leader_reward_and

An examination of subordinate-perceived relationships between leader reward and punishment behavior and leader bases of power
Human Relations
New York
Jul 1994

--------------------------------------------------------------------------------

Authors: Hinkin, Timothy R

Authors: Schriesheim, Chester A

Volume: 47

Issue: 7

Start Page: 779

ISSN: 00187267

Subject Terms: Subordinates
Studies
Statistical analysis
Perceptions
Hypotheses
Human relations
Behavioral sciencesSupervision
Social psychology
Human resource management
Behavior

Classification Codes: 9190: US
9130: Experimental/theoretical treatment
2500: Organizational behavior

Geographic Names: US

Abstract:

A study examined the relationships among subordinate perceptions of supervisory
(leader) reinforcement bahaviors (contingent and noncontingent reward and
punishment behavior) and the 5 French and Raven (1959) bases of power (reward,
coercive, legitimate, expert, and referent power). Based upon the literature,
5 sets of hypothesized relationships were developed, relating the perceived
power bases to the perceived reward and punishment behaviors. Bivariate
correlation, stepwise multiple regression, and hierarchical multiple regression
analyses were performed on data obtained from surveys conducted in 2 large
samples. The results supported the positive effects of subordinate-perceived
supervisor contingent reward and punishment behavior and the negative effects
of noncontingent punishment on subordinate perceptions of their supervisor's
power. Not supported were the moderating effects of time spent in interaction
or time under the direction of the supervisor on these relationships.

Copyright Plenum Publishing Corporation Jul 1994

Full Text:

Timothy R. Hinkin(1,3) and Chester A. Schriesheim(2)

INTRODUCTION

Power is the ability of one person (the agent) to cause another person
(the target) to act in accordance with the agent's wishes (Bass, 1990).
Power is unobservable (it is a "construct") and, consequently, it can only
be measured in an abstract sense. However, the acquisition and maintenance
of power is often seen as one of the most important processes that occur
in organizations (McClelland & Burnham, 1976). Two primary sources of power
exist in organizations, one that is granted by a person's role or position
and a second that is personal and therefore unique to the individual (Biggert
& Hamilton, 1984; Pfeffer, 1981). Inherent in every position in an organization
is a certain degree of power, and individuals have the potential to either
enhance or decrease their power by virtue of their on-the-job behavior
(Kotter, 1985).

Despite the centrality of power to supervisory or "leadership" processes,
little is known about the effects of managerial behaviors on subordinate
perceptions of a leader's power (Hollander, 1985; Yukl, 1989a). Although
it has been suggested that an individual's behavior may have a strong effect
on perceptions of power (Gioia & Sims, 1983), little research has been
done on such antecedents of power (Yukl, 1989b). As Mintzberg (1983) points
out, researchers hesitate to venture into organizations to study power
directly and, instead, tend to examine power from a "macro" perspective.
As a result, much of the research on power has focused on controlling strategic
contingencies or obtaining a critical position in the organization (e.g.,
Brass, 1984; Hickson, Hinings, Lee, Schneck, & Pennings, 1971; Pfeffer
& Salancik, 1978).

Although the ability to exercise power over others throughout an organization
is very important, of possibly greater importance to both organizations
and their managers are the power relationships which exist between managers
and their immediate "direct reports" or "subordinates." This is because
managers must first obtain cooperation and compliance from their subordinates
in order to be effective (Yukl, 1989a). Since subordinates react to the
behaviors and attributes of their managers, as they perceive them, it seems
reasonable to expect that perceived supervisory behaviors will thus affect
subordinates' perceptions of their supervisor's power.

Calder (1977) suggests that to better understand the leadership process
it is important to understand the perceptions of followers. Consequently,
he develops a set of theoretical propositions to examine subordinate attributions
based on leader behavior. He proposes that leaders possess qualities that
are unobservable to followers (e.g., power), and that followers therefore
make attributions about these qualities on the basis of their perceptions.
Subordinates, by observing and cognitively processing the behavior of their
superiors, make attributions about the individual characteristics of their
leaders that influence future subordinate behaviors and attitudes. Raven
(1990) has developed a similar model that deals specifically with perceptions
of power.

Consistent with the Calder and Raven models, perceived leader behavior
has been theorized by others to have a significant effect on subordinate
perceptions of leader power, although much of this work proposes no specific
relationships (French & Raven, 1959; Gioia & Sims, 1983; Herold, 1977;
Kaplowitz, 1978; Levinger, 1959; Schopfler & Layton, 1974; Tjosvold, 1985;
Yukl, 1989a). The above considerations suggest that it would be both desirable
and worthwhile to conduct research that links perceived supervisory behavior
with perceived supervisory power constructs which are currently being used
in leadership research. Such research would be useful because the development
and exploration of relationships among variables is essential to increasing
our understanding of all phenomena, including leadership (Nunnally, 1978;
Schwab, 1980). It is therefore the purpose of this paper to extend this
earlier work and to begin integrating existing theory and research by investigating
relationships between perceived supervisory reinforcement behavior and
subordinate perceptions of supervisory power. To guide the current investigation,
five hypotheses are developed concerning the variables of interest, after
pertinent literature is first briefly reviewed.

REVIEW OF THE LITERATURE AND HYPOTHESES

REWARD AND PUNISHMENT BEHAVIOR

Although relatively new, one approach to leadership that has received increasing
attention involves conceptualizing leaders as agents or mediators of rewards
and punishments to subordinates. This approach has generated considerable
theoretical and empirical interest and measures with sound psychometric
properties have been developed and used in a number of studies (e.g., Podsakoff,
Todor, Grover, & Huber, 1984; Podsakoff, Todor, & Skov, 1982). A reasonable
amount of evidence has recently developed concerning relationships between
perceived supervisory reward and punishment behavior and subordinate satisfaction
and performance (Williams & Podsakoff, 1988). To date, however, few attempts
have been made to examine how these reinforcement behaviors relate to other
independent variable constructs that are commonly used in leadership research,
and virtually no research has explored linkages between supervisory reinforcement
behaviors and constructs in the domain of social or interpersonal power.

