Confounding variables and multiple regressions Suppose you are interested in studying the effect of education on salary. You propose the following regression of hourly wage on education: wage=0 1 educ u where wage = hourly wage educ = years of education u = residual Failing to account for industry, which may influence hourly wage and may be positively correlated with education, could lead to estimate of the effect 1 . This issue of confounding variables would likely bias the estimate of 1