Asian Journal of Business Ethics

, Volume 3, Issue 1, pp 57–64

Authentic leadership questionnaire (ALQ) psychometrics

Authors

Article

DOI: 10.1007/s13520-013-0031-2

Cite this article as:
Roof, R. Asian J Bus Ethics (2014) 3: 57. doi:10.1007/s13520-013-0031-2

Authenticity, the concept of self-awareness acted out (Caza et al. 2010) or the functioning in a manner aligned with values (Leroy et al. 2012), has been the subject of great interest recently (Nelder and Schriescheim 2011), and is considered by many to be a “root construct” (Avolio and Gardner 2005) forming the foundation for all contemporary positive leadership theories (Qian et al. 2012) including transformation, ethical, servant, spiritual, and even charismatic leadership (Avolio and Gardner 2005; Nelder and Schriescheim 2011). While the early stage nature of authentic leadership has resulted in multiple conceptualizations (Caza et al. 2010), and there is contention over whether the construct includes moral and ethical prescriptives (Peus et al. 2012), the subject of this research note, the authentic leadership questionnaire (ALQ), was designed to include the ethical dimension of leadership (Walumbwa et al. 2008). My focus within this research note is not on the current theoretical debate, but only exploring whether the ALQ exhibits adequate validity and reliability to suggest its use in appropriate research.

Quality leadership research relies on sound, theoretically driven research design, and a critical element is the use of reliable and valid instruments (Girden and Kabacoff 2011) that can be applied across cultures to support global research (Walumbwa et al. 2008). Since the ALQ is widely applied in leadership research to measure the genuine nature of leadership, this research note consolidates validity and reliability testing from existing studies to inform researchers considering using the scale to assess leader authenticity including moral or ethical characteristics. A search of literature in Academic Search Complete revealed that over 138 journal articles discussed both authentic leadership and ethics suggesting significant interest in the connection of leader authenticity to ethical and moral organizational behavior. To evaluate the cultural transferability of the ALQ, the instrument’s reliability and validity characteristics were considered across a wide range of countries including the US, China, Kenya, New Zealand, Belgium, Norway, Germany, Iran, Taiwan, and Canada.

Creation of the authentic leadership questionnaire

The authentic leadership questionnaire instrument development effort was published by Walumbwa et al. (2008) and leveraged the authors’ earlier theoretical work on authentic leadership (Avolio and Gardner 2005; Gardner et al. 2005). The instrument reflected the earlier theoretical definition and dimensions of authentic leadership as follows:

a pattern of leader behavior that draws upon and promotes both positive psychological capacities and a positive ethical climate, to foster greater self-awareness, an internalized moral perspective, balanced processing of information, and relational transparency on the part of leaders working with followers, fostering positive self-development. (Walumbwa et al. 2008, p.94).

The initial instrument publication included a detailed description of the instrument, initial development steps, and researchers’ efforts to confirm reliability and validity through the use of five samples drawn from the US, China, and Kenya (Walumbwa et al. 2008). Walumbwa et al. (2008), in developing the instrument, drew from prior theoretical literature, dissertations, and expert input, applying a combination of deductive and inductive approaches to include the refinement of the items through a process of secondary expert reviews. Relating items to specific authentic leadership dimensions was performed theoretically rather than statistically through exploratory factor analyses as is typical in scale development, and validity of the theoretical structure was confirmed during subsequent analyses (Walumbwa et al. 2008).

The resulting instrument was comprised of 16 items reflecting four dimensions or subscales of authentic leadership as follows: self-awareness reflecting an understanding of how leaders interpret the world and their own strengths and weaknesses (4 items), relational transparency demonstrating how the leader presents a true and genuine self (5 items), internalized moral perspective reflecting self-regulation and actions based on personal moral standards (4 items), and balanced processing which indicates how leaders solicit and objectively consider alternative views (3 items) (Nelder and Schriescheim 2011; Walumbwa et al. 2008). Responses for the ALQ were designed to be collected on a 5-point Likert scale with answers from never to almost always.

Three surveys were conducted during the first stage of the scale development effort, two independent US samples totaling 224 full-time employees from a high-technology manufacturing firm in the northeast, and one sample of 212 full-time employees from a state-owned firm in Beijing, the People’s Republic of China (Walumbwa et al. 2008). The first stage was intended to confirm the structure of the questionnaire items and factors, and develop initial support for reliability and validity of the measurement instrument. The three samples of stage one successfully demonstrated the four factor structure, verified the interdependency of the four factors of authentic leadership, confirmed that a second-order factor accounted for the connection between the four dimensions, and developed advance support for the reliability and validity of the instrument (Walumbwa et al. 2008).

