Re: [xwiki-devs] GSOC 2009 project idea "Survey Application" Clearing some doubts in statistical analysis

Hi, I'm not exactly sure what you are trying to say concerning the hypothesis. There is something called "weighting" when you are running statistics over a population. This is the way to correct the answers you received because you received too many answers from certain type of people This is a very well known process (I use to work in stastistics) but it requires initial population references (like number of men/women, number of people by age, by profession, etc..). This is too complex for the project as it usually requires a special algorithm running in special software. Most survey software don't do it themselves except that they allow to export the data so that you can calculate it in a separate tool and allow to load in your tool a "weight" for each response (this is a floating number close to 1). You then use this number to show your statistical result. Instead of doing Reponse 1 to question 1 : count(response 1) Reponse 2 to question 1 : count(response 2) You do Reponse 1 to question 1 : sum(weight * response 1) Reponse 2 to question 1 : sum(weight * response 2) Ludovic Chathura Prabuddha a écrit : -- Ludovic Dubost Blog: http://blog.ludovic.org/ XWiki: http://www.xwiki.com Skype: ldubost GTalk: ldubost