The Authentic Leadership Inventory: measurement invariance in selected industries in South Africa
Abstract
A need exists for validating the authentic leadership construct across different contexts and ensuring measurement invariance of the Authentic Leadership Inventory (ALI) in different groups. Despite this need, few attempts have been made in a South African context to evaluate the psychometric properties of this instrument and to demonstrate whether it is invariant. This is important given the fact that South Africa differs significantly from Western countries where the majority of theorising and empirical testing of the construct was performed. These differences may influence the way in which individuals interpret observed indicators. Furthermore, leadership is socially constructed and context-dependent, influencing both the interpretation of leadership behaviour as well as the effectiveness thereof in different organisational contexts.
The aim of this study was to evaluate measurement invariance of the ALI in two organisations from different industries, after the construct and concurrent validity of the instrument had been investigated. A cross-sectional survey with two convenience samples, namely mining (N = 244) and public health care (N = 633), was used. The ALI, a sub-scale of the Workplace Trust Scale (WTS), and a biographical questionnaire were administered. The results showed that a one-factor model fitted the data best in the mining organisation. This is contradictory to the original conceptualisation of authentic leadership which is considered to be a higher-order construct comprising four lower-order dimensions, namely self-awareness, balanced processing, internalised moral perspective and relational transparency. Results indicated that authentic leadership significantly predicted followers’ trust in the leader; thereby establishing the concurrent validity of the instrument. The instrument was also found to be reliable (i.e. internally consistent) in the two organisations.
Measurement invariance by gender and organisational groups was evaluated and in both instances only partial (configural and scalar) invariance could be established, indicating the biasness of certain items in the measure. Finally, no significant differences existed in terms of the mean levels of authentic leadership for the respective gender and organisational groups. The results highlighted the importance of evaluating the equivalence of an instrument in different contexts, especially instruments measuring leadership behaviour.