The control group should be smokers not non-smokers and if all the cocaine using group are also smokers the difference should be the effect of the cocaine.