Professional Documents
Culture Documents
Self Leadership
Self Leadership
Journal of
Managerial The revised self-leadership
Psychology
17,8
questionnaire
Testing a hierarchical factor structure
672 for self-leadership
Received July 2001
Revised May 2002
Jeffery D. Houghton
Accepted July 2002 Department of Management, Lipscomb University, Nashville,
Tennessee, USA and
Christopher P. Neck
Department of Management, Virginia Polytechnic Institute and State
University, Blacksburg, Virginia, USA
Keywords Leadership, Measurement, Self-efficiency, Factor analysis, Models
Abstract Despite the popularity and potential of self-leadership strategies in modern
organizations, no acceptably valid and reliable self-leadership assessment scale has heretofore
been developed. The present study tests the reliability and construct validity of a revised self-
leadership measurement scale created on the basis of existing measures of self-leadership. Results
from an exploratory factor analysis (EFA) demonstrate significantly better reliability and factor
stability for the revised scale in comparison to existing instruments. Further, results from a
confirmatory factor analysis (CFA) utilizing structural equation modeling techniques
demonstrate superior fit for a higher order factor model of self-leadership, thus providing
evidence that the revised scale is measuring self-leadership in a way that is harmonious with self-
leadership theory. Based on these results, the revised scale appears to be a reasonably reliable and
valid instrument for the measurement of self-leadership skills, behaviors, and cognitions.
Implications for future empirical self-leadership research are discussed.
Self-leadership (Manz, 1983, 1986, 1992; Manz and Neck, 1999; Manz and Sims,
2001) is a process through which people influence themselves to achieve the
self-direction and self-motivation necessary to behave and perform in desirable
ways. Self-leadership is rooted in several related theories of self-influence
including self-regulation (Kanfer, 1970; Carver and Scheier, 1981), self-control
(Cautela, 1969; Mahoney and Arnkoff, 1978, 1979; Thoresen and Mahoney,
1974), and self-management (Andrasik and Heimberg, 1982; Luthans and
Davis, 1979; Manz and Sims, 1980). Self-leadership is generally portrayed as a
broader concept of self-influence that subsumes the behavior-focused strategies
of self-regulation, self-control, and self-management, and then specifies
additional sets of cognitive-oriented strategies derived from intrinsic
motivation theories (e.g. Deci, 1975; Deci and Ryan, 1985), social cognitive
theory (Bandura, 1977, 1986, 1991), and positive cognitive psychology (Beck et
al., 1979; Burns, 1980; Ellis, 1977; Seligman, 1991). Thus, drawing from these
well-established theoretical foundations, self-leadership comprises specific sets
Journal of Managerial Psychology,
Vol. 17 No. 8, 2002, pp. 672-691.
# MCB UP Limited, 0268-3946
Paper presented at the Annual Meeting of the Southern Management Association, New Orleans,
DOI 10.1108/02683940210450484 Louisiana, USA.
