239814_en-US

liliconcom

Difference between ANOM with Levene(ADM) and Unequal Variance Levene Test

Hello,

I am attempting to test homogeneity using the Levene's test and am running into a discrepancy between the Analysis of Means for Variances-Levene (ADM) and the Unequal Variances test wherein the p-value for the Levene test is significant. What is the difference between these two tests and why would one come back significant while the other shows that the limit is not exceeded? Which is more important when analyzing homogeneity, the test for unequal variances? 

 

Thank you very much,

Nick Christy

 

Re: Difference between ANOM with Levene(ADM) and Unequal Variance Levene Test

The underlying hypothesis tests are different.  For unequal variance you are testing if the variances are all equal (null) against at least one is different (alternative).  See, https://www.itl.nist.gov/div898/handbook/eda/section3/eda35a.htm.

 

For ANOM with Levene you are testing to see if any group standard deviation is different from the "overall" standard deviation. See: https://www.jmp.com/support/help/en/15.0/#page/jmp/analysis-of-means-for-scale.shtml