Twin Research (1998) 1, 216–223

© 1998 Stockton Press All rights reserved 1369–0523/98 $12.00

http://www.stockton-press.co.uk/tr

Nature vs nurture: Are leaders born or made? A behavior

genetic investigation of leadership style
Andrew M Johnson1, Philip A Vernon1, Julie M McCarthy1, Mindy Molson2, Julie A Harris1 and
Kerry L Jang3
1
Department of Psychology, University of Western Ontario, London, Ontario, Canada
2
Department of Psychology, University of California, Los Angeles, USA
3
Department of Psychiatry, University of British Columbia, Vancouver, Canada

With the recent resurgence in popularity of trait theories of leadership, it is timely to consider the
genetic determination of the multiple factors comprising the leadership construct. Individual
differences in personality traits have been found to be moderately to highly heritable, and so it
follows that if there are reliable personality trait differences between leaders and non-leaders,
then there may be a heritable component to these individual differences. Despite this connection
between leadership and personality traits, however, there are no studies of the genetic basis of
leadership using modern behavior genetic methodology. The present study proposes to address the
lack of research in this area by examining the heritability of leadership style, as measured by self-
report psychometric inventories. The Multifactor Leadership Questionnaire (MLQ), the Leader-
ship Ability Evaluation, and the Adjective Checklist were completed by 247 adult twin pairs
(183 monozygotic and 64 same-sex dizygotic). Results indicated that most of the leadership
dimensions examined in this study are heritable, as are two higher level factors (resembling
transactional and transformational leadership) derived from an obliquely rotated principal
components factors analysis of the MLQ. Univariate analyses suggested that 48% of the variance
in transactional leadership may be explained by additive heritability, and 59% of the variance in
transformational leadership may be explained by non-additive (dominance) heritability. Multi-
variate analyses indicated that most of the variables studied shared substantial genetic
covariance, suggesting a large overlap in the underlying genes responsible for the leadership
dimensions.

Keywords: leadership style, behavior genetics, transformational leadership, transactional

leadership, heritability

A great deal of behavioral genetic research has been uals whom he considered to be ‘great men’, insofar
conducted into various personality traits, resulting as they had attained eminence in their field to an
in the consensus that personality is dependent on extent realized by only 1 in 4000 individuals. He
both genetic and environmental influences. The then undertook to examine the pedigree of these
majority of twin studies has demonstrated moderate men, concluding that, since ‘greatness’ appeared to
to large genetic contributions to many personality be more prevalent within the family history of these
dimensions.1,2 On average, individual differences in subjects than would be expected in the public at
personality have been found to be approximately large, ‘greatness’ is wholly due to the action of genes.
40% heritable.2 It was a small leap from this to trait theories, and the
Current scientific thought on leadership may be ‘great man’ theory of William James.4 James believed
traced to Galton,3 who conducted the first study of that individuals are chosen by the situation, due to
the genetic basis of leadership. The main thesis of some intrinsic quality that makes them suitable to
Galton’s work was not, however, leadership, nor did lead or to ‘initiate movement’.
he address the heritability of leadership in a modern Scientists began to research leadership with the
sense, owing to the underdeveloped genetic method- intention of discovering which personality traits
ology of his time. Galton’s subjects were 100 individ- distinguish those with this ability from those with-
out. To this end, researchers compiled lists of traits
that had been associated with leadership, typically
Correspondence: Andrew M Johnson, Department of Psychology, through observations of the characteristics of pub-
University of Western Ontario, London, Ontario N6A 5CZ,
Canada. Tel: 519 680 5997; Fax: 519 661 3961; E-mail:
licly visible leaders. Stogdill5 surveyed hundreds of
ajohnson@julian.uwo.ca articles, and concluded that the most important
Received 6 October 1998; accepted 14 October 1998 traits were (in order of importance): originality,
 Born to lead
AM Johnson et al
217

