on 2/2/01 1:15 AM, [EMAIL PROTECTED] wrote: > Regarding John's comments on multiple regression: > > 1. Matrix algebra is not necessary, and for purposes such as stepwise multiple > linear regression will actually get in the way. > -In your opinion, what type of statistics are better suited for matrix > algebra? Multiple regression, ANOVA, ANCOVA, PCA, etc., where there is no stepwise selection of variables. > > 2. I have a little treatise I wrote on stepwise multiple linear regression in > meta/hypercard, including the relevant meta/hypertalk code. > -In this stack, you stated that the simple multiple regression routines used > here are based on two simple, but elegant routines that are at the heart of > virtually any multiple regression package commercially available: the > recursive SSCP algorithm and the Sweep algorithm. As I understand , logistic > regression is just a transformation of the dependent variable to the log odds > ratio, after which the usual regression procedures are followed. Can your > stack/routines be modified to perform logistic regression? > ie. determine Coefficients and Standard Errors...Odds Ratios and 95% > Confidence Intervals... In a word, yes, although I would preferentially use a maximum likelihood rather than a least-squares algorithm for logistic regression (i.e., multiway frequency analysis). > mike -- "The danger in writing science fiction is that, like masturbation, if you do too much of it you may begin to mistake it for the real thing." -- Robert Park John R. Vokey, Ph.D. Chair Department of Psychology and Neuroscience University of Lethbridge Archives: http://www.mail-archive.com/metacard@lists.runrev.com/ Info: http://www.xworlds.com/metacard/mailinglist.htm Please send bug reports to <[EMAIL PROTECTED]>, not this list.
