The measurement invariance and measurement equivalence of the sources of work stress inventory (SWSI) across gender groups in South Africa

Davis, Samantha (2014-12)

Thesis (MCom)--Stellenbosch University, 2014.

You will be needing the program SPSS in order to read the .spv files

Thesis

ENGLISH ABSTRACT: The primary goal of an organisation, in a capitalistic system, is the maximisation of profit. The task of the human resource function in organisations is to affect the work performance of working man to the advantage of the organisation and in a manner that adds value to the organisation. The management of employee wellbeing/psychological health is one of the human resource interventions with which the human resource function pursues this objective. It is imperative for organisations to be aware of, and sensitive to, negative factors in the workplace, such as occupational stress, that influence employees’ health and wellbeing and have a significant effect on job satisfaction and performance (Hamidi & Eivazi, 2010). Prevailing stress levels need to be monitored regularly if escalating stress levels are to be detected in time to prevent serious personal and organisational problems from developing. The Sources of Work Stress Inventory (SWSI) is an instrument developed in South Africa specifically for this purpose (De Bruin & Taylor, 2005). The inappropriate use of occupational stress assessments across genders can seriously jeopardize the extent to which occupational stress assessments, and the decisions based on them, achieve their intended objectives. In order to avoid making widespread generalisations and untested assumptions which will eventually do a disservice to the field of psychology, the absence of measurement bias (i.e. invariance and equivalence) should be demonstrated instead of simply assumed (Van de Vijver & Tanzer, 2004). Establishing the measurement invariance and equivalence of an instrument across groups should be a prerequisite to conducting substantive cross-group comparisons (Dunbar, Theron & Spangenberg, 2011). It is imperative to empirically ascertain whether the instruments that are used are free of cultural, language, gender, age and racial bias, not only because it is prohibited by the Employment Equity Act 55 of 1998, but also as it is in the interest of good workmanship. Bias is indicated as nuisance factors that threaten the validity of cross-group (cultural) comparisons (Van de Vijver & Leung, 1997). These nuisance factors could be due to construct bias, method bias and/or item bias. Due to the importance of the decisions made, it would seem essential that the information provided by test results apply equally across different reference groups. In this study the specific measurement invariance and equivalence sequence of tests set out by Dunbar et al. (2011) was used to answer a sequence of research questions that examine the extent to which the SWSI multi-group measurement model may be considered measurement invariant and equivalent or not, and to determine the source of variance if it existed (Vandenberg & Lance, 2000). Upon investigating the measurement model fit of the SWSI, the results indicated that support was found for the hypotheses that the measurement model fits the data of both gender samples independently. Furthermore, support was found for the configural and weak invariance model. However, due to not meeting the requirements for metric equivalence, partial measurement invariance and equivalence was explored. The SWSI multi-group measurement model met the requirements of partial complete invariance and partial full equivalence, and the non-invariant items were identified in the process. The implications of the results are discussed, limitations are indicated and areas for further research are highlighted.

AFRIKAANSE OPSOMMING: Die kerndoelwit van enige organisasie, veral in ‘n kapitalistiese stelsel, is om optimale wins te genereer. Die taak van die menslike hulpbronbestuurfunksie binne organisasies is om die werksverrigting van die werkende mens te beïnvloed tot voordeel van die organisasie en terselfdetyd waarde tot die organisasie toe te voeg. Die bestuur van ‘n werknemer se welstand / sielkundige gesondheid is een van die menslike hulpbron-iintervensies waarmee die menslike hulpbronfunksie hierdie doelwit nastreef. Dit is uiters belangrik vir organisasies om bewus te wees van, asook sensitief te wees vir, negatiewe faktore soos werkstres, wat werknemers se gesondheid en welsyn beïnvloed en wat 'n beduidende invloed op werkstevredenheid en prestasie het (Hamidi & Eivazi, 2010). Heersende stresvlakke moet gereeld gemonitor word om tydig stygende stresvlakke te bespeur ten einde ernstige persoonlike en organisasieverwante probleme te verhoed. Die Bronne van die Werkstres-inventaris (BWSI) is in Suid-Afrika spesifiek vir hierdie doel ontwikkel (De Bruin & Taylor, 2005). Die ontoepaslike gebruik van werkstresmetings oor geslagte kan egter die mate waartoe beroepstresmetings en die besluite wat daarop gebaseer word hul oogmerke bereik ernstig benadeel. Die afwesigheid van metingsydigheid (bv. invariansie en ekwivalensie) moet dus empiries gedemonstreer word, in stede daarvan dat die afwesigheid daarvan eenvoudig aanvaar word (Van de Vijver & Tanzer, 2004). Die afwesigheid van hierdie informasie kan lei tot wydverspreide veralgemenings en ongetoetsde aannames wat die Sielkunde professie ernstige skade kan berokken. Die meetings-invariansie en -ekwivalensie van 'n instrument oor groepe is 'n voorvereiste vir substantiewe kruis-groepvergelykings (Dunbar, Theron & Spangenberg, 2011). Dit is noodsaaklik om empiries te bepaal of die instrumente wat gebruik is vry is van kulturele-, taal, geslag-, ouderdom- en rasse-sydigheid, nie net omdat dit verbied word deur die Wet op Diensbillikheid 55 van 1998 nie, maar ook omdat dit in die belang van goeie vakmanskap is. Sydigheid is sistermatiese steurnisse wat die geldigheid van die kruis-groep (kulturele) vergelykings (Van de Vijver & Leung, 1997) bedreig. Hierdie steurnisse kan wees as gevolg van konstruk-, metode- en/of itemsydigheid. Gegewe die belangrikheid van die besluite wat geneem word gebaseer op die metings is dit noodsaaklik dat die inligting vergelykbaar oor die verskillende verwysingsgroepe is. Die studie het die stel metingsinvariansie en -ekwivalensie toetse wat deur Dunbar et al. (2011) gebruik om 'n reeks van navorsingsvrae te beantwoord. Daar is ondersoek gestel na die mate waartoe die BWSI multi-groep metingsmodel as invariant of ekwivalent beskou kan word, en die bron van variansie te bepaal as dit sou bestaan (Vandenberg & Lance, 2000). In die ondersoek na die metingsmodel passing van die BWSI, is daar ondersteuning gevind is vir die hipoteses dat die metingsmodel beide van die geslagsteekproewe goed pas. Steun is ook gevind vir die konfigurale en swak invariansie modelle. Aangesien slegs beperkte steun vir metriese ekwivalensie gevind is, is ondersoek na die parsiële metriese invariansie en ekwivalensie ingestel. Die BWSI multi-groep metingsmodel het voldoen aan die vereistes van parsiële volledige invariansie en parsiële volle ekwivalensie, en die nie-invariante items is deur die proses geïdentifiseer. Die implikasies van die resultate word bespreek, beperkinge word aangedui en areas vir verdere navorsing word uitgelig.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/95741
This item appears in the following collections: