Date: Thu, 17 Sep 1998 07:54:12 -0600
Reply-To: "Douglas J. Anderson" <Douglas.Anderson@ENMU.EDU>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: "Douglas J. Anderson" <Douglas.Anderson@ENMU.EDU>
Subject: Q-Q Plots
Content-Type: text/plain; charset="us-ascii"
I would appeciate hearing some users' experiences with Q-Q Plots. I was
using this technique to check my data for normality and then using the
transform that yielded the "best" transform to a normal distribution, for
observations that were not normal. Before proceding on this course, I
researched the literature and found that normality is being statistically
assessed by a number of statistics: Wilk's W, K-S, and the correlation
coefficient. Several ranking procedures are also used including Hazel's
and Blom's. I decided to use the correlation coefficient in determining
the best transform.
When I presented the results of my research to date to members of my thesis
committee, I was roundly criticized for using a procedure for which I
discovered they were not familiar (the Q-Q plots, determining the best
transform by these methods, etc.). There was great suspicion about what I
was doing and serious questions about the validity of my approach. Not
being a mathematician, I could only present a defense based on the research
I had done. After two hours, the members relented a bit and decideded to
let my work stand but they want copies of my references to this procedure.
Any testimonials that I might receive from those who have used Q-Q Plots
and the validity of this method would be appreciated.