Obtained Relationships. The evidence that has accumulated suggests that
perceived leader behavior which provides subordinates with rewards contingent
on their performance ("contingent reward behavior"; CR) is strongly and
positively associated with desirable affective and behavioral states (e.g.,
subordinate satisfaction and performance), while noncontingent reward behavior
(NCR) has a small positive or no relationship. Punishment that is contingent
upon poor subordinate performance ("contingent punishment behavior"; CP)
has a small positive or, more commonly, no relationship with these dependent
variables,. while noncontingent punishment (NCP) tends to be very negatively
related (Podsakoff & Schriesheim, 1986; Williams & Podsakoff, 1988).

POWER

Power has been examined from a wide variety of perspectives (e.g., Kipnis,
1984; Pfeffer, 1977), but perhaps the most influential framework is that
of French and Raven (1959; Mintzberg, 1983). It has been proposed by French
and Raven (1959) that persons may possess five types of interpersonal power:
reward, coercive, legitimate, referent, and expert. As mentioned above,
some researchers have speculated that supervisory behavior might impact
upon subordinate perceptions of supervisory power, yet few specific predictions
have been proposed and almost no empirical work has been done to explore
such relationships. This may be because of researchers' hesitancy to empirically
examine power at the "micro" level in organizations (as suggested by Mintzberg,
1983) or because psychometrically sound measures of the bases of power
were not available until recently (Hinkin & Schriesheim, 1989; Podsakoff
& Schriesheim, 1985; Schriesheim, Hinkin, & Podsakoff, 1991a).

Obtained Relationships. Perceived referent and expert power have been shown
to be very positively related to subordinate affective states, while perceived
coercive power has been very negatively related. The relationships for
perceived reward and legitimate power have generally been positive or nonsignificant
(Hinkin & Schriesheim, 1989; Podsakoff & Schriesheim, 1985; Schriesheim
et al., 1991a).

REWARD AND PUNISHMENT BEHAVIOR AND POWER

Obtained Relationships. The paucity of research notwithstanding, a few
studies have examined relationships between managerial behaviors and subordinate
perceptions of power. One recent study found that managerial attempts to
gain subordinate compliance that are viewed as rational and job-related
were associated with increased perceived legitimate, expert, and referent
power. Behaviors viewed as involving personal assertiveness, or "strong-arm"
attempts to gain compliance were positively associated with perceived coercive
power and negatively associated with perceived expert and referent power.
On the other hand, the punitive use of organizational sanctions was negatively
associated with perceived legitimate, expert, and referent power (Hinkin
& Schriesheim, 1990). In a complementary earlier study, Biggart and Hamilton
(1984) found that managerial influence attempts that were perceived by
subordinates as not legitimate may result in decreases in perceived power.

Raven (1988) discusses various "stage setting devices," which are behaviors
designed to increase perceptions of power. For example, a compliment regarding
a subordinate's work may have the effect of increasing the manager's perceived
reward power. In this regard, it seems reasonable to expect that the use
of some types of reinforcement would be associated with an increase in
perceived leader power, and others with a decrease. Supporting this position
is previous research which found that leaders who reward subordinates contingent
on their performance will enhance their power (Marak, 1964). Greene and
Podsakoff (1981) also found that withdrawing performance-contingent rewards
from subordinates was associated with perceptions of reduced leader power,
and Gioia and Sims (1983) found managerial reward and punishment behavior
to be strongly related to perceived power. Finally, a number of studies
have found that subordinate satisfaction with performance-contingent feedback
is associated with the perceived credibility of the source (Bannister,
1986; Ilgen, Fisher, & Taylor, 1979).

HYPOTHESES

Reward Power. The ability to administer to another things he/she desires
or to remove or decrease things he/she does not desire (French & Raven,
1959). For supervisors in an organizational setting, it is the perceived
ability to provide subordinates with positively valued outcomes (Hinkin
& Schriesheim, 1989). Previous research has found that observed managerial
reward behavior is strongly associated with perceived reward power (Gioia
& Sims, 1983) and that the withdrawal of contingent rewards is associated
with reduced reward power (Greene & Podsakoff, 1981). It is therefore likely
that those managers who are seen as rewarding subordinates contingent on
their performance will also be perceived as possessing reward power. Consequently,
Hypothesis 1 is that:

Hypothesis 1. Perceived supervisory contingent reward behavior will have
a positive relationship with perceived supervisory reward power.

Coercive power. The ability to administer to another things he/she does
not desire or to remove or decrease things he/she does desire (French &
Raven, 1959; Hinkin & Schriesheim, 1989). Coercive Power has been shown
to be associated with punitive behavior that may be outside normal role
expectations (Hinkin & Schriesheim, 1989). Coercive power has also been
positively associated with generally punitive behavior and negatively related
to contingent reward behavior (Gioia & Sims, 1983). However, the positive
relationships that have been shown to exist between contingent punishment
and subordinate affective and behavioral responses (e.g., Podsakoff et
al., 1484) would suggest that no relationship should exist between contingent
punishment and coercive power. Thus, Hypothesis 2 is that:

Hypothesis 2. Perceived supervisory contingent reward behavior will have
a negative relationship with perceived supervisory coercive power, while
perceived supervisory noncontingent punishment will have a positive relationship
with perceived supervisory coercive power. (No relationship is hypothesized
for perceived contingent punishment behavior.)

