In regards to normalizing variables in a repeated measures design, I recall having a discussion about the proper strategy. For example if there are four repeated measures, and one of the measures does not approximate the Gaussian distribution, then if normalization is appropriate, one colleague suggested that the same transformation needs to take place for ALL of the other repeated measures so as to maintain the same metric. However, I found this advice confusing in that (1) the other variables may be fine in their distribution properties and (2) ultimately it is the orthonormal variates (i.e., difference scores) that are the focus of analysis. So, quite frankly, I have never been sure about the proper strategy when one or a couple of the variables in a repeated measures design are nonnormal. And a search in multivariate texts does not even discuss violations of normality in a repeated measures context.
As for missing data techniques in repeated measures, though I believe some strides have been made in likelihood/multiple imputation approaches (contingent on source of missingness..i.e, MCAR, MAR, etc.) isn't imputing or using likelihood approaches in a repeated measures context a bit tricky given the possibility of nonindependence?............... Dale N. Glaser, Ph.D. Pacific Science & Engineering Group 6310 Greenwich Drive; Suite 200 San Diego, CA 92122 Phone: (858) 535-1661 Fax: (858) 535-1665 http://www.pacific-science.com -----Original Message----- From: Robert Ehrlich [mailto:[EMAIL PROTECTED] Sent: Wednesday, July 09, 2003 7:55 AM To: [EMAIL PROTECTED] Subject: Re: oultliers : replacing with the mean Deleting or altering measured values from a data set is serious business including moral as well as analytical aspects. As previous writers have said, You have to make a strong case for doing so in any report on results. Are the outliers so large that they are impossible (e.g. child's' height = 30 m). Are they large (or small) with respect to the Gaussian assumption. After taking care to document your reasons, you may substitute the mean only as the beginning of an imputation procedure. You also may reduce the effect of outliers on data summaries by using the median rather than the mean. Ref: Statistical analysis with Missing Data; 2nd ed. Roderick Little and Donald Rubin this ref discusses how you handle empty cells in a data table--it looks to me that using the mean replacement for "outliers" falls in the same category. Most stat packages have capability to impute data. Good luck on your thesis, you have a golden oportunity to teach your advisor some useful concepts. Dennis wrote: >Hi all > >I would like to remove outliers from my repetitive measures design, however >making it missing removes the whole case of the subject. >I've heard that it's possible to replace outliers with the mean of the >group. I am wondering if it's a standard practice (to use for my thesis), >and are there any good references? >If it is acceptable, how should I compute the mean if there are several >outliers in one group/DV, (or variable in SPSS)? > >Dennis > > > > . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . ================================================================= . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
