Solution to quality assurance challenge 2.

To validate the HPLC determination of the active pharmaceutical ingredient TAMO in tablets a first calibration set, I, (Table 1, 2nd line) was measured. The experimentally obtained counts are listed in Table 1, 3rd line. In this challenge we tested whether you could detect an outlier in the linear regression function. An outlier shows an unusually high or low residual. The residuals ei are the deviations of measurements yi from their values predicted by the regression line ŷi : ei 1⁄4 yi ŷi with ŷi 1⁄4 a0 þ a1 xi: The calibration data for calibration set I given in Table 1, calculated by use of MS Office Excel spreadsheet for a straight line regression, are: intercept a0=−828.56 counts, slope a1=2191.27 counts·L g , calibration error sy.x= 347.87 counts, degrees of freedom df=8. The residuals are listed in Table 1 (last line) and are presented in Fig. 1. As Fig. 1 shows, the major residual belongs to the standard 10 which should be tested for an outlier. First, let us use the F-test [1]. The test value TV is calculated by use of Eq. 1: