Job candidates often
are faced with the conflicting goals of conveying their
Pfeffer et al. / OVERCOMING THE SELF-PROMOTION DILEMMA 3

competence and desirability while also appearing like-
able and modest, and benefits such as extra help that
result from liking can be crucial in determining future
success. Furthermore, research has demonstrated that
job candidates often employ self-promoting tactics, with
varying degrees of success (Higgins & Judge, 2004), but
the effects of promotion by third-party intermediaries
has yet to be investigated.
Although likeability is certainly important in a job
interview scenario, perceptions of competence can be
of equal, if not greater, importance. As we have already
noted, there are several reasons why representation by
an agent might positively influence perceptions of com-
petence. Because perceptions of competence seem to
be more directly relevant to market value and are essen-
tially assessments of quality of the product being pur-
chased, we would expect that perceptions of competence
should be more strongly related to financial outcomes
than perceptions of likeability. Thus, we propose a sep-
arate causal pathway for the relationship between per-
ceptions of competence and financial outcomes such
that the relationship between use of an agent and sub-
sequent financial benefits is mediated by competence.
Although likeability should mediate the effects of an
agent on others’ willingness to offer additional help,
people who are represented by agents will receive more
financial benefits than those who represent themselves
as a result of increased perceptions of their compe-
tence and value.
Method
Participants. One-hundred and twenty-five students
(65 women, 60 men) in an introduction to manage-
ment course at a major university in the United States
received extra credit for participation in this study—
50% of the participants were between the ages of 21
and 23; 36% were between 18 and 20; and 13% were
older than 23. Almost half of the participants identified
as European American (47%), a third identified them-
selves as Asian American (30%), 2% as Hispanic, 1% as
African American, and the remaining 17% were classi-
fied as Other.
Materials and procedure. Participants were told that the
university administration was asking for their help in
selecting a new director of student life. It was plausible to
ask for student input in this decision because the job
involved interaction with students. Participants were told
that different students would be evaluating transcripts
from interviews with different candidates to fill the posi-
tion. Because we were interested in increasing the
involvement of the students and the psychological real-
ism of our materials, we described the job as an on-cam-
pus position that involved interaction with students. The
job description for the position included duties such
as “advising/counseling students individually through
in-person, phone, and e-mail communications.”
Participants were randomly assigned to either the
agent condition or the no-agent condition. The inter-
view transcripts were virtually identical; however, in the
agent condition, the introductory statement included
an additional passage that read as follows:
The candidate has hired a professional from a recruit-
ing firm to represent him in this interview. In this indus-
try, it is quite common to hire such agents to represent
candidates, and the agent will receive a fee, which is
commensurate with the salary offered to the candidate.
Either having or not having an intermediary could be
considered acceptable and normative in the job-finding
context. According to a study by Coopers and Lybrand
(Kennedy Information Online, 2004), 64% of executive
positions are filled through recruiters. Therefore, either
being represented or not would be considered appropri-
ate and there should be no effect of being perceived as
counternormative. More important, however, this state-
ment makes it clear that the intermediary has a financial
interest in the candidate being perceived positively.
Participants in the no-agent condition read the same
transcript except that the candidate represented him-
self and the wording of the candidate’s responses was
modified from third person for the agent to first person
for the candidate. Because the context is a job inter-
view, the dialogue consisted of questions and answers
about the candidate’s relevant job experience, strengths
and weaknesses, and other background details. Given
that the participants in the no-agent condition read the
candidate’s responses in first person, there was, of
course, a significant tone of self-aggrandizement that
was not present in the agent condition. One example of
the candidate’s response was, “Anyone who has worked
with me would say that I am a natural leader,” whereas
in the agent condition, this read as, “Anyone who has
worked with him would say that he is a natural leader.”
After reading the transcript, participants completed
a packet of questionnaires in which they provided, in
order, a salary recommendation, person perception rat-
ings that included competence and likeability ratings,
and ratings of their willingness to provide extra help.
Dependent Variables
Extra help. To explore whether having an agent pro-
duces more willingness to offer extra help, participants
indicated their willingness on a 10-point scale (ranging
from 1 = very unwilling to 10 = extremely willing) to alert
this candidate about a problem in student life, help
him distribute a survey to other students, and work in
4 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

his office on a project for no pay. These three items
were averaged to create a scale of extra help (α=.74).
Likeability and competence. Using a 10-point scale
(ranging from 1 = not at all to 10 = extremely), the survey
asked participants to rate the candidate and the agent
(if applicable) on a number of traits. A likeability scale
was composed of ratings of the degree to which the job
candidate was perceived as polite, pleasant, likeable,
disagreeable (reverse-coded), sincere, honest, devious
(reverse-coded) and cunning (reverse-coded; α=.59).
It should be noted that the reliability for this measure
is below the conventional standards of what is consid-
ered to be a good scale and this fact could have poten-
tial implications for the interpretations of our results.
However, because the reliability of this same scale is
higher in the next two studies, we were not overly con-
cerned with the low reliability in this one instance. The
measure of competence was composed of ratings of the
candidate’s perceived talent, competence, intelligence,
and ability (α=.82).
Because we collected several measures from a single
respondent at the same time, we wanted to ensure that
we were measuring distinct constructs and capturing
more than common method variance. In addition, it was
particularly important to determine whether our extra
help and likeability constructs were theoretically and
empirically distinct. Therefore, we conducted confirma-
tory factor analyses to compare our model, which had
the three independent constructs of extra help, compe-
tence, and likeability, to a series of other possible models,
including a highly restricted single-factor structure and
two different two-factor models. As Table 1 shows, our
model provided the best fit to the data. Chi-square dif-
ference tests for the nested models were consistently
large and significant, showing that improvements in fit
were gained as we moved from restricted models to the
target model. These results mean that the three scales
that we created do in fact tap separate constructs.