Following completion of the stage-1 work, stage 2 further examined construct validity by evaluating dimensionality and internal instrument consistency, testing convergent validity across alternative similar constructs of ethical and transformational leadership, and analyzing discriminant and predictive validity through examining nomological validity (Walumbwa et al. 2008). Two independent samples were drawn from employed students at a large university in the southwestern US, resulting in 178 usable surveys (Walumbwa et al. 2008). Study 2 research further supported the relationship between the four sub-constructs of the ALQ, and explored the connection between authentic leadership and both ethical leadership and transformational leadership (Walumbwa et al. 2008). The ALQ correlations with ethical and transformational leadership in development study 2 did not indicate redundancy between authentic leadership and the leadership constructs, but suggested the theories were complimentary overlapping concepts (Walumbwa et al. 2008). Relationships between authentic leadership as measured by the ALQ and other expected positive outcomes were also confirmed supporting the scales predictive validity (Walumbwa et al. 2008).

Study three of the Walumbwa et al. (2008) ALQ development effort involved a sample of 478 usable participants from 11 different US-based companies in Kenya with a sample population comprised almost exclusively of Africans. Two data collections efforts were performed 6 weeks apart with demographics and ALQ information gathered at time 1, and job satisfaction and job performance data collected later with the longitudinal separation intended to minimize bias (Walumbwa et al. 2008). In addition to deriving further support for the model structure, study 3 demonstrated predictive validity and additional cultural applicability of the instrument (Walumbwa et al. 2008).

The multistage development and testing work of the original Walumbwa et al. (2008) research provided an initial foundation upon which independent follow-on research further evaluated reliability, validity, and cultural generalizability of the ALQ. Specific details pertaining to the initial Walumbwa et al. (2008) research are incorporated with the reviews of the follow-on research studies to offer a more comprehensive perspective on the instrument’s psychometric properties.

Reliability evidence across research

Reliability of the ALQ was evaluated during the instrument development effort and repeated by a large number of follow-on studies with consistently acceptable results. Reliability refers to the consistency of measurement, and the ability to measure with minimum error (Cabanda et al. 2011), and while reliability is not sufficient to establish validity, it is a necessary element (Hair et al. 2010). Reliability is typically the first measure examined by researchers to ensure measurement consistency, and while it can be one initial indicator of convergent validity (Hair et al. 2010), it does not assure that the construct of interest is the phenomenon being measured (Cabanda et al. 2011). Reliability can also be confirmed through test–retest correlations (Cabanda et al. 2011), but the research reviewed for the ALQ relied exclusively on internal consistency evaluations.

Of 11 studies (Darvish and Rezaei 2011; Hsiung 2012; Leroy et al. 2012; Nielsen et al. 2013; Peus et al. 2012; Qian et al. 2012; Walumbwa et al. 2008; Wang and Bird 2011; Wong and Laschinger 2012) reviewed that employed the ALQ instrument, all tested for reliability using the internal consistency approach only, and all studies reported good Cronbach’s alpha values. Table 1 summarizes the eight studies that evaluated the second tier authentic leadership construct and the subscales for authentic leadership. All studies yielded alpha values greater than .70, indicating respectable reliability, and many measured alpha values greater than .80 demonstrating very good reliability (DeVellis 2012). Two additional studies tested only the second order overall authentic leadership measure; Hsiung (2012) in Taiwan found an alpha value of .96, and Leroy et al. (2012) in a Belgium study reported .95 for the scale. Nielsen et al. (2013), in studying a Norwegian population, also reported the second-order authentic leadership construct for reliability analyses, and confirmed internal reliability for the four subscales finding Cronbach’s alphas of .82 to .95 without reporting specific values by factor. The good Cronbach’s alpha values across a variety of study populations, cultures, and languages provided consistent and broad support for the reliability characteristics of the ALQ.
Table 1

Reliability coefficients (Cronbach’s alpha) values for studies employing the authentic leadership questionnaire (ALQ)

Study authors

Cronbach’s alpha values

Country of participants

Self-awareness

Relational transparency

Internalized moral perspective

Balanced processing

ALQ

Walumbwa (2008)

      

 Study 1

US/China

.92

.87

.76

.81

NR

 Study 2, population 1

US

.85

.74

.82

.74

NR

 Study 2, population 2

US

.85

.78

.78

.77

NR

 Study 3

Kenya

.73

.77

.73

.70

NR

Darvish and Rezaei (2011)

Iran

.85

.87

.82

.80

NR

Peus et al. (2012)

German

.86

.81

.85

.78

.94

Qian et al. (2012)

China

.89

.89

.86

.84

.96

Wang and Bird (2011)

US

.92

.87

.76

.81

NR

Wong and Laschinger (2012)

Canada

.93

.88

.89

.86

.97

NR not reported

Authentic leadership questionnaire validity support

Validity of an instrument such as the ALQ is a central issue in determining the measure’s usefulness (Girden and Kabacoff 2011). Validity refers to whether the instrument measures the characteristics intended (Cabanda et al. 2011), and assessing validity is a cumulative ongoing process (DeVellis 2012). To examine evidence of validity for the ALQ, I considered the theoretical face and content validity, convergent and discriminant validity characteristics, and nomological validity. The intent was to consider not only the statistical evidence, but the theoretical foundation for the ALQ, the soundness of the items, factors and construct, and the expected relationships with other measures and outcomes (Cabanda et al. 2011; Hair et al. 2010).