popularity, sociability, judgement, aggressiveness, Method

desire to excel, humor, cooperativeness, liveliness,
and athletic ability. To simplify classification further,
he proposed a rough taxonomy of leadership, with Participants were part of a large behavior genetic
six dimensions, five of which were personality traits investigation spanning multiple personality con-
(capacity, achievement, responsibility, participation, structs.10 The subjects were 247 adult twin pairs:
and status). The sixth dimension was the situation. 183 pairs of monozygotic (MZ) twins (149 female
It was Stogdill’s5 reference to the importance of the pairs, mean age = 45.1 years, SD = 16.5; 34 male
pairs, mean age = 45.1 years, SD = 15.8), and
situation that prompted the greatest reaction from
64 same-sex dizygotic (DZ) twin pairs (55 female
the scientific community. As a consequence of this
pairs, mean age = 42.8 years, SD = 17.6; 9 male
article, there arose an increased level of research
pairs, mean age = 33.9 years, SD = 8.9). The subjects
activity into the situation as the primary determinant
represent a wide variety of backgrounds and levels of
of emergent leadership ability.6 Trait theories are,
education.
however, seeing a resurgence in popularity among
Subjects completed three self-report question-
researchers. As is often the case when theoretical
naires assessing different facets of leadership behav-
models swing from one extreme to another, the best
ior: the Multifactor Leadership Questionnaire
available answer may be a compromise between the (MLQ),11 the Leadership Ability Evaluation (LAE),12
two. Although it is becoming increasingly clear that and the Adjective Checklist (ACL).13 These ques-
there are systematic trait differences between leaders tionnaires were mailed to subjects, along with
and non-leaders,6,7 there are situational, organiza- instructions for their completion and the promise of
tional, and motivational characteristics that affect a chance to win one of 10 cash prizes of $100 in
the success of the individual within the leadership return for filling in the questionnaires. Completed
role.8 One widely accepted model of leadership style questionnaires were returned in stamped, pre-
is the transactional–transformational model pro- addressed envelopes. Subsequently, 10 subjects
posed by Burns.9 Transactional leadership refers to a were randomly selected from among those who had
leadership style in which the leader offers promises returned their completed questionnaires and these
of rewards and benefits in exchange for fealty. subjects were sent the prize money.
Transformational leadership, on the other hand, The MLQ is an 80-item measure requiring subjects
involves the use of inspirational techniques, to to rate the applicability of items to their own
inspire followers to suppress their own interests in behavior, using a 5-point scale. The test consists of
favor of the long-term benefit of the group.9 nine measures of leadership behavior: attributed
If individual differences in the personality charac- charisma, idealized influence, inspirational motiva-
teristics of leaders may be determined, it seems tion, intellectual stimulation, individualized con-
logical to determine the extent to which these sideration, contingent reward, active management-
characteristics, and the overall construct of leader- by-exception, passive management-by-exception,
ship, are heritable. This notwithstanding, in over and laissez-faire leadership. Attributed charisma
2000 citations since 1990 involving a leadership refers to an individual’s perceived charisma in non-
construct taken from Psychological Abstracts, none behavioral situations, and may be thought of as a
has employed a modern behavior genetic approach. physical charisma. Idealized influence refers to the
This study was designed to address this issue, individual’s charisma in behavioral situations. The
applying a twin study paradigm to several psycho- inspirational motivation scale measures the leader’s
metric measures of leadership. tendency to cause followers to respond to their task
Examination of the heritability of leadership is an on an emotional level. Scores on the intellectual
important contribution to leadership research for a stimulation variable reflect the degree to which the
number of reasons. The most immediate benefit of leader provides tasks or subtasks that represent an
such an analysis is to provide some initial answers to appropriate level of mental challenge to their fol-
the question of whether leaders are ‘born or made’. lowers. The individualized consideration scale is a
Another important feature of genetic analyses, how- measure of how well the leader provides personal
ever, is the examination of contributions due to an attention to each member of the target group.
individual’s unique experiences, a component of Contingent reward represents the exchange of value
environmental variance that can only be properly for performance. The two forms of management-
identified with data collected in a twin (or similarly by-exception (active and passive) are quite similar,
genetically informative) sample. When considering a with both forms involving a laissez-faire leadership
personality construct such as leadership style, the style until the situation commands their attention (in
magnitude of this environmental component may other words, until the situation demands correction).
give some indication of the degree to which training To remedy a problem situation, however, an active
would be effective for any given dimension. leader will set performance standards and monitor
 Born to lead
AM Johnson et al
218

subordinates carefully, whilst the passive leader will original format of the ACL would yield a more
suggest that the situation must be remedied, but reliable measure than having subjects simply tick
provide no concrete method for the amelioration of any adjectives they felt were descriptive of them.
the problem. Finally, individuals who are com- The 47 adjectives judged to be relevant to leadership
pletely non-involved with the follower group behavior were aggregated to form an adjectival
(except in an advisory capacity) typify the laissez- leadership measure.
faire leadership style. The MLQ also contains the In total, 16 measures of leadership were obtained,
criterion scale ‘extra effort’, which measures the the reliabilities of which ranged from moderate (LAE
degree to which the individual inspires others to Autocratic–Aggressive, α = 0.55), to high (ACL
work beyond the immediate requirements of the task Leadership, α = 0.91); median α = 0.78. Reliabilities
at hand, and perform ‘above and beyond’ the call of for all personality scales may be found in Table 2.
the situation. Although subordinate ratings of leader Subjects also completed a zygosity questionnaire,15
performance are preferable, the extra effort scale which has a reported accuracy of 93% in comparison
provides a useful comparison variable for the other with the results of blood-typing.16
leadership variables.11 Bass14 provides a good
description of the construction methods and
assumptions underlying the MLQ. Table 1 Adjectives contained in the ACL leadership scale
The LAE is a 50-item questionnaire designed to Aggressive Frank
measure the decision pattern or social climate in Alert Hard-working
which the leadership behavior is likely to take place. Appreciative Having initiative
The individual is presented with 50 hypothetical Authoritative Honest
Business-oriented Independent
situations and is required to identify with the leader, Capable Industrious
choosing the decision mode that he or she believes is Caring Inspiring
best for the described situation. Each situation Charismatic Leaderlike
describes four types of decision: laissez-faire, demo- Civilized Likeable
Conscientious Methodical
cratic–cooperative, autocratic–submissive, and auto- Co-operative Moderate
cratic–aggressive. Laissez-faire leadership is typified Decisive Motivated to achieve
by a lack of involvement with group members Deliberate Persevering
beyond the role of advisor or mentor, whilst demo- Demanding Powerful
Dependable Responsible
cratic–cooperative leadership emphasizes attention Driven Self accepting
to the group dynamic among followers, with deci- Dynamic Self monitoring
sions being made through the consensus of the Efficient Strong
follower group. Both autocratic–submissive and Empathic Thoughtful
Enterprising Tolerant
autocratic–aggressive leadership styles involves the Enthusiastic Trustworthy
direction of the group according to the leader’s plan, Entrepreneurial Versatile
and allow little deviation from the leader’s pre- Extraverted Well adjusted
conceived notion of what should be done. Auto- Farsighted
cratic–submissive leaders, however, allow followers
to determine how they wish to achieve the objective,
whilst autocratic–aggressive leaders define the proc- Table 2 Reliabilities of the MLQ, LAE, and ACL-r leadership
scales
ess as well as the objective.12 One can also calculate
a total leadership score by weighting the decision Scale α
mode scores to discriminate optimally between LAE Autocratic–aggressive 0.55a
leaders and non-leaders. Cassel and Stancik12 recom- LAE Autocratic–submissive 0.59a
mend that an individual’s total leadership score on LAE Laissez-faire 0.58a
LAE Democratic–cooperative 0.91a
the LAE be one tenth of the sum of their laissez-faire LAE Total 0.90a
score (multiplied by seven), their democratic–coop- MLQ Attributed charisma 0.74
erative score, and their autocratic–submissive score MLQ Idealized influence 0.82
(multiplied by four). The autocratic–aggressive score MLQ Inspirational motivation 0.84
does not enter into calculations of the total leader- MLQ Intellectual stimulation 0.82
MLQ Individualized consideration 0.87
ship score. MLQ Contingent reward 0.83
The ACL is a list of 300 descriptive adjectives, 47 MLQ Management-by-exception, active 0.65
of which were judged by the present authors to be MLQ Management-by-exception, passive 0.69
relevant to leadership behavior (see Table 1). Sub- MLQ Laissez-faire 0.71
MLQ Extra effort 0.81
jects were to respond to the full adjective checklist ACL Leadership 0.91
on a 5-point Likert scale. It was hoped (and subse- a
From Cassel R, Stancik E.12
quently confirmed) that this modification to the
 Born to lead
AM Johnson et al
219

Results high positive loading from contingent reward (0.80),

but this does not necessarily pose a problem for
Preliminary analyses interpretation, given that transformational and trans-
actional leadership are not mutually exclusive lead-
Means were computed for each raw scale score based ership styles, and the use of contingent rewards is a
on the whole sample and within each kinship group. valid extension of transformational leadership
A one-way analysis of variance was performed on behavior.18 The second rotated factor of the solution
each variable to compare the means between kinship accounts for 20.9% of the variance, and is identifi-
groups, using each member of each twin dyad as an able as transactional leadership, possessing high
independent replication. None of the variables dem- positive loadings from active management-
onstrated a significant mean difference between MZ by-exception (0.73), passive management-
and DZ twins, and tests for heterogeneity of variance by-exception (0.84), and laissez-faire leadership
indicated that there were no significant differences (0.72). This factor analysis is similar to analyses that
in variances between kinships. have been undertaken in previous analyses, and the
As was demonstrated by McGue and Bouchard,17 factors arrived at in this fashion may be seen to be
the presence of age and sex effects on a trait score representative of the higher-level leadership factors
can seriously bias estimates of genetic and environ- of transactional and transformational
mental components. To avoid these confounds, leadership.11,18
corrections for age and sex effects were made by Because the LAE was not designed to tap higher
computing completely standardized residual scores level factors, it was not reduced beyond the scale
from the multiple regression of each score on age and scores, nor was it included in a factor analysis with
sex. All further analyses are based on these trans- the MLQ scale scores.
formed scores.

Factor analyses Univariate genetic analyses

Because the MLQ was designed to tap multiple Using LISREL 8,19 univariate genetic analyses were
facets of a two-dimensional factor space (ie trans- conducted to assess the relative contributions of
formational and transactional leadership), higher genetic and environmental effects to individual
order factors were extracted from a principal compo- differences on the 16 leadership variables and the
nents factor analysis of the MLQ, and the resulting two MLQ factor scores (transformational and trans-
factors were obliquely rotated to facilitate inter- actional leadership). For all variables, a full ACE
pretation. The resulting factor matrix is presented in model was fit first, to determine the proportion of
Table 3. Two factors were extracted, based on an variance that is attributable to additive genetic
examination of the scree plot. The first rotated factor effects (A), common environmental effects (C), and
of the solution obtained in this fashion accounts for specific environmental effects (E). If the presence of
50.9% of the total variance in the sample space, and non-additive genetic effects was indicated (ie when
is identifiable as transformational leadership, as it the MZ correlation was more than twice the DZ
has high positive loadings from attributed charisma correlation), then an ADE model was applied to the
(0.84), idealized influence (0.88), inspirational moti- data, examining the proportion of variance attributa-
vation (0.88), intellectual stimulation (0.82), and ble to additive genetic effects, non-additive (dom-
individualized consideration (0.87), all of which are inance) genetic effects (D), and specific environ-
considered to be the scales of the MLQ that designate mental effects. These full models were then
transformational leadership.11 Factor one also has a systematically decomposed into three models, AE,
CE, and DE, comprising only two sources of varia-
Table 3 Varimax rotation of principal components solution for tion each. The final model fit to the data was an E
the MLQ
model, comprising only one source of variation,
I II namely specific environmental variance. To deter-
Attributed charisma 0.84 –0.08 mine which model afforded the ‘best fit’ to the data,
Idealized influence 0.88 –0.01 the χ2 obtained from each model was divided by its
Inspirational motivation 0.88 –0.17 degree of freedom, and the model with the lowest
Intellectual stimulation 0.82 0.05 χ2:df ratio was considered to be the best fit to the
Individualized consideration 0.87 –0.11
Contingent reward 0.80 0.22 data. Maximum likelihood estimates resulting from
Management-by-exception, active 0.35 0.73 this analysis were squared to produce estimates of
Management-by-exception, passive –0.13 0.84 the variance due to each of the effects specified by
Laissez-faire –0.36 0.72 the model.20 The best fitting model for each variable
Eigenvalues 4.58 1.88
Percentage of variance 50.90 20.90
may be found in Table 4. The best-fit model for each
of the variables demonstrated a good fit to the data.
 Born to lead
AM Johnson et al
220