Legitimate Power. The ability to administer to another feelings of obligation
or responsibility (Hinkin & Schriesheim, 1989). Rewarding and punishing
subordinates is generally seen as a legitimate part of the formal or appointed
leadership role and most managerial positions in work organizations carry
with them some degree of expected reward and punishment (Bass, 1990), Perceived
legitimate power has also been shown to be positively associated with perceptions
of rational leader behavior that is within role expectations (Hinkin &
Schriesheim, 1990). Thus, it may be predicted that supervisors who are
seen as differentially acting toward subordinates on the basis of their
performance are more likely to be perceived as having legitimate power.
Consequently, Hypothesis 3 is that:

Hypothesis 3. Both perceived supervisory contingent reward and contingent
punishment behavior will be positively associated with perceived supervisory
legitimate power.

Expert Power. The ability to administer to another information, knowledge,
or expertise (French & Raven, 1959; Hinkin & Schriesheim, 1989). It would
be expected that leaders who possess expert power would have an understanding
of various organizational tasks and functions, and therefore be able to
acknowledge and reward good performance and punish poor performance. Consequently,
leaders who demonstrate contingent reward and punishment behaviors are
more likely to be viewed as possessing at least some expert power. At the
same time, leaders who punish noncontingently would more likely be seen
by subordinates as lacking expertise. Thus, Hypothesis 4 is that:

Hypothesis 4. Perceived supervisory contingent reward and contingent punishment
behaviors will be positively associated with perceived supervisory expert
power, while perceived supervisory noncontingent punishment will have a
negative relationship with perceived supervisory expert power.

Referent Power. The ability to administer to another feelings of personal
acceptance or approval (Hinkin & Schriesheim, 1989). Leaders who possess
referent power may have subordinates who identify strongly with their leader,
see him or her as role model, and emulate the leader's behavior (Raven,
1988). Prior research has shown perceived referent power to have a strong
positive relationship with perceptions of logical and reason-based leader
behavior, and a strong negative association with perceived punitive behavior
(Hinkin & Schriesheim, 1990). However, since noncontingent punishment behavior
seems strongly disliked by subordinates (Williams & Podsakoff, 1988), it
seems unlikely that leaders who are seen as engaging in noncontingent punishment
behavior will be seen as possessing referent power. Thus, Hypothesis 5
is that:

Hypothesis 5. Perceived contingent supervisor reward behavior will have
a positive relationship with perceived supervisory referent power, while
perceived supervisory noncontingent punishment will have a negative relationship
with perceived supervisory referent power.

METHOD

SAMPLES AND PROCEDURES

Samples. Two samples, one intra-and one inter-organizational, were utilized
to enhance the generalizability of results and to reduce the likelihood
of obtaining artifactual findings. In both samples, the respondents were
asked to answer survey questionnaires to report (1) brief demographic information,
(2) their perceptions of their immediate or direct supervisor's reward
and punishment behavior, (3) their perceptions of their immediate or direct
supervisor's bases of power, and (4) their assessment of their own satisfaction
with supervisor human relations and technical abilities. The first sample
consisted of 375 full-time employees of a state-operated southern U.S.
hospital. Job categories included nurses, clinical technicians, clerical
staff, and administrators. Seventy-three percent were female, the average
time worked under the present supervisor was approximately 2 years. Over
90% were high school graduates, and 40% had college degrees or graduate
training. The second sample consisted of 297 mid-level managers employed
at over 100 U.S. financial institutions. The average age of these respondents
was 32.3 years old and 69% were male. These respondents' average tenure
in the profession was 4.3 years, and they had worked with their current
supervisors slightly less than 1 year. Over 90% of these respondents were
college graduates, and all had at least some college education.

Procedures. Survey questionnaires were pretested in both samples, using
small groups of respondents (about one dozen in each sample; the pretest
participants did not participate in the final data collection). As a consequence
of the pretesting, relatively minor modifications were made in the written
instructions and in several of the demographic items. The revised survey
was then administered to the hospital respondents in their natural work
settings, during normal work hours. The banking respondents completed their
questionnaires at the beginning of a 2-week residential industry-sponsored
management effectiveness program. For both samples, the same written instructions,
along with brief oral presentations, were given to assure the respondents
complete anonymity protection and to explain (in broad terms) the purpose
of the research. The participants were all given the opportunity to ask
questions and were encouraged to answer the surveys honestly; anonymity
was guaranteed and no names or other identifying information was asked.

MEASURES

Leader Reward and Punishment Behavior. Perceived leader reward and punishment
behavior was assessed by using the 23-item instrument developed by Podsakoff
and Skov (1980; Podsakoff, Todor, Grover, & Huber, 1984). It employs a
7-point Likert response format (from "Very strongly agree" to "Very strongly
disagree") and consists of four subscales that appear to be the psychometrically-soundest
measures that are currently available (Podsakoff & Schriesheim, 1986);
this measure has demonstrated a very stable factor structure and acceptable
levels of internal consistency reliability in many previous samples (Podsakoff
et al., 1984; Podsakoff & Schriesheim, 1986; Schriesheim et al., 1991b).
Contingent reward behavior (CR, ten items) measures the degree to which
a leader is perceived as utilizing positive reinforcers, such as recognition
or commendation, contingent on desired subordinate performance. Noncontingent
reward behavior (NCR, three items) is a measure of the degree to which
a leader is seen as rewarding subordinates regardless of their performance.
Contingent punishment (CP, five items) measures the degree to which a leader
is viewed as using punishment, such as disapproval and reprimands, contingent
upon poor performance. Noncontingent punishment (NCP, five items) assesses
the degree to which a leader is perceived as administering punishment independent
of subordinate performance.

Leader Power. The respondents' perceptions of leader power were assessed
by Hinkin and Schriesheim's (1989) measures of the five French and Raven
(1959) bases of power. These scales employ a 5-point Likert response format
(from "Strongly agree" to "Strongly disagree") and contain four items each;
these measures have thus far demonstrated content validity, discriminant
validity, excellent factor structures, good internal consistency reliabilities,
and significant concurrent validities (Hinkin & Schriesheim, 1989).

Satisfaction. Satisfaction has been an important dependent variable in
leadership studies, and relationships with satisfaction were investigated
for the perceived leader reinforcement behaviors, as well as for the perceived
bases of power, to assess the consistency of findings with prior research.
Thus, satisfaction with supervision was assessed by the supervisor-technical
and the supervisor-human relations satisfaction subscales of the Minnesota
Satisfaction Questionnaire (MSQ; Weiss, Dawis, England, & Lofquist, 1967).
The MSQ has been carefully developed and refined, and substantial evidence
exists on its validity and reliability (Dunham, Smith, & Blackburn, 1977;
Gillet & Schwab, 1975; Weiss et al., 1967); it employs a 5-point response
scale which ranges from "Very satisfied to "Very dissatisfied."

Potential Moderator Variables. It has been suggested that perceptions may
change over time and that time spent with subordinates may affect the relationships
that exist between managers and subordinates (Ashford & Tsui, 1991). Time
worked under a manager and percent of work time spent in direct contact
might thus potentially affect perceptions and therefore moderate at least
some of the hypothesized relationships. Consequently, both variables were
assessed using 5-point scales, with time under the manager scaled at 1
= 0-3 months, 2 = 3-6 months, 3 = 6 months-1 year, 4 = 1-2 years, and 5
= greater than 2 years. Time in contact was assessed in increments of 20%
from 0-100% (i.e., 1 = 0-20%, 2 = 21-40%, etc.). Although allowing open-ended
answers and asking the respondents for exact times and percentages might
have been preferable, these two questions were included in the demographic
section of the surveys (which appeared at the beginning), and the pretests
in the two samples suggested that the respondents might have felt threatened
if they had been asked for more precise (and possibly identifying) answers.
Thus, the five response categories which were used were developed from
these pretests so as to obtain about equal numbers of respondents in each
of the five categories (the full-sample results did yield approximately
equal numbers of respondents selecting each of the five responses to these
two questions).

ANALYTICAL PROCEDURES

Factor Analyses. Both the reinforcement and power instruments have demonstrated
excellent factor structures in previous studies (Hinkin & Schriesheim,
1989; Podsakoff et al., 1984; Schriesheim et al., 1991b). To examine factor
structure in the current study, the items from both instruments were separately
factor-analyzed (in each sample separately), using principle axis factoring,
with squared multiple correlations as initial communality estimates and
a varimax rotation. In both samples, most of the items had loadings on
their appropriate factors in excess of .60; all were in excess of the .40
criterion level which appears most commonly used to interpret factor loadings
as meaningful (Ford, MacCallum, & Trait, 1986). Additionally, none of the
items had meaningful cross-loadings on other factors (none were greater
than .36), and in both samples over 70% of the total item variance was
explained by the four-factor solution for the 23 leader reward and punishment
items and by the five-factor solution for the 20 bases of power items.

Internal Consistency Reliability. Coefficient alpha internal consistency
reliability estimates were also computed for each individual scale and,
as a set, the scales demonstrated acceptable reliabilities (Nunnally, 1978).
They were as follows, for the bank and hospital samples, respectively:
reward power .72, .73; coercive power .85, .86; legitimate power .87, .85;
expert power .83, .87; referent power .82, .91; contingent reward .91,
.92; contingent punishment .80, .75; noncontingent reward .61, .63; noncontingent
punishment .64, .79; human relations satisfaction .91, .90; technical satisfaction
.91, .91.

Zero-Order Correlation Analyses. The relationships between satisfaction
with supervision and perceived leader reinforcement behavior and bases
of power were first assessed by computing variable means, standard deviations,
and zero-order Pearson intercorrelations. This was done to examine the
degree of similarity between the two samples and to compare the obtained
relationships with satisfaction with those found in prior research. Then,
the five hypotheses were each examined by the zero-order Pearson coefficients
between the perceived leader reinforcement behaviors and the perceived
leader bases of power.

Stepwise Regression Analyses. To further examine the hypotheses, a series
of exploratory multivariate analyses were performed using stepwise multiple
regression (because no a priori entry order could be theoretically justified
for the independent variables). For each of the five hypothesis, a separate
stepwise regression was performed, regressing those perceived leader reinforcement
behaviors which were involved in the hypothesis on each perceived base
of supervisory power; the criterion set for the addition of variables to
the regression was statistical significance at the .05 level.

Hierarchical Regression Analyses. After the stepwise regression analyses
were completed, a set of second exploratory regressions were conducted
for each hypothesis. These examined the potential moderating effects of
percent of time in contact with the supervisor or length of tenure under
the supervisor on relationships between perceived supervisory reinforcement
behavior and power. Conventional procedures for the testing of simple moderator
variable effects require a three-step hierarchical analysis (Cohen & Cohen,
1983). In the first step, the independent variable is entered in the regression,
followed by adding the moderator variable in the second step. The third
and final step then adds to the prediction equation a multiplicative cross-product
term which reflects the interaction of the independent and moderator variables.
Moderator effects are interpreted only if a significant increase in explained
variance is obtained at the third step (over that obtained at the second
step; Cohen & Cohen, 1983). This procedure was employed for each moderator
in each of the two samples (i.e., four sets of analyses were conducted).
However, a more complex moderator design was employed because the earlier
stepwise analyses had indicated that, in several instances, the best "predictor"
of a particular perceived supervisory power base was more than just one
of the perceived supervisory reinforcement behaviors (cf. Cohen & Cohen,
1983). Consequently, for these moderator analyses, the first step involved
entering into the regression a set of independent variables (as a block),
rather than one independent variable alone. This set consisted of those
perceived supervisory reinforcement behaviors which had been significant
predictors of the perceived power base in the earlier stepwise analyses.
The second and third steps then followed conventional simple moderator
effects procedures and added the main effect of the moderator variable
(step 2) and the interaction of the moderator and one of the independent
variables (step 3) (as further clarification, it should be highlighted
that in those analyses where more than one independent variable was entered
in the first step, all the different combinations of independent-moderator
variable interactions were tested at step 3). In keeping with conservative
data-analytic procedures, significance at step three was assessed at the
.01 level; this was to reduce the likelihood of finding significant moderator
effects due simply to the number of moderator analyses performed (Cohen
& Cohen, 1983).

RESULTS

SAMPLE SIMILARITY AND CONGRUENCE

Table I presents the means, standard deviations, and intercorrelations
of the independent and dependent variables. The means for reward power
and coercive power are larger for the banking sample (p < .01), but there
are no significant mean differences on any other measures. All the correlations
between perceived leader reinforcement behavior and power and subordinate
satisfaction are consistent with those of prior research, with the exception
of perceived noncontingent reward behavior (NCR) in the hospital sample
(cf. Podsakoff & Schriesheim, 1985; Williams & Podsakoff, 1988). Overall,
the correlation coefficients tend to be slightly higher for the hospital
sample, but the general correlational patterns of the two samples appear
remarkably similar.

BIVARIATE RESULTS

The Pearson correlations shown in Table I support Hypothesis 1, with perceived
supervisory contingent reward behavior (CR) and perceived supervisory reward
power (REWP) being positively correlated (.30 and .25, both p < .01) in
the bank and hospital samples (respectively).

Hypothesis 2, on the other hand, is only partly supported. As predicted,
perceived supervisory contingent reward behavior (CR) is negatively correlated
with perceived supervisory coercive power (COEP) in both samples (-.18
and -.33, p < .05 and p < .01 for the bank and hospital, respectively).
However, perceived supervisory noncontingent punishment (NCP) is unrelated
to perceived supervisory coercive power (COEP) in the bank sample (.07,
ns), while it is positively related (as hypothesized) in the hospital (.26,
p < .01).

The data are fully supportive with respect to Hypothesis 3. As can be seen
in Table I, perceived supervisory contingent reward behavior (CR) is positively
correlated with perceived supervisory legitimate power (LEGP) (.19 and
.33, both p < .01) in the bank and hospital, and perceived supervisory
contingent punishment behavior (CP) has a positive relationship with perceived
supervisory legitimate power (LEGP) as well (.20 and .34, both p < .01).

The more complex Hypothesis 4 is also fully supported. Here, Table I shows
that perceived supervisory contingent reward behavior (CR) is positively
related to perceived supervisory expert power (EXPP) in both the bank and
hospital (.34 and .64, both p < .01). Furthermore, perceived supervisory
contingent punishment behavior (CP) is significantly correlated with perceived
supervisory expert power (EXPP) in both samples as well (.16 and .31, p
< .05 and p < .01, respectively). Finally, perceived supervisory noncontingent
punishment behavior (NCP) is negatively related to perceived supervisory
expert power (EXPP) in the bank and hospital samples (-.31 and -.32, both
p < .01).

The final hypothesis, Hypothesis 5, is fully supported by the bivariate
analysis. Perceived supervisory contingent reward behavior (CR) has a positive
relationship with perceived supervisory referent power (REFP) in both samples
(.36 and .59, both p < .01), while perceived supervisory noncontingent
punishment (NCP) has negative relationships (-.32 and -.40, both p < .01).

STEPWISE REGRESSION RESULTS

The stepwise multiple regression analyses further suggest that Hypotheses
1, 3, and 5 have strong support. As shown in Table II, the regressions
for the dependent variables of perceived supervisory Reward Power, Legitimate
Power, and Referent Power show that their hypothesized perceived leader
reinforcement "predictors" are significant correlates when examined in
a multivariate framework. Furthermore, the signs of all the betas are in
the directions predicted by Hypotheses 1, 3, and 5, thus lending them further
support.

Table II shows Hypotheses 2 and 4 to be less supported, however, by the
stepwise regression analyses. For the bank sample, perceived supervisory
noncontingent punishment behavior is not a significant positive "predictor"
of perceived supervisory coercive power (Hypothesis 2), nor does perceived
supervisory contingent punishment serve as a significant positive "predictor"
of perceived supervisory expert power (Hypothesis 4). Supporting this later
finding is the hospital sample results, which do not show perceived supervisory
noncontingent punishment behaviors to negatively "predict" perceived supervisory
expert power. These results, then, mirror the bivariate findings in that
Hypotheses 1, 3, and 5 are well-supported and Hypothesis 2 is only partly
supported. The bivariate and multivariate results, on the other hand, are
discrepant with respect to Hypothesis 4, and suggest that future research
needs to begin to explore these relationships more thoroughly by routinely
employing multivariate analyses (such as multiple regression or, perhaps,
structural equations modeling; cf. Joreskog & Sorbom, 1989).

HIERARCHICAL REGRESSION RESULTS

None of the hierarchical regression analyses showed significant incremental
variance for the time interaction terms, thus clearly indicating that neither
of the two variables (time spent working for a manager and amount of manager-subordinate
contact time) could be considered moderators of the relationships specified
in Hypotheses 1-5. Consequently, the implications of this finding are briefly
discussed in the next section, along with those which follow from the bivariate
and stepwise analyses presented earlier. (Full tabular results reporting
the analyses with the two time variables are available from the authors
upon request.)

DISCUSSION

The focus of this study was on exploring relationships between supervisory
reinforcement behaviors perceived by subordinates and the subordinates'
perceptions of their supervisor's bases of power. Although the research
design which was employed does not allow the drawing of causal conclusions,
previous research evidence does, in fact, suggest that perceived supervisory
behavior impacts upon subordinate perceptions of a supervisor's power (e.g.,
Gioia & Sims, 1983; Greene & Podsakoff, 1981). Of course, further research
on the causal interactions of these two sets of variables appears desirable,
since it would also appear unwarranted to assume that such relationships
are simple and unidirectional.