Furthermore, the chi-square test of model fit was 146.14,
df = 87, p < 0.001, with a Comparative Fit Index (CFI)
of .91 and a standardized root mean square residual
(SRMR) of .08. In addition, the ratio of chi-square to
degrees of freedom is 1.96. Good fit is characterized by
CFI values near or greater than 0.95 and a SRMR of less
than .06 (Hu & Bentler, 1999) and value of less than 3 for
the ratio of chi-square to degrees of freedom (Carmines
& McIver, 1981). Taken together, these results indicate
that the model provides an adequate fit to the data.
Financial outcomes. After participants read the inter-
view transcript, they were asked to indicate the salary that
they thought the school should offer the candidate on a
10-point scale that ranged from 1 (less than $63,000) to
10 (more than $85,000). The midpoint (5) was anchored
with the label $74,000, which is the average salary for this
type of position, and the points on the scale increased or
decreased from this number by $4000, or about 5%.
Results and Discussion
Table 2 presents the means and standard deviations of
the relevant dependent variables by condition. Job candi-
dates represented by an intermediary were perceived as
more likeable, t(123) = 6.06, p < .001, d = .95, Confidence
Interval (CI)(diff) = .84-1.66, than those who represented
themselves. Participants also were more willing to offer
extra help to candidates who were represented by an
agent than candidates who represented themselves,
t(123) = 4.40, p < .001, d = .74, CI(diff) = .71-1.87.
Candidates in the agent condition also were perceived
as more competent than those in the no agent condi-
tion, t(123) = 4.92, p <.001, d = .81, CI(diff) = .58-1.36,
Pfeffer et al. / OVERCOMING THE SELF-PROMOTION DILEMMA 5
TABLE 1: Nested Models Assessing Discriminant Validity
Model CFI ∆CFI χ
2
∆χ
2
1. Three-factor model: “likeability,” “competence,” and “extra help” .91 146.14
2. Single-factor model .77 245.79
Difference between models 2 and 1 .14 99.65***
3. Two-factor model combining “likeability” and “extra help” .85 193.46
Difference between models 3 and 1 .06 47.32***
4. Two-factor model combining “likeability” and “competence” .84 199.47
Difference between models 4 and 1 .07 53.33***
NOTE: CFI = Comparative Fit Index.
***p < .001.
TABLE 2: Means (Standard Deviations) From Study 1
Dependent Variable No Agent (n = 61) Agent (n = 63)
Salary (financial outcomes) 5.20
a
(1.39) 5.79
b
(1.27)
Likeability scale 5.73
a
(1.19) 6.98
b
(1.12)
Competence scale 7.55
a
(1.24) 8.52
b
(.95)
Extra help scale 4.99
a
(1.66) 6.29
b
(1.62)
NOTE: Within each row, means with different subscripts are signifi-
cantly different at p < .05.

and candidates who were represented by an agent
received higher salary recommendations, t(122) = 2.50,
p < .01, d = .46, CI(diff) = .124-1.07, than candidates who
represented themselves.
Mediational analysis. To test the mediating effects of
increased likeability and competence, we conducted a
series of mediational analyses as directed by Baron and
Kenny (1986). Baron and Kenny designate full media-
tion as occurring when the effect of the independent
variable is no longer significant when entered with the
mediating variable, whereas the effect of the mediating
variable remains a significant predictor. Full mediation
indicates that the mediating variable has explained all of
the variance initially associated with the independent
variable. Partial mediation occurs when the predictive
power of the mediating variable remains high while the
predictive power of the independent variable is dimin-
ished but remains statistically significant. In the case of
partial mediation, other factors also may be influencing
the relationship between the independent and depen-
dent variable. Because we believed that other factors
(such as market forces, individual differences, or other
situational factors) may influence the effect of agents on
obtaining extra help or salary, we expected that percep-
tions of likeability and competence would be partial
mediators of the relationship between being repre-
sented by an intermediary and our dependent variables.
In all of our mediational analyses, the independent
variable was the agent condition (coded as 0 if the can-
didate represented himself or 1 if the candidate was rep-
resented by an agent). The dependent variable was
either the extra help scale or the salary item, and we
tested both competence and interpersonal likeability as
potential mediators. As expected, condition (agent or
no-agent) was a significant predictor of both of the
dependent variables of interest: extra help (β=.37, p <
.001) and salary (β=.22, p < .05). Condition also signif-
icantly predicted both likeability (β=.48, p < .001) and
competence (β=.42, p < .001), and likeability (β=.48,
p < .001) and competence (β=.42, p < .001) signifi-
cantly predicted extra help. However, only competence
was a significant predictor of salary (β=.33, p < .001).
Perceptions of likeability partially mediated the
effects of having an agent on offering extra help
because the effect of condition on extra help was
reduced (β=.18, p < .05) when simultaneously entered
with the likeability score (β=.40, p < .001). A Sobel
(1982) test indicated that the association between the
independent variable and the dependent variable was
significantly reduced when the mediating variable was
included (Z = 4.31, p <.001). In addition, competence
partially mediated the effects of having an agent on
being willing to offer extra help; the effect of condition
on extra help was reduced (β=.24, p < .05) when
entered simultaneously with competence (β=.32, p <
.001) and the Sobel test indicated that this mediation
was statistically significant (Z = 3.56, p < .001).
Perceptions of competence fully mediated the
effects of having an agent on salary because the effects
of condition on salary became nonsignificant (β=.11,
ns) when entered simultaneously with competence (
β=
.28, p < .01), and the Sobel test indicated that these
mediated effects were significant (Z = 3.02, p < .01).
These findings imply that, contrary to our initial expec-
tation of partial mediation, perceptions of competence
fully explained the variance associated with agent rep-
resentation in affecting salary. In addition, perceptions
of likeability did not mediate the effects of having an
agent on salary. Figure 1 depicts all of the significant
mediational pathways described.
These results show that using an agent leads to
increased likeability, even though the intermediary is
not a neutral or disinterested third party. Furthermore,
this increased likeability contributes to more willing-
ness to help the job candidate in the future. In addi-
tion, an agent can make a client appear more competent
even when saying the identical things that the candi-
date might say, and this increased perceived compe-
tence leads to a higher recommended salary for the job
candidate.
STUDY 2
In Study 2, our primary goal was to see if the inter-
personal effects of having an agent would generalize
across a different experimental context. We chose the
domain of book publishing because (a) agents are
sometimes, but not invariably, used in placing books so
6 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
Agent
vs.
No Agent
Extra Help
Likeability
SalaryCompetence
Figure 1 Summary of mediational analyses for Study 1.

both having an agent and representing oneself would
both be plausible and normative; (b) the scenario situ-
ation in book publishing did not require a lot of spe-
cialized technical knowledge that participants would
be unlikely to possess; and (c) both the economic terms
of the deal and also the quality of the relationship
between the parties are important outcomes.
For this study, we also wanted to examine the possi-
bility that intermediaries may be useful not just because
they provide third-person advocacy but also because
they provide social proof of their client’s likeability and
competence. To begin to explore this possibility, we var-
ied the number of agents to see if having two rather
than one affected the results. Previous research has
demonstrated that as the number of individuals who
share the same opinion grows, so does the perception
of social proof (Milgram, Bickman, & Berkowitz, 1969).
Finally, although the effects of having an intermedi-
ary on perceptions of competence and through com-
petence on salary revealed in Study 1 were theoretically
interesting, our primary focus is on the interpersonal
and social effects of having a third party sing one’s
praises. Therefore, we chose an experimental context
and constructed a scenario where we could establish
the competence of our author and standardize percep-
tions of success of the individual across conditions to,
in effect, hold competence constant. We also were
interested in exploring the possibility that the effect of
being represented by an intermediary comes from the
status or prestige associated with retaining an agent, an
idea that is separate from but related to perceptions of
competence. Consequently, we included two new, addi-
tional measures of perceptions of status and prestige in
this study as manipulation checks.
Method
Participants. Forty-four students from a major univer-
sity in the United States were recruited from an elec-
tronic mailing list that advertises behavioral studies to
people who have indicated that they are interested in
participating in them (M age = 21.66, SD = 2.93; 16
women, 19 men; demographic information for 8 par-
ticipants was not collected). In this case, participants
responded to an advertised study on decision making
and were paid $10 for their participation.
Procedure. Participants were randomly assigned to
read and respond to one of three vignettes. In all cases,
participants were instructed to imagine themselves in
the role of a senior editor at a successful book-publish-
ing house considering a deal with an experienced
author named Michael Green. In all conditions, partici-
pants were told that the author with whom they are
negotiating had already been successful in the academic
book market and was now thinking about moving into
the popular book market. Therefore, the competence
of the author was constant and presumed to be equal
across all three conditions. Participants in the control
condition then read excerpts from a conversation with
the author asking for a sizeable advance owing to the
quality of his reputation, talent, and vision for the new
book. Participants in the agent conditions read the iden-
tical speech as being delivered entirely by one agent or
split between two agents.
Dependent Variables
Extra help. Because the situation is different in Study 2,
we modified the extra assistance measure to reflect help
that could be plausibly given by a publisher to an author.
Participants indicated on a 10-point scale ranging from 1
(extremely unwilling) to 10 (extremely willing) their desire to
invite the author to dinner at their home and to favor
him over other authors in granting requests for special
help. These two items were averaged to create a scale of
extra help that accrues to the client (α=.61).
Likeability. The identical likeability scale used in
Study 1 was used to assess participants’ impressions of
the author and his agent or agents; in this instance, the
scale achieved a high degree of reliability (α=.83).
Financial outcomes. After reading the scenario, partic-
ipants indicated how much they would offer the author
for the advance on his proposed book. The scale for the
financial offer ranged from 1 (less than $25,000) and
increased in amounts of $5,000 until 10 (more than
$60,000).
Manipulation checks. To ensure that perceptions of
competence were, in fact, equivalent across all three
conditions, we used the same four-item competence
scale from Study 1 (α=.88). We also included two addi-
tional items to explore whether agent representation or
number of agents influenced perceptions of status or
prestige. Participants were asked to answer the ques-
tions, “How much status do you think Michael Green
has as an author?” and “How prestigious do you think
Michael Green is as an author?” on 10-point scales rang-
ing from 1 (very low status/very unprestigious) to 10
(extremely high status/extremely prestigious).
Results and Discussion
Table 3 presents the means and standard deviations
for the economic and likeability outcomes by condition.
Effects of social proof. Pairwise comparisons revealed
that there were no differences between having one or
two agents on any of the dependent variables of inter-
est (ts ranged from .02 to 1.00, ns, ds ranged from 0 to
.22). Null results are, of course, difficult to interpret
Pfeffer et al. / OVERCOMING THE SELF-PROMOTION DILEMMA 7

and may be specific to this research design, measure-
ment, or particular stimulus materials. Nonetheless, the
absence of any significant differences between having
one or two agents suggests that the observed effects of
using an intermediary are unlikely to be solely because
of social proof processes and that other factors, such as
perceptions of likeability, are more potent in this par-
ticular research setting. Moreover, considering the dif-
ferences in the means between the one- and two-agent
conditions, it would take a very large number of agents
to have much greater effect than one agent alone.
Because there was no difference between the one- and
two-agent conditions, for all subsequent analyses, we
collapsed across this variable.
Perceptions of competence, status and prestige. As intended
in our design, condition did not influence ratings of the
client’s competence, t(42) = 1.11, ns, d = .36, CI(diff) =
–1.78-3.34. We presume that this intended null result is
because of the statement regarding the author’s previ-
ous success, a statement that was equivalent across con-
ditions, but the absence of any differences in perceived
competence also could be from the difference in the set-
ting (publishing vs. job interviewing) or other factors.
The data also show that the presence or absence of an
agent did not influence perceptions of the author’s
status, t(42) = .25, ns, d = .08, CI(diff) = –3.81-4.35, or
prestige, t(42) = .79, ns, d = .25, CI(diff) = –4.89-3.18.
Financial outcomes. There was a marginally significant
difference in participants’ recommended advance when
the author was represented by an agent (or two) than
when he was alone, t(42) = 1.85, p = .07, CI(diff) = –5.25-
1.73, although the Cohen’s d for this effect size was .60,
which is traditionally considered a medium effect. In
combination with the results from Study 1, these results
could be interpreted as evidence that perceptions of
competence influence how much money people decide
to offer, making approximately the same offers when
competence is judged to be about equal (as in Study 2)
and offering more when they perceive greater compe-
tence (as we observed in Study 1).
Perceptions of likeability and extra assistance. Consistent
with the results of Study 1, the author was perceived
more favorably, as measured by the likeability scale,
when he was represented by an agent (or two) than
when he represented himself, t(42) = 7.58, p < .001, d =
2.48, CI (diff) = –2.61 to –1.51. As we expected, authors
were more likely to receive extra assistance when they
were represented by an agent than when they repre-
sented themselves, t(42) = 6.10, p < .001, d = 1.99,
CI(diff) = –3.80 to –1.89.
Mediational analysis. For the mediational analyses,
the independent variable was coded 0 if the author rep-
resented himself and 1 if he was represented by one or
two agents. Once again, as the pattern of coefficients
depicted in Figure 2 demonstrates, likeability partially
mediates the effect of having an agent on obtaining
extra help from the publisher, and a Sobel test indi-
cated that the mediated effects were significant (Z =
2.48, p < .05). Just as in Study 1, likeability ratings did
not mediate the effects of having an agent on financial
outcomes.
The results of Study 2 indicate once again that agents
can provide important person perception benefits even
when people know the agents’ economic interest in
making their clients look good. In this case, authors who
used agents to tout their positive qualities were more
likeable than those who presented the favorable infor-
mation themselves, and this increased likeability con-
tributed to willingness to offer extra assistance.
Furthermore, these results occurred in a negotiation
context where the client’s competence was not in ques-
tion, and because there was no evidence that agents cre-
ated more status or prestige for the author, and there
was no difference between having one or two agents, we
can, at least in this instance, rule out some plausible
alternative interpretations for the observed results.
Taken together, these results are consistent with our
arguments about the self-presentation dilemma and the
8 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
TABLE 3: Means (Standard Deviations) From Study 2
No Agent 1 agent 2 agents
Dependent Variable (n = 15) (n = 14) (n = 15)
Book advance offer 5.60
a
(1.55) 6.43
a
(1.65) 6.53
a
(1.36)
Likeability scale 4.45
a
(0.96) 6.40
b
(0.70) 6.62
b
(0.90)
Competence scale 7.70
a
(1.42) 7.96
a
(0.57) 8.21
a
(1.16)
Status 7.00
a
(1.81) 6.85
a
(1.46) 6.87
a
(2.00)
Prestige 6.67
a
(2.06) 6.93
a
(1.64) 7.27
a
(1.53)
Extra help scale 3.67
a
(1.57) 6.36
b
(0.72) 6.67
b
(1.87)
NOTE: Within each row, means with different subscripts are signifi-
cantly different at p < .05.
No agent
vs.
1 or 2 agents
Extra Help
Likeability
• = .70** (separate model)
• = .43* (combined model)
• = .76***
• = .69*** (separate model)
• = .36* (combined model)
Figure 2 Mediational analyses for Study 2: Likeability mediates the
relationship between having an agent and obtaining extra
help.
*p < .05. **p < .01. ***p < .001.

effect of intermediaries in overcoming it. Having an
agent changed perceptions of likeability even when, as
in the first study, the agent obviously had a financial
interest in the client’s success.
STUDY 3
In Study 3, our objective was first to replicate the
results of Study 2 using an even more realistic video
vignette methodology that would enhance the motiva-
tional involvement and engagement of the study partic-
ipants. Second, this method also allowed us to explore
the effects of the client being physically present with the
agent, which puts the effect of situational constraints on
interpersonal perceptions to a very stringent test. One
possible benefit of using an intermediary is that the
client can be physically absent when positive informa-
tion is presented on that person’s behalf. This fact of not
being visible when claims of competence are being
made might decrease negative attributions or associa-
tions that come from self-promotion. Even though
people know that the agent is acting on the client’s
behalf, the fact that the client is not present may make
this association less salient and reduce the perceived
connection between the two parties. However, it also
may be the case that as long as the words don’t come out
of the client’s mouth, the person being represented may
escape the social opprobrium that results from appear-
ing self-aggrandizing even if the client is seen to be com-
plicit in the agent’s behavior.
One limitation of our first two studies was that we
could not disentangle whether it was the agent making
the statements or the client’s physical absence that
shielded the client from negative interpersonal percep-
tions. In Study 3, we manipulated both agent represen-
tation and physical presence separately to examine the
effects of these two factors.
Method
Participants. Eighty-five participants (53 women, 32
men; M age = 22.36, SD = 5.05) were recruited from the
same electronic mailing list as described in Study 2. In
this sample, 51.8% identified themselves as White or
Caucasian, 36.5% identified themselves as Asian/Asian
American, 3.5% were Black/African American, and
8.2% were Hispanic/Latin American. Half of our par-
ticipants stated that their major was in science and/or
engineering, whereas the others identified themselves
as humanities majors (13.6%), social sciences majors
(25.8%), other (1.5%), or undecided (9.1%); 27.4% of
our participants were in the master’s of business admin-
istration (MBA) program, which means that they had at
least some work experience.
Procedure. Participants were told that the researchers
were interested in their reactions to differing types of
negotiation tactics. They were told to imagine them-
selves as a senior editor in the videotaped scene that
they were about to watch and that the scene was of a fic-
tional negotiation with actors playing the parts of an
author or an agent. To help them prepare, participants
read a background information sheet prior to watching
the videotape. The sheet contained a description of the
scenario used in Study 2. They were then shown one of
four randomly assigned videotaped scenes of the nego-
tiation and then completed a series of questionnaires.
Stimulus materials. Four different videotaped scenar-
ios were created. A professional actor was hired to
deliver the short speech used in Study 2. The same actor
delivered identical lines in all conditions. In all condi-
tions, the actor (either alone or accompanied by a con-
federate in a nonspeaking role) entered an office set, sat
down at the desk, and delivered his lines. In the agent
alone condition, the actor was alone on the camera and
played the part of the agent. In the author alone condi-
tion, the actor also was alone but this time played the
role of the author representing himself. In a third con-
dition (author speaks/agent silent), the actor played the
author and was accompanied by a confederate, whom
he introduced as his agent. The confederate did not say
anything and only his arm and shoulder were in view
while the author made his speech. Finally, in the agent
speaks/author silent condition, the actor played the
agent and the confederate was introduced as his client.
Again, only the confederate’s arm and shoulder were vis-
ible during the delivery of the lines that followed.
Dependent Variables
Complicity measures .The measures collected were
identical to those from Study 2, with an important addi-
tion of complicity measures. In addition to the variables
from Study 2, we also included measures that assessed
the extent to which the participants felt that the author
was complicit in the agent’s statements and behavior.
These questions were answered using a 10-point scale
that ranged from 1 (no knowledge at all/not at all approv-
ing/no control at all) to 10 (complete knowledge/extremely
approving/complete control). Because these questions
referred to the agent’s statements and actions, they
were only collected in the agent alone and agent
speaks/author silent conditions.
Results and Discussion
Table 4 contains the means and standard deviations
for all variables of interest in Study 3. For all of the
dependent variables in this study, we conducted 2
(speaker alone vs. accompanied) × 2 (agent speaks vs.
Pfeffer et al. / OVERCOMING THE SELF-PROMOTION DILEMMA 9

author speaks) ANOVAs to disentangle the differential
effects of speaking, physical presence, and the interac-
tion of the two.
Manipulation checks. Our analyses revealed a margin-
ally significant main effect for being alone or accompa-
nied, such that when the agent or author was alone in
the meeting, the author was perceived as more compe-
tent (M = 7.64) than when the agent or author was
accompanied (M = 7.00), F(1, 59) = 2.92, p < .10, η
2
=
.05. As expected, there was no significant main effect of
speaker, F(1, 59) = .94, ns, η
2
= .02, or a significant inter-
action effect, F(1, 59) = .32, ns, η
2
= .01, on ratings of
the author’s competence.
In addition, there were no main effects for speaker,
being alone or accompanied, or the interaction between
the two on ratings of the author’s status or prestige (all
Fs < 2.11, ns, all η
2
< .04).
Financial outcomes. There were no significant main or
interaction effects on the size of the author’s advance
(all Fs < 1.81, ns, all η
2
< .03). Taken as a whole, these
results correspond with the findings in Study 2 that
show that when competence is not in question and is
equal across conditions, using an agent does not neces-
sarily lead to better financial outcomes.
Complicity measures. As expected, an independent
samples t test demonstrated that participants in the
agent speaks/author silent condition (where the author
was physically present) rated the client as having more
knowledge about his agent’s behaviors, t(40) = 3.26,
p < .01, d = 1.50, CI(diff) = –3.56 to –0.84, and believed
that the author approved of his agent’s behavior to a
greater extent, t(40) = 2.32, p < .05, d = 1.07, CI(diff) =
–2.60 to –1.67, than those in the agent alone condition.
In addition, participants in the agent speaks/author
silent condition rated the author as having marginally
more control over his agent’s behavior, t(40) = 1.90,
p <.06, d = .87, CI(diff) = –2.84 to .09, than those in the
agent alone condition. Thus, physical presence did, as
we had intended, serve as a reminder that the agent was
complicit and in a relationship with the client. However,
there were no correlations between the complicity mea-
sures and the extra assistance scale or the likeability
scale (rs ranged from .00-.31, ns).
Extra assistance and likeability. A 2 × 2 ANOVA demon-
strated that there was no significant main effect for
whether the author or agent was alone or accompa-
nied, F(1, 59) = .51, ns, η
2
= .01, and there was no sig-
nificant interaction effect, F(1, 59) = .00, ns, η
2
= .00. Of
importance, however, there was a significant effect for
speaker, F(1, 59) = 13.03, p < .01, η
2
= .19, such that the
author received significantly greater extra assistance
when the agent spoke (M = 5.90) than when he spoke
himself (M = 3.96).
Results for ratings of the author’s likeability fol-
lowed the identical pattern we have seen in the previ-
ous studies. There was no significant main effect for
whether the author or agent was alone or accompa-
nied, F(1, 59) = .09, ns, η
2
= .00, and there was no sig-
nificant interaction effect, F(1, 59) = .36, ns, η
2
= .00,
but there was a significant effect for speaker, F(1, 59) =
12.31, p < .01, η
2
= .18. The author was better liked
when his agent sang his praises (M = 6.18) rather than
when he praised himself (M = 5.13).
These results show that the differences in person per-
ceptions and extra assistance that come from having an
intermediary are not simply because the client is physi-
cally absent. Instead, the fact that someone other than
the client makes claims of competence permits the
client to avoid being seen as self-promoting, thereby
providing the benefits of being offered extra assistance.
Although participants were aware that the author was
more complicit in the statements that the agent made
when the author was present, this awareness did not
influence how favorably the author was perceived or his
likelihood of receiving extra help. These results indicate
that just as Jones and Harris (1967) and others (e.g.,
Ginzel, 1994) have found, people are not able to fully
10 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
TABLE 4: Means (Standard Deviations) Fom Study 3
Agent Alone Author Alone Author Speaks/Agent Silent Agent Speaks/Author Silent
Dependent Variable (n = 22) (n = 19) (n = 21) (n = 23)
Book advance offer (financial outcomes) 5.77 (1.63)
a
5.00 (1.94)
a,c
5.06 (1.51)
a,c
4.74 (1.60)
b,c
Likeability scale (? = .76) 6.14 (.99)
a
5.16 (.98)
b
4.54 (.99)
b
6.22 (1.06)
a
Competence scale (? = .94) 7.85 (1.48)
a
7.37 (1.01)
a,b
6.46 (1.57)
b
7.17 (2.07)
b
Status 6.18 (2.13)
a
6.21 (1.51)
a
5.71 (1.71)
a
5.61 (2.21)
a
Prestige 6.22 (2.07)
a
6.26 (1.52)
a
5.86 (1.80)
a
5.30 (2.27)
a
Extra help scale (? = .63) 5.98 (2.08)
a
3.95 (1.96)
b
3.33 (1.69)
b
5.70 (1.47)
a
Knowledge 5.59 (2.17)
a
— — 7.79 (2.12)
b
Approve 5.77 (2.00)
a
— — 7.16 (1.80)
b
Control 4.72 (2.31)
a
6.11 (2.31)
a
NOTE: Within each row, means with different subscripts are significantly different at p < .05.

correct their interpersonal judgments and perceptions
for knowledge of situational constraints and influence.
The results also show that the Inman et al. (2004) find-
ings on the benefits of having a third party sing one’s
praises persist even when that third party is seen to be
associated with and closely tied to—even under the con-
trol of—the person being praised.
Mediational analyses. Once again, we conducted medi-
ational analyses, which are shown in Figure 3. For these
analyses, the independent variable was whether the
author spoke (i.e., in the author alone and author
speaks/agent silent conditions) or the agent spoke (i.e.,
agent alone and agent speaks/author silent conditions).
These analyses revealed that in this study, likeability fully
mediated the effects of who spoke (agent or author) on
whether the author received extra assistance (see Figure
3 for relevant coefficients). In addition, the results of
the Sobel (1982) test indicated that this mediation was
statistically significant (Z = 2.23, p < .01).
The results of Study 3 provide important insights
into the mechanisms that create the benefits of agent
representation by investigating the effects of perceived
complicity and physical salience as well as speaking on
behalf of another. Study 3 revealed that the positive
results associated with using an agent to sing one’s
praises occur even when participants are aware that the
agent is complicit with and somewhat under the control
of the client because the client is actually physically pre-
sent while the agent speaks. These findings suggest that
it is who actually makes the positive statements, not
physical salience or even perceived control, that
accounts for the effects of using an agent on increased
likeability and obtaining extra help.
GENERAL DISCUSSION
The results of the three studies paint a consistent pic-
ture: Using an intermediary to make positive comments
shields the individual being promoted from the adverse
consequences of being seen as self-aggrandizing. People
whose praises are sung by others are perceived as more
likeable, and this likeability leads, in turn, to a greater
willingness to do favors or expend extra effort on behalf
of that individual. What is particularly remarkable is that
these effects occurred not just in situations such as those
described by Inman et al. (2004), in which neutral or
disinterested third parties are doing the talking, but
even when the intermediary had a financial stake in the
client’s outcomes, even when others knew about this
financial stake (because they were explicitly told about
it), and even when the client was physically present and
seen to be complicit in the intermediary’s statements on
his behalf. As such, the results not only show that agent
representation provides a resolution to the self-promo-
tion dilemma but they also provide another vivid illus-
tration of the fact that individuals are unable to fully
discount situational factors in forming social judgments
and making decisions. So, even though participants
were told (and in Study 3 actually saw) the connection
between the intermediary and his client, they nonethe-
less perceived the client to be more likeable and were
willing to accord him more assistance just because
another sang his praises.
