4ddada400547
TRANSCRIPT
-
8/10/2019 4DDADA400547
1/13
Perceived Leader Behavior as a
Function of Personality Characteristics
of upervisorsand Subordinates^
D O U G L A S E . D U R A N D
University of Missouri St. Louis
WALTER R. NORD
Washington University
Subordinate perceptions of supervisory initiation of
structure and consideration were studied as a function of
locus of control and Machiavellianism Persona lity was
found to be a major factor in predicting leader behavior
as perceived by subordinates Locus of control may be
an important personaiity dimension influencing thelead
ership process in organizations
The view of leadership as an interactional process (Gibb, 1969) has
achieved paradigmatic status in organizational psychology. While most of
the interactionist perspectives stress that leadership is a function of both
situational and personality factors, the strength of the reaction against the
trait app roa ch appears to have suppressed the study of personality factors.
Recently, Fleishman (1973) and Kerr, Schriesheim, Murphy and Stog-
dill (1974) have pointed to the need for more study of personality factors
in leadership. Several researchers, including Beer (1 9 6 6 ), Dessler (1 97 4)
and Evans (1974), have reported that personality characteristics of sub-
ordinates may act as moderator variables in the relationship of initiation of
structure to performance. While these studies have demonstrated the im-
portance of the personality characteristics of subordinates, only the study
by Evans provided any information about the effect of the subordinates'
personality characteristics on their perception of the leaders' styles.
A review of the literature revealed only four studies where the followers'
personalities have been related to their perceptions of a leader. First,
Haythorn, Couch, Haefner, Langham, and Carter (1956) focused directiy
-
8/10/2019 4DDADA400547
2/13
-
8/10/2019 4DDADA400547
3/13
1976 Volume 19 Number 3 429
Little empirical work about the influence of locus of control on the
supervisor relationship is available. In addition to the studies of Pryer and
Distefano and Evans, only a laboratory study by Goodstadt and Hjelle
(1 97 3) and a field study by Run yon (1 97 3) provided add itional direct
insight into the influence of locus of control on the supervisor-subordinate
relationship.
The second personality dimension chosen for study was Christie and
Geis' (1 9 7 0 ) construct of M achiavellianism. They described the ideal
type
Machiavellian as being cool, detached, logically oriented, and likely
to disregard the affective states of both himself and others. Moreover,
Machiavellian people are prone to establish structure in loosely structured
situations, advocate the use of guile and deceit in interpersonal relation-
ships,and hold an unflattering view of hum an natu re. W hile little use has been
made of Christie and Geis' scales in organizational research, Gemmill and
Heisler (1972) did find that Machiavellianism was negatively related to
the perceived opportunity to exercise formal control.
Perceived leader behavior was measured by the form XII version of
Stogdill 's (1963) leadership behavior description questionnaire (LBDQ).
Two facts about this instrument should be kept in mind. First, this instru-
ment measures a leader's structure and consideration by asking the sub-
ordinates to describe the behavior of their leader. Consequently, character-
istics of the followers can have very strong effects on what the LBDQ
measures. In fact, in a different context, Stogdili (20, p. 141) observed
tha t . . . the significance of con sidera tion and stru cture is to be exp lained,
not in terms of leadership, bu t in terms of followership. Second, as Evans
(1973) has observed, the measures of both structure and consideration
yielded by form XII are narrower in scope than results based on the
original LBDQ.
For the most part, LBDQ scores have been averaged to measure super-
visory behavior; possible systematic differences have been ignored. While
this procedure may be useful for many purposes, Blanchard (see Fleishman,
1973) suggests that averaging LBDQ scores may be misleading because it
obscures the influence of such things as differential treatment of individuals
by the same supervisor. The tendency to average across subordinates could
also obscure the effects of personality differences among subordinates on
the perception of leadership behavior. Consequently, in this study the
relationship of each supervisor-subordinate dyad was taken as the relevant
unit for analysis. In other words, it was assumed that, due to personality
and other differences among subordinates, each supervisor reacted dif-
ferently to each subordinate.
Moreover, the supervisor-subordinate relationship develops over time.
-
8/10/2019 4DDADA400547
4/13
-
8/10/2019 4DDADA400547
5/13
1976
Volume 19, Num ber 3
431
This procedure should have yielded 48 dyads. However, due to person-
nel changes, 14 m anagers did n ot have the same supervisor six mo nths after
completing the personality scales. Consequentiy, data from 34 supervisor-
subordinate dyads were useable for testing the hypotheses.
The LBDQ and mach V instruments were scored according to the
conventional scoring system for each instrument. The I-E scale was scored
using Rotter's (1966) procedure and the subscales of control ideology
(CI) and personal control (PC) of Gurin et al. Personal control is mea-
sured by items 9, 13, 15, 25, and 28 of Rotter's instrument. The salient
feature of these items is that they concern one's personal life (e.g., I have
influence over . . . ) rathe r than people in general (e.g., people have influence
over . . .). Control ideology includes items 6, 7, 10, 11, 16, 18, 20, and 23
of the Rotter scale. The CI subscale refers to a general belief or ideology
in which hard work, skill and ability are important determinants of success
in life (e.g., becoming a success is a matter of hard work, luck has littie
or nothing to do with it).
