Testing of alpha-stable distributions with the characteristic function

We describe a test for the alpha-stable (/spl alpha/S) distribution based on the characteristic function (cf) by drawing on a cf based Gaussianity test and the parametric bootstrap procedure. High rejection rates are achieved against most alternative distributions, except the t/sub 2/ distribution which is found to be similar to a stable distribution. The development of this procedure was motivated by the desire to use the /spl alpha/S distribution to model heart rate variability data. Results show potential in this approach and clinical studies are being undertaken.