The proportion of the cohort identifying as Aboriginal is low within the cohort, so the variable is more likely to be affected by issues such as random misclassification (for example, due to data entry errors or misunderstanding of the questionnaire item). http://www.healthyminimarket.com/bio-x4/