Thesis (MComm)--University of Stellenbosch, 2007. / ENGLISH ABSTRACT: Understanding and measuring job performance constructs enables organisations to utilise their human resources more effectively. The job performance of workers can be conceptualised on two levels, namely on a job result or outcome level (e.g. sales figures, units produced) and on a behavioural level (e.g. planning a budget, attentiveness to customers). To the extent that the former is within the control of the worker responsible, its success is a function of the worker’s behaviour on the job. Successful performance on the job outcome level for which the job exists thus requires specific levels of performance (i.e., competence) on the behavioural competencies in that the latter complexly determines the former. The level of performance achieved on the outcome level could, however, also recursively feed back onto the level of performance reached on the behavioural competencies. A complex network of causal influences thus exist between the job outcomes for which the job exists and the latent behavioural competency variables. Although this multi-dimensional structural model between job behaviour and job outcomes are widely recognised in theory, it is not often developed or tested in practice. Such models will give credence to performance criteria used as part of performance management and will also assist organisations in selecting and evaluating job success predictors. In addition, these models can serve as diagnostic tools for organisational development. This research study develops and evaluates a performance or competency structural model for branch managers in the clothing retail industry. The results seem to suggest a reasonable good fit for the exogenous model (i.e. competency measurement model), but a poor fit for the endogenous model (i.e. job outcomes measurement model). Due to estimation problems with the endogenous model, multiple regression analysis is used instead of the more appropriate analysis in this case, structural equation modelling, for evaluating the structural model. The regression results confirm the importance of certain competencies in terms of unit performance and provide understanding of the rather complex performance domain. / AFRIKAANSE OPSOMMING: Die verstaan en meet van posprestasie stel organisasies in staat om hul menslike hulpbronne meer effektief te bestuur. Die posprestasie van werkers kan op twee vlakke gekonseptualiseer word, naamlik op ‘n posuitkomsvlak (bv. verkoopssyfers, eenhede geproduseer) en op ‘n gedragsvlak (bv. beplanning van ‘n begroting, oplettendheid teenoor klante). Sukses op die eersgenoemde vlak, in die mate waartoe dit binne die beheer van die verantwoordelike werker is, is ‘n funksie van die werker se gedrag in die werk. Suksesvolle prestasie op die posuitkomsvlak, waarvoor die pos bestaan, vereis dus spesifieke vlakke van prestasie (dws. bevoegdheid) op die gedragsvlak deurdat die laasgenoemde die eersgenoemde op ‘n komplekse wyse bepaal. Die vlak van prestasie behaal op die uitkomsvlak sou egter ook ’n terugvoereffek kon hê op die vlak van prestasie op die gedragsbevoegdhede. ‘n Komplekse netwerk van kousale invloede bestaan dus tussen die posuitkomse waarvoor die pos bestaan en die latente gedragsbevoegdhede veranderlikes. Alhoewel hierdie multi-dimensionele strukturele model tussen werksgedrag en posuitkomste wyd erken word in die teorie, word dit nie algemeen ontwikkel of getoets in die praktyk nie. Sulke modelle sal geloofwaardigheid bied aan prestasiekriteria wat aangewend word as deel van prestasiebestuur en sal ook bydra tot organisasies se seleksie en evaluasie van voorspellers vir possukses. Hierdie modelle kan verder as diagnostiese instrumentasie dien vir organisasieontwikkeling. Hierdie navorsingstudie behels die ontwikkeling en evaluering van ‘n strukturele prestasie- of bevoegdheidsmodel vir takbestuurders in die klere-kleinhandelindustrie. Die resultate blyk ‘n redelike goeie passing aan te dui vir die eksogene (bevoegdheids-) model, maar ‘n swak passing vir die endogene (posuitkoms-) model. As gevolg van beramingsprobleme met die endogene model word meervoudige regressieontleding gebruik in plaas van strukturele vergelykingsmodellering wat ‘n meer toepaslike analise in hierdie geval sou wees. Die regressieresultate bevestig die belangrikheid van sekere bevoegdhede in terme van eenheidsprestasie en werk mee om ‘n redelike komplekse prestasiedomein meer verstaanbaar te maak.