ABSTRACT
It has recently been argued that much health research has incorporated one of two types of gender bias (Ruiz and Verbrugge 1997). The first type of bias is the assumed equality or similarity of men and women. This is reflected in the way in which results from studies of men are extrapolated in an unquestioning way on to women. This is particularly the case for diseases which are often assumed (sometimes wrongly) to be more prevalent in men, such as heart disease; clinical trials and epidemiological studies of heart disease have commonly been undertaken in male-only samples, and the results then generalized to the total population. However this type of bias also applies to studies which include both men and women but which then treat gender as a control variable, a form of noise in the system, to be controlled away in multivariate analysis so that the supposedly true effect of a risk factor or treatment can be more correctly assessed.
