Basic test statistic

The Anderson–Darling test assesses whether a sample comes from a specified distribution. It makes use of the fact that, when given a hypothesized underlying distribution and assuming the data does arise from this distribution, the cumulative distribution function (CDF) of the data can be assumed to follow a uniform distribution. The data can be then tested for uniformity with a distance test (Shapiro 1980). The formula for the test statistic ${\displaystyle A}$ to assess if data ${\displaystyle \{Y_{1}<\cdots <Y_{n}\}}$ (note that the data must be put in order) comes from a CDF ${\displaystyle F}$ is

${\displaystyle A^{2}=-n-S\,,}$

where

${\displaystyle S=\sum _{i=1}^{n}{\frac {2i-1}{n}}\left[\ln(F(Y_{i}))+\ln \left(1-F(Y_{n+1-i})\right)\right].}$

The test statistic can then be compared against the critical values of the theoretical distribution. In this case, no parameters are estimated in relation to the cumulative distribution function ${\displaystyle F}$.

Test for normality

Empirical testing has found^[5] that the Anderson–Darling test is not quite as good as the Shapiro–Wilk test, but is better than other tests. Stephens^[1] found ${\displaystyle A^{2}}$ to be one of the best empirical distribution function statistics for detecting most departures from normality.

The computation differs based on what is known about the distribution:^[6]

• Case 0: The mean ${\displaystyle \mu }$ and the variance ${\displaystyle \sigma ^{2}}$ are both known.
• Case 1: The variance ${\displaystyle \sigma ^{2}}$ is known, but the mean ${\displaystyle \mu }$ is unknown.
• Case 2: The mean ${\displaystyle \mu }$ is known, but the variance ${\displaystyle \sigma ^{2}}$ is unknown.
• Case 3: Both the mean ${\displaystyle \mu }$ and the variance ${\displaystyle \sigma ^{2}}$ are unknown.

The n observations, ${\displaystyle X_{i}}$, for ${\displaystyle i=1,\ldots n}$, of the variable ${\displaystyle X}$ must be sorted such that ${\displaystyle X_{1}\leq X_{2}\leq ...\leq X_{n}}$ and the notation in the following assumes that X_i represent the ordered observations. Let

${\displaystyle {\hat {\mu }}={\begin{cases}\mu ,&{\text{if the mean is known.}}\\{\bar {X}}={\frac {1}{n}}\sum _{i=1}^{n}X_{i},&{\text{otherwise.}}\end{cases}}}$

${\displaystyle {\hat {\sigma }}^{2}={\begin{cases}\sigma ^{2},&{\text{if the variance is known.}}\\{\frac {1}{n}}\sum _{i=1}^{n}(X_{i}-\mu )^{2},&{\text{if the variance is not known, but the mean is.}}\\{\frac {1}{n-1}}\sum _{i=1}^{n}(X_{i}-{\bar {X}})^{2},&{\text{otherwise.}}\end{cases}}}$

The values ${\displaystyle X_{i}}$ are standardized to create new values ${\displaystyle Y_{i}}$, given by

${\displaystyle Y_{i}={\frac {X_{i}-{\hat {\mu }}}{\hat {\sigma }}}.}$

With the standard normal CDF ${\displaystyle \Phi }$, ${\displaystyle A^{2}}$ is calculated using

${\displaystyle A^{2}=-n-{\frac {1}{n}}\sum _{i=1}^{n}(2i-1)(\ln \Phi (Y_{i})+\ln(1-\Phi (Y_{n+1-i}))).}$

An alternative expression in which only a single observation is dealt with at each step of the summation is:

${\displaystyle A^{2}=-n-{\frac {1}{n}}\sum _{i=1}^{n}\left[(2i-1)\ln \Phi (Y_{i})+(2(n-i)+1)\ln(1-\Phi (Y_{i}))\right].}$

A modified statistic can be calculated using

${\displaystyle A^{*2}={\begin{cases}A^{2}\left(1+{\frac {4}{n}}-{\frac {25}{n^{2}}}\right),&{\text{if the variance and the mean are both unknown.}}\\A^{2},&{\text{otherwise.}}\end{cases}}}$

If ${\displaystyle A^{2}}$ or ${\displaystyle A^{*2}}$ exceeds a given critical value, then the hypothesis of normality is rejected with some significance level. The critical values are given in the table below for values of ${\displaystyle A^{*2}}$.^{[1] [7]}

Note 1: If ${\displaystyle {\hat {\sigma }}}$ = 0 or any ${\displaystyle \Phi (Y_{i})=}$(0 or 1) then ${\displaystyle A^{2}}$ cannot be calculated and is undefined.

Note 2: The above adjustment formula is taken from Shorack & Wellner (1986, p239). Care is required in comparisons across different sources as often the specific adjustment formula is not stated.

Note 3: Stephens^[1] notes that the test becomes better when the parameters are computed from the data, even if they are known.

Note 4: Marsaglia & Marsaglia^[7] provide a more accurate result for Case 0 at 85% and 99%.

Case	n	15%	10%	5%	2.5%	1%
0	≥ 5	1.621	1.933	2.492	3.070	3.878
1			0.908	1.105	1.304	1.573
2	≥ 5		1.760	2.323	2.904	3.690
3	10	0.514	0.578	0.683	0.779	0.926
	20	0.528	0.591	0.704	0.815	0.969
	50	0.546	0.616	0.735	0.861	1.021
	100	0.559	0.631	0.754	0.884	1.047
	${\displaystyle \infty }$	0.576	0.656	0.787	0.918	1.092

Alternatively, for case 3 above (both mean and variance unknown), D'Agostino (1986) ^[6] in Table 4.7 on p. 123 and on pages 372–373 gives the adjusted statistic:

${\displaystyle A^{*2}=A^{2}\left(1+{\frac {0.75}{n}}+{\frac {2.25}{n^{2}}}\right).}$

and normality is rejected if ${\displaystyle A^{*2}}$ exceeds 0.631, 0.754, 0.884, 1.047, or 1.159 at 10%, 5%, 2.5%, 1%, and 0.5% significance levels, respectively; the procedure is valid for sample size at least n=8. The formulas for computing the p-values for other values of ${\displaystyle A^{*2}}$ are given in Table 4.9 on p. 127 in the same book.

Tests for other distributions

Above, it was assumed that the variable ${\displaystyle X_{i}}$ was being tested for normal distribution. Any other family of distributions can be tested but the test for each family is implemented by using a different modification of the basic test statistic and this is referred to critical values specific to that family of distributions. The modifications of the statistic and tables of critical values are given by Stephens (1986)^[2] for the exponential, extreme-value, Weibull, gamma, logistic, Cauchy, and von Mises distributions. Tests for the (two-parameter) log-normal distribution can be implemented by transforming the data using a logarithm and using the above test for normality. Details for the required modifications to the test statistic and for the critical values for the normal distribution and the exponential distribution have been published by Pearson & Hartley (1972, Table 54). Details for these distributions, with the addition of the Gumbel distribution, are also given by Shorack & Wellner (1986, p239). Details for the logistic distribution are given by Stephens (1979). A test for the (two parameter) Weibull distribution can be obtained by making use of the fact that the logarithm of a Weibull variate has a Gumbel distribution.