A moderator (or effect modifier) is a variable that influences the very association between the x-variable and the y-variable. Thus, the association between x and y looks different depending on the value of z. Suppose that we are interested in the association between unemployment (x) and ill-health (y). Here, it could be reasonable to assume that men’s and women’s health is affected differently by unemployment – in that case, gender (sex) would be a moderating variable (z).

Interaction analysis

In data analysis, moderating variables are examined through something called interaction analysis (see Interaction analysis).