The GEH has become the standard statistic for evaluating model error in the United Kingdom and the MAPE has become a common statistic in many other forms of computer modeling. The GEH has the attractive property of having similar ranges of "good" values (< 5) across a wide range of volumes. Unlike the RMSE and MAPE statistics, the GEH statistic is generally used for individual observations rather than groups of observations. Validation criteria are generally specified as a percentage of observations with a good GEH (e.g., 85% of observations with a GEH < 5).