Watson's U2 statistic

©1990-1998 Pierre A. Zippi. All rights reserved.

Calculates Watson's non-parametric two sample U2 statistic for circular-normal data.
 
Watson's non-parametric two sample U2 statistic provides a criteria to test whether 2 samples differ significantly from each other. The test is robust. The difference is not specified; it may be in the mean, the angular variance, or something else. 
Data should be two independent random samples of circular observations drawn from populations with a continuous distribution.
Data should not be grouped, or if it must be grouped, class intervals should be no larger than 5 degrees.
For large samples of grouped data (where there are likely to be many ties), the test may not be applicable. 

U2=Watson's U2 test statistic;
n=sample size of first data set;
m=sample size of second data set. 

 Null hypothesis Ho: The two samples belong to the same parent population.
Decision rule: If U2 < critical value: the null hypothesis cannot be dismissed. If U2 > critical value: reject the null hypothesis, and conclude that the two samples differ significantly.
The larger the value for U2, the more likely that the 2 samples belong to different populations. The smaller the value for U2, the more likely that the 2 samples belong to same population. 

All Macintoshes. System 7.x and MacOS-8 & 9 compatible.

Price: $75 US.

NOTE: This test is included as a function in VectorRose 3.0


Pierre A. Zippi
7518 Twin Oaks Court
Garland, Texas 75044

email: paz@pazsoftware.com

See also, VectorRose 3.0 calculates circular-normal statistics and creates several types of vector diagrams.

How to order

Go to: PAZ Software Homepage