MR Dictionary

Most MR studies assume a linear (dose-response) effect of the risk factor on outcome. Methods for exploring non-linear effects in one-sample and two-sample MR have been developed but require very large numbers to be adequately powered. 

There may be lack of statistical power to detect non-linear effects. 


Other terms in 'Sources of bias and limitations in MR':