Estimating and controlling for the effects of volunteer bias with pairs of relatives