|08-13-2012, 03:57 PM||#1|
Test for statistically significant different distributions?
My statistics background is slowly improving and have turned to the forum for help.
I have a several distributions that I've applied a formula to in hopes to decrease the variation of each distribution. After applying the formula to a distribution, I refer to the distribution as being “corrected”.
Is there a statistical test that allows me to determine if the formula changes the distributions so that the corrected distributions have significantly less variation than the uncorrected distributions?
I’d like to be able to have one group of corrected distributions and a separate group of uncorrected distributions, then be able to compare the two distributions to determine any significant difference.
What I’ve done so far is calculate the deciles in each distribution, then calculate RSD for each decile for the corrected distributions and the RSD for each decile of the uncorrected distributions. By comparing the RSD’s of each decile I can determine which group of distributions (corrected or uncorrected) has a greater variation within a specific decile, but I cannot determine if any difference in variation is statisically significant.
Thank you for helping.