Using a kurtosis statistic as a dependent variable

[jdreg]

Hi everyone,

I am working on a research project in public policy and I am testing kurtosis for on the year-to-year budget changes (as %) in different policy areas. In the literature we know there will more than likely be L-kurtosis so I want to take it one step further.

I would like to use the kurtosis statistic as a dependent variable in a regression in order to test the significance and magnitude of how different policy factors impact the normality of each policy area. So far I haven't found anywhere that says I couldn't use the kurtosis statistic as a dependent variable but I wanted to get a second opinion.

Thanks,

Darrin

 

[hlsmith]

Why wouldn't you just use the variable that the kurtosis value is based on? What if from year to year the kurtosis is the same, but the central tendency of the variable changes?