As is apparent from Table 4, virtually every leader- analyses conducted on the scale scores. The results
ship scale demonstrated evidence of heritability, of these analyses are also contained in Table 4. The
with the exceptions of MLQ contingent reward, MLQ best fitting model for both transactional and trans-
passive management-by-exception, and MLQ lais- formational leadership indicated the presence of
sez-faire. Among the scales that evidence genetic genetic effects, with additive effects for transactional
determinism, there is roughly an even split between leadership, and non-additive effects for transforma-
additive and non-additive sources of variance, with tional leadership.
effects ranging from 0.30 to 0.59. Because contingent
reward, passive management-by-exception, and lais- Multivariate genetic analyses
sez-faire leadership (the only univariate models not
displaying heritable components) represent two Table 5 contains phenotypic correlations between
thirds of the transactional leadership dimension, as the scales of the LAE, the ACL leadership scale, the
proposed by Avolio, Bass, and Jung,21 this might two MLQ leadership factors (transformational and
indicate that individual differences in this form of transactional leadership), and MLQ extra effort.
leadership are predominantly environmentally Correlations in bold are significant at P < 0.01. As
determined. Indeed, the other component of transac- one would expect, given that these variables are
tional leadership, active management-by-exception, purported to measure the same general construct,
possesses the smallest amount of additive genetic most are highly intercorrelated. Within the LAE, it
variance, at 0.30. would appear that the overall scale score for the LAE
The heritability of the MLQ factor scores was (LAE total) is predominantly measuring a laissez-
examined with univariate genetic models by analyz- faire leadership style. Democratic–cooperative lead-
ing the factor scores in a fashion similar to the ership might be considered to be the best LAE

