Basic and Applied Social Psychology

ISSN: 0197-3533 (Print) 1532-4834 (Online) Journal homepage: https://www.tandfonline.com/loi/hbas20

Faith in Supervision and the Self-Enhancement

Bias: Two Psychological Reasons Why Managers
Don't Empower Workers

Jeffrey Pfeffer , Robert B. Cialdini , Benjamin Hanna & Kathleen Knopoff

To cite this article: Jeffrey Pfeffer , Robert B. Cialdini , Benjamin Hanna & Kathleen Knopoff
(1998) Faith in Supervision and the Self-Enhancement Bias: Two Psychological Reasons Why
Managers Don't Empower Workers, Basic and Applied Social Psychology, 20:4, 313-321, DOI:
10.1207/s15324834basp2004_8

To link to this article: https://doi.org/10.1207/s15324834basp2004_8

Published online: 07 Jun 2010.

Submit your article to this journal

Article views: 370

View related articles

Citing articles: 29 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=hbas20
BASIC AND APPLIED SOCIAL PSYCHOLOGY, 20(4), 313 – 321
Ó
Copyright 1998, Lawrenc e Erlbaum Associates, Inc.

Faith in Supervision and the Self-Enhancement Bias:

Two Psychological Reasons Why Managers
Don’ t Empow er Workers PFEFFER, CIALDI
FAIT
NI, HHANNA,
IN SUPERVI
KNOPOF
SION
F

Jeffrey Pfeffer
Graduate School of Business
Stanford University

Robert B. Cialdini
Department of Psycholo gy
Arizona State University

Benjamin Hanna and Kathleen Knopoff

Graduate School of Business
Stanford University

This study provides evidence for 2 psychological processes that may help explain managers’ re-
luctance to use worker empowerment practices such as delegation or self-managing teams: (a) a
faith in supervision effect, which reflects the tendency of observers to see work performed under
the control of a supervisor as better than identical work done without as much supervision; and
(b) a self-enhancement effect, which reflects the tendency of mangers to evaluate a work prod-
uct more highly the more self-involved they are in its production. Because empowerment prac-
tices dilute managers’ individualized supervision of work, they can also reduce managers’ bi-
ased perceptions of work quality. Support for this argument was obtained in that (a) participants
assigned higher quality to the identical work product as supervisory involvement increased, (b)
they did so at elevated levels when they had more self-involvement in supervising the work, and
(c) a team-based, empowerment orientation curtailed both biases.

There is accumulating evidence that organizational practices participative practices (Chalykoff & Kochan, 1989). Indeed,
that empower employees by increasing their participation in after an extensive review of the literature on participation, Le-
decisions about the work process typically result in greater vine and Tyson (1990) reported the disproportionately posi-
worker productivity, morale, and organizational commitment tive impact of participation on productivity and asked a pro-
(Arthur, 1994 ; Chapman, Elhance, & Wenum, 1995 ; Kochan vocative question concerning employee empow erment,
& Osterman, 1994 ; Lawler, Mohrman, & Ledford, 1992 ; ª Why don’ t we see more of it? º (p. 184).
MacDuffie, 1995 ; O’ Reilly, 1989 ; Pfeffer, 1994). Puzzling Many commentators have sought to answer this question
then is the relative resistance to such empowerment practices by pointing to various market or workplace features that
by American managers in favor of more hierarchical com- would constrain tendencies to delegate authority to work-
mand-and-control approaches (Pfeffer, 1994), especially ersÐ features such as generic problems inherent in the diffu-
given evidence of the substantially reduced costs to managers sion and implementation of any administrative innovation
in time and energy following implementation of these (Johns, 1993); general market or institutional conditions that
make empowerment unattractive (Levine & Tyson, 1990);
manager recognition that increasing employee participation
Requests for reprints shoul d be sent to Jeffrey Pfeffer, Graduate School of can limit managerial power, status, and jobs (Womack,
Business , Stanford University, Stanford, CA 94305 –5015, E-mail: 1996); and lack of trust in employees’ abilities, motives, and
pfeffer_jeffrey@ gsb.stanford.edu; or to Robert B. Cialdini, Department of
integrity (Schoorman, Mayer, & Davis, 1996). In addition,
Psychology , Arizona State University , Tempe, AZ 85287 – 1104, E-mail: rob-
ert.cialdini@ asu.edu empowerment practices do not appear well suited to the
314 PFEFFER, CIALDINI, HANNA, KNOPOFF

