Hello,

I am currently validating an analytical method and more specifically the robustness of the method. We should have been evaluate the robustness of the method during developement but for this old one, it has not been performed.

To evaluate the robustness of the method, we perform a placket burman design with severals HPLC factors and we get as response the concentration of the analyte.

To study if the method is robust we use Monte Carlo simulation setting parameters to the expected range and specification around 2 % of the target concentration (we considered that the result is simailar if we have less than 2.0 % variation).

We start the simulation with 20 000 runs and we obtain a defect rate of 7 %. My problem is that the R² of the model is round 0.9 and then only 90 % of my results could be explain by the model.

Thus my question is: Could it be possible taht the 6 % of efect rate cousl be explain by the lack of precision of the model? In case, how interpret this defect rate well interpret this defact rate? Is there any statistical criteria to take into account?

Thank you very much for your help