Levene's test of equality of variances

Discuss statistics related things

by jantonie » Fri Mar 29, 2019 9:50 pm

Hello,

Jamovi is a very intuitive software. Thank you so much. It allows, among other things, to test the equality of the variances using the Levene's test (independent samples T Test). The statistic of this test is a Fisher's F and I propose that you add in the output the second degrees of freedom corresponding to the sample size less the number of groups (as it is done for the same test of Levene but in the one-way anova menu).

A thousand thanks in advance

Jantonie
jantonie
 
Posts: 16
Joined: Tue Oct 17, 2017 11:58 am

by jonathon » Mon Apr 01, 2019 6:51 am

hey jantonie,

i've added this now, and it will be in the next release (0.9.6.6)

with thanks

jonathon
User avatar
jonathon
 
Posts: 986
Joined: Fri Jan 27, 2017 10:04 am

by vmoreno » Wed Nov 20, 2019 7:04 am

Hi,
I know that Levene's test is less sensitive than Bartlett's test to deviations of normality.
However, it would be useful to have both tests performed. I have a data scenario in which doubling the variance in one group vs the other, Levene's p = 0.36 while Bartlett or the result of var.test (2 groups) p = 0.0088.
These data are skewed, but with relatively large sample size, and Levene's test is not useful to advise the use of Welch's correction for t-test.

Thanks

Victor
vmoreno
 
Posts: 16
Joined: Mon Jun 05, 2017 6:04 pm


Return to Statistics