Two-dimensional goodness-of-fit testing in astronomy

Abstract

This paper deals with the techniques available to test for consistency between the empirical distribution of data points on a plane and a hypothetical density law. Two new statistical tests are developed. The first is a two-dimensional version of the Kolmogorov–Smirnov test, for which the distribution of the test statistic is investigated using a Monte Carlo method. This test is found in practice to be very nearly distribution-free, and empirical formulae for the confidence levels are given. Secondly, the method of power-spectrum analysis is extended to deal with cases in which the null hypothesis is not a uniform distribution. These methods are illustrated by application to the distribution of quasar candidates found on an objective-prism plate of the Virgo Cluster.