of behavioral and cognitive strategies designed to shape individual Revised self-
performance outcomes. leadership
Self-leadership strategies may be divided into three general categories: questionnaire
behavior-focused strategies, natural reward strategies, and constructive
thought pattern strategies (Anderson and Prussia, 1997; Manz and Neck, 1999;
Prussia et al., 1998). Behavior-focused strategies are aimed at increasing self-
awareness, leading to the management of behaviors involving necessary but 673
perhaps unpleasant tasks (Manz, 1992; Manz and Neck, 1999). These strategies
include self-observation, self-goal setting, self-reward, self-correcting feedback,
and practice. Self-observation of one’s own behavior can lead to an awareness
of when and why one engages in certain behaviors. This heightening of self-
awareness can, in turn, lead to the identification of specific behaviors that
should be changed, enhanced, or eliminated (Mahoney and Arnkoff, 1978, 1979;
Manz and Sims, 1980; Manz and Neck, 1999). Based on this foundation of self-
assessment, the individual can effectively set personal goals that may lead to
improved performance (Manz, 1986; Manz and Neck, 1999; Manz and Sims,
1980). A multitude of research has shown that the act of setting and accepting
challenging and specific goals can have a dramatic effect in motivating
individual performance (Locke and Latham, 1990). In addition, self-rewards can
be effectively used to reinforce desirable behaviors and goal attainments
(Mahoney and Arnkoff, 1978, 1979; Manz and Sims, 1980; Manz and Neck,
1999). Self-rewards can be something tangible, like a nice restaurant meal or a
weekend vacation following the completion of a difficult project at work, or
self-rewards can be something abstract and simple such as congratulating
oneself or mentally visualizing a favorite place or experience. Like self-rewards,
self-correcting feedback can also be used to shape desirable behaviors
effectively. An introspective yet positively framed examination of negative
behaviors or performance failures can be more effective in correcting
performance than excessive self-punishment based on habitual guilt and self-
criticism (Manz and Sims, 2001). Finally, the rehearsal or practice of desired
behaviors before actual performance can allow for the correction of problems
and the avoidance of costly miscues (Manz, 1992; Manz and Neck, 1999; Manz
and Sims, 1980; Thoresen and Mahoney, 1974). In short, behavior-focused self-
leadership strategies are designed to encourage positive, desirable behaviors
that lead to successful outcomes, while suppressing negative, undesirable
behaviors that lead to unsuccessful outcomes.
Natural reward strategies emphasize the enjoyable aspects of a given task or
activity. Natural or intrinsic rewards result when incentives are built into the
task itself and a person is motivated or rewarded by the task itself (Manz, 1992;
Manz and Neck, 1999). Naturally rewarding activities tend to foster feelings of
increased competence, self-control, and purpose (Manz, 1986, Manz and Neck,
1999). Natural reward strategies include efforts to incorporate more pleasant
and enjoyable features into a given task or activity and efforts to change
perceptions of an activity by focusing on the task’s inherently rewarding
aspects (Manz and Neck, 1999). For instance, a person might attempt to create a
Journal of more enjoyable work environment by playing soft music, hanging pictures, or
Managerial adding other personal touches. Alternatively, a person could shift attention
Psychology toward job features that they particularly enjoy, such as working outdoors or
engaging customers in conversation. Through natural reward strategies such
17,8 as these, an individual can increase performance levels by focusing on the
pleasant aspects of the job or task.
674 Constructive thought pattern strategies involve the creation and maintenance
of functional patterns of habitual thinking (Manz and Neck, 1991, 1999; Neck
and Manz, 1992, 1996). Specific thought-oriented strategies include the
evaluation and challenging of irrational beliefs and assumptions, mental
imagery of successful future performance, and positive self-talk. Dysfunctional
thought processes, a common and serious hindrance to individual performance,
generally result from underlying dysfunctional beliefs and assumptions that
are often triggered by stressful or troubling situations (Burns, 1980; Ellis, 1977).
Through a process of self-analysis, individuals may identify, confront, and
replace dysfunctional beliefs and assumptions with more rational ones (Burns,
1980; Ellis, 1977; Manz and Neck, 1999; Neck and Manz, 1992). In a similar
manner, negative and destructive self-talk should be replaced by more positive
and constructive self-dialogues. Self-talk is defined as what we covertly tell
ourselves (Ellis, 1962; Neck and Manz, 1992, 1996). Self-dialogues usually take
place at unobservable levels as individuals evaluate, instruct, and mentally
react to themselves (Ellis, 1962, 1977; Manz and Neck, 1991; Neck and Manz,
1992). Through an analysis and evaluation of self-talk patterns, an individual
can learn to suppress and discourage negative and pessimistic self-talk while
fostering and encouraging optimistic self-dialogues (Seligman, 1991).
Finally, mental imagery is generally defined as the symbolic, covert, mental
invention or rehearsal of an experience or task in the absence of actual, overt
physical muscular movement (see Driskell et al., 1994; Finke, 1989). Through
the use of mental imagery it may be possible to create and symbolically
experience behavioral outcomes prior to actual performance (Manz and Neck,
1991; Neck and Manz, 1992, 1996). This technique has also been variously
referred to as imaginary practice (Perry, 1939), covert rehearsal (Corbin, 1967),
symbolic rehearsal (Sackett, 1934), and mental practice (Corbin, 1972). Those
individuals who envision the successful performance of a task or activity
beforehand are much more likely to perform successfully when faced with the
actual situation (Manz and Neck, 1999). Empirical research evidence provides
support for this assertion. Indeed, a recent meta-analysis of 35 empirical studies
(Driskell et al., 1994), suggests that mental practice generally has both a
positive and significant effect on individual performance outcomes.
In recent years, self-leadership concepts have gained considerable
popularity as evidenced by the large number of practitioner-oriented books and
articles on the subject (e.g. Blanchard, 1995; Cashman, 1995; Manz, 1991; Manz
and Neck, 1999; Manz and Sims, 2001; Sims and Manz, 1996; Waitley, 1995) and
by coverage in an increasing number of management and leadership textbooks
(e.g. Ivancevich and Matteson, 1999; Kreitner and Kinicki, 2001; McShane and
Von Glinow, 2000; Nahavandi, 2000). Given the popularity of self-leadership Revised self-
concepts and the recent emphasis on employee empowerment (e.g. Conger and leadership
Kanungo, 1988; Thomas and Velthouse, 1990) and self-managing work teams questionnaire
(e.g. Cohen and Ledford, 1994; Hackman, 1986), self-leadership appears to have
impressive potential for application in today’s dynamic organizations. Indeed,
self-leadership has often been presented as a primary mechanism in both
empowerment (e.g. Anderson and Prussia, 1997; Manz, 1992; Prussia et al., 675
1998; Shipper and Manz, 1992) and the successful implementation of self-
managing work teams (e.g. Neck et al., 1996; Manz and Sims, 1986, 1987).
While a plethora of conceptual (non-empirical) self-leadership research
exists, a sparse amount of empirical research has examined self-leadership
theory and its application in organizational settings (Anderson and Prussia,
1997). One possible explanation for this lack of empirical research is that no
valid scale of self-leadership has heretofore been developed. The lack of a
proven scale for the assessment of self-leadership skills makes it difficult to
advance empirical research in this promising area. Until self-leadership and its
component dimensions can be effectively measured, empirical research efforts
in the area of self-leadership will remain scarce. Indeed, Markham and
Markham (1995, 1998), in offering an agenda for future self-leadership research,
call for the construction and validation of self-leadership scales.
Self-leadership measurement
While no psychometrically-sound self-leadership measurement scale has
previously been developed, two noteworthy preliminary attempts have been
made to develop a self-leadership questionnaire (SLQ). Both of these efforts
utilized a prototype created by Manz and Sims and rooted in the self-leadership
literature (Manz, 1986, 1992; Manz and Sims, 1987, 1991) as a basis for the
development of a more advanced instrument. Cox (1993) developed and
assessed a 34-item SLQ. Unrestricted factor analysis led to an eight-factor
solution with factors labeled as: self-problem solving initiative, efficacy,
teamwork, self-reward, self-goal setting, natural rewards, opportunity thought,
and self-observation/evaluation. Alpha coefficients (Cronbach, 1951) for the
Cox (1993) SLQ sub-scales ranged from 0.69 to 0.93. Mean James coefficients
(assessing interrater consensus; James et al., 1984) of 0.82 or better were
obtained for all eight SLQ behavior dimensions. Based on these assessments,
Cox’s (1993) SLQ has shown some preliminary potential as a self-leadership
assessment scale. Indeed, Roberts and Foti (1998) recently employed the Cox
scale as a measure of self-leadership in a field study. They reported a coefficient
alpha of 0.91 for the Cox SLQ, thus providing additional preliminary evidence
of the scale’s psychometric properties.