The items included in Rotter's total score but not in either of the sub-
scales were treated as a third subscale and will be referred to as I-E residual
(I-E res). The meaning of I-E res is less clear-cut than the other subscales.
Many of the items seem to be concerned with what Gurin et al. described
as system mod ifiability degree of belief that an investmen t of effort can
make a difference in the social system. While only two I-E res items (items
3 and 17) appeared on Gurin et al.'s factor of system modifiability, a
number of the other items seem to be concerned with the same type of
issues. For example, item 12 asked about the influence of the average
citizen on governmental decisions; item 22 asked whether political corrup-
tion could be eliminated by concerned people; item 29 asked whether the
people were really responsible for bad government. Thus, it appeared that
the residual items were tapping something akin to system modifiability.
Since multiple regression techniques were to be used, all analysis of the
locus of control data was based on the two subscales and the residual.
Simultaneous use of the total score and the subscales would, of course,
have resulted in a double-weighting being assigned to subscale items.
Analysis
Simple and multiple correlation coefficients were computed to test the
hypotheses. Hypothesis 3 was nondirectional; consequentiy a two-tailed
test of significance was appropriate. One-tailed tests were used for hy-
potheses 1, 2, and 4.
For the exploratory analysis to determine what combination of supervisor
and subordinate variables best predicted the LBDQ scores assigned to a
-
8/10/2019 4DDADA400547
6/13
432 Academy of Managem ent Journal September
The procedure is repeated until the addition of another independent variable
fails to significantly improve the explanatory power of the relationship.
Based on Blanchard's (Fleishman, 1973) comments on the problems
of averaging LBDQ scores and the assumption that a leader responds
differently to each subordinate, each superior-subordinate dyad was treated
as the relevant unit for analysis. Consequently, the measures of each super-
visor's personality were paired with each of the LBDQ scores given by his
subordinates. This procedure is more conservative than employing mean
LBDQ scores because the residual variation in the relationship between
supervisor me asure a nd subo rdinate score has a lower limit; 7? has an
upper limit less than one. In effect, the use of several scores has introduced
a noise eleme nt into the analysis which increases the susceptibility of the
tests to type I error.
It should be emphasized that the analysis in which a supervisor's per-
sonality score appears as an independent variable and a dimension from
the LBDQ is the only dependent variable yields a conservative result be-
cause the authors have chosen the supervisor-subordinate interaction as the
unit of analysis. Since each supervisor has only one personality score, the
variance in the independent variable is restricted. The effect of this restric-
tion is to reduce the predictive power of this variable. Consequently, the
tests of the hypotheses are susceptible to type I error.
RESULTS
Before testing the hypotheses, the correlations between the conceptually
dependent variables, structure and consideration, were calculated. These
variables were no t significantly correla ted {r .05 ,p .05); consequently
each of the hypotheses could be tested independently.
Hypotheses 1 and 2 predicted that initiation of structure would be posi-
tively related to both locus of control and Machiavellianism. The two
hypotheses were tested with the multiple regression equation. Since hy-
pothesis 1 related initiation of stru ctu re to locus of con trol and hypo thesis 2
similarly related Machiavellianism to initiation of structure, they must be
tested simultaneously. Presumably the respective explanatory variables
of these hypotheses are neither statistically nor theoretically independent.
If, for example, the association between locus of control and structure were
measured in a simple regression or correlation format, the relationship
would be misspecified, and the effect of Machiavellianism on structure
which was falsely attributed to structure would be proportionate to the
strength of association between Machiavellianism and locus of control.
Therefore, the appropriate medium for testing these two hypotheses is an
estimation equation specifying structure to be determined by the simulta-
-
8/10/2019 4DDADA400547
7/13
1976
Volume 19, Number 3
433
TABLE 1
Correlation Coefficients of Supervisor Locus of Control
and M achiavellianism with Initiation of Structure
Independent Variables
R
R
Change Simple r
Supervisor personal control
Supervisor control ideology
Supervisor I-E residual
Supervisor Machiavellianism
38
4
42
42
.14
.16
.18
.18
.14*
.02
.02
.00
.38*
.23
.18
.09
Supervisor initiation of structure was the depend ent variable.
* p < .05 (one-tailed)
visory locus of personal control and structure (r = .3 8, p < .02 5) was
statistically significant. Supervisors whose locus of personal control was
external tended to be perceived as initiating more structure than did in-
ternal supervisors. Personal control by itself accounted for 14 percent of
the variation in structure. Addition of the other two portions of the I-E
scale strengthened the prediction only minimally. Machiavellianism ac-
counted for almost no additional variation. These results provided some
support for hypothesis 1, but none for hypothesis 2.
Hy pothes is 3 pred icted an association (n o directio n specified) betw een
locus of control and consideration; hypothesis 4 predicted a negative re-
lationship between Machiavellianism and consideration. These hypotheses
were also tested in a multiple regression format; the results are summarized
in Table 2. The simple correlation coefiScients between consideration and
personal control and I-E res were positive; however, only the association
of cons ideration with the I-E res was statistically significant (r = .3 6, p