Robust estimation in pulse fluorometry. A study of the method of moments and least squares.

Most laboratories use least-squares iterative reconvolution (LSIR) as a routine method for estimating decay parameters in pulse fluorometric data. It is shown here, however, that LSIR is very sensitive to small amounts of error in the data whenever two decays become too close to one another, or whenever analyses of three decays are attempted. In such cases, inferior methods of estimating integrals, small zero point shifts, or small errors in the measured exciting light will result in failures of least squares, where the method of moments, with moment index displacement and lambda invariance testing, will succeed. The method of moments is therefore robust with respect to such errors while least squares is not.