ProcessMA Resource


 

 

Test of Equal Variances

F-TEST

The F-test is used to test if the standard deviations of two populations are equal. In order for the test to be reliable, the populations from which the samples were obtained must be normally (even if the data displays only modest departures from the normal distribution, the test is unreliable and should not be used)

BARLETT’S TEST

The Bartlett's test is used to test if two or more samples have equal variances. It is a chi-square statistic with (k-1) degrees of freedom, where k is the number of categories in the iindependent variable. It can be used to determine is variances across samples are equal, to see if data is suitable for parametric tests like ANOVA. In order for the test to be reliable, the populations from which the samples were obtained must be normal.

LEVENE’S TEST

The Levene's test is used to test if two or more samples have equal variances. This test is an alternative to the Bartlett test. The Levene test is less sensitive to departures from normality. It can be used to determine is variances across samples are equal, to see if data is suitable for parametric tests like ANOVA. If data is normal, the Bartlett's test performs better than the Levene’s test.