Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: Descriptive statistics and normality tests

From   "Nick Cox" <>
To   <>
Subject   RE: st: Descriptive statistics and normality tests
Date   Tue, 21 Sep 2004 17:16:47 +0100

I dissent here. The Jarque-Bera test is based on 
the same underlying idea of a test statistic 
based on the moment measures of skewness and kurtosis. 
Why Jarque and Bera get the credit I do not know, 
beyond the fact that they published a paper. The 
main idea is decades older. 

What is more, -jb6- is a very inflexible program, with no 
support for -if- and -in-. 

For this and other reasons, 
I see no advantage whatsoever in -jb6- over -sktest-. 

Early issues of the Stata Technical Bulletin featured
several contributions to a debate, at times rather 
heated, on tests in this area. -sktest-, and also 
-swilk- and -sfrancia-, emerged from this debate. 

The list can be seen on

sg3: Skewness and kurtosis test of normality. W. Gould.
sg3.1: Tests for departure from normality. P. Royston.
sg3.2: Shapiro-Wilk and Shapiro-Francia tests ado-files. P. Royston.
sg3.3: Comment on sg3.1. R. D'Agostino, A. Belanger, and R. D'Agostino Jr.
sg3.4: Summary of tests of normality. W. Gould and W. Rogers.
sg3.5: Common on sg3.4 and an improved D'Agostino test. P. Royston.
sg3.6: Comment on sg3.3. P. Royston.
sg3.7: Final summary of tests of normality. W. Gould.

and the whole can be read comfortably in your bath or bed 
after taking advantage of the 50% discount on STB Reprints
currently available. 


by the way: jb6 provides an alternative tests for normality (jarcque-bera test)

it is a user written ado-file and can be downloaded

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index