Data classification methods: percent black population in Escambia County, Florida

This week we used Census 2000 population data for Escambia County, Florida, and looked at the percent black population in each census tract, using four different data classification methods.

I think quantiles is the most appropriate method (see the larger map) because it shows in greatest detail the low end of the scale, where 40% of the population is classified in two low-percentage categories. Equal interval does not represent the data well because it ignores its uneven distribution and merely divides the number of census tracts by five and organizes the categories accordingly. I think standard deviation doesn't represent the data well either in this case. It is better at showing the extreme outliers, which in this case are only the small number of heavily black census tracts, than it is at expressing the percent black distribution across the county. Natural breaks has a lot of similarities to the quantiles display, but it displays in greater detail the high end of the scale, which has only a small proportion of the census tracts, at the expense of showing in detail the low end of the scale, which contains a much greater proportion of the census tracts.

Good Map(s) example: from an Icelandic mystery novel

My Good Map example is actually a set of maps at the start of a novel by the Icelandic mystery writer Arnaldur Indridason. The maps are drawn by Robert Guillemette. I like several things about the map set:

1) They are logical: first you see Iceland (inset), then the area around Reykjavik, then this area progressively enlarged.

2) They are simple: they show names of places that appear in the book, and they show, for instance, the main roads between towns, but they are minimalist. They are not even in color. The largest-scale map does not name most streets, but the idiosyncratic nature of the street length and relative positions make this less important, and there are strategically labeled important (for a mystery novel) buildings, such as the morgue.

3) Finally, I like this set of maps because - for all my admiration of the things you can do with ESRI's ArcMap - they look gracefully different from ArcMap maps. I'm not sure whether the maps are hand drawn but I would not be surprised. The outlining around the coast and lining the lakes makes me think of historic or fantastical maps, as do the little mountains on the small-scale map at the top. Overall the maps look enticingly unusual, but absolutely clear. And I can easily find places in the book on the maps - which is the point, after all.

Bad Map example: from a Google search for images of Durham, NC

This is my Bad Map example: a map that is intended to focus on Durham, NC. I don't think it is absolutely terrible, but it is a mediocre map. You can tell that there are towns here, but it's hard to decipher some of the names. There are roads, and some of them are numbered, but some are not and without numbers they are impossible to identify (without a better map). County boundaries are shown, but they could just as easily be roads - there's nothing to say they're not, and they're about the same thickness as the roads, and some of them are just as crooked and bendy. There is no scale. There's no legend to distinguish freeways from smaller roads from county boundaries. There's a red pin head in the middle, probably to tell us that Durham is Right There - but it's still hard to tell that the focus of the map is Durham, even if one could read the word. And but how do you get there? It's Spaghetti Junction with a few towns thrown in for good measure. This is the sort of map that makes me think good cartographers are badly needed!