Investigating the impact of tolerance-affected parameters on the performance of electric machines gets evermore important. A major reason is that designs are optimized and thus are on or close to boundaries regarding performance constraints. Moreover, highly-utilized designs are obtained due to optimization of cost or specific measures like torque or power density following increased tolerance sensitiveness. To find the best tradeoff regarding rated performance and robustness, appropriate measures should be incorporated to optimization problems. Often, computational effort acts as barrier for succesful implementation, as numerous design parameters feature considerable variation. To minimize this effort, this work is about screening and quantifying the importance of tolerance-affected parameters. By contrast to classical sensitivity analysis, the likelihood of parameters to change must be considered. Different measures are introduced and compared by applying them to an electric machine scenario. The detailedness, accuracy, and fluctuation of the prediction is evaluated. At the best, the analysis does not require specific data point arrangements, e.g, orthogonal sets. The derived results can be utilized for determining most influential parameters and to focus on those during optimization. The overall computational effort can thus be signficantly reduced and tolerance analyses can be more easily incorporated to optimization problems.