Much too late, I do have a better answer now, Geocommons which (sigh, why would any site put a description of their service as a freakin graphic! let me copy paste!):

delivers visual analytics through maps; enabling non-technical professionals to view multiple datasets, draw conclusions, make decisions and solve problems without traditional GIS overhead

More or less, by clicking, you can select from a library of different data sets, and then layer them on a map (There are ways of adding your own data to the Finder).

Some quick examples:

• Prevalence of Bike Commuters in Less Obese States http://maker.geocommons.com/maps/566: So does one cause the other??
• Wait Time At Polls, November 4, 2008 http://maker.geocommons.com/maps/1334: using data from Twitter Vote Report
• Number of Facebook Users in the US http://maker.geocommons.com/maps/83?page=1: Is it for city folk only?
• Freshmen are Criminals http://maker.geocommons.com/maps/352?page=1: Not a conclusion I would make- looks like someone else was playing with data!

I gave it the 20 minute spin this morning, creating a world map that attempts to conclude that countries in violent conflict don’t do well in the Olympics (duh, it is just an experiment):

http://maker.geocommons.com/maps/1475

There was another layer I had tossed in to map cities with the tallest skyscrapers, which was meaningless and made more clutter than interest.

I poked around the sets of data until I found two things that maybe were not directly associated. Data sets can be..data associated with place, but also include things like roadways and geographic boundaries (like school districts, maritime regions). There are display options for the way data is distributed and displayed (colors, circles or squares).

The data is tagged, and you can combine tags and keywords in the searches- the results are a bit tedious in that they are paged 4 at a time. It would also have been nice if once I had made base map on a world scale, if I could filter out data sets that were at a different scale (like crop yields in Iowa). The data is also spotty, in some places there are pages and pages of EPA data, but the geography boundary category only has 2 items. It would also be nice if the results had an embed code.

But this is minor compared to the kinds of things you can create by connecting data and maps, and to visualize relationships. You may not achieve the energy of Hans Rosling, but I see a lot of potential here, be it in learning how to look at data (and then research whether there is a real meaning or just accidental correlation). Or an activity to add your own geodata to the collection.

Mashing up data and maps is so easy a dog can do it!