Of the ten hypothesized relationships which were embedded in Hypotheses
1-5, five were fully supported by both the bivariate and at least partial
support. Furthermore, the results of the stepwise regression analyses demonstrated
that some of the perceived bases of power appear to be affected by several
perceived leader reinforcement behaviors, and the hierarchical moderator
regressions indicated that time as a manager's direct report and frequency
of supervisor-subordinate interaction were not moderators in the current
sample.

TIME AS A MODERATOR

Implications of the unmoderated findings are further considered below,
but the lack of time-as-a-moderator effects result suggests that subordinates
may form quick and relatively stable impressions about their managers,
a finding consistent with other literature on impression formation (e.g.,
Asch, 1946; Jones & Nisbett, 1971). If this is true, these results would
appear to suggest that managers should be concerned about the early impressions
which their subordinates form about them and therefore be attentive to
managing the impression formation process as best possible (as discussed
below, perhaps by emphasizing the use of contingent reward and punishment
behaviors). However, as a cautionary note, it should be mentioned that
both of the time variables were assessed in the current study by single-item
measures (measured on 5-point scales), so that range restriction and/or
measurement unreliability remain rival explanations for the obtained time-as-moderator
results. Replication of these findings is therefore clearly necessary before
one can feel confident about them. However, they do point to the importance
of early impressions in the development of supervisor-subordinate relations.

PERCEIVED CONTINGENT REWARD BEHAVIOR

Consistent with the hypotheses, the use of contingent rewards had the greatest
effect on enhancing all five bases of power and, particularly, on enhancing
what previous research would label as the more positively perceived bases
of power--expert and referent power (i.e., those perceived power bases
most positively correlated with subordinate affect; Podsakoff & Schriesheim,
1985). Since prior research has found contingent reward behavior to have
strong positive effects on subordinate satisfaction and performance (Williams
& Podsakoff, 1988), the current results clearly extend these findings and
further suggest that the use of contingent rewards can also have strong
effects on "positive power perceptions." In fact, what may be most surprising
is the potential magnitude of the effect of contingent reward behavior
on expert and referent power, since the amount of variance explained in
both samples of the current study was moderately high. The stronger relationships
that existed in the hospital sample may reflect a difference in organizational
culture, where subordinates are very sensitive to the feedback that they
receive and form strong impressions from their interactions with managers.
This is not surprising in a hospital setting, where quick decisions often
need to be made and performance outcomes may be more directly observable.

Perceived supervisory contingent reward behavior was also negatively related
to the perception of coercive power, thereby suggesting that not rewarding
good performance may be interpreted by subordinates as being punitive and
consequently as producing a negative impact on subordinates' perceptions
of their supervisors. This, again, highlights the importance for managers
of providing subordinates with contingent rewards, particularly since coercive
power is generally negatively viewed by subordinates (as evidenced by its
typically negative correlations with affective outcome variables; Podsakoff
& Schriesheim, 1985).

PERCEIVED CONTINGENT PUNISHMENT BEHAVIOR

As predicted, a supervisor's perceived contingent punishment behavior was
associated with enhanced perceptions of supervisory legitimate power, but
it did not have the predicted relationship with expert power for the bank
sample (although it nearly attained statistical significance). Punishment,
then, appears to enhance legitimate power when it is contingent on behavior;
this suggests that supervisors may not need to refrain from punishing poor
performances, subject to the caveat that punitive behavior must clearly
be seen by subordinates as resulting from their poor performances. However,
since previous research also suggests that perceived supervisory contingent
punishment behavior has only, at most, very modest positive relationships
with subordinate affect (Williams & Podsakoff, 1988), contingent punishment
behavior would appear to be best used by leaders in a judicious manner.

PERCEIVED NONCONTINGENT PUNISHMENT BEHAVIOR

The use of perceived noncontingent punishment had the predicted negative
relationship with referent power in both samples, but the hypotheses were
supported for decreasing expert power only in the bank sample; furthermore,
noncontingent punishment did not have the predicted positive relationship
with coercive power in the bank sample. Previous research has shown perceived
supervisory noncontingent punishment to have a very negative relationship
with subordinate affect (Williams & Podsakoff, 1988). However, it would
appear that the use of perceived supervisory noncontingent punishment may
have less consistent relationships with perceptions of power than one might
expect. This might be due to a number of factors, including the possibility
that on some occasions, noncontingent punishment may appear warranted to
subordinates (if it is displayed after they perform poorly) and at other
times it may appear unwarranted (if it is displayed after they perform
well). However, since perceived noncontingent punishment most often results
is strong negative affective and behavioral responses on the part of subordinates
(Williams & Podsakoff, 1988), the current research extends earlier findings
by suggesting that its effects on perceived power are either negative (as
in the case for referent power) or inconsistent (as for coercive and expert
power). Managers would therefore seem best advised to monitor subordinate
perceptions of their behavior closely and to refrain from displaying what
is perceived as noncontingent punishment behavior.

PERCEIVED NONCONTINGENT REWARD BEHAVIOR

No hypotheses were developed for perceived supervisory noncontingent reward
behavior. It has been previously shown that noncontingent reward behavior
has little effect on subordinate outcomes (Podsakoff et al., 1984), and
there was no reason to expect noncontingent rewards to have any effect
on perceptions of power. In the current study, perceived supervisory noncontingent
reward behavior had no significant correlations with the perceived power
bases in the bank sample (see Table I) and only modest correlations (the
highest being .21) in the hospital. These results, coupled with those reported
earlier in the literature therefore suggest that the use of noncontingent
reward behavior is not advisable for supervisors who wish to enhance their
effectiveness. Simply put, leaders or supervisors seem best advised to
concentrate on the display of performance-contingent behavior, particularly
on behavior which results in increased subordinate perceptions of their
contingent reward behavior.