More recently, Anderson and Prussia (1997) have presented an alternative
preliminary effort toward the developmentof a self-leadership scale. In this study,
a 90-item SLQ prototype, based on earlier work by Manz (1992) and Manz and
Sims (1991), was reduced to 50 items through a sorting process that assessed the
agreement of 18 category judges. The 50-item SLQ was then administered to a
Journal of sample of 194 students. Factor analysis of the resulting data yielded ten unique
Managerial factors. Six factors (self-goal setting, self-reward, self-punishment, self-
Psychology observation, self-cueing, and self-withholding) assess self-leadership’s behavior
focused strategies, one factor (focusing thoughts on natural rewards) represents
17,8 self-leadership’s natural reward strategies, and three factors (visualizing
successful performance, self-talk, and evaluating beliefs and assumption)
676 evaluate self-leadership’s constructive thought pattern strategies. Alpha
coefficients (Cronbach, 1951) ranged from 0.69 to 0.91 for each of the ten scale
sub-dimensions. However, several items loaded on the wrong factor and/or
demonstrated minimally acceptable factor loadings (greater than 0.30) on more
than one factor. The Anderson and Prussia (1997) SLQ was then administered to a
second sample to further assess the construct validity of the scale. In the second
administration, one sub-scale (i.e. focusing thoughts on natural rewards) dropped
below Nunnally’s (1978) recommended scale reliability threshold of 0.70 to an
alpha of 0.62, thus indicating significant instability across samples for this
dimension of the SLQ factor structure (Anderson and Prussia, 1997). Another sub-
scale (i.e. self-observation) was marginal with an alpha of 0.70. While the efforts of
Cox (1993) and Anderson and Prussia (1997) represent significant and valuable
progress in the development of a valid self-leadership scale, these efforts are
clearly preliminary. Further development of a psychometrically acceptable self-
leadership scale is a pressing concern in the advancement of self-leadership
research.
The purpose of the present study is to test the construct validity of a revised
self-leadership measurement scale, created on the basis of the two existing
measures of self-leadership (i.e. Anderson and Prussia, 1997; Cox, 1993). In short,
exploratory factor analysis (EFA) was used to examine to the factor structure of
the revised scale and to facilitate comparisons with existing measures of self-
leadership, while confirmatory factor analysis (CFA) utilizing structural equation
modeling techniques was used to examine the extent to which the revised scale
fit a hierarchical model of self-leadership as specified by self-leadership theory.
Method
Sample and procedure
The data were collected from two independent samples of students in two
introductory management courses at a large southeastern university in the
USA (Sample 1: n = 477; Sample 2: n = 381; 60 percent male, 40 percent female,
mean age = 21.12 across both samples). Listwise deletion for missing data
resulted in final sample sizes of 442 and 357 for Samples 1 and 2, respectively.
As part of a class lecture on individual differences, students completed the
revised self-leadership instrument. The questionnaires were completed
anonymously and participation was voluntary.
Instrumentation
Self-leadership was measured using the revised self-leadership questionnaire
(RSLQ). The RSLQ (shown in the Appendix) consists of 35 items in nine
distinct sub-scales representing the three primary self-leadership dimensions. Revised self-
Table I provides a summary of the relationships between the nine RSLQ sub- leadership
scales and the three self-leadership dimensions. The behavior-focused questionnaire
dimension is represented by five sub-scales labeled:
(1) self-goal setting (five items);
(2) self-reward (three items); 677
(3) self-punishment (four items);
(4) self-observation (four items); and
(5) self-cueing (two items).
A single sub-scale consisting of five items represents the natural rewards
dimension. The constructive thought dimension is represented by three sub-
scales labeled:
(1) visualizing successful performance (five items);
(2) self-talk (three items); and
(3) evaluating beliefs and assumptions (four items).
The RSLQ was developed by building on the previous versions of self-
leadership questionnaires (e.g. Anderson and Prussia, 1997; Cox, 1993)
described above. Anderson and Prussia’s (1997) 50-item SLQ served as the
primary basis for the RSLQ. To create the RSLQ, 17 ambiguous items were
deleted from the Anderson and Prussia (1997) scale and two items were added
from the Cox (1993) instrument. Specifically, items with primary loadings on an
incorrect factor (i.e. loading on a dimension the item was not intended to
represent), items with low primary factor loadings on the correct factor, and/or
items with high cross-factor loadings (i.e. high secondary loadings on an
incorrect factor) were dropped from the scale. For example, the item ``I try to
build activities into my work that I like doing’’ was intended to assess the
``focusing on natural rewards’’ dimension. However, in an earlier EFA, this item
demonstrated a primary loading of 0.59 on the ``visualizing successful
Factor
Dimensions Sub-scales Scale items number
Analysis
The data were analyzed in two primary stages. First, in order to make a fair
and consistent comparison between the psychometric properties of the RSLQ
and the Anderson and Prussia (1997) SLQ, an exploratory factor analysis
(EFA) with principle components extraction and varimax rotation was applied
to the data from Sample 1. Consistent with Anderson and Prussia (1997), a
critical value of 0.35 was chosen as the cut-off point in determining whether an
item defined a factor. In addition, the ``eigenvalue greater than one test’’ and the
scree test were used to define factors (Gorsuch, 1974). Second, in an effort to
examine the extent to which the RSLQ effectively represents self-leadership
theory, a second-order confirmatory factor analysis (CFA) model of self-
leadership (see Figure 1) was tested on the data from Sample 2 through an
analysis of covariance structures using LISREL 8 (Jöreskog and Sörbom, 1993).
Established item-parceling procedures (e.g. Collins and Gleaves, 1998) were
utilized to create composite indicators. Items in each of the sub-scales were
summed and averaged to create five composite indicators for the behavior-
focused dimension and three composite indicators for the constructive thought
dimension. Three single items served as indicators for the natural reward
dimension.
In order to assess the relative fit of the theoretically-based hierarchical model
of self-leadership (Figure 1), a one-factor model (i.e., all indicators loading on a
single factor; see Figure 2) and a three uncorrelated factors model (see Figure 3)
were tested for comparison. In accordance with the recommendations of Hoyle
and Panter (1995), the following fit indexes were used to assess the fit of the
Revised self-
leadership
questionnaire
679
Figure 1.
Second order factor
model of self-leadership
and its primary
dimensions with
standardized parameter
estimates (all parameter
loadings were
significant at the ¬ =
0.05 level)
three models: Chi-square (À2, e.g. Bollen, 1989a), the goodness-of-fit-index (GFI,
Jöreskog and Sörbom, 1981), the non-normed fit index (NNFI, Bentler and
Bonnett, 1980), the incremental fit index (IFI, Bollen, 1989b), and the
comparative fit index (CFI, Bentler, 1990). The use of multiple fit indexes is
generally advisable in order to provide convergent evidence of model fit. The
values of GFI, NNFI, IFI, and CFI range from 0 to 1.0, with values closer to 1.0
indicating a well-fitting model (Bentler and Bonnett, 1980; Hoyle and Panter,
1995).
Results
Exploratory factor analysis: Sample 1
Based on this sample, the RSLQ demonstrated significantly better reliability
and factor stability in comparison to the Anderson and Prussia (1997)
instrument. Coefficient alphas for each of the nine RSLQ sub-scales are shown
in Table II. For comparison purposes, coefficient alphas from the Anderson and
Prussia (1997) instrument are also shown in parentheses. As demonstrated in
the table, alphas for the RSLQ either remained relatively unchanged or showed
Journal of
Managerial
Psychology
17,8
680
Figure 2.
One-factor model of self-
leadership with
standardized parameter
estimates (all parameter
loadings were
significant at the
¬ = 0.05 level)
681
Figure 3.
Three uncorrelated
factors model of self-
leadership (all parameter
loadings were
significant at the
¬ = 0.05 level
684
Table III.
