A1346
Title: Benford's law(s) and power law distributions
Authors: Claudio Lupi - University of Molise (Italy) [presenting]
Abstract: Benford's law predicts that the relative frequency of the first significant digits in many variables in both the natural and social sciences is not uniformly distributed but rather follows a distribution in which smaller numbers are more likely to occur than larger ones. Although the reasons for the emergence of Benford's law in observed data are still not entirely clear, its empirical validity has been confirmed in many different disciplines. This empirical regularity is often exploited to investigate data quality and the possible presence of data manipulations, e.g. in finance, economics, accounting, and epidemiology. The purpose is to show that Benford's law and power laws are strictly connected. Many Benford random variables display a power law distribution in a delimited range. It should, therefore, come as no surprise that the fields of application of Benford's law and power laws are often very similar. It also shows how to simulate Benford random variables with characteristics (range, quantiles, mean) similar to those observed but not necessarily Benford data. This can prove useful in assessing the Benfordness of the observed data.
