You are on page 1of 6

Package Name: Normtest

Author: IHS EViews


Date: 2010/09/07

Description: These add-ins allow you to perform various techniques for assessing a data
set for goodness-of-fit to univariate/multivariate normal distribution. There are three
add-ins; one for testing a single series of data, one for testing a group of data, and one for
assessing time-series data.

When you run these add-ins, a dialog will appear asking you to select a type of add-in
programs to be installed:

Add-Ins: UniNorm
MVNorm
TSNorm

1
UniNorm

Add-in Type: Series


Default Proc Name: uninorm
Default Menu Text: The Shapiro–Wilk and Shapiro–Francia tests

Description: This add-in computes the test statistics and P-values of the Shapiro–Wilk
and Shapiro–Francia tests for the composite hypothesis of normality, based upon the
formulas given in Royston (1995) and Royston (1993), respectively.

The results will be displayed in a box as follow:

References:

Patrick Royston (1993) “A pocket-calculator algorithm for the Shapiro-Francia test for
non-normality: an application to medicine”, Statistics in Medicine 12, 181–184.

Patrick Royston (1995) “Algorithm AS R94”, Applied Statistics 44, 547–551.

2
MVNorm

Add-in Type: Group


Default Proc Name: mvnorm
Default Menu Text: Multivariate normality tests

Description: This add-in provides test statistics (along with some scaled statistics) and
P-values for assessing multivariate normality of a group of series. The currently available
test techniques are:

(a) Multivariate Shapiro-Wilk (Royston, 1983) test,


(b) Henze-Zirkler (1990) test,
(c) Mardia’s skewness and kurtosis test (Mardia, 1970),
(d) Adjusted Mardia’s skewness test (Rencher, 2002), and
(e) Doornik-Hansen omnibus test (Doornik and Hansen, 2008).

The results will be displayed in a box as follow:

References:
Doornik, Jurgen A. and H. Hansen (2008) “An omnibus test for univariate and multivariate
normality”, Oxford Bulletin of Economics and Statistics 70, 927–939.

3
Henze, Norbert and B. Zirkler (1990) “A class of invariant consistent tests for multivariate
normality”, Communications in Statistics, Theory and Methods 19, 3595–3617.

Mardia, K.V. (1970) “Measures of multivariate skewness and kurtosis with applications”,
Biometrika 57, 519–530.

Rencher, A. C. (2002) “Methods of Multivariate Analysis”, Wiley, New York.

Royston, Patrick (1983) “Some techniques for assessing multivariate normality based on
the Shapiro-Wilk W”, Applied Statistics 32, 121–133.

4
TSNorm

Add-in Type: Series


Default Proc Name: tsnorm
Default Menu Text: Time-series normality, skewness and kurtosis test
Interface: CLI and GUI

Description: This add-in computes the time series skewness, kurtosis and normality tests,
as described in “Tests for Skewness, Kurtosis and Normality for Time Series Data”, Jushan
Bai and Serena Ng, 2005, Journal of Business Statistics.

Upon running the add-in you’ll be prompted to select a type of long-run covariance method.
This is a subset of the choices available in the standard EViews long-run covariance methods
- see the EViews documentation for an explanation of the methods. Selecting to perform
prewhitening elects to do prewhitening with automatic lag selection.

Note that the Add-in folder includes a small program, called example.prg, that replicates
the first row of Table 5 in the Bai and Ng paper. The results are not identical to those given
in the paper, but I suspect this is due to small differences in the data.

5
Command line:
Syntax: equation.tsnorm(options)

Options:

df degree-of-freedom adjust the long-run covariance


kern=int choose the kernel (1=Bartlett, 2=Parzen, 3=Quadratic Spectral)
bw=int choose the bandwidth type (1=fixed NW, 2=Andrews, 3=User-specified)
bwv=val set the user-specified bandwidth value
pw use prewhitening.

You might also like