Instemmingsgeneigdheid en verskillende item- en responsformate in 'n gesommeerde selfbeoordelingskaal

Abstract

This study examines the degree of acquiescence present when the item and response formats of a summated rating scale are varied. It is often recommended that acquiescence response bias in rating scales may be controlled by using both positively and negatively worded items. Such items are generally worded in the Likert-type format of statements. The purpose of the study was to establish whether items in question format would result in a smaller degree of acquiescence than items worded as statements. the response format was also varied (five- and seven-point options) to determine whether this would influence the reliability and degree of acquiescence in the scales. A twenty-item Locus of Control (LC) questionnaire was used, but each item was complemented by its opposite, resulting in 40 items. The subjects, divided randomly into two groups, were second year students who had to complete four versions of the questionnaire, plus a shortened version of Bass's scale for measuring acquiescence. The LC version were questions or statements each combined with a five- or seven-point respons format. Partial counterbalancing was introduced by testing on two separate occasions, presenting the tests to the two groups in the opposite order. The degree of acquiescence was assessed by correlating the items with their opposite, and by correlating scores on each version with scores on the acquiescence questionnaire. No major difference were found between the various item and response format in relation to acquiescence.

Opsomming
Hierdie ondersoek is uitgevoer om te bepaal of die mate van instemmingsgeneigdheid deur die item- en responsformaat van 'n gesommeerde selfbeoordelingskaal beinvloed word. Daar word dikwels aanbeveel dat die gebruik van positief- sowel as negatiefbewoorde items in 'n vraelys instemmingsgeneigdheid beperk. Suike items word gewoonlik in die tradisionele Likertformaat as stellings geformuleer. Die doel van die ondersoek was om te bepaal of items in vraagformaat minder instemmingsgeneigdheid as die stellingformaat tot gevolg het. Die responsformaat is ook gevarieer (vyf- en sewepunt opsies) om te bepaal of die betroubaarheid en graad van instemmingsgeneigdheid daardeur beinvloed word. 'n Twintigitem Lokus van Beheer (LvB) vraelys is vir hierdie doel gebruik, maar vir eike item is "n inverse gekonstrueer, sodat die uiteindelike vraelys uit 40 items bestaan het. Die steekproef wat ewekansig in twee groepe verdeel is, was tweedejaarstudente wat elk vier weergawes van die vraelys asook "n verkorte weergawe van Bass se mstemrningsgeneigdheidsvraelys moes voltooi. Die LvB-weergawes was vrae of stellings wat elk met 'n vyf- of sewepuntresponsformaat gekombineer is. Gedeeltelike teenbalansering is toegepas deur twee toetsgeleenthede waarin die vraelys aan die onderskeie groepe in die teenoorgestelde volgorde geadministreer is, te gebruik. Die teenwoordigheid van instemmingsgeneigdheid is bepaal deur die items met hul inverses te korreleer, asook om eike vraelys met die instemmmgsgeneigdheidsvraelys te korreleer. Geen noemenswaardige verskille is tussen die verskillende item- en responsformaat, met betrekking tot instemmingsgeneigdheid, gevind nie.