Furthermore, the results of Study 1 showed that in
circumstances when competency may be in question,
having an agent make favorable statements can have
the added benefit of increasing perceptions of compe-
tency or ability above the individual claiming compe-
tency for himself, and these perceptions of competency
can then positively influence financial outcomes.
We have argued that having another sing one’s
praises leads to greater likeability and, as a conse-
quence, a greater willingness on the part of others to do
favors for the individual. This causal model was tested
through our mediation analyses and we found either
partial or full mediation of the effect of using an agent
on obtaining extra help occurring through the effect of
an agent on likeability in each instance. However, there
are other plausible causal pathways not tested or rele-
vant in the present study. For example, it is possible that
offering extra help could, through the mechanism of
cognitive consistency, cause those who offered the extra
help to see the individual to whom the help was offered
as more likeable or more competent. Future research
should examine additional causal pathways, such as cog-
nitive consistency, to develop a more complete under-
standing of the benefits of agent representation.
Our results are theoretically important because they
highlight the utility of employing an agent in affecting
the interpersonal climate of negotiations and transac-
tions. Intermediaries can help clients build and maintain
positive relationships with others, a social psychological
Pfeffer et al. / OVERCOMING THE SELF-PROMOTION DILEMMA 11
Speaker (Agent
promotes versus
author promotes self)
Extra Help
Likeability
• = .63** (separate model)
• = .47* (combined model)
• = .43**
• = .43** (separate model)
• = .23+ (combined model)
Figure 3 Mediational analyses for Study 3: Likeability mediates the
relationship between self-promotion or agent promotion
and obtaining extra help.
†p < .10. **p < .01. ***p < .001.

effect that may help to explain why intermediaries are so
pervasive in their use. Our results serve to substantially
broaden the exploration of the effects of agents on trans-
actions from the narrow focus on the possible divergence
of interests between agents and their clients to a broader
consideration of other dimensions of the interaction—
for instance, interpersonal affect—and also to consider
the effects of having an intermediary on the other party
to the transaction. Our results also show that the positive
benefits of having a third party sing one’s praises hold
even when that third party stands to benefit from the
positive portrayals being offered.
Directions for Future Research
A number of questions remain for subsequent
research to more fully understand the mechanisms and
circumstances under which using an agent helps over-
come the self-promotion dilemma. An important dimen-
sion of the experimental situations we have used is the
absence of any prior history or relationship between
the people involved in the transactions. One possible
moderator to the relationships we observed could be
whether there was a prior history of interactions between
any of the negotiating parties. Prior history could con-
ceivably make self-promotion more tolerable and less
unacceptable because an existing favorable relation-
ship could produce idiosyncrasy credits—permission to
deviate from socially expected and accepted behavior
(Hollander, 1958)—that mitigate the negative effects of
self-promotion.
The particular mechanism we have explored is also
something that will be more or less important in differ-
ent contexts. When interpersonal relations are an
important part of an ongoing transaction, the effects
we have discussed should be more important in pre-
dicting outcomes. In other, more one-off transactions
with less interpersonal content, such as real estate trans-
actions, the sorts of effects we have demonstrated here
should have much less importance, whereas financial
outcomes and perceptions of competence may become
more important. Consequently, one extension of the
present research is to systematically study the effects of
intermediaries in transactions that presumably vary
along both dimensions of prior history and the impor-
tance of interpersonal versus financial outcomes. When
financial outcomes are important, a more in-depth
understanding of the mediational relationship between
agent representation, competence, and financial out-
comes would likely be beneficial, and more work is nec-
essary to determine potential moderators of this effect.
It would be useful to understand more fully the con-
ditions under which interpersonal perceptions (of like-
ability) fully mediate the effects of using an intermediary
and when the mediation is partial. Full mediation was
observed only in the case of Study 3, which is also the
instance of the greatest emotional and motivational
engagement because of the use of the video vignette
methodology. It is possible that full mediation through
likeability occurs only when the intermediary is present
or absent in more psychologically involving or dynamic
situations.
Although there are a number of possible extensions
of the present research and a need to extend the argu-
ment to other negotiating situations, it does seem clear
that the general situation we have explored—having
another represent an individual in some transaction—
is quite important as the use of intermediaries is wide-
spread. Therefore, it is both theoretically and substantively
useful to explore the effects of various representation
arrangements as well as the social psychology that
underlies those effects. Going beyond the principal-
agent literature with its assumptions of conflicting
interests to investigate other social psychological mech-
anisms that affect interpersonal interaction seems to be
a useful avenue to pursue.
REFERENCES
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator distinc-
tion in social psychological research: Conceptual, strategic, and sta-
tistical considerations. Journal of Personality and Social Psychology, 51,
1173-1182.
Baumeister, R. F., & Ilko, S. A. (1995). Shallow gratitude: Public and
private acknowledgement of external help in accounts of success.
Basic and Applied Social Psychology, 16, 191-209.
Bazerman, M. H., Neale, M. A., Valley, K. L., Zajac, E. J., & Kim, M. Y.
(1992). The effect of agents and mediators on negotiation out-
comes. Organizational Behavior and Human Decision Processes, 53,
55-73.
Carmines, E., & McIver, J. (1981). Analyzing models with unobserved
variables: Analysis of covariance structures. In G. Bohrnstedt & E.
Borgatta (Eds.), Social measurement: Current issues (pp. 65-115).
Beverly Hills, CA: Sage.
Cialdini, R. B. (2001). Influence: Science and practice (4th ed.). Boston:
Allyn & Bacon.
Coser, L. A., Kadushin, C. K., & Powell, W. W. (1982). Books: The culture
and commerce of publishing. Chicago: University of Chicago Press.
Forsyth, D. R., Berger, R., & Mitchell, T. (1981). The effects of self-
serving vs. other-serving claims of responsibility on attraction and
attribution in groups. Social Psychology Quarterly, 44, 59-64.
Giacalone, R. E., & Rosenfeld, P. (Eds.). (1989). Impression manage-
ment in the organization. Hillsdale, NJ: Lawrence Erlbaum.
Ginzel, L. E. (1994). The impact of biased inquiry strategies on per-
formance judgments. Organizational Behavior & Human Decision
Processes, 57, 411-429.
Higgins, C. A., & Judge, T. A. (2004). The effect of applicant influ-
ence tactics on recruiter perceptions of fit and hiring recommen-
dations: A field study. Journal of Applied Psychology, 89, 622-632.
Higgins, C. A., Judge, T. A., & Ferris, G. R. (2003). Influence tactics
and work outcomes: A meta-analysis. Journal of Organizational
Behavior, 24(2), 89-106.
Hollander, E. P. (1958). Conformity, status, and idiosyncrasy credit.
Psychological Review, 65, 117-127.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in
covariance structure analysis: Conventional criteria versus new
alternatives. Structural Equation Modeling, 6, 1-55.
12 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

Inman, M. L., McDonald, N., & Ruch, A. (2004). Boasting and first-
hand and secondhand impressions: A new explanation for the
positive teller-listener extremity effect. Basic and Applied Social
Psychology, 26, 59-75.
Jonas, E., Schulz-Hardt, S., & Frey, D. (2005). Giving advice or making
decisions in someone else’s place: The influence of impression,
defense, and accuracy motivation on the search for new informa-
tion. Personality and Social Psychology Bulletin, 31, 977-990.
Jones, E. E., & Harris, V. A. (1967). The attribution of attitudes.
Journal of Experimental Social Psychology, 3, 1-24.
Jones, E. E., & Wortman, C. (1973). Ingratiation: An attributional
approach. Morristown, NJ: General Learning Press.
Kennedy Information Online. (2004). Executive recruiter database.
Retrieved August 17, 2004, from http://www.kennedyinfo.com/db/
db_der_bas.html.
Kenrick, D. T., Neuberg, S. L., & Cialdini, R. B. (2002). Social psychol-
ogy: Unraveling the mystery. Boston: Allyn & Bacon.
Kurtzberg, T., & Medvec, V. H. (1999). Can we negotiate and still be
friends? Negotiations Journal, pp. 355-361.
Kurtzberg, T., Moore, D., Valley, K., & Bazerman, M. H. (1999). Agents
in negotiations: Toward testable propositions. In R. H. Mnookin &
L. E. Susskind (Eds.), Negotiating on behalf of others (pp. 283-298).
Thousand Oaks, CA: Sage.
Levitt, S. D., & Syverson, C. (2004). Market distortions when agents are
better informed: The value of information in real estate transactions.
Unpublished manuscript, University of Chicago.
Milgram, S., Bickman, L., & Berkowitz, O. (1969). Note on the draw-
ing power of crowds of different size. Journal of Personality and
Social Psychology, 13, 79-82.
Mnookin, R. H., & Cohen, J. R. (1999). Introduction. In R. H.
Mnookin & L. E. Susskind (Eds.), Negotiating on behalf of others (pp.
1-20). Thousands Oaks, CA: Sage.
Rubin, J. Z., & Sander, F. E. A. (1988). When should we use agents:
Direct vs. representative negotiation. Negotiation Journal, 4, 395-401.
Schotter, A., Snyder, B., & Zheng, W. (1995). Bargaining through agents:
An experimental study of delegation and commitment (C. V. Starr Center
for Applied Economics Research Report #95-27). New York: New
York University.
Shell, G. R. (1999). Bargaining for advantage. New York: Viking.
Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects
in structural equation models. In S. Leinhaidt (Ed.), Sociological
methodology (pp. 290-312). Washington, DC: American Sociological
Association.
Wortman, C., & Linsenmeier, J. (1977) Interpersonal attraction and
techniques of ingratiation in organizational settings. In B. M. Staw
& J. R. Salancik (Eds.), New directions for organizational behavior
(pp. 133-178). Chicago: St. Clair Press.
Wosinska, W., Dabul, A. J., Whetstone-Dion, R., & Cialdini, R. B.
(1996). Self-presentational responses to success in the organiza-
tion: The costs and benefits of modesty. Basic and Applied Social
Psychology, 18, 229-242.
Yavas, A., Miceli, T. J., & Sirmans, C. F. (2001). An experimental analy-
sis of the impact of intermediaries on the outcome of bargaining
games. Real Estate Economics, 29, 251-276.
Received July 29, 2005
Revision accepted April 13, 2006
Pfeffer et al. / OVERCOMING THE SELF-PROMOTION DILEMMA 13