Journal of
deviations, and
(sample 2 data)
Managerial
Psychology
Means, standard
indicator variables
intercorrelations among
Indicator
variable M SD 1 2 3 4 5 6 7 8 9 10 11
1. C2 3.94 0.675 –
2. C4 3.99 0.960 0.252** –
3. C6 3.81 0.862 0.288** 0.184** –
4. C7 4.04 0.637 0.577** 0.219** 0.360** –
5. C9 3.98 1.06 0.319** 0.137** 0.151** 0.124* –
6. V26 3.68 0.976 0.372** 0.235** 0.034 0.225** 0.139** –
7. V32 4.00 0.895 0.492** 0.301** 0.098 0.400** 0.241** 0.461** –
8. V35 3.87 0.917 0.445** 0.319** 0.055 0.225** 0.152** 0.454** 0.418** –
9. C1 3.62 0.816 0.492** 0.299** 0.113* 0.279** 0.143** 0.344** 0.412** 0.348** –
10. C3 4.04 0.927 0.296** 0.129* 0.145** 0.189** 0.191** 0.190** 0.183** 0.256** 0.385** –
11. C5 3.77 0.698 0.437** 0.289** 0.259** 0.354** 0.177** 0.296** 0.317** 0.390** 0.451** 0.320** –
Notes: N = 357. C1 = visualizing successful performance item composite; C2 = self-goal setting item composite; C3 = self-talk item composite;
C4 = self-reward item composite; C5 = evaluating beliefs and assumptions item composite; C6 = self-punishment item composite; C7 = self-
observation item composite; C8 = natural rewards item composite; C9 = self-cueing item composite; V26 = item 26 – RSLQ; V32 = item 32 –
RSLQ; V35 = item 35 – RSLQ. * p < 0.05 (two-tailed); ** p < 0.01 (two-tailed)
rewards, and constructive thought factors have a higher order factor; namely, Revised self-
self-leadership. leadership
questionnaire
Discussion
The results of this study provide support for the validity and reliability of the
RSLQ as an acceptable measure of self-leadership skills and behaviors.
Reliability estimates for the RSLQ improved significantly or remained 685
relatively stable in comparison to Anderson and Prussia’s (1997) SLQ. Most
notably, the scale reliability estimate for the ``focusing on natural rewards’’
dimension of the RSLQ increased above the commonly recommended level
(0.70) to a coefficient alpha of 0.74, a significant improvement over the
reliability estimates (0.69 and 0.62) reported by Anderson and Prussia (1997). In
addition, EFA results suggest a remarkably stable factor structure for the
RSLQ. As expected, nine interpretable factors emerged representing distinct
self-leadership dimensions as specified by self-leadership theory. Behavior-
focused strategies were represented by five factors, natural reward strategies
were represented by one factor, and the constructive thought pattern strategies
were represented by three factors. The RSLQ items loaded strongly on the
correct factors with virtually no significant cross-factor loadings.
The construct validity of the RSLQ was further examined through a CFA
that examined the fit of a theoretically-based hierarchical model of self-
leadership to the data from a separate large sample. The superior fit of this
second order factor model over competing one-factor and three-factor models
suggests that the RSLQ is measuring self-leadership in a way that is
harmonious with the specifications of self-leadership theory, thus providing
additional evidence of the RSLQ’s construct validity. Based on the results of
both the EFA and CFA, it appears that the RSLQ is a fairly reliable and valid
measurement instrument that effectively reflects self-leadership theory in the
assessment of self-leadership skills, behaviors, and cognitions.
This study contributes to the self-leadership literature in at least three
important ways. First, the RSLQ, a revised self-leadership measurement scale,
was presented and described. Second, evidence of the reliability and construct
validity of the revised scale was demonstrated across two large independent
samples using EFA and CFA techniques. Third, a theory-based hierarchical
model of self-leadership was examined empirically for the first time. The results
of this study provide support for this model of self-leadership and for the RSLQ