Robustness is the ability of a process to continue delivering effective performance even when the assumptions and variables based on which the process runs are altered to some extent.
Statistical methods are used to evaluate various processes and to draw certain estimates. The distribution that is generally used is normal distribution. These evaluations and estimations often require a great number of assumptions. Without the assumptions it would be difficult to come upon reasonable solutions. However, in reality many of these assumptions aren’t met.
Many a time the entire calculation falls apart due minor and sometimes major variations in the values of the variables used and departures from the assumptions made. This especially happens when there are outliers in the sample of values being used. For instance, if there are 2 outliers in a sample of 100 with values very different from the rest of the sample, the calculation produces a result that is very different from the real picture. Robustness is that quality of the approach used by which the large changes in assumptions do not affect the performance of the process gravely.