Hi,
This is my first time posting to this forum, but I was wondering if the community might be able some information regarding the circle comparison graph on a student's t-test performed on an XY chart. First off, let me state that I am not a statistical expert and have been only using JMP for a short time now. In the process of trying to convince more people to use it, there have been a few colleagues who question its practicality and the conclusions that we can draw from the data. I was wondering if the community might be able some information regarding the circle comparison graph on a student's t-test performed on an XY chart.
The skepticism was generated from a data set consisting of a table with 500 samples represented by two columns, column (A) being a number between zero and one(ie: 0.4876). Column (B) was a choice of 6 separate conditions (ie: X/Y/Z or X/X/Y). We then generated a "Plot Y by X" and associated column (A) with the "Y" choice in the graph interface and column(B) with with "X" choice so that the data points were grouped into 6 columns on the graph. We then performed a student's t-test and showed the comparison circles on the right of the plot.
Upon presenting the data, there was quite extensive discussion as to the meaning of the circles and if groups could be distinguished and/or indistinguishable. Namely, the size of the circles, their overlap and the math behind everything was put into question. I was wondering if anybody out there in the JMP community could provide some sort of meaningful logic behind the interpretation of such a graph beyond the "95% confidence interval" that JMP gives in the guide book. Specifically, what kind of confidence can we obtain (as observers) from such a comparison and how much rigor should be put into understanding the meaning of circle overlap and size for the interpretation of data.
Thanks in advance,
Tei
