The development and validation of a generic non-managerial performance measure

Myburgh, Hester Magdalena (2013-03)

Thesis (MComm)--Stellenbosch University, 2013.

To access data please use IBM SPS Statistics software.

Thesis

ENGLISH ABSTRACT: The human resource function attempts to proactively and reactively affect the work performance of employees through advice, diagnosis and interventions in a manner that will benefit the quality of the product and service an organisation provides to the market. In this study a generic performance construct (that can be applied to non-managerial, individual positions), was defined based on previous studies in the area of generic performance. A South African performance measure was designed accordingly that can be used to obtain multi-rater assessments of the generic, non-managerial, individual performance construct. This questionnaire was validated by evaluating the fit of the measurement model, using confirmatory factor analysis. Most of the data for this study was obtained from OK Furniture. A few questionnaires were completed by a non-probability sample of non-managerial personnel from a variety of organisations. The total sample size comprised of 205 respondents. Item and dimensionality analysis was performed on the 12 subscales to assess the success with which they represented the underlying performance constructs. In the item analysis two items were identified as somewhat problematic but both were judged not to be sufficiently problematic to delete them from the GPQ. Results from the dimensionality analysis showed satisfactory evidence that the items of each subscale reflect a common underlying factor. A spectrum of goodness-of-fit statistics was used to assess the measurement model fit. The hypothesis of exact fit was rejected but the hypothesis of close fit could not be rejected (p>.05). The position that the measurement model fits the data closely in the population was found to be a tenable position. The fit indices reflected good model fit in the sample. The measurement model parameter estimates indicated that the indicator variables represented the latent performance dimensions satisfactorily. Discriminant validity was investigated. Mixed evidence on discriminant validity was obtained. The sample size of this study was satisfactory when viewed from the perspective of statistical power given the method of item parcelling that was used, but a larger and more representative sample would have been preferable in that it would have allowed the GPQ measurement model to be fitted with individual items as indicator variables. Another limitation seemed to have been the language of the questionnaire. Informal feedback suggested that some respondents, especially those that are not fluent in English, struggled to fully understand all the questions. Recommendations for future research are made.

AFRIKAANSE OPSOMMING: Die menslike hulpbronfunksie in organisasies poog om die kwaliteit van die produk of diens wat die organisasie aan die mark lewer te verbeter deur werknemers se werksprestasie proaktief en reaktief te beïnvloed deur middel van menslike hulpbronbestuursadvies, -diagnose en -intervensies. In hierdie studie is 'n generiese prestasiekonstruk gedefinieer (gerig op individuele nie-bestuursposte) gebaseer op vorige studies van generiese prestasie. Dit is hoofsaaklik gemik op individuele nie-bestuursvlak poste. ‘n Suid Afrikaanse prestasiemeetinstrument is ontwikkel wat gebruik kan word om multi-beoordelaar assesserings van die generiese nie-bestuursprestasiekonstruk te verkry. Die vraelys is gevalideer deur, met behulp van bevestigende faktorontleding, die pasgehalte van die metingsmodel te ondersoek. Die meerderheid data vir hierdie studie is verkry vanaf die OK Furniture-kettingwinkelgroep. Enkele vraelyste is ook voltooi deur 'n nie-waarskynlikheid steekproef van nie-bestuurspersoneel van ander organisasies. Die totale steekproefgrootte het bestaan uit 205 respondente. Item- en dimensionaliteit-analise is uitgevoer op die 12 subskale om die sukses waarmee hulle die onderliggende prestasiekonstrukte verteenwoordig te evalueer. In die item-analise is twee items geïdentifiseer wat moontlik problematies kan wees maar nie sodanig problematies dat dit hul verwydering uit die GPQ regverdig het nie. Die resultate van die dimensionaliteit-analise het aangetoon dat die items van elke subskaal inderdaad redelik bevredigend ’n gemeenskaplike onderliggende faktor reflekteer. 'n Spektrum van pasgehaltestatistiek is gebruik om die pasgehalte van die metingsmodel te beoordeel. Die hipotese dat die model die data perfek pas is verwerp, maar die hipotese van benaderde passing kon nie verwerp word nie (p>.05). Die pasgehalte maatstawwe dui op goeie modelpassing in die steekproef. Die skattings van die metingsmodelparameters dui daarop dat die aanwyserveranderlikes die prestasie-dimensies bevredigend verteenwoordig. Diskriminantgeldigheid is ondersoek. Gemengde bewyse van diskriminantgeldigheid is verkry. Beskou vanuit die perspektief van statistiese krag was die steekproefgrootte van hierdie studie bevredigende (gegewe die gebruik van item pakkies) ‘n Groter en meer verteenwoordigende steekproef sou egter wenslik gewees het insoverre dit die passing van die GPQ metingsmodel met individuele items sou toelaat. ‘n Verdere oënskynlike beperking was die vraelys se taalgebruik. Informele terugvoer dui daarop dat sommige respondente, veral dié wat nie vlot Engels praat nie, gesukkel het om sommige vrae te verstaan. Aanbevelings vir verdere navorsing word gemaak.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/80083
This item appears in the following collections: