DagosPtest

D'Agostino-Pearson's K2 test for assessing normality of a data vector.

Description

The D'Agostino-Pearson's K2 test is used to assessing normality of data using skewness and kurtosis. It approximates to the chi-squared distribution.

File needs to input the data vector and significance level (default = 0.05).
It outputs whether or not the normality is met.

• MATLAB release: MATLAB 5.3 (R11)

[Faraaz Yusufi]

I read about this test in Zar's book and wanted to use it in matlab. Thanks for implementing this in matlab, it works fine and is very useful.

[C Lynam]

Works perfectly. nb, Calls on a few other files such as chi2cdf, distchck and chi2test.

[enrique buendia]

is x2=2.284957205
not x2=3.1397
in your example code.

[Leandro Navarro Pablo]

Thanks a lot. I have implemented it in my scripts and works nicely.

[Bahriye Basturk Akay]

Thanks for implementing this function. it helped me a lot.

[Joseph Boomer]

Downloaded and ran it today. Many thanks.

[A. M. Winkler]

Hi,

This is a nice tool. However, I'm afraid there may be a small, but important error in the code. On the expression that defines H, line 117:

H = ((n-2)*(n-3)*abs(g2))/((n+1)*(n-1)*sqrt(G));

it should instead be:

H = ((n-2)*(n-3)*g2)/((n+1)*(n-1)*sqrt(G));

There is no need to use abs() here, as it will remove the sign and make the test produce incorrect results in approximately 50% of the cases.

If in doubt, please see the Equation 16 on D'Agostino RB, Belanger A & D'Agostino Jr RB, 1990. The American Statistician, 44(4), pp.316-21, and also the Equation 4 on D'Anscombe FJ & Glynn WJ, 1983. Biometrika, 70(1), p.227-34. This second paper is where the actual equation is coming from.

The correct results for your example are, therefore, K^2 = 4.3931, with a p-value of 0.1112.

Hope this helps!

A.