Step 1: PLOT THE DATA! Of course. You knew that, so I take back the yelling. A plot of frequency of complaints/month vs. frequency of refunds/month. visually, do you 'see' anything? Not likely, based on your comments.
Step 2: Calculate the correlation coefficient, and confidence intervals. If the coef. is well off 0 - either toward -1 or +1, and the conf. interval does not include 0, then you probably have something. (A t-test for r = 0, with conclusion, no, it is not, will give the same results as the conf. interval.)
Step 3: Look closely at the original data, on time line basis, and other ways, to see if there is a time based (and other) trend involved. Did number of complaints jump at one point, and maybe refunds concurrently, or lagged by a month or 2? Lots of other possible questions here - you are in 'exploration' mode.
Cheers, Jay
Smooth wrote:
My company has a consumer hotline which we field comments(usually complaints) from consumers of our products. I want to be able to compare the monthly frequency of these calls to the actual product refund frequency. I've already plotted both and it appears that there is little, if any, ability to predict future product losses based upon the hotline calls. However, I want to be able to demonstrate this mathematically. What is the best statistical way of accomplishing this?
Thanks,
SRR . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
-- Jay Warner Principal Scientist Warner Consulting, Inc. 4444 North Green Bay Road Racine, WI 53404-1216 USA
Ph: (262) 634-9100 FAX: (262) 681-1133 email: [EMAIL PROTECTED] web: http://www.a2q.com
The A2Q Method (tm) -- What do you want to improve today?
. . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
