A New Surrogate for Experimental Data Analysis
Kevin T. Dolan · AIP conference proceedings · 2002
We show here that the commonly used surrogate generating techniques based on phase randomization produce a population of surrogates that are not consistent with the null hypothesis that they are designed to test. We present a new surrogate generating method, based on digital filtering techniques. This surrogate algorithm has significant advantages over the most commonly used techniques, in that it provides a more robust statistical test by producing an entire population of surrogates that are consistent with the null hypothesis. The new surrogate is tested against an autoregression process and the Rössler system.