?

Log in

No account? Create an account
http://www.time.com/time/covers/20061030/where_we_live/ via digg - Nate Bunnyfield [entries|archive|friends|userinfo]
Nate Bunnyfield

[ userinfo | livejournal userinfo ]
[ archive | journal archive ]

Links
[Links:| natehaas.com onetake (my experimental music podcast) ]

[Oct. 30th, 2006|09:52 pm]
Nate Bunnyfield

http://www.time.com/time/covers/20061030/where_we_live/

via digg
LinkReply

Comments:
[User Picture]From: lemur68
2006-10-31 07:06 am (UTC)
The map is neat, but I am SO not in the mood to read about America's Long Immigration Debate or religious demographics.
(Reply) (Thread)
From: natebunnyfield
2006-10-31 08:39 am (UTC)
Semi-related question:

Do you have any plans to leave Ohio?
(Reply) (Parent) (Thread)
[User Picture]From: eyeboogies
2006-10-31 02:54 pm (UTC)
That is the scariest map of the US ever.
(Reply) (Thread)
[User Picture]From: lazyman
2006-11-03 08:28 am (UTC)
I've always found these pretty frightening.
(Reply) (Parent) (Thread)
[User Picture]From: lazyman
2006-10-31 05:52 pm (UTC)
Cool map, but I wish there was a way to show it without resorting to logarithmic plotting.
(Reply) (Thread)
From: natebunnyfield
2006-10-31 07:21 pm (UTC)
I can't say I follow you here.

Linear plotting would eliminate everything but the metropolii.

Logarithmic plotting is my first choice as we are trying to communicate relative populations visually, not statistically or "realistically".

Know what I mean?
(Reply) (Parent) (Thread)
[User Picture]From: lazyman
2006-11-01 03:56 am (UTC)
I realize that it has to be done logarithmically because the range is so great. But people tend to "think" linearly, so it would be nice if there was some way to show it that way. Confusion about log plots often distorts perceptions.
(Reply) (Parent) (Thread)
From: natebunnyfield
2006-11-01 06:26 am (UTC)

Distortion is another word for communication.

I can see why you'd say headcounts should be linear.

But if you're trying to communicate where Americans live and still show which cities are the biggest, a logarthmic-type scale is one of the better ways to do it. I'm not going to address psychological scaling or how we group stuff - the purpose behind the map is what's important.

Basically, with this scale, we're able to both see where nobody lives and find cities with similar populations of any size (e.g. San Juan and Miami, Anchorage and Billings). The metropolii still tower over the nation. No one in the audience is trying to estimate the exact population of a town using this map.

All that being said, you have to wonder why Houston and Philly claim to have the same population, but dramatically different heights.

By the way, the University of Kansas pioneered this field of cartographic research. It's the only thing that could get me to move to Kansas.
(Reply) (Parent) (Thread)
[User Picture]From: lazyman
2006-11-02 09:17 am (UTC)
My first comment was just a lament that humans don't generally have a good way of perceiving data with a large dynamic range, especially visual data. Logarithms are a way around that, but I think it most cases it's not intuitive.

However, the more I look at this particular image, the more I have a problem with it. Most significantly, we both made a mistake of thinking it was plotting population, misled on by the data labels which discuss population numbers -- but looking at the legend, it turns out it's plotting population density. Density is a significantly different measure than total number, and that we were both misled suggests the chart doesn't make that sufficiently clear.

Combined with the potential confusion caused by the log scale, and I think the map image overly distorts the information, and not in a helpful way. The reason I'm harping a bit on the log scale for this particular plot is that the urbanization of America is an important fact. I think people should realize just how insignificant the population of many states are compared to just a handful of cities. This is an even more important fact on the eve of an election. Putting numbers on a log scale gives the visual impression that some of these places are in the same league, when they really aren't. Using densities is almost dishonest, since it doesn't directly support the statements accompanying the map.
(Reply) (Parent) (Thread)
From: natebunnyfield
2006-11-02 07:00 pm (UTC)
You have to use density with mapping.

I guess you could define specific cities, towns and metropolitan areas, then estimate their population. That's quite a bit of work and you'd end up with something very subjective and much lower resolution.
(Reply) (Parent) (Thread)
[User Picture]From: lazyman
2006-11-03 07:29 am (UTC)
A map could show either density or population. The relation between the quantities is only land area, which should be apparent from the map itself. It's just a question of if you want the viewer to integrate over area or divide by area to recognize the non-charted quantity. (Here is a map showing population and land area simultaneously, although it has plenty of other problems.)

Something that bugs me about the Time Magazine map is that spikey graphics mean the spatial area used for the density calculation isn't clear. Without it, the relation between population and density isn't apparent, and the text surrounding the map isn't supported by the mapped data. All the text refers to population totals, but the data displayed claims to be density measurements. A high density area does not necessarily mean large population if it's not a big region and vice versa.

In fact, I'm increasingly less sure I understand what is being plotted on this map. You pointed out some irregularities, and I also noticed that Los Angeles has a spike about twice as high as Boston, although the Boston area is a far more dense.
(Reply) (Parent) (Thread)
From: natebunnyfield
2006-11-03 07:40 am (UTC)
You're mixing up area data and point data.

You can make a map of population, but it would be points of groups people (cities, towns, individuals, &c.).

This map is an areal map, so it has to be of density.

Hope this clears it up.
(Reply) (Parent) (Thread)
[User Picture]From: lazyman
2006-11-03 08:01 am (UTC)
But by adding spikes, it starts to imply pointwise data. Moreover, their text and "data labels" on the spikes have pointwise values, which further confuse the issue.

More conventional area-wise density maps like I linked below are much more clear. If they had just used one of those, I wouldn't complain.
(Reply) (Parent) (Thread)
From: natebunnyfield
2006-11-03 08:21 am (UTC)
Yeah, at this scale the US Census tracts really look like points.

Yay mapreading!
(Reply) (Parent) (Thread)
[User Picture]From: lazyman
2006-11-03 08:34 am (UTC)
It looks like he grabbed the data from SEDAC, which means the census tract data has been resampled to a raster grid. Should have just left it that way.
(Reply) (Parent) (Thread)
From: natebunnyfield
2006-11-03 08:40 am (UTC)
You'd be hard pressed to tell a difference at web graphic resolution.
(Reply) (Parent) (Thread)
[User Picture]From: lazyman
2006-11-03 08:46 am (UTC)
I mean he shouldn't have added spikey things with some not-really-log height scale. The flat raster map with a log color scale is plenty clear.
(Reply) (Parent) (Thread)
From: natebunnