When conducting an experiment with a new compound X, including a control group, to determine whether the compound X has a negative or positive effect, we want to estimate the minimum detectable dose (MDD), which is estimated from the root mean square error (RMSE) obtained from previous experimental data. While this seems like a sample size design issue in various experimental studies, I understand it to be a question of determining whether compound X is negative or not by referencing the MDD obtained from previous experimental data, assuming a standard experimental design with a fixed sample size. By explaining how to use the newly added "Sample Size Explorer" in JMP's DOE, experimental researchers will likely be able to easily estimate complex MDDs. However, since a typical previous experiment was a 4x4 Latin square experiment, is there a way to estimate the MDD of a new compound X using the RMSE obtained from previous experimental data? The "Design of Experiments" method is limited to one-way layouts, and does not support Latin squares. We demonstrate that this can be estimated using the "power of two independent sample means" by adjusting the sample size of the two groups so that it is equal to the degrees of freedom of the residuals in the analysis of variance table.
