Lars, Would suggest you google 'statistics' 'survey' ('guidelines' &/or 'best practices') 'analysis' for more complete information on this subject.
That being said however, yes I'd say a group of 25 (per week) is fairly representative of 3500. Any undergraduate statistic course/book will give an algorithm for defining how big any survey set should be to represent a group of a certain size (this way you don't under-represent, nor waste extra time and money). An example would be watching US News (CNN) surveys on a given subject. They usually involve surveying around 1,000 (nore or less) people to represent the 'average' person's opinion. After you collect your surveys and create the data tables, how you analyze them is solely up to you and your/company priorities. Next, yes Excel is decent for displaying data (in a pivot table). If you already own MS Office, then their is no need to re-create the wheel and/or purchase separate software. Excel has in-depth help files on how to create a pivot table, as well as a wizard that will help you set one up. Personnally, reports (that are being sent up to managers) are better created via reporting engines, but this can leave you with a static view for a single particular question. Pivot tables allow you to dynamically change your view based on different criteria. (Not a great definition, the one in MS Excel is more thorough. Finally, tracking historically past surveys (or just accumulated results) can not only show trends (I like to use not only quarterly for fiscal planning, but monthly/weekly to show detailed peaks and valleys that can equate w/ seasonal or holidays), but also be a good basis for justifying fiscal issues (bigger budget?, more work force?). Hope this helps, -Chris _______________________________________________________________________________ UNSUBSCRIBE or access ARSlist Archives at http://www.wwrug.org