Paul E Johnson
2001-Jul-18 21:25 UTC
[R] does this sound like any kind of R procedure to you?
I apologize that this is not precisely about R, but it does fall into the "what would an R user do?" category of questions. My colleague came to me and said "I have data on 20 countries, 4 elections in each one. I have measures for each election on a number of things, and I want to know if the change from one election to the next follows a pattern. For example, is the measure always rising, or falling, or rising and then falling, or what?" If we had a specific hypothesis about which countries should "go together", I think I could push this into a cross-sectional timeseries framework and design a test. But, there is no apriori hypothesis of that sort, we don't know ahead of time whether there will be 20 different patterns or 1 pattern in common across all countries. I think the best start is to get plots of these to see if there is anything obvious, but after that, what would you suggest? I was thinking that we could take the data from the 4 elections, and then calculate the differences, so for each country we have a vector (x1,x2,x3). We could then plot these in a 3d scatter, and maybe something would jump out at us. Also we can calculate the differences between these vectors, and use them as inputs into one of the clustering programs available in the R community. I hoped that somebody here had seen something like this and could give me a pointer. -- Paul E. Johnson email: pauljohn at ukans.edu Dept. of Political Science http://lark.cc.ukans.edu/~pauljohn University of Kansas Office: (785) 864-9086 Lawrence, Kansas 66045 FAX: (785) 864-5700 -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Stephane Dray
2001-Jul-19 08:23 UTC
[R] does this sound like any kind of R procedure to you?
Hello Paul, I think that K-Tables ordination methods can help you to solve your problem. Theses methods consists in the ordination of several tables. The method finds a common structure of all tables and then ordinate the variations of the different tables in comparison with the common structure. This family of methods have been developped in ecology when you measure abundances of species in different sites (one table) at different dates (k tables). In ecology, we want to know if the distibution of species across sites changes (or not) along the different dates. If you are interested, K-Tables methods are available in ade-4 software (for mac and windows) which can be freely downloadable at: http://pbil.univ-lyon1.fr/ADE-4/ADE-4.html. Most of the documentation is in french but some chapters have been translate in english and there is a mailing list where you can find help. Sincerely.>I apologize that this is not precisely about R, but it does fall into >the "what would an R user do?" category of questions. > >My colleague came to me and said "I have data on 20 countries, 4 >elections in each one. I have measures for each election on a number of >things, and I want to know if the change from one election to the next >follows a pattern. For example, is the measure always rising, or >falling, or rising and then falling, or what?" > >If we had a specific hypothesis about which countries should "go >together", I think I could push this into a cross-sectional timeseries >framework and design a test. But, there is no apriori hypothesis of that >sort, we don't know ahead of time whether there will be 20 different >patterns or 1 pattern in common across all countries. > >I think the best start is to get plots of these to see if there is >anything obvious, but after that, what would you suggest? I was >thinking that we could take the data from the 4 elections, and then >calculate the differences, so for each country we have a vector >(x1,x2,x3). We could then plot these in a 3d scatter, and maybe >something would jump out at us. Also we can calculate the differences >between these vectors, and use them as inputs into one of the clustering >programs available in the R community. > >I hoped that somebody here had seen something like this and could give >me a pointer.-- St?phane DRAY --------------------------------------------------------------- Biom?trie et Biologie ?volutive - Equipe "?cologie Statistique" Universite Lyon 1 - Bat 711 - 69622 Villeurbanne CEDEX - France Tel : 04 72 43 27 56 Fax : 04 78 89 27 19 04 72 43 27 57 E-mail : dray at biomserv.univ-lyon1.fr --------------------------------------------------------------- ADE-4 http://pbil.univ-lyon1.fr/ADE-4/ADE-4F.html --------------------------------------------------------------- -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._