Construct validation of Joubert's value oriëntation measure

Abstract

The purpose of this research was to determine empirically the construct validity of Joubert's typology of value orientations in the rapidly changing social context in South Africa. The validation sample consisted of students from historically disadvantaged communities in South Africa who had registered for fulltime degree studies at a multi-campus university in South Africa for the first time. The construct validity of the typology was examined through submitting the items of each scale to a test of internal consistency. All of the 16 scales had alpha coefficients less than .43. Therefore, Joubert's theoretically derived Value Orientation measure had insufficient support for its construct validity in the sample used. A principal factor analysis with varimax rotation was performed and squared multiple correlations were employed as the initial communality estimates which were then iterated. The factor analysis produced only 4 correlations exceeding .30 and this fact in conjunction with the low squared multiple correlations suggested that the factors were not well defined.

Opsomming
Die doel van hierdie navorsing was om empirics die konstrukgeldigheid van Joubert se waarde orientasie tipologie in die vinnig veranderende Suid-Afrikaanse sosiale konteks te bepaal. Die geldigheid is getoets op 'n steekproef studente vanaf historiesafgeblewene gemeenskappe in Suid-Afrika wat vir die eerste keer vir voltydse graadstudies geregistreer het by 'n multikampus Universiteit in Suid-Afrika. Die konstrukgeldigheid van die tipologie is ondersoek deur die items van eike skaal te onderwerp aan 'n toets van interne konsekwentheid. Al 16 skale is gekenmerk deur alpha koeffisiente van minder as .43. Dus het Joubert se teoreties ontwikkelde Waarde Orientasie-instrument ongenoegsame konstrukgeldigheid getoon vir die huidige steekproef. 'n Hooffaktoranalise met varimaxrotasie is uitgevoer en gekwadreerde veelvuldige korrelasies is gebruik as die aanvanklike gemeenskaplike beramings wat daarna in die iterasies gebruik is. Die faktoranalise het slegs 4 korrelasies gelewer wat .30 oorskry het, en die feit met inagneming van die lae gekwadreerde veelvuldige korrelasies, suggereer dat die faktore nie goed gedefinieer is nie.