Conover's Two-Sample Squared Ranks Test for Equality of Variance

Performs two-sample squared ranks test for equality of variance.

Description

Performs the nonparametric squared ranks test on quality of variance for two samples. Based on Conover (1981), 'Practical Nonparametric Statistics', pp. 239-248.

Uses normal approximation for for m >10 (MATLAB fxn norminv or table if norminv unavailable). Uses normal approximation if ties exist. Uses tabulated critical values from Conover if small samples sizes and no ties.

Input:
x vector;
y vector;
type I error rate;
alternative hypothesis type (upper, lower, two-sided)

Output:
squared ranks test statistic if no ties;
squared ranks test statistic if ties;
decision (1 reject; 0 fail to reject);
Type I error rate;
alternative hypothesis type;
lower side critical value;
upper side critical value.

• Required Products: Statistics Toolbox
• MATLAB release: MATLAB 6.5 (R13)
• Other requirements: If norminv is unavailable and ties occur, a set of standard normal z-values is used as an approximation.

[Ben Corden]