CONCLUSION

The results of this study provide support for the general proposition by
Bass (1990) that perceived leader behavior and perceived leader power would
demonstrate significant associative relationships. The fact that the relationships
generally demonstrated consistency across bivariate and multivariate analyses
and two very diverse samples permits one to view the findings with some
degree of confidence. Although replication of all research results is certainly
desirable, the current study seems to highlight that there are behaviors
which managers can exhibit to provide desirable outcomes for subordinates
and, at the same time, enhance their power in the subordinates' eyes as
well. The results of this study therefore suggest that it would be beneficial
for managers to take advantage of the opportunities to provide contingent
rewards, to avoid the use of noncontingent punishments, and to not be hesitant
to use contingent punishment.

In addition to the above results, one general finding probably merits further
brief elaboration here, to highlight possible avenues for future research.
The importance of perceived legitimacy is an idea that has been supported
by several researchers and theorists (e.g., Barnard, 1938; Biggart & Hamilton,
1984; French & Raven, 1959). In a recent study, Yukl and Falbo (1991) found
that legitimate power is a very important source of influence in organizations.
Thus, investigations into the effects of the abuse or illegitimate use
of power, and further attempts to theoretically link research in this subdomain
to research in other subdomains appears desirable. In particular, it would
seem that future attempts to develop linkages between perceived power variables
and different streams of research could result in producing a body of knowledge
that would be of great value in better explaining and predicting interpersonal
relationships and behavior in organizations, as well as in providing us
with increased faith in the validity of the variables and measures that
we employ.

ACKNOWLEDGMENTS

The first author was supported during this research by a grant from the
Marriott Corporation; the second author was supported by the Summer Awards
in Business and the Social Sciences Program, University of Miami, and by
the Corporate Affiliates Program, School of Business Administration, University
of Miami.

NOTES

1 School of Hotel Administration, Cornell University, Ithaca, New York
14853-6901.

2 Management Department, School of Business Administration, 414 Jenkins
Building, University of Miami, Coral Gables, Florida 33124-9145.

3 Requests for reprints should be addressed to Timothy R. Hinkin, School
of Hotel Administration, Cornell University, Ithaca, New York 14853-6901.

REFERENCES

ASCH, S. E. Forming impressions of personality. Journal of Abnormal and
Social Psychology, 1946, 41, 259-290.

ASHFORD, S. J., & TSUI, A. S. Self regulation for managerial effectiveness:
The role of active feedback seeking. Academy of Management Journal, 1991,
34, 251-280.

BANNISTER, B. D. Performance outcome feedback and attributional feedback:
Interactive effects on recipient responses. Journal of Applied Psychology
1986, 71, 203-210.

BERNARD, C. The functions of the executive. Cambridge. MA: Harvard University
Press, 1938.

BASS, B. M. Bass & Stogdill's handbook of leadership (3rd ed.). New York:
Free Press, 1990.

BIGGERT, N. W., & HAMILTON. G. G. The power of obedience. Administrative
Science Quarterly, 1984, 29, 540-549.

BRASS, D. J. Being in the right place: A structural analysis of individual
influence in an organization. Administrative Science Quarterly, 1984, 29,
518-539.

CALDER, B. J. An attribution theory of leadership. In B. M. Staw and G.
R. Salancik (Eds.), New directions in organizational behavior. Chicago:
St. Clair Press, 1977.

COHEN, J., & COHEN, P. Applied multiple regression/correlation analysis
for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum
and Associates, 1983.

DUNHAM, R. B., SMITH, F. J., & BLACKBURN, R. S. Validation of the index
organizational reactions with the JDI, the MSQ, and Faces scales. Academy
of Management Journal, 1977. 20, 420-432.

FORD, J. K., MacCALLUM, R. C., & TAIT, M. The application of exploratory
factor analysis in applied psychology: A critical review and analysis.
Personnel Psychology, 1986, 39, 291-314.

FRENCH, J. R. P., RAVEN, B. The bases of social power. In D. Cartwright
and A. Zander (Eds.), Group dynamics. New York: Harper & Row, 1959.

GILLET, B., & SCHWAB, D. P. Convergent and discriminant validities of corresponding
Job Descriptive Index and Minnesota Satisfaction Questionnaire scales.
Journal of Applied Psychology, 1975, 60, 313-317.

GIOIA, D. A., & SIMS, H. P. Perceptions of managerial power as a consequence
of managerial behavior and reputation. Journal of Management 1983, 9, 7-26.

GREENE, C. N., & PODSAKOFF, P. M. Effects of withdrawal of a performance-contingent
reward on supervisory influence and power. Academy of Management Journal,
1991, 24, 527-542.

HEROLD, D. M. Two-way influence processes in leader-follow dyads. Academy
of Management Journal, 1977, 20, 224-237.

HICKSON, D. J., HININGS, H. R., LEE, C. E., SCHNECK, R. E., & PENNINGS,
J. M. A strategic contingencies theory of interorganizational power. Administrative
Science Quarterly, 1971, 16, 216-229.

HINKIN, T. R., & SCHRIESHEIM, C. A. Development and application of new
scales to measure the French and Raven (1959) bases of social power. Journal
of Applied Psychology, 1989 74, 561-567.

HINKIN, T. R., & SCHRIESHEIM, C. A. Relationships between subordinate perceptions
of supervisor influence tactics and attributed bases of power. Human Relations,
1990, 43, 221-237.

HOLLANDER, E. P. Leadership and power. In G. Lindzey and E. Aronson (Eds.),
Handbook of social psychology (3rd ed., Vol. 2). New York: Random House,
1985.

ILGEN, D., FISHER, C., & TAYLOR, S. Consequences of individual feedback
on behavior in organization. Journal of Applied Psychology, 1979, 64, 349-371.

JONES, E. E., & NIESBETT, R. E. The actor and the observer: Divergent perceptions
of the cases of behavior. Morristown, NJ: General Learning Press, 1971.

