On Fri, 28 Jul 1995 Wippen@aol.com wrote: > > > Hey Fred, > > Was it Mark Twain that said"......there's lies,damn lies, and statistics"? > > Just wondering.[ (BG ;-)] > > Paul Dempsey > Marshall University > wippen@aol.com I think from a statistical viewpoint that all of the approaches deserve attention. It is true that many of the multi-variable designs have a bad reputation, as Paul points out, mainly because so many people (including well-meaning researchers) have either selected the wrong statistical design or have purposely slanted the numbers to obtain results more in favor with management (often the granting agency, $$$). Altering one variable at a time, as most of us do, is still using statistics. Whenever more than one variable needs to be studied, say on a grand action or whatever, then it can save lots of time (# of readings) to to use an appropriate experimental design. If one runs into an action or mechanism where we might have to vary more than one thing at a time, then an appropriate multiple regression or ANOVA might be best. Experimental designs (statistics) can be pretty powerful tools since they can help save up to 90 to 95% of your time in a multi-variable study. Piano techs. in the universities might have chances to take tuition-free real-world statistics courses, and to count something other than pigs, sheep, and barley which Fisher used in the 1930's. Statistics is supposed to make life alot easier, as long as it's used correctly. Fred Scoles
This PTG archive page provided courtesy of Moy Piano Service, LLC