1.. Many procedures in epidemiology are aimed at deriving the correct amount of error usually measured as the variance or standard error.Would a more conservative estimate of the error have a larger or smaller standard error?

2.Confounding is one obvious source of bias.If you stratify a crude relationship with a third variable, perform a Mantel Haenszel procedure, and find that the MH OR is LOWER than the crude OR, then is the crude OR negatively or positively biased?