Substance abuse has been shown to be associated with reduced academic attainment, anti-social behaviour and psychological health problems, although the causal pathway for this association has not been appropriately established. Causal inference for substance abuse has to be determined to discover the reason why it is such a common place problem, allowing us to understand what environmental factors precedes and proceeds the abuse of substances in a causal chain. Within the literature there are several schools of thought for the causal inference between substance use and education variables. These are a direct cause between substance abuse and education, either as the use of substances effects education or that academic underachievers are more likely to use substances15. The other school of thought is that there is an indirect cause, perhaps academic underachievement and substance abuse is a common syndrome of problem behaviours4. The main problem the literature has with defining causal inferences between substances and academics is the problems with uncontrolled confounding1. Confounding may be underestimated or perhaps not taken in to account (such as genetics).

The overall aim for this research is to evaluate causal inference using different statistical techniques and culminating in a Mendelian Randomisation approach to reduce the effect of confounding to infer causation between education and substance abuse.