structure of some industries (Kenney & Florida, 1993). Our stituted authorities, not simply because of their greater power
view is that each of these alternatives provides a partial an- in the situation but because of greater assumed wisdom and
swer to the question of why participation-based employee expertise (Cialdini, 1993). Moreover, for a pair of reasons,
empowerment approaches have not spread rapidly through observers may justifiably assume that superiors will be moti-
U.S. corporate practice and why they are frequently vated to assure high-quality performance in the tasks they su-
short-lived once adopted (Drago, 1988). However, we wish pervise. First, individual superiors will be held responsible
to consider the relevance of factors of a different sortÐ those for the successes and failures of their group by virtue of the
grounded in features of individual psychology rather than in romance of leadership effect w e have already discussed. Sec-
organizational or market structure. ond, such accountability does normally spur people to better
thinking and harder work (Simonson & Nye, 1992 ; Skitka,
Mosier, & Burdick, 1996 ; Tetlock, 1985); and this is espe-
THE FAITH IN SUPERVISION EFFECT cially so as the responsible person becomes more individu-
ally accountable (Karau & Williams, 1993 ; Szymanski &
Meindl (Meindl, 1993; Meindl & Ehrlich, 1987 ; Meindl, Harkins, 1987), as is the case in hierarchical organizations.
Ehrlich, & Dukerich, 1985 ) coined the term romance of lead- Thus, in the workplace, traditional forms of managerial
ership to apply to the tendency of observers to assign oversight, surveillance, and control, which elevate a man-
greater-than-w arranted causal w eight to the actions of lead- ager’ s individualized supervisory role in the work process,
ers in determining the outcomes of their organizations. In a should also increase the presumption and perception of resul-
variety of business contexts, Meindl and his associates pro- tant work quality. That is, with greater managerial supervi-
vided compelling evidence that organizational performance, sion, observers will bias upward their view of the work
especially when notably good or notably poor, is unduly at- productÐ the basic faith in supervision effect. Conversely,
tributed to the role of the leader. Because of the hierarchical employee empowerment practices, which reduce a man-
character of most U.S. business organizations, w hich sepa- ager’ s individualized role, should reduce such overvaluation.
rates and elevates management from workers, it is under- According to this analysis, then, one reason that managers
standable that individual leaders and managers would be fo- may resist and seek to overturn w orker participation schemes
cal points for causal attributions, as much evidence indicates is that, under these schemes, the quality of the work appears
that in the eyes of observers what is focal is causal (Fiske & to observers inferior to the quality of the work accomplished
Taylor, 1994). under hierarchical control. Although this reasoning sug-
The same hierarchical structure and belief in leadership gested to us that a faith in supervision bias may well operate,
produces a phenomenon that is related to but fundamentally it did not strike us as likely to have notable influence on man-
different from that of the romance of leadershipÐ one that can agers’ perceptions when it operated alone. It seemed more
be labeled the faith in supervision effect. Meindl (1990, p. 165) likely that the faith in supervision effect would have a potent
proposed that the romance of leadership can lead to a transmis- impact on such perceptions when it was combined with an-
sion of value from cause to effect, such that a valued causal fac- other, more fundamental and more powerful human
tor, such as leadership, will logically be expected to produce a biasÐ the tendency to promote the self to the self.
valued outcome. Following this reasoning, Meindl proposed
that an outcome would be evaluated differently depending on
the causes that were believed to have produced it. Consistent SELF-ENHANCEMENT
with this prediction, Meindl and Ehrlich (1987) found in an ex-
periment that identical company performance was evaluated One of the most widely documented effects in social psy-
more favorably w hen leadership factors accounting for the chology is the preference of most people to see themselves in
performance were made more salient. a self-enhancing fashion. As a consequence, they regard
Consistent with this research, we suggest that there is a themselves as more intelligent, skilled, ethical, honest, per-
tendency for observers to presume that the greater the degree sistent, original, friendly, reliable, attractive, and
of supervisory involvement, the better the work product. fair-minded than their peers or the average person; they even
Consequently, particularly in the case of work output that has consider themselves better and safer drivers than average af-
some degree of uncertainty associated with its quality, the ter having been involved in a serious automobile accident
greater the supervisory involvement, the more favorable w ill (see Myers, 1996 , for a review of much of this work). On the
be the assessment of that output. There are numerous reason- job, approximately 90% of managers and workers rate their
able bases for observing a result of this sort. For example, it is performances as superior to their peers’ (French, 1968 ;
sensible to believe that someone put in charge of an operation Headey & Wearing, 1987). Unrealistic self-conceptions of
was given the assignment because of skills or capacities for this sort are fostered by the tendency of most individuals to
enhancing its function; this belief reflects the rationale for process and interpret incoming information in ways that pro-
any merit-based, hierarchical system. Indeed, considerable mote the self. For instance, as a rule, people assign more in-
research attests to the deference accorded legitimately con- telligence to one who accepts rather than resists their persua-
 FAITH IN SUPERVISIO N 315

sive arguments (Cialdini, Braver, & Lewis, 1974 ; Cialdini & evaluated the quality of a print advertisement developed by a
Mirels, 1976), attribute their successes to internal causes new coworker. The advertisement w as produced under one
such as ability but their failures to external causes such as of three levels of supervisory involvement in the process. At
luck (Brown & Rogers; 1991 ; Fletcher & Ward, 1988 ; Miller the lowest level, the manager saw only the finished draft. At
& Ross, 1975), and ascribe more validity to tests on which an intermediate level, managers were able to monitor the
they have done well versus poorly (Pyszczynski, Greenberg, worker’ s progressÐ they saw an initial draft and indicated
& Holt, 1985; Shepperd, 1993). areas for improvement; however, they knew that the worker
In general, the more the self is invested in activities, ob- would not be shown this managerial assessment while devel-
jects, or outcomes, the more positively they will be perceived oping the final version. At the highest level of involvement,
to maintain or bolster a favorable self-view (Sedikides, the manger both saw an initial draft and provided direct feed-
1993). Thus, for instance, experimental (Bazerman, Beekun, back. In all cases, participants were shown the identical final
& Schoorman, 1982) and field data (Schoorman, 1988 ) dem- product and were asked to evaluate its quality. Figure 1 pres-
onstrate that if supervisors have had positive input into a ents the initial and final drafts. We varied a cross-cutting fac-
prior hiring or promotion decision, they give the person tor by having participants perform their functions under one
higher subsequent performance appraisals and monetary re- of two forms of managerial orientation. The first, designed to
wards than if they were not involved in that prior decision. represent the conventional, hierarchically structured ap-
This propensity for self-enhancement, when added to the proach, emphasized the superior– subordinate character of
faith in supervision effect, could generate potent workplace the manager’ s relationship to the worker, which not only ele-
consequences in that the tendency for an observer to over- vates the managerial role but individualizes it by separating it
value supervised work should be greatly amplifiedÐ at all from that of the worker. The second, designed to represent
levels of supervisory involvementÐ when the supervisor the employee empowerment approach, emphasized a
himself or herself is that observer. It is this supplemented team-like relationship, which dilutes the individualized char-
form of the faith in supervision phenomenon, operating at acter of the managerial role. Finally, to vary the extent to
meaningfully high degrees of self-delusion, that we feel has
the potential to make managers reluctant to empower em-
ployees in the work process. That is so because such empow-
erment practices, which reduce managers’ individualized
supervisory involvement in the work process, should simul-
taneously deflate managers’ exagg erated evaluations of
work quality, rendering the work output subjectively inferior
in the process.
Three hypotheses emerged from this reasoning. First, as a
result of the faith in supervision effect, we would expect that
work produced with greater traditional supervisory involve-
ment (i.e., hierarchically based oversight, surveillance, and
intervention) would be regarded by managers and observers
as superior to identical work produced w ith less such in-
volvement. Second, because of the impact of
self-enhancement tendencies, (self-invested) managers
should rate work outcomes more positively at every supervi-
sory level than should observers. Third, we would expect that
strategies designed to decentralize control over the work pro-
cess would, by reducing the manager’ s individualized role,
limit the faith in supervision effect, even for self-implicated
managers. The prototypical instance of a decentralizing strat-
egy is the work team, in which the manager’ s role is to sup-
port and facilitate the efforts of team members rather than to
direct and control those efforts (Guzzo & Salas, 1995 ;
Katzenbach & Smith, 1993; Kinlaw, 1991).

A TEST

In a laboratory study w e designed to test these hypotheses,

experimental participants acting as marketing managers FIGURE 1 Initial and final drafts of print advertisement.
316 PFEFFER, CIALDINI, HANNA, KNOPOFF

which self-enhancement processes would operate, we had Managerial Role) completely between-subjects design. Al-
half of our participants take the manager’ s role in the study though we had no a priori predictions regarding it, we also
while the other half took an observer’ s role. examined a fourth factor, participant gender, for exploratory
In addition to assessing participants’ evaluations of the fi- purposes. Analyses revealed that gender had a main effect on
nal advertisement, we measured their judgments of the ability our major dependent variable, evaluation of the advertise-
of the worker who produced it. With this measure we hoped to ment, such that female participants rated the advertisement
inquire into whether managers’ tendencies to overestimate the more negatively than men. However, gender did not interact
quality of work produced under their supervision would ex- with any of our other independent variables, indicating that
tend to their supervisees. Tw o opposite possibilities seemed the theoretically relevant effects we obtained occurred equiv-
plausible. In the first, managers would see credit for a work alently for men and women; consequently, we do not address
product as a zero-sum proposition: The more supervisory in- the issue of gender further.
volvement they had in a perceived positive outcome, the more
credit they would assign to their own abilities and the less they Procedure and Independent Variables
would assign to the w orker’ s. Such a possibility is consistent
with early research (e.g., Strickland, 1958 ) indicating that an Participants arrived for the study in pairs and were separated
individual supervised more closely came to be seen by super- into individual rooms where they were told that the study in-
visors as requiring their continued supervision to achieve a volved how differing management styles and levels of infor-
comparable outcome. However, a second possibility exists in mation affect the quality of a finished work product. To liken
which managers would see credit in nonzero-sum terms: The the situation to a business setting, participants were told to
greater the amount of success achieved, the greater the as- assume that they were working for a marketing and advertis-
sumed abilities of those responsible for it, w orkers included. A ing company and that a European wristwatch manufacturer
finding of this sort w ould be congruent with balance theory had asked the company to design a print ad campaign to in-
(Heider, 1958 ; Insko, 1984), which predicts that entities asso- troduce a new model watch to a young target market in
ciated with a positive outcome should also be viewed posi- America. Through a rigged drawing, all participants took the
tively. We had no strong expectations as to which of the two role of a manager who had been with the firm for 5 years
possibilities was more likely to be supported; therefore, no ex- since earning an MBA; they w ere informed that the partici-
plicit predictions were made in this regard. pant in the other room would take the role of a new comer to
Finally, we included a rating of managerial effectiveness the company who had been assigned to the project to develop
to explore the possibility that a team-based orientation might a print advertisement. By assigning the worker newcomer
lower perceptions of manager competence. That is, it seemed status in the organization, we sought to control for the poten-
conceivable that by reducing managers’ centrality in the su- tially confounding influence of such considerations as the
pervisory processes, a team-based orientation could under- history of the manager–w orker relationship and of the
mine managers’ views of their own effectiveness. Should worker’ s prior performance. No participant questioned this
this be the case, organizations would have legitimate grounds aspect of the procedure or suggested that it was unrealistic.
for resisting the implementation of such approaches, as man-
agers who felt that they were not effective in their jobs would Managerial orientation. Participants were assigned to
be likely candidates for the costly problems of absenteeism, one of two managerial orientations, hierarchical or
low job satisfaction, and turnover. team-based. In the Hierarchical Orientation condition, they
were referred to as ª managerº or ª supervisorº throughout the
METHOD experiment, w hereas the worker was referred to as an ª em-
ployee.º In addition, they were told that they would be ª re-
Participants sponsible for supervising while the other person prepares a
solution.º In the Team-Based Orientation condition, they
Tw o hundred and eighty-two first- and second-year MBA were told that the company stressed a team approach; they
students at a leading business school were paid $10 to partici- were referred to as ª team leader,º whereas the worker was
pate in an experiment on management and leadership style. called a ª team memberº throughout. In addition, they were
Most participants had previous work (M = 5.08 years) and told that they would be ª responsible for assisting while the
management (2.05 years) experience and therefore provided other person prepares a solution.º Note that the only differ-
a population that had good ecological validity for a study of ence across these conditions was in the language employed to
managerial involvement. describe them. Although this may seem a relatively subtle
manipulation of managerial orientation, prior research had
Design demonstrated that assigning controlling labels to oversight
responsibilitiesÐ such as ª managing directorº or ª group su-
The experimental conditions formed a 3 × 2 × 2 (Supervi- pervisorº Ð led to more self-oriented actions (e.g., allocating
sory Involvement × Managerial Orientation × Self/Other more rewards to oneself) than did assigning noncontrolling
 FAITH IN SUPERVISIO N 317

labelsÐ such as ª group leaderº or ª group guideº Ð to the final draft and the evaluation scales. The identical final ad-
same responsibilities (Messe, Kerr, & Sattler, 1992 ; vertisement w as shown to all participants.
Samuelson & Allison, 1994).
Self/other managerial role. Half of the participants
Supervisory involvement. Participants were told that played active managerial roles in the project. The other half
because of other demands on their time, the opportunity to in- served as observers to another’ s active management. That is,
teract with the worker during the development of the adver- they were told that two MBAs like themselves had worked on
tisement would be limited. One group of participants (in the the project, were asked to study copies of the instructions and
Lowest Involvement condition) had neither surveillance nor materials relevant to one or another of the experimental con-
feedback opportunity. They learned that they would be able to ditions, and were asked to fill out the evaluation scales based
see and judge the quality of the advertisement only after it was on their own views of the final product.
completed. A second group (in the Intermediate Involvement
condition) had surveillance but no feedback opportunity. Baseline control condition. Finally, a group of 15
They w ere able to monitor the product development process additional participants served as a baseline control. They
somewhat by seeing a first draft of the adÐ an event the em- were informed that the final advertisement had been created
ployee realized would occurÐ but they were not able to offer by an MBA like themselves over a 40-min period, but they
feedback to the person on that first draft. These participants were told nothing of the managerial circumstances under
did, however, fill out a form providing feedback, but were which it w as produced before they evaluated it.
told that it was not possible for the person working on the ad-
vertisement to see it. A third group (in the Highest Involve- Dependent Variables
ment condition) had both surveillance and feedback opportu-
nity. They learned that they would see a first draft and w ould Participants responded to a questionnaire that included several
be able to provide the individual feedback on it from a stan- scales measuring three major dependent variablesÐ assess-
dard checklist of suggestionsÐ for example, ª provide pricing ments of the final advertisement, of the worker’ s ability, and
information,º or ª show people from the target group.º of managerial performance.
It is possible that final advertisements that were more
ª responsiveº to the intermediate feedback would be evalu- Evaluation of the advertisement. On four 7-point
ated more highly simply for this reason. To check for this scales ranging from 1 (poor) to 7 (outstanding), participants
possibility and ensure that it was not affecting our results, rated (a) the creativity and originality, (b) interest level, (c)
we redid the analyses to be reported hereafter including an- demonstrated ª business sense,º and (d) overall quality of the
swers to a question assessing how responsive the final ad- final advertisement. Responses to these items were combined
vertisement was to feedback as a covariate (obviously this to form an overall evaluation index that had a Cronbach’ s al-
question was asked only for those participants who were pha of .72.
permitted to provide intermediate feedback). There were no
significant effects. This is possibly because the final adver- Evaluation of worker ability. The worker’ s ability to
tisement was intentionally constructed to be improved in develop marketing concepts was rated on a 7-point scale
some fundamental ways from the intermediate advertise- ranging from 1 (very low ) to 7 (very high).
ment along dimensions that most subjects would pay at
least some attention to, which restricted the variation in the Evaluation of managerial performance. Manage-
suggestions offered. rial effectiveness was assessed on a 7-point scale ranging
During the time that the worker was supposedly generat- from 1 (very ineffective) to 7 (very effective).
ing the ad, participants provided demographic data on them-
selves and filled out a variety of questionnaires unrelated to
the study. After 40 min, participants in the Lowest Involve- RESULTS
ment condition were shown the final advertisement and
asked to evaluate it on a series of scales. For those in the In- To assess the degree of confirmation of our major hypothe-
termediate Involvement condition, the experimenter showed ses, we first performed an analysis of variance (ANOVA) on
them the first draft after approximately 20 min, had them re- the evaluation of advertisement index. That analysis pro-
spond to it using the standard checklist, then approximately duced only three effects that approached conventional signif-
20 min later returned with the final draft and the evaluation icance, each providing support for one of those hypotheses.
scales. For those in the Highest Involvement condition, the First, the existence of a faith in supervision bias in the data
experimenter showed them the first draft after approximately was clearly demonstrated by a main effect for Supervisory
20 min and had them respond to it using the standard check- Involvement, F(2, 264) = 12.67 , p < .001, reflecting increas-
list, which w as then ª deliveredº to the worker in the other ingly positive evaluations of advertisement quality as the de-
room. About 20 min later, the experimenter returned with the gree of supervisory involvement increased.
318 PFEFFER, CIALDINI, HANNA, KNOPOFF

