Tests for Departure from Normality. Empirical Results for the Distributions of b_2 and \sqrt{b_1} *Biometrika*, Vol. 60, No. 3. (1973) by Ralph D'Agostino, Egon S. Pearson

@article{dagostino-and-pearson-1973, abstract = {This paper is a preliminary to a detailed survey of the relative powers of a number of omnibus and directional tests of nonnormality. The probability integrals of √ b1 and b2, the standardized third and fourth moment statistics, are found for random samples from a normal distribution. Main attention is given to b2. Extensive computer simulation and curve fitting have been used to provide charts of probability levels out to the 0.1\% point, for 20 ⩽ n ⩽ 200. For √ b1, the parameters of Johnson's symmetrical {SU} approximation are tabled for values of n between 8 and 1000. An illustration is given of two `omnibus' tests applying the charts and table, involving the joint use of √ b1 and b2.}, author = {D'Agostino, Ralph and Pearson, Egon S.}, citeulike-article-id = {13230183}, citeulike-linkout-0 = {http://www.jstor.org/stable/2335012}, journal = {Biometrika}, number = {3}, posted-at = {2014-06-16 16:13:27}, priority = {2}, publisher = {Biometrika Trust}, title = {Tests for Departure from Normality. Empirical Results for the Distributions of \$b\_2\$ and \$\sqrt{b\_1}\$}, url = {http://www.jstor.org/stable/2335012}, volume = {60}, year = {1973} }

See the CiteULike entry for more info, PDF links, BibTex etc.

Computer simulations have been used as early as at least the 1960s to study problems in statistics.⇒