TOPICS
Search

Anderson-Darling Statistic


The Anderson-Darling statistic measures discrepancies between an empirical distribution function and a specified continuous distribution function F, giving extra weight to the tails. For a sample of size n, let u_i=F(x_((i))), where x_((i)) is the ith order statistic. Then

 A^2=-n-1/nsum_(i=1)^n(2i-1)[lnu_i+ln(1-u_(n+1-i))].

Under the null hypothesis with F fully specified, the transformed observations are uniform on (0,1). Marsaglia and Marsaglia (2004) compute the limiting distribution using a series with two-term recurrences and give a simulation-based correction for finite samples.


See also

Anderson-Darling Test, Kolmogorov-Smirnov Test, Kuiper Statistic

Explore with Wolfram|Alpha

References

Marsaglia, G. and Marsaglia, J. "Evaluating the Anderson-Darling Distribution." J. Stat. Softw. 9, No. 2, 2004. https://doi.org/10.18637/jss.v009.i02.Press, W. H.; Flannery, B. P.; Teukolsky, S. A.; and Vetterling, W. T. Numerical Recipes in FORTRAN: The Art of Scientific Computing, 2nd ed. Cambridge, England: Cambridge University Press, p. 621, 1992.

Referenced on Wolfram|Alpha

Anderson-Darling Statistic

Cite this as:

Weisstein, Eric W. "Anderson-Darling Statistic." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/Anderson-DarlingStatistic.html

Subject classifications