Industrial psychology is the study of working man, both as individual employee and as a collective, that attempts to explain and potentially change his behaviour. However, the behaviour of man is complexly determined and therefore never fully predictable; it is widely accepted in the field (e.g. Cilliers, 1998) that the pursuit of a perfect, complete explanation of working man’s behaviour is an ideal that is to be considered unobtainable.
That is not to consider the study of human nature and working man to be fruitless. As a science, we have succeeded and continue to succeed in gaining approximations of the nomological net of latent variables that underpin behaviour that are to be considered satisfactory for the objective of influencing the behaviour of working man. Man can, however, only improve his ability to derive interventions that will successfully affect the performance of working man to the extent that he can extend his understanding of the vast and complex nomological net that regulates the level of performance that individual employees and organisational work units achieve. More penetrating insight into the various constructs which form the psychological mechanism that regulates the level of performance has to be sought as well as the nature of the structural relationships that exist between the various constructs so as to understand the working of the psychological mechanism in its entirety. Of importance is the fact that the understanding of the psychological mechanism lies spread across all the components (i.e. constructs) and structural relations comprising the mechanism (Cilliers, 1998). Omission of components and/or structural relations therefore invariably mean a loss in meaning or understanding and a concomitant loss in control.
The study of the social sciences has created and adopted many procedures and methods that allow the researcher the greatest opportunity to gain an understanding of human behaviour. Structural equation modelling (Gefen, Straub & Boudreau, 2000; Hox & Bechger. 1998) is one of these aforementioned procedures that is widely employed and has been proven to provide a large amount of insight. The value of structural equation modelling lies firstly in the fact that it promotes the construction of psychological mechanisms in explanatory research. Its value secondly lies in the fact
that it permits the empirical evaluation of complex hypotheses on the nature of the psychological mechanism as integrated entities. Structural equation modelling therefore explicitly acknowledges that the understanding of the psychological mechanism lies spread across all the components (i.e. constructs) and structural relations comprising the mechanism (Cilliers, 1998). In addition, it allows complex hypothesised relationships between latent variables to be studied rather than the corresponding relationships between fallible measures of the latent variables (Diamantopoulos & Siguaw, 2009). Acknowledging this success and these advantages, it will be used extensively in this research.
Essentially, a structural model is used to explain a phenomenon that is observed in reality (i.e., in Babbie and Mouton’s (2001) World 1). It is appreciated that activities and events, especially in the social sciences and industrial psychology, are not random but are as a result of a large number of inter-related cause and effect relationships. These relationships, many of which are unknown to man, are essential to the understanding of the phenomenon of interest (Babbie & Mouton, 2001; Diamantopoulos & Siguaw, 2009). A structural model is therefore essentially a stance on what the nomological net underpinning performance looks like, or stated differently, an illustration of a psychological mechanism that is capable of regulating the levels of the latent variables constituting the performance construct5.
A structural model consists of a large number of constructs (or latent variables) that represents each factor in the previously mentioned cause and effect relationships that comprise the psychological mechanism that regulates the phenomenon of interest. These constructs can be categorised into exogenous variables, namely variables that purely act as causes in the structural model, and endogenous variables that act both as cause and dependent variables in the structural model (Diamantopoulos & Siguaw, 2009). A major advantage of structural equation modelling as a statistical technique, is that it allows the empirical testing of hypotheses that postulate structural (i.e., causal) relationships between endogenous latent variables (Diamantopoulos & Siguaw, 2009).
5 It is acknowledged, given Popper’s (1972) falsification principle, that a definite claim on the nature of the
nomological net underpinning performance is not possible. At best it can be claimed that a specific stance on what the nomological net looks like is valid (i.e. permissible) because it survived an opportunity to be falsified.
Competency modelling, as interpreted in the current study, is a form of structural modelling that describes what is meant to successfully complete a job and that explains what determines the degree of success that is achieved (Myburgh, 2013). In addition to explicating the internal structure of the performance construct, the model suggests the necessary person qualities and environmental conditions that are required to adequately complete a job. These previously explicated components of a competency model can be categorised into four differing structurally inter-related domains namely; situational latent variables, competency potential latent variables, competency latent variables and outcome latent variables.
For the purpose of the research proposal, the conceptualisation by SHL (2011, p. 6) of a competency model will be utilised as basis of the current study’s conceptualisation of a competency model:
A model of performance at work that defines the relationship between competency potential, competency requirements and competencies themselves. ‛Competencies’ are defined as behaviours that support the attainment of organisational objectives. ‛Competency potential’ is seen to derive from individual dispositions and attainments and ‛competency requirements’ refer both to facilitators of and barriers to effective performance in the workplace.
The SHL 2011 and Bartram (2005 & 2006) conceptualisation of a competency model essentially thought of the various domains as comprising of lists of characteristics, behaviours and outcomes. The emphasis that the current study places on the structural relations existing between various elements within a specific domain as well as between domains did not form part of the original SHL thinking on competency models. Therefore, using the above definition as basis, but integrating it with the concept of structural equation modelling (Diamantopoulos & Siguaw, 2009; Du Toit & Du Toit, 2001), a competency model is a set of structurally related competency potential latent variables that affect a set of structurally related competency latent variables, which in turn, affect a set of structurally related outcome latent variables. Further, the effect of the competency potential variables on the competency latent variable and the competency latent variable on the outcome latent variable is moderated by situational characteristics. Situational characteristics are finally also allowed to exert main effects on the competency potential latent variables and latent outcome variables.
Competency modelling refers to the act of developing and empirically testing the validity of the model by testing the fit of the model and the statistical significance of the hypothesised structural linkages.
Whilst the latent variables in the competency model can be categorised into four differing domains, it is the fact that these latent variables within and across the domains are structurally related in a richly interconnected manner in which they impact on each other in cause and effect relationships that prevents the location of the understanding of performance in any specific latent variable or structural linkage. The totality of a model is of critical importance to fully understand a phenomenon as each latent variable and each structural linkage is a necessary but not sufficient component to describe the working of the psychological mechanism that plays a unique role in the dynamic of the mechanism. Thus, the entirety, structural interrelatedness and comprehensiveness of the competency model is what gives the model its value. A basic representation of this relationship is presented below in Figure 2.1.