Face and content validity

Hair et al. (2010) suggested that in determining validity, the theoretical soundness of the measure must be considered before the statistical assessments. While individual users of the research instrument necessarily assess the theoretical fit of the measure within their intended research purpose, such as the theoretical evaluation demonstrated by Darvish and Rezaei (2011) for their authentic leadership research, establishing face and content validity depends primarily on the scale developers forming a clear definition of the concept and developing items that properly reflect the construct (DeVellis 2012). So face and content validity is initially and primarily addressed during scale development (DeVellis 2012). Walumbwa et al. (2008) developed the authentic leadership construct definition from prior research literature which guided the follow-on scale development. The definition incorporated the four dimensions that would become the instrument’s factors as follows: balanced processing of information, self awareness, an internalized moral perspective, and relational transparency (Walumbwa et al. 2008).

Walumbwa et al. (2008) did not use the more common exploratory factor analysis to define the construct and refine the items, but used literature, theory, a variety of experts, and both inductive and deductive methodologies to generate items for measuring authentic leadership and the innate dimensions. In addition, qualitative descriptions of authentic leaders were solicited from a group of doctoral students, and their responses were found to be similar to the theoretical categories and the four domains of authentic leadership established for the ALQ (Walumbwa et al. 2008). The researchers also compared authentic leadership theory with ethical and transformational leadership concepts for overlaps and distinctions (Walumbwa et al. 2008). To further support content and face validity as well as confirm the subdimensions of the construct, the initial pool of items was then independently assigned to one of the four categories of authentic leadership by a group of faculty and doctoral students (Walumbwa et al. 2008). Items that were linked to the proper category by 80 % of the reviewers were retained resulting in retention of 16 of the 22 items reviewed (Walumbwa et al. 2008). The approach of developing the sound theoretical construct definitions followed by broad expert involvement in the generation and confirmation of dimensions and items offered strong support for the face and content validity of the ALQ (Walumbwa et al. 2008). Walumbwa et al. (2008) posited that the four factors resulting from initial development efforts did not describe distinct constructs, but rather the various dimensions that combine to define authentic leadership. Such interrelationships of the subscales should have been evident in later research, and in fact, later studies generally did supported the initial theoretical design.

Convergent and discriminant validity

Support for convergent and discriminant validity of the ALQ was provided initially through the confirmative factor analysis (CFA) performed by Walumbwa et al. (2008), and later reinforced by other researchers who, when applying the ALQ instrument in their research, confirmed the validity characteristics. Walumbwa et al. (2008) and others analyzed the one-factor model, where all individual items were fit to the overall authentic leadership construct. Additionally, later research widely supported the second-order factor model whereby the items were loaded onto their individual factors and the four authentic leadership dimension’s factors loaded onto the second-order authentic leadership factor (Caza et al. 2010; Walumbwa et al. 2008). For clarity, Table 2 depicts only the CFA model fit statistics for the second-order models of authentic leadership (Hair et al. 2010; Walumbwa et al. 2008). As reported, consistent reliability of the ALQ also supported convergent validity (Hair et al. 2010). Hair et al. (2010) emphasized that for interpreting CFA as depicted in Table 2, there are no specific rules or singular indexes that can determine whether a model demonstrates acceptable fit, but rather researchers should consider a variety of measures including the χ2 value, an additional absolute index such as the RMSEA or SRMR, and an incremental index such as the CFI. The fit statistics must also be considered based on the characteristics of the research (Hair et al. 2010). The results depicted in Table 2 generally and consistently support reasonable fit of the data to the second-order construct of the ALQ.
Table 2

CFA validity measures for studies employing the authentic leadership questionnaire (ALQ)—second-factor models

Study authors

Country

n

χ2

df

CFI

RMSEA

SRMR

Factor loadings

Walumbwa (2008)

 Study 1, population 1

US

224

234.70***

98

.97

.05

NR

.66–.93***

 Study 1, population 2

China

212

176.03***

98

.95

.06

NR

.62–.78***

 Study 3

Kenya

478

247.97**

95

.97

.06

NR

62–.78NR

Caza, et al. (2010)

New Zealand

960

1,833.89NR

682

.97

.065

.06

.55–.81 NR

Leroy (2012)

Belgium

252

133.41*

100

.97

.08

.06

NR

Peus, et al. (2012)