Next, the magnifying impact of self-enhancement pro- outcomes of that analysis closely mirrored those of the analy-
cesses on these quality judgments was evident in a large main sis of advertisement quality: There were significant main
effect for Self/Other Managerial Role, F(1, 265) = 64.41 , p < effects for both the Supervisory Involvement factor, F(2,
.001, which showed a strong tendency for participants to ele- 263) = 5.48, p < .01, and the Self/Other Managerial Role fac-
vate their views of the ad when theyÐ rather than an- tor, F(1, 264) = 31.03, p < .001, plus a marginally significant
otherÐ had supervised its production. What is more, this interaction between Supervisory Involvement and Manage-
tendency was present in every relevant comparison within rial Orientation, F(2, 263) = 2.21 p < .12. Figure 3 presents
our design. That is, the mean evaluation index score for each the means relevant to these analyses. No other effects proved
of the Self conditions was significantly higher than its coun- significant.
terpart Other condition mean score, all ps < .03. A third ANOVA was conducted on our measure of man-
Finally, the capacity of employee empow erment prac- agerial effectiveness. It produced only one significant out-
tices to limit these biases was demonstrated in a marginally comeÐ a main effect for the Supervisory Involvement
significant Supervisory Involvement × Managerial Orienta- factor, F(2, 262) = 45.68 , p < .001, such that ratings of
tion interaction, F(2, 264) = 2.33, p < .10. The form of that managerial effectiveness were higher as supervisory in-
interaction is depicted in Figure 2. It shows that the faith in volvement increased. How ever, as can be seen in Figure 4,
supervision effect was curtailed (for both observers and this effect was due almost entirely to the more favorable
managers) at the highest level of supervisory involvement ratings at the highest level of supervisory involvement. No
only when the supervision occurred within a team-based other effect was significant.
managerial context. An examination of Figure 2 suggests
that this curtailment of the faith in supervision effect was
stronger in the Self- than the Other Managerial Role condi-
tions. To assess the reliability of this difference, we per-
formed an orthogonal comparison that tested the interaction
between the Self/Other Managerial Role factor and the
Managerial Orientation factor solely within the highest
level of supervisory involvement. That comparison proved
significant, F(1, 265) = 4.70, p < .05. Simple effects tests
within this interaction demonstrated that the curtailment ef-
fect was robust primarily when raters were actively in-
volved in the supervision of the work. That is, the
difference between the hierarchical and team-based mana-
gerial orientation cells was significant for the Self condition
participants, F(1, 265) = 4.34, p < .05, but not for the Other
condition participants, F < 1.
No other main or interaction effects were significant.
However, an orthogonal comparison testing the Self/Other
FIGURE 2 Mean ratings of final draft advertisement quality.
Managerial Role × Managerial Orientation interaction solely
within the highest level of supervisory involvement was sta-
tistically significant, F (1, 268) = 4.09, p < .05. Simple tests
within this interaction demonstrated that the difference be-
tween the hierarchical and team-based managerial cells was
significant for the Self condition participants, F (1, 268) =
4.86, p < .05 but not for the Other condition participants, F <
1. These results precisely parallel those for the advertisement
quality ratings, and provide additional evidence that the cur-
tailment of the faith in supervision effect under more
team-like conditions occurs primarily in the Self Managerial
Role, which is w here the self-enhancement and involvement
bias is the strongest.

Secondary Analyses

To explore the impact of the experimental procedures on par-

ticipants’ evaluations of worker competence, we performed a
similar three-way ANOVA on ratings of worker ability. The FIGURE 3 Mean ratings of worker’ s ability.
 FAITH IN SUPERVISIO N 319

assessed through an examination of the baseline control

condition of our study. Recall that this condition was designed
to provide a relatively dispassionate evaluation of the final ad-
vertisement by asking for quality judgments from participants
who knew nothing about the circumstances under which the ad
had been produced except that an MBA student like them-
selves had generated it over a 40-min period.
As can be seen in Figure 2, virtually to a one, the ad
quality ratings of experimental participants were displaced
upward and aw ay from those made from the more objective
perspective of the control condition participants. Plainly,
the bias associated with greater supervisory involvement in
our data is toward exaggerated quality appraisals. Notable
is the size of the difference between the control group’ s
mean judgment and that of the participants most actively
and individually involved in supervisionÐ those in the self
FIGURE 4 Mean ratings of managerial effectiveness. role, highest supervisory involvement, hierarchical orienta-
tion condition. The magnitude of the distortion is in excess
of 100% (2.34 – 5.03 on a 7-point scale), changing an objec-
DISCUSSION tively weak performance into a subjectively strong one!
This finding points to the central importance of using inde-
Several conclusions seem warranted from the data of this pendent or objective measures of work quality whenever
study. First, as managers become more involved in controlling possible. For accuracy’ s sake, the more managers are in-
the work process (through traditional forms of oversight such volvedÐ and therefore investedÐ in the work, the more
as surveillance and feedback), crucial evaluations of the re- they should be separated from the critical evaluation of it.
sulting work product will likely be distorted upward. In this Third, the introduction of employee empowerment ori-
experiment, as managers’ individualized supervisory respon- entation reduces the biased upward valuation of work.
sibilities seemed to provide them with progressively greater Thus, at the highest level of supervisory involvement, ex-
involvement in the development of a print advertisement, they perimental participants who encountered a team-based
and external observers saw the identical final ad as increas- (rather than a hierarchical) orientation to managerial duties
ingly better. We take this to be evidence of a basic faith in su- resisted the progressive pull of the faith in supervision ef-
pervision effect, in which work performed under supervision fect, even and especially when the self w as implicated. It is
is presumed superior to that done without such oversight. worth noting that the circumstances of our highest supervi-
Ironically, this faith is frequently misplaced in that creative sory involvement condition are closest to those of typical
and intellectual activity is often undermined by close supervi- managerial responsibility, which require that the manager
sion (Deci & Ryan, 1985). provide final evaluations as well as prior surveillance and
Second, the overvaluation of individually supervised work feedback concerning the work. This pattern of findings of-
is greater for those who provide the supervision than for those fers one possible explanation for the puzzling reluctance of
who merely witness it. Our participants who actively super- U.S. companies to replace traditional command and control
vised a worker’ s efforts biased their perceptions of the out- managerial approaches with empowerment practices that
come of those efforts to a much greater degreeÐ at all levels of delegate more responsibility for work to those who perform
supervisionÐ than did external observers. Self-aggrandizing it. By virtue of psychological propensities to overvalue
perceptions of this sort are perhaps not surprising given exten- work done under focused supervision, especially one’ s own
sive evidence of self-enhancement tendencies in much of hu- supervision, traditional practices that concentrate responsi-
man functioning (Myers, 1996). bility in the hands of managersÐ and therefore provide
Still, a critic might argue that the case w e have made for them with greater feelings of individual responsibility and
self-promotional bias is not complete. Throughout the presen- connection to the workÐ will lead them to heightened valu-
tation of results, w e have discussed the inflation of perceptions ations of the w ork product. It is understandable, then, that
of work quality. That is, we have implied that the participants managers would resist procedures that reduce their individ-
with more firsthand involvement and with more individualized ualized involvementÐ as those procedures are likely to de-
supervisory involvement overvalued the resultant work. But, it flate their honestly felt, but misguided, perceptions of work
is possible thatthese participants actually valued the work prod- quality.
uct more accurately, and that it was the participants with less in- We also examined the impact of our experimental ma-
volvement who distorted their judgmentsÐ downwardÐ from nipulations on perceptions of worker ability, which proved
an objective appraisal of work quality. Such a possibility is best highly similar to those of work product quality. Apparently,
320 PFEFFER, CIALDINI, HANNA, KNOPOFF

