Distributional assumptions + case studies (was: Random or Fixed effects appropriate?)
Hi Reinhold, thanks very much! Your paper is eminently suitable, especially insofar as it captures the interplay between model choice and statistical outcome. I do suggest that you make whatever alterations you deem suitable to preclude any problems with your future publisher, and if possible, provide some informal commentary on the structure of the analysis - eg how do you interpret the graphics that you produced, what motivated your decisions, etc. Your abstract already includes a description of the characteristics that makes sthis study interesting as a case study, so that's very convenient. In general, my plan is to focus on case studies for which the data are unencumbered and the authors don't mind providing a detailed explanation of their analyses. The sort of thing that I'm envisioning is along the lines of a cleaned up version of the analysis from p 116 to 137 of this document: http://www.ms.unimelb.edu.au/~andrewpr/r-users/icebreakeR.pdf So, much more detail about the process of the analysis than would be in a published paper (hopefully therefore side-stepping any copyright issues), but much less detail about the context. However, these ideas are not set in stone. I suppose that much of data analysis is a question of style, so we can't afford to be dogmatic. In the unlikely event that we get an overwhelming response then we might invoke some kind of filter. Ideally a submission would be a Sweave file and a data file, so the analysis gets dsiscussed in the context of the code that is being run. I'm happy to provide advice and/or templates for using Sweave, which I have found invaluable. Warm regards, Andrew
On Thu, Apr 10, 2008 at 05:51:36PM +0200, Reinhold Kliegl wrote:
Hi Andrew, The manuscript (Kliegl, R., Masson, M.E.J., & Richter, E.M. (2007). Fixed and random effects of word frequency and masked repetition priming: A linear mixed-effects model perspective) is available as PDF at the top of my publications page here: http://www.psych.uni-potsdam.de/people/kliegl/personal/pubs-e.html I will send you a LaTeX version later this week. What all do you need? For case studies, it may make sense to include data and R-scripts. Is this your plan? My co-authors and I realize that the manuscript is in need of an overhaul with respect to the precision of some of the arguments (especially with respect to justifications of data transformation--another red herring in experimental psychology, aside from p-values); we already have a very helpful set of reviews from a first submission. Not sure yet, where we will go next with it. Suggestions? Thanks and all the best, Reinhold On Thu, Apr 10, 2008 at 12:06 AM, Andrew Robinson <A.Robinson at ms.unimelb.edu.au> wrote:
> > In analyses of reaction times (using subjects and items as crossed > random factors; carried out with Mike Masson and Eike Richter, 2007), > model-based estimates of correlations among random effects revealed > "clearer" patterns than the correlations between means and effects > computed for each subject (as they should, given that they were > corrected for unreliability). Unlike for fixed-effects estimates, > however, estimates of correlations among random effects were quite > susceptible to violations of distributional assumptions for the > residuals--up to a change in the sign of the correlation!
This is a very interesting observation, and one that I suspect should not be buried in an email. Can you tell us more about it? In my workshops, I spend a lot of time focusing on the use of diagnostics to check distributional assumptions. It would be fabulous to be able to identify a case study in which getting the distributional assumptions was so clearly important. More generally, I wonder if it might be worth collecting such a set of case studies with clear and thorough analyses and wrapping them in a document. It seems to me that it would answer the request made by Iasonas Lamprianou recently. I'd be happy to coordinate such an effort, so long as the contributions were in LaTeX and Sweave. I know my students would benefit from it :) Is there any interest in such an idea, from potential conributors or (equally importantly) potential users? Cheers Andrew
Andrew Robinson Department of Mathematics and Statistics Tel: +61-3-8344-6410 University of Melbourne, VIC 3010 Australia Fax: +61-3-8344-4599 http://www.ms.unimelb.edu.au/~andrewpr http://blogs.mbs.edu/fishing-in-the-bay/