Is DIGINAV the best site for nautical charts?

Edward Tufte: What is best practice for index charts on web sites?

• In some datasets, graphing data in comparison to an average value (or to more fairer comparisons) is better than graphing its raw values. For lack of a better term, I'm calling this an index chart. An example is Quantcast's demographics bar charts, like http://www.quantcast.com/gawker.com/demographics where the portion of the bar to the right of the dotted line (colored blue) is how many more members of a demographic visit that site than on average for all web sites. The average "index" seems to be an arbitrary value of 100, so if grad. schoolers had an index of 129, the site has an above average grad. school audience (29% more?). I think this is more interesting than the site's audience being 19% grad. schoolers, a demographic I might otherwise dismiss for being so small. So if I'm an advertiser and I want to reach grad. schoolers, I might advertise here. However, I've always had trouble explaining these charts. The index is a powerful concept, but many do not get it. Rather than charts around raw values, are there any recommendations on communicating this index and designing charts around it, a la Edward Tufte? Can we do better than Quantcast's example? Is "index" what it is even called?

• Answer:

  Take a look at Mosaic Plots [1] and [2]. They are a nice way to show interaction between two categorical variables. In order to show the average of a variable, you will need to create a recoded variable, with counts above and below the average. [1] http://www.childrensmercy.org/stats/definitions/mosaic.htm [2] http://rosuda.org/~unwin/Japan2003/UnwinISMTokyoNov03mosaic.pdf