Germany

306

251.15***

100

.94

.07

NR

NR

Qian et al. (2012)

China

237

191.29 NR

50

.94

.08

NR

NR

Wang and Bird (2011)

US

917

705.20 NR

100

.98

.08

.04

.55–.90 NR (note)

NR not reported

*p < .05; **p < .01; ***p < .001

Wang and Bird factor loadings mostly >.70, one item in transparency shown on chart as .21 was suspect

Evaluating discriminant validity across similar constructs, Leroy et al. (2012) confirmed discriminant validity in comparing the ALQ with behavioral integrity, and Walumbwa et al. (2008), in study 2, confirmed that authentic leadership, while related to both ethical and transformational leadership measures, was distinguishable in measurement. Caza et al. (2010) performed perhaps the most detailed discriminant analyses between the factors by examining confidence intervals and found that many of the first-order or subfactors of authentic leadership failed to demonstrate adequate discriminant validity. Combined, the Caza et al. (2010) findings and the results of Walumbwa et al.’s (2008) initial development both of which demonstrated convergence between the four dimensions within the ALQ, and “suggest that it might not be reasonable to conceptualize the measures as assessing entirely separate and distinct constructs” (Walumbwa et al. 2008, p.101). That is, the second-order authentic leadership construct may be a better predictor than the four individual dimensions (Caza et al. 2010). The variety of studies, populations, cultures, and settings offer early stage support the convergent and discriminant validity of the ALQ, specifically at the second-order construct level pending additional research and confirmation.

Nomological validity

As an emergent concept, nomological validity of the ALQ, the demonstration that the measure relates as expected to other theoretical constructs, is an important characteristic of the instrument (DeVellis 2012; Hair et al. 2010) and must be demonstrated to confirm validity of the construct (Walumbwa et al. 2008). The research reviewed consistently supported theoretical relationships between authentic leadership as measured by the ALQ and a variety of expected positive outcomes including organizational citizenship behavior (Walumbwa et al. 2008), organizational or team commitment (Darvish & Rezaei 2011; Leroy et al. 2012; Walumbwa et al. 2008), satisfaction with supervisor (Peus et al. 2012; Walumbwa et al. 2008), behavioral integrity (Leroy et al. 2012), job performance (Leroy et al. 2012; Peus et al. 2012; Walumbwa et al. 2008; Wong and Laschinger 2012), employee voice behavior (Hsiung 2012), job satisfaction (Darvish and Rezaei 2011; Walumbwa et al. 2008; Wong and Laschinger 2012), perceived team effectiveness (Peus et al. 2012), trust, engagement, empowerment (Wang and Bird 2011), feedback seeking behavior (Qian et al. 2012), and safety climate (Nielsen et al. 2013). While further nomological or predictive validity testing is needed, since validation is an ongoing process (DeVellis 2012), the consistent validation in recent research supports the predictive validity of the ALQ.

Conclusion

Authentic leadership theory has emerged primarily over the last 10 years (Gardner et al. 2005), and initial development of the ALQ was completed even more recently (Walumbwa et al. 2008), so the construct and measurement instrument are both in early stages of their evolution. While a variety of research studies conducted in that short time span across many cultures have yielded support for the reliability and validity of the ALQ, especially for the second-order model (Caza et al. 2010; Darvish and Rezaei 2011; Hsiung 2012; Leroy et al. 2012; Nielsen et al. 2013; Peus et al. 2012; Qian et al. 2012; Walumbwa et al. 2008; Wang and Bird 2011; Wong and Laschinger 2012), further exploration of the four-dimension structure is needed to determine if the first order constructs demonstrate discriminant validity, that is, whether they are unique (Caza et al. 2010; Walumbwa et al. 2008). Discriminant validity and theoretical development has demonstrated how authentic leadership, as conceptualized and measured by the ALQ, contributes to understanding leadership beyond related positive contemporary leadership theories such as ethical and transformational leadership (Walumbwa et al. 2008). In addition to future evaluation of the construct and predictive validity of the ALQ across additional cultures and populations, exploring the correlation and discrimination with other contemporary theories, such as spiritual leadership and servant leadership in a manner similar to that of Walumbwa et al. (2008) performed for transformational and ethical leadership, would also be helpful in defining the overlaps and unique elements of authentic leadership.

Results to date have been encouraging, especially in terms of predictive validity and reliability, and early global research has supported the generalizability of the ALQ across a variety of cultures and languages. While Hair et al. (2010) suggested that for each research design, content, convergent, discriminant, and nomological validity and reliability should all be evaluated prior to scale use, even for preexisting scales, such thorough evaluation is especially necessary for early-stage instruments such as the ALQ. As DeVellis (2012) reminds, validity is not a static process but cumulative and ongoing. So it is for the ALQ.

Copyright information

© Springer Science+Business Media Dordrecht 2013