<P> If the dependent variable in a regression is measured with error, regression analysis and associated hypothesis testing are unaffected, except that the R will be lower than it would be with perfect measurement . </P> <P> However, if one or more independent variables is measured with error, then the regression coefficients and standard hypothesis tests are invalid . </P>

Difference between constant and proportional error in analytical chemistry