Sense of personal control: Can it be assessed culturally unbiased across Aboriginal and non-Aboriginal Australians?

Abstract

In recent decades, several studies have emphasized sense of personal control as a prominent aspect of Aboriginal health. However, one limitation is that instruments available to measure personal control were originally developed in western countries and validation for Aboriginal Australians has not been conducted. The aims of the current study were to evaluate whether the Sense of Personal Control Scale (SPCS) can be used to obtain culturally unbiased measurement of personal control across Aboriginal and non-Aboriginal Australians and to assess the psychometric properties of the SPCS for Aboriginal and non-Aboriginal Australian.

Methods: The current study utilized two Australian subsamples retrieved from the Teeth Talk Study (n = 317) and the National Survey of Adult Oral Health 2004-2006 (n = 3,857) in which the SPCS was included. Graphical Loglinear Rasch Models (GLLRM) were used to fulfill the aims of the study.

Results: The Perceived Constraints subscale fitted a GLLRM for Aboriginal Australians after the exclusion of three items, while fit to any Rasch model (RM) or GLLRM model could not be found in the non-Aboriginal sample. The Mastery subscale fitted a GLLRM in the non-Aboriginal sample after the exclusion of one item. In the Aboriginal sample, two items of the Mastery subscale fitted the RM, however, two items cannot be considered as a scale.

Conclusion: In the present study, we showed that the development of new items is crucial before the revised SPCS might constitute a valid and reliable measure of sense of personal control in both Aboriginal and non-Aboriginal Australian populations, and it is possible to assess whether the SPCS can be measured without bias across these two populations.

Fig 1. Resulting models for the Perceived Constraints and Mastery subscales.

Note. Graphical loglinear Rasch model for the Perceived Constraints subscale for Aboriginal Australians (top left), Rasch model for the Mastery subscale for Aboriginal Australians (top right) Graphical loglinear Rasch model for the Mastery subscale for non-Aboriginal Australians (bottom right). Disconnected nodes indicate that variables are conditionally independent and partial gamma coefficients (Goodman & Kruskal’s γ) informs the magnitude of the LD and DIF.

Fig 2. Item map of the PC subscale and MA subscale for Aboriginal and non-Aboriginal Australians.

Note. The orange bars display the person parameters (WML estimates). The grey bars display the population distribution under the assumption of normality. The red bars display the item thresholds and the green line is the Fisher’s information function.