participants viewed credit for seemingly positive work out- ACKNOWLEDGMENT

comes in nonzero-sum termsÐ not expropriating credit in-
creasingly for themselves but assigning greater credit to the We gratefully acknowledge the support of the Human Re-
worker’ s ability as their estimates of work product quality sources Initiative at the Graduate School of Business, Stan-
rose. These highly parallel data patterns suggest that when ford University.
considering the question of worker ability, our participants
focused more on the perceived final outcome of the work REFERENCES
than on the quality of managerial supervision associated
with it. Indeed, evaluations of worker ability were much Arthur, J. B. (1994). Effects of human resourc e systems on manufacturing
performanc e and turnover. Academy of Management Journal, 37,
more related to evaluations of the work product (r = .79)
670 – 687.
than to evaluations of managerial effectiveness (r = .14). Bazerman, M. H., Beekun, R. I., & Schoorman, F. D. (1982). Performanc e
As can be seen in Figure 4, the ratings of managerial ef- evaluation in a dynamic context: A laboratory study of the impact of prior
fectiveness did not track those of our other measures. In- commitment to the ratee. Journal of Applied Psychology, 67, 873 – 876.
stead, they showed a strong jump at the highest level of Brown, J. D., & Rogers, R. J. (1991). Self-serving attributions: The role of physi-
ological arousal. Personality and Social Psychology Bulletin, 17, 501 –506.
supervisory involvement, where participants in all condi-
Chalykoff, J., & Kochan, T. A. (1979). Computer-aide d monitoring : Its in-
tions were equally favorable about the level of demonstrated fluence on employee job satisfaction and turnover. Personnel Psychol-
managerial effectiveness (all Fs < 1.03). Proponents of em- ogy, 42, 807 –829.
ployee empowerment approaches should find these data en- Chapman, M. L., Elhance, A. P., & Wenum, J. D. (1995). Mitsubishi Motors
couraging in that assigning participants team leader roles in Illinois: Global strategies, local impacts. Westport, CT: Quorum.
Cialdini, R. B. (1993). Influence: Science and practice. New York:
may have reduced their centrality in the managerial process,
HarperCollins.
but did not undercut their perceived effectivenessÐ an im- Cialdini, R. B., Braver, S. L., & Lewis, S. K. (1974). Attributional bias and
portant component of morale and job satisfaction. the easily persuade d other. Journal of Personality and Social Psychology,
30, 631 – 637.
Cialdini, R. B., & Mirels, H. L. (1976). Sense of personal control and attribu-
tions about yielding and resisting persuasion targets. Journal of Personal-
ity and Social Psychology, 33, 395 – 402.
CONCLUSION Deci, E. L., & Ryan, R. M. (1985). Intrinsic interest and self-determination
in human behavior. New York: Plenum.
Most of the literature on the effects of supervision and sur- Drago, R. (1988). Quality circle survival: An exploratory analysis. Industrial
veillance has focused on the recipients of those practices Relations, 27, 336 –351.
(e.g., Deci & Ryan, 1985; Enzle & Anderson, 1993 ; Lepper Enzle, M. E., & Anderson, S. C. (1993). Surveillant intentions and intrinsi c
motivation. Journal of Personality and Social Psychology, 64, 257 – 266.
& Greene, 1975 ; but see Kipnis, 1993). Similarly, the focus
Fiske, S. T., & Taylor, S. E. (1994). Social cognition. New York:
of most studies of work practice change has been on the ef- McGraw-Hill.
fects of such change on the productivity, turnover, attitudes, Fletcher, G. J. O., & Ward, C. (1988). Attribution processes : A cross-cultural
or performance of those who experience the change. Al- perspective . In M. H. Bond (Ed.), The cross cultural challenge to social
though important, such a focus is seriously incomplete. If we psychology (pp. 230 – 244). Newbury Park, CA: Sage.
French, J. R. P. (1968). The conceptualization and the measurement of men-
are to understand why in most workplaces intensive manage-
tal health in terms of self-identity theory . In S. B. Sells (Ed.), The defini-
rial supervision persists in the face of a growing body of evi- tion and measurement of mental health (pp. 30 – 48). Washington, DC:
dence attesting to its ineffectiveness and dysfunctionality, Department of Health, Education, and Welfare.
then we need to understand the effects of various supervisory Guzzo, R. A., & Salas, E. (1995). Team effectiveness and decision-making in
practices on those who administer them. organizations. San Francisco: Jossey-Bass .
Headey, B., & Wearing, A. (1987). The sense of relative superiority Ð central
This study represents a first step toward such an under-
to well-being. Social Indicators Research, 20, 498 – 516.
standing. One reason it may be difficult for managers to dele- Heider, F. (1958). The psychology of interpersonal relations. New York:
gate responsibility is that they may perceive work produced Wiley.
without their direct intervention as inferior to comparable Insko, C. A. (1984). Balance theory, the Jorden paradigm, and the Wiest tet-
work produced under their direction. In this study, participants rahedron. In L. Berkowitz (Ed.), Advances in experimental social psy-
chology (Vol. 18, pp. 89 – 140). New York: Academic.
viewed the same print advertisementas superior when they ap-
Johns, G. (1993). Constraints on the adoption of psychology-base d personne l
parently had more individualized supervisory involvement in practices: Lessons from organizational innovation. Personnel Psychol-
