After estimating coefficients using your sample of data, the residuals capture: O the average level of difference between the predicted level of your dependent variable and the actual level the minimum of the squared differences between the actual level of your dependent variable and the predicted level the difference between the actual level of your dependent variable and the estimated value for an observation i the sum of only the non-random component of the error term O the sum of the errors aggregated over any time dimensions in your data set