Abstract:

Training, as a core HR function in organizations, concerns competency models, which are theoretical frameworks used to build job descriptions, to plan professional objectives and, thus, to design effective training courses. Therefore, it is crucial to analyze the way these theoretical tools are built, because their practical application has tangible consequences on the way employees are trained, evaluated and/or promoted.
According to our experience and studies in this field, these models have many weak points, because of the bad methodology approaches they are developed with. Above all, we notice that: 1) they are permeated by theory and benchmark models which are mostly developed in remote organizational and academic contexts; 2) they are rigid and static tools in a changeable world, unable to interpret the present or to forecast the continuous changes of professional profiles; 3) they consider competencies as they were tangible objects, each one indipendent from another.
We believe that a bottom-up perspective may tackle many of these problems. In the paper, we propose an alternative model to analyze the validity of the competency frameworks in organization. We have tested this method on an actual case, a big organization which has developed a competency model during a research project lasted 5 years and used it to design a one-year long training course for a professional profile.
At first, we collected the documents concerning the competency model elaborated by the organization and translated them in terms of single atomic elements (as behaviors). Then, we carried out a survey of a significant sample of people belonging to that profile/role, to get a self-evaluation on the listed behaviors. The main output of this phase was a matrix, with a series of multidimensional vectors, which is the mathematical representation of the profile.
In the second phase, we generated the Kohonen maps thanks to “SOM Analyzer” software, which classifies the observed elements in a three-dimensional space. Another software, in the third phase, helped us to find similarities among these maps and identify any emergent clusters: indeed, these group formations could be considered competencies. Then, we labelled them to allow a comparison between this data-driven configuration and the organizational competency model.
According to the results, we believe that this method presents several advantages: 1) it is completely customized on the organizational reality and allows to assess people according to a familiar and effective standard; 2) it is not constrained by repacked and strict categories; 3) it is a flexible and dynamic tool, independent of any theoretical point of view; 4) it is empirically adequate; 5) it is rigorous and reproducible.
Moreover, this methodology may be used in different organizational contexts, for different purposes, i.e. to create or validate competency models, to improve them or make them consistent with a variable and uncertain environment.

Keywords:

Competency model, Training, Organization, Technology, Kohonen map, Neural network, Evaluation.