Table 4 MZ and DZ correlations and genetic analyses for LAE, ACL, and MLQ scales
a2 d2 c2 e2
Variable MZ DZ (SE) (SE) (SE) (SE) c2(df)
LAE Autocratic–aggressive 0.32 0.09 — 0.33 — 0.67 3.59 (4)
(0.062) (0.040)
LAE Autocratic–submissive 0.31 0.05 — 0.31 — 0.69 0.99 (4)
(0.066) (0.042)
LAE Laissez-faire 0.41 0.29 0.42 — — 0.58 1.25 (4)
(0.057) (0.038)
LAE Democratic–co-operative 0.35 0.15 0.36 — — 0.64 0.44 (4)
(0.061) (0.040)
LAE Total 0.36 0.28 0.38 — — 0.62 2.58 (4)
(0.060) (0.040)
ACL Leadership 0.50 0.16 — 0.49 — 0.51 1.95 (4)
(0.054) (0.037)
MLQ Attributed charisma 0.50 0.13 — 0.49 — 0.51 0.84 (4)
(0.053) (0.037)
MLQ Idealized influence 0.50 0.18 0.48 — — 0.52 2.92 (4)
(0.054) (0.036)
MLQ Inspirational motivation 0.54 0.20 — 0.55 — 0.45 2.49 (4)
(0.051) (0.034)
MLQ Individualized consideration 0.50 0.20 — 0.52 — 0.48 1.71 (4)
(0.053) (0.036)
MLQ Intellectual stimulation 0.47 0.20 — 0.47 — 0.53 1.16 (4)
(0.054) (0.037)
MLQ Contingent reward 0.25 0.27 — — 0.25 0.75 0.11 (4)
(0.067) (0.040)
MLQ Management-by-exception, active 0.31 0.11 0.30 — — 0.70 3.52 (4)
(0.068) (0.042)
MLQ Management-by-exception, passive 0.31 0.31 — — 0.31 0.69 2.86 (4)
(0.060) (0.038)
MLQ Laissez-faire 0.28 0.33 — — 0.29 0.71 3.60 (4)
(0.061) (0.037)
MLQ Extra effort 0.48 0.05 — 0.48 — 0.52 0.66 (4)
(0.056) (0.038)
MLQ Factor I 0.58 0.21 — 0.59 — 0.41 1.16 (4)
Transformational leadership (0.051) (0.035)
MLQ Factor II 0.47 0.33 0.48 — — 0.52 4.09 (4)
Transactional leadership (0.055) (0.037)
 Born to lead
AM Johnson et al
221

exemplar of transformational leadership, whilst lais- bivariate Cholesky decompositions were performed
sez-faire leadership is probably the best LAE indica- on all possible pairs of LAE subscales, the MLQ
tor of transactional leadership. The ACL leadership factors, the ACL leadership scale, and the MLQ extra
scale is also likely to be a good indicator of effort scale. Tables 6 and 7 contain estimates of the
transformational leadership, as it is highly positively genetic and environmental correlations (respec-
correlated with MLQ factor 1. tively) between the LAE scales, MLQ factors, ACL
Having noted that the putative model of trans- leadership scale, and the MLQ extra effort scale,
formational/transactional leadership appears to cut with correlations significant at P < 0.05 in bold. (In
across the three psychometric measures used in the the interest of brevity, only reduced scores are
study, it is interesting to note the extent to which the reported at a multivariate level. The complete matrix
scales share common genetic variance. Using Mx,22 of phenotypic and genetic correlations is available

Table 5 Phenotypic correlations between transformational/transactional leadership, MLQ extra effort scale, leadership ability
evaluation scales, and adjective checklist leadership scale
LAE LAE LAE LAE
Autocratic– Autocratic– Democratic— Laissez- LAE MLQ ACL MLQ
aggressive submissive co-operative faire Total Extra effort Leadership Factor 1
LAE Autocratic–aggressive
LAE Autocratic–submissive 0.02
LAE Democratic–co-operative –0.40 –0.29
LAE Laissez-faire –0.14 –0.28 –0.65
LAE Total –0.22 0.03 –0.73 0.92
MLQ Extra effort 0.00 –0.06 0.27 –0.24 –0.27
ACL Leadership –0.04 –0.06 0.22 –0.18 –0.21 0.50
MLQ Factor 1 –0.04 –0.08 0.32 –0.26 –0.29 0.76 0.60
MLQ Factor 2 0.07 0.08 –0.33 0.28 0.28 –0.18 –0.27 –0.09
Correlations in bold are significant at p<0.01; MLQ Factor 1 = Transformational Leadership; MLQ Factor 2 = Transactional Leadership