JORESKOG, K. G., & SORBOM, D. Lisrel 7 user's guide. Mooresville, IN: Scientific
Software, 1989.

KAPLOWITZ, S. A. Towards a systematic theory of power attribution. Social
Psychology, 1978, 41, 131-148.

KIPNIS, D. The use of power in organizations and interpersonal settings.
Applied Social Psychology Annual, 1984, 5, 179-210.

KOTTER, J. P. Power and influence: Beyond formal authority. New York: Free
Press, 1985.

LEVINGER, G. The development of perceptions and behavior in newly-formed
social relations. In D. Cartwright (Ed.), Studies in social power. Ann
Arbor: University of Michigan, Institute for Social Research, 1959.

MARAK, G. E. The evolution of leadership structure. Sociometry, 1964, 27,
74-182.

McCLELLAND, D. C., & BURNHAM, D. H. Power is the great motivator. Harvard
Business Review, 1976, 54, 100-110.

MINTZBERG, H. Power in and around organizations. Englewood Cliffs, NJ:
Prentice-Hall, 1983.

NUNNALY, J. C. Psychometric theory (2nd ed.). New York: McGraw-Hill, 1978.

PFEFFER, J. The ambiguity of leadership. Academy of Management Review,
1977, 2, 104-112.

PFEFFER, J. Power in organizations. Marshfield, MA: Pitman, 1981.

PFEFFER, J., & SALANCIK, G. R. The external control of organizations. New
York: Harper & Row, 1978.

PODSAKOFF, P. M., & SCHRIESHEIM, C. A. Field studies of French and Raven's
bases of power: Critique, reanalysis, and suggestions for future research.
Psychological Bulletin, 1985, 97, 387-411.

PODSAKOFF, P. M., & SCHRIESHEIM, C. A. Leader Reward and Punishment Behavior:
A Conceptual, Methodological, and Integrative Review. Unpublished paper.
Graduate School of Business, Indiana University, 1986.

PODSAKOFF, P. M., & SKOV, R. Leader Reward and Punishment Behavior Scales.
Unpublished research, Graduate School of Business, Indiana University,
1980.

PODSAKOFF, P. M., TODOR, W. D., GROVER, R. A., & HUBER, V. L. Situational
moderators of leader reward and punishment behavior: Fact or fiction? Organizational
Behavior and Human Performance, 1984, 34, 21-63.

PODSAKOFF, P. M., TODOR, W. D., & SKOV, R. B. Effects of leader contingent
and noncontingent reward and punishment behaviors on subordinate performance
and satisfaction. Academy of Management Journal, 1982, 25, 810-821.

RAVEN, B. H. French and Raven 30 Years Later: Power/Interaction and Interpersonal
Influence. Paper presented at the XXIV International Congress of Psychology,
Sydney, Australia, 1988.

RAVEN, B. H. Political applications of the psychology of interpersonal
influence and social power. Political Psychology, 1990, 11, 493-520.

SCHWAB, D. P. Construct validity in organizational behavior. In B. M. Staw
and L. L. Cummings (Eds.), Research in organizational behavior (Vol. 2).
Greenwich, CT: JAI Press, 1980, pp. 3-43.

SCHOPFLER, J., & LAYTON, B. Attributions of interpersonal power. In J.
T. Tedeschi (Ed.), Perspectives in social power. Chicago: Aldine, 1974.

SCHRIESHEIM, C. A., HINKIN, T. R., & PODSAKOFF, P. M. Can ipsative and
single-item measures produce erroneous results in field studies of French
and Raven's (1959) five bases of social power? An empirical investigation.
Journal of Applied Psychology, 1991, 76, 106-114.(a)

SCHRIESHEIM, C. A., HINKIN, T. R., & TETRAULT, L. A. The discriminant validity
of the leader reward and punishment questionnaire (LRPQ) and satisfaction
with supervision: A two-sample, factor analytic investigation. Journal
of Occupational Psychology, 1991, 64, 159-166.(b)

TJOSVOLD, D. Power and social context in superior-subordinate interaction.
Organizational Behavior and Human Decision Processes, 1985, 35, 281-293.

WEISS, D. J., DAWIS, R. V., ENGLAND, G. W., & LOFQUIST, L. H. Manual for
the Minnesota Satisfaction Questionnaire. Minneapolis: Industrial Relations
Center, University of Minnesota, 1967.

WILLIAMS, M. L., & PODSAKOFF, P. M. A meta-analysis of attitudinal and
behavioral correlates of leader reward and punishment behaviors. Southern
Management Association Proceedings, 1988, 161-163.

YUKL, G. A. Leadership in organizations (2nd ed.). Englewood Cliffs, NJ:
Printice-Hall, 1989. (a) YUKL, G. A. Managerial Leadership: A review of
theory and research. Journal of Management, 1989, 15, 251-290. (b)

YUKL, G. A., & FALBO, C. M. Importance of different power sources in downward
and lateral relations. Journal of Applied Psychology, 1991, 76, 416-423.

BIOGRAPHICAL NOTES

TIMOTHY R. HINKIN is an Associate Professor in the School of Hotel Administration
at Cornell University, Ithaca, New York. He holds an M.B.A. with specialization
in human resources management from Michigan State University and a PhD
in organizational behavior from the University of Florida. He has published
articles in the Journal of Applied Psychology and the Academy of Management
Journal, among other outlets, and his current research interests are in
power and influence in organizations.

CHESTER A. SCHRIESHEIM is Distinguished Professor of Management and Rosa
R. and Carlos M. de la Cruz Leadership Scholar at the University of Miami,
Miami, Florida. He has an MBA from Michigan State University and a PhD
in organizational behavior from the Ohio State University. He has published
articles in such outlets as the Journal of Applied Psychology and Psychological
Bulletin, and his primary research interests are in the study of leadership.