The reliability and validity of a social identity inventory in the South African tertiary education context.
Social Identity Theory’s (SIT) explanations of psychological and group phenomena have been endorsed repeatedly through numerous laboratory experiments (to cite a few Tajfel, 1978; Tajfel, 1982; Turner, & Oakes, 1997; Oakes, Haslam, & Turner, 1994; Ellmers, Spears & Doosje, 2002; Hogg & Cooper, 2007). However many studies applying the framework include only a small subset of the key SIT constructs that form an integral part of the overall model (Abrams & Brown, 1978; Turner & Brown, 1978). Consequently, there are only a few measures of these socio-structural variables that are widely available internationally and this problem is more pronounced in the local South African context. This research study thus aimed to construct a reliable and valid measure of fundamental SIT constructs and pilot them on naturally occurring groups within the tertiary education context. The SIT constructs included in the scale were: in-group/ out-group closeness; in-group/out-group identification; group permeability; stability; legitimacy; conflict; intergroup differentiation; and in-group/out-group homogeneity. A sample of n = 510 university students were recruited for participation in one of two study conditions. In condition one participants’ sex (male or female) was the salient social identity for intergroup comparison. In condition two participants’ student identity (undergraduates or postgraduates) was made salient. These study conditions were chosen because the groups that were used were naturally occurring groups, to elaborate, the two groups that were compared in condition one had impermeable boundaries and the status hierarchy was illegitimate; whilst in condition two the two groups had permeable boundaries and legitimate categories for comparison. A reliability analysis was then conducted in order to examine the reliability of the scale as well as to improve the scales by weeding out poor items. A confirmatory factor analysis was then performed in order to confirm the independence and statistical coherence / logic of the constructs included in the inventory. Finally four hypotheses based on SIT literature were tested to partially test the construct validity for a subset of the subscales. These procedures resulted in subscales that loaded independently and predictably on coherent factors and had an acceptable to good reliability as research instruments, but not for psychometric or testing purposes. Finally the hypothesis tests confirmed that, in accordance with theoretical predictions from SIT there were significant relationships between: (1) in-group identification and group status; (2) permeability, status and in-group identification; (3) in-group identification and in-group homogeneity. However, a fourth and more complex hypothesis, namely that low status group members in groups with low legitimacy would express more conflict than members of low status groups with high legitimacy, was not confirmed. The successful hypothesis tests indicate that the in-group/out-group identification, status, permeability, and in-group/out-group homogeneity sub-scales were able to correctly replicate the theoretical predictions that were drawn from Social Identity Theory. These findings are useful indicators of the construct and criterion validity of these subscales. Based on these results, one can conclude that the Social Identity Inventory has a reasonable reliability and there are some indications of validity. However additional research is needed to further explore the reliability and validity of the scale using a more representative sample of the general population and with the recommended final versions of the scales.