Table 6 Genetic correlations between transformational/transactional leadership, MLQ extra effort scale, leadership ability evaluation
scales, and adjective checklist leadership scale
LAE LAE LAE LAE
Autocratic– Autocratic– Democratic— Laissez- LAE MLQ ACL MLQ
aggressive submissive co-operative faire Total Extra effort Leadership Factor 1
LAE Autocratic–aggressive
LAE Autocratic–submissive 0.34
LAE Democratic–co-operative –0.38 –0.20
LAE Laissez-faire –0.25 –0.45 –0.72
LAE Total –0.23 –0.21 –0.78 0.97
MLQ Extra effort 0.06 –0.22 0.39 –0.29 –0.34
ACL Leadership –0.05 –0.13 0.38 –0.27 –0.33 0.85
MLQ Factor 1 –0.04 –0.25 0.47 –0.30 –0.35 0.90 0.88
MLQ Factor 2 –0.36 0.08 –0.42 0.48 0.56 –0.31 –0.33 –0.21
Correlations in bold are significant at p<0.05; MLQ Factor 1 = Transformational Leadership; MLQ Factor 2 = Transactional Leadership

Table 7 Environmental correlations between transformational/transactional leadership, MLQ extra effort scale, leadership ability
evaluation scales, and adjective checklist leadership scale
LAE LAE LAE LAE
Autocratic– Autocratic– Democratic— Laissez- LAE MLQ ACL MLQ
aggressive submissive co-operative faire Total Extra effort Leadership Factor 1
LAE Autocratic–aggressive
LAE Autocratic–submissive 0.10
LAE Democratic–co-operative –0.38 –0.35
LAE Laissez-faire 0.09 –0.20 –0.62
LAE Total –0.21 0.14 –0.72 0.89
MLQ Extra effort –0.08 0.07 0.21 –0.20 –0.19
ACL Leadership –0.02 0.02 0.10 –0.10 –0.11 0.18
MLQ Factor 1 –0.04 0.06 0.23 –0.23 –0.23 0.61 0.29
MLQ Factor 2 0.26 0.08 –0.22 0.09 0.06 –0.05 –0.20 0.05
Correlations in bold are significant at p<0.05; MLQ Factor 1 = Transformational Leadership; MLQ Factor 2 = Transactional Leadership
 Born to lead
AM Johnson et al
222

on request.) Most of the phenotypic correlations may should be noted that due to an imbalance in sex
be decomposed to find a significant genetic compo- composition within the sample, all effects of sex
nent, indicating that there is a strong common source were removed from the analysis. Future research
of genetic variation underlying the different dimen- may be directed at examining leadership in a sample
sions of leadership. of adult twins containing a more substantial number
of male subjects, to allow for the study of sex
effects.
Discussion This study represents the first behavior genetic
study to examine psychometric indices of leadership
and, as such, the results are exploratory and require
Univariate genetic analyses revealed that psycho- replication. Despite this fact, it is encouraging to
metric measures of leadership demonstrate moderate note that the results appear to be generalizable,
to large heritabilities in most dimensions of the
insofar as they are consistent across multiple meas-
construct. Although transformational leadership
ures of leadership. It is unlikely that this study will
demonstrated a clearly genetic determination, fur-
end all debate on the topic of whether leaders are
ther examination of the transactional leadership
born or made. It does, however, provide preliminary
domain is warranted, given that several of the
transactional leadership scales yielded models that empirical data regarding the contributions of both
did not include additive or non-additive genetic genetic and environmental factors to individual
effects. differences in leadership style.
Higher level leadership factors, constructed from
the MLQ, represent transformational and transac-
tional leadership. Univariate genetic analyses indi-
cated that both of these factors were heritable, with
References
transformational leadership demonstrating non-
additive heritability, and transactional leadership 1 Loehlin JC. Genes and Environment in Personality Develop-
demonstrating additive heritability. This is an inter- ment. Sage Publications: London, 1992.
esting finding in itself, because it suggests that 2 Plomin R, DeFries JC, McClearn GE. Behavioral Genetics: A
Primer, 2nd edn. WH Freeman: New York, 1990.
transformational leadership might be of greater 3 Galton F. Hereditary Genius. Appleton: New York, 1869.
evolutionary significance, as it has been proposed 4 James W. Great men, great thoughts, and their environment.
that traits expressing dominance heritability pat- Atlantic Monthly 1880; 46: 441–459.
terns may be more adaptive to the organism.23 In 5 Stogdill R. Personal factors associated with leadership: A
comparing these factors to the scales from the LAE survey of the literature. J Psychol 1948; 25: 35–71.
6 Bass B. Bass and Stogdill’s Handbook of Leadership, 3rd edn.
and the ACL, through multivariate genetic analyses, Free Press: New York, 1990.
it becomes clear that a great deal of the variance 7 Lord RG, DeVader CL, Alliger GM. A meta-analysis of the
shared by these variables is attributable to common relation between personality traits and leadership percep-
genes, because the correlations among the majority tions: An application of validity generalization procedures. J
of the variables contain significant common genetic Appl Psychol 1986; 61: 402–410.
8 Kirkpatrick S, Locke E. Leadership: Do traits matter? In: Steers
influences.
RM, Porter LW, Bigley GA (eds). Motivation and Leadership at
One limitation of the present study lies in the Work, 6th edn. McGraw-Hill: New York, 1996, pp 186–199.
nature of the leadership measures: they are exclu- 9 Burns JH. Leadership. Harper & Row: New York, 1978.
sively self-report questionnaires. It would be inter- 10 Vernon PA, McCarthy JM, Johnson AM, Jang KL, Harris JA.
esting to examine a sample of behavioral measures of Individual differences in multiple dimensions of aggression: A
leadership within a twin sample, or perhaps to send univariate and multivariate genetic analysis. (Twin Research,
in press).
an evaluation version of leadership batteries to 11 Bass B, Avolio B. Multifactor Leadership Questionnaire (Form
someone in a subordinate position to the twins, so 5X – Self). Consulting Psychologists Press: Palo Alto, CA,
that this individual might evaluate the twins’ leader- 1991.
ship performance. In a similar vein, it would be 12 Cassel R, Stancik E. The Leadership Ability Evaluation –
interesting to examine individual perceptions of an Revised. Western Psychological Services: Los Angeles, CA,
1982.
ideal leader within a sample of twins. Bass14 13 Gough H, Heilbrun A. The Adjective Checklist Manual.
addresses the idea that people tend to think of their Consulting Psychologists Press: Palo Alto, CA, 1983.
ideal leader in similar terms, and that this ideal 14 Bass B. Does the transactional-transformational leadership
leader is typically a transformational leader. paradigm transcend organizational and national boundaries?
Although this question has been approached from a Am Psychologist 1997; 52(2): 130–139.
15 Nichols R, Bilbro W. The diagnosis of twin zygosity. Acta
psychometric standpoint, it has yet to be examined
Genet Stat Med 1966; 16: 265–275.
from a behavior genetic stance, and there is sub- 16 Kasriel J, Eaves L. The zygosity of twins: Further evidence on
stantial information to be had from such an analysis, the agreement between diagnosis by blood groups and written
as is illustrated by the present study. Finally, it questionnaires. J Biosoc Sci 1976; 8: 263–266.
 Born to lead
AM Johnson et al
223

17 McGue M, Bouchard TJ Jr. Adjustment of twin data for the 21 Avolio B, Bass B, Jung D. Construct validation and norms for
effects of age and sex. Behav Genet 1984; 14: 325–343. the Multifactor Leadership Questionnaire (MLQ – Form 5X).
18 Bass B. Leadership: Good, better, best. Organiz Dynam 1985; CLS Report 95-4. Binghamton University: Center for Leader-
13(3): 26–40. ship Studies, 1995.
19 Jöreskog K, Sörbom D. LISREL 8: A Guide to Program and 22 Neale MC. Mx: Statistical Modeling (4th ed.). Box 126 MCV,
Applications, 3rd edn. Scientific Software: Mooresville, IN, Richmond, VA 23298. Department of Psychiatry, 1997.
1993. 23 Jensen AR. The g Factor: The Science of Mental Ability.
20 Neale MC, Cardon LR. Methodology for Genetic Studies of Praeger: Westport, CT, 1998.
Twins and Families. Kluwer Academic: Dordrecht, 1992.