What does this have to do with R?! This is nonsense. -- Bert
On Sun, Dec 2, 2012 at 6:34 PM, Jim Lemon <j...@bitwrit.com.au> wrote: > On 12/02/2012 08:18 PM, Solmaz Filiz KARABAÄ wrote: > >> Dear R user! >> I have a small question! >> I have calculated the relative importance of the variables. >> >> However I would like to compare the relative importance of two different >> groups of variables (i.e Strategy and industry) >> >> For example let me say that strategy has 2 sub varialbes and industry has >> four different variables! >> >> Can I simply add the importance of those four industry variables >> importance >> over each other and say that the importance level of industry is the >> total >> of those four varibales' importance? >> Can I also do the same thing and add the importance of two strategic >> variables and have a strategic level importance? >> >> After these simple calculation, can I compare the importance of those >> groups? >> >> Hi Solmaz, > There are two ways to combine related variables that are generally > accepted. The cold, hard, arms-length method is to see whether those > variables are covarying to the extent that we can legitimately infer that > an underlying variable is responsible for that covariance. Say that your > strategy measures 1) how long you spent developing that strategy and 2) how > many sources of information you consulted. These two measures are likely to > involve the underlying behavior of extensive preparation for developing a > strategy rather than just having a couple of beers and flipping a coin. So > the beer-flippers are likely to score low on both measures and the slow > swots are likely to score high and principal components analysis or similar > will get you through. > > The second method is to convince people that they go together. Instead of > applying the black box of mathematic analysis, one shines the clear light > of logic upon the problem. It is apparent to anyone with the normal quota > of neurons that expended time and verified sources of information are more > likely to be applied together in developing a good strategy and so on. If > you are important or persuasuve enough, you may get away with mere > assertion. If not, you must appeal to the authority of others, particularly > those who have already demonstrated some quantitative association between > the measures. > > Reality usually involves performing the first method, and if this does not > produce the desired result, trying to find support in the literature for > the result you would like. You can of course just baldly state that you are > combining the variables in a particular way beacuse you think it makes > sense and apply the empirical test of whether anyone buys your story. > > Jim > > ______________________________**________________ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/**listinfo/r-help<https://stat.ethz.ch/mailman/listinfo/r-help> > PLEASE do read the posting guide http://www.R-project.org/** > posting-guide.html <http://www.R-project.org/posting-guide.html> > and provide commented, minimal, self-contained, reproducible code. > -- Bert Gunter Genentech Nonclinical Biostatistics Internal Contact Info: Phone: 467-7374 Website: http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm [[alternative HTML version deleted]]
______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
