Perhaps your data is of the form which I observed some years ago. We were looking at the effect of moisture in a gas (fuel) feed on the porosity in a weld. Lots of water in the gas means lots of pores, or holes, in the weld area. Not a good plan for strong welds.
We made a series of welds, each time raising the water content of the gas stream slightly. Each time we raised the gas content, we made a pair of welds, first the new conditions, then the old conditions. The new conditions were distinctly superior, at all moisture levels. Great cheers & rejoicing! then we found out that as you increased the moisture content, a good deal of that moisture adhered to the walls of the gas tube (plastic). So when the moisture was raised, the next weld actually experienced less moisture than the second one. If you randomized the order of new vs. old weld conditions, the positive effects went away. Bummer! Embarrassment! Loss of face! In this case the auto correlation was inherent to the manner of performing the experiment, and would not be removed by changing the order of listing the measurements. the detected auto correlation was part of the data collection. Do you have a similar situation, with data that appears to be auto correlated if you report in time sequence of observation? Maybe it would help if you went into a bit more detail of your experimental conditions. Cheers, Jay PS Does your posted email address work? A msg directly to you bounced. Mountain Bikn' Guy wrote: > As I read these replies and think about them, it is starting to seem more > and more clear that I am not dealing with a time series, even though > everyone keeps arguing that it is a true time series. I look forward to more > comments. > > In my case, the order of the data does not matter in terms of making > predictions about new cases: > [snip] -- 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/ . =================================================================