the process that produced it. Similarly, they evaluated their ogy, 46, 569 –592.
supervisee’ s abilities and their own managerial effectiveness Karau, S. J., & Williams, K. D. (1993). Social loafing: A meta-analytic re-
more highly as well. If, when engaging in more intervention view and theoretical integration. Journal of Personality and Social Psy-
chology, 65, 681 – 706.
and supervision, one thinks the w ork is better, one’ s subordi-
Katzenbach, J. R., & Smith, D. K. (1993). The wisdom of teams: Creating the
nates are better, and one’ s management efforts are better, it is high performance organization. Boston: Harvard Business School Press.
scarcely surprising that the simple prescription ª delegate Kenney, M., & Florida, R. (1993). Beyond mass production: The Japanese
moreº is likely to have limited behavioral impact. system and its transfer to the U.S. New York: Oxford University Press.
 FAITH IN SUPERVISIO N 321
Kinlaw, D. C. (1991). Developing superior work teams. Lexington, MA: O’ Reilly, C. (1989). Corporations , culture, and commitment: Motivation and
Lexington. social control in organizations . California Management Review, 31, 9 –25.
Kipnis, D. (1993). Unanticipate d consequence s of using behavioral technol- Pfeffer, J. (1994). Competitive advantage through people. Boston: Harvard
ogy. Leadership Quarterly, 4, 149 –171. Business School Press.
Kochan, T., & Osterman, P. (1994). The mutual gains enterprise. Boston: Pyszczynski, T., Greenberg , J., & Holt, K. (1985). Maintaining consistenc y
Harvard Business School Press. between self-serving beliefs and available data. Personality and Social
Lawler, E. E., III, Mohrman, S. A., & Ledford, G. E., Jr. (1992). Employee in- Psychology Bulletin, 11, 179 –190.
volvement and total quality management. San Francisco: Jossey-Bass . Samuelson, C. D., & Allison, S. T. (1994). Cognitive factors affecting the use
Lepper, M. R., & Greene, D. (1975). Turning play into work: Effects of adult of social decision heuristic s in resource-sharin g tasks. Organizational Be-
surveillanc e and extrinsic rewards on children’ s intrinsi c motivation. havior and Human Decision Processe s, 58, 1 – 27.
Journal of Personality and Social Psychology, 31, 479 – 486. Schoorman, F. D. (1988). Escalation bias in performanc e appraisals: An un-
Levine, D. I., & Tyson, L. D. (1990). Participation, productivity , and the intended consequenc e of supervisor participation in hiring decisions.
firm’ s environment. In A. S. Blinder (Ed.), Paying for productivity: A Journal of Applied Psychology, 73, 58 – 62.
look at the evidence (pp. 183 – 237). Washington, DC: Brookings . Schoorman, F. D., Mayer, R. C., & Davis, J. H. (1996, April). Empowerment in vet-
MacDuffie, J. P. (1995). Human resourc e bundle s and manufacturing perfor- erinary clinics: The role of trust in delegation. Paper presented at the meeting of
mance: Flexible production systems in the world auto industry . Industrial the Society for Industrial and Organizational Psychology, San Diego, CA.
and Labor Relations Review, 48, 197 –221. Sedikides, C. (1993). Assessment, enhancement, and verification determi-
Meindl, J. R. (1990). On leadership: An alternative to the conventional wis- nants of the self-evaluation process. Journal of Personality and Social
dom. In B. M. Staw & L. K. Cummings (Eds.), Research in organiza- Psychology, 65, 317 –338.
tional behavior (Vol. 12, pp. 159 – 203). Greenwich, CT: JAI. Shepperd, J. A. (1993). Student derogation of the Scholastic Aptitude Test:
Meindl, J. R. (1993). Reinventing leadership: A radical, social psychological Biases in perceptions and presentations of college board scores. Basic and
approach. In J. K. Murnighan (Ed.), Social psychology in organizations Applied Social Psychology, 14, 455 – 473.
(pp. 89 –118). Englewood Cliffs, NJ: Prentice Hall. Simonson, I., & Nye, P. (1992). The effects of accountability on susceptibil -
Meindl, J. R., & Ehrlich, S. B. (1987). The romanc e of leadership and the ity to decision errors . Organizational Behavior and Human Decision Pro-
evaluation of organizational performance . Academy of Management cesses, 51, 416 – 446.
Journal, 30, 91 – 109. Skitka, L. J., Mosier, K., & Burdick, M. D. (1996, July). Accountability and
Meindl, J. R., Ehrlich, S. B., & Dukerich, J. M. (1985). The romanc e of lead- automation bias. Paper presente d at the American Psychological Society
ership. Administrative Science Quarterly, 30, 78 –102. convention, San Francisco.
Messe, L. A., Kerr, L., & Sattler, D. N. (1992). ª But some animals are more Strickland, L. H. (1958). Surveillance and trust. Journal of Personality, 26, 200–215.
equal than othersº : The supervisor as a privileged status in group contexts. Szymanski. K., & Harkins, S. G. (1987). Social loafing and self-evaluation with a
In S. Worchel, W. Wood, & J. A. Simpson (Eds.), G roup processes and social standard. Journal of Personality and Social Psychology, 53, 891 –897.
productivity (pp. 118 – 131). Newbury Park, CA: Sage. Tetlock, P. E. (1985). Accountability and complexity of thought. Journal of
Miller, D. T., & Ross, M. (1975). Self-serving bias in the attribution of cau- Personality and Social Psychology, 45, 74 – 83.
sality: Fact or fiction. Psychological Bulletin, 82, 213 – 225. Womack, J. P. (1996). The psycholog y of lean production. Applied Psychol-
Myers, D. G. (1996). Social psychology. New York: McGraw-Hill . ogy: An International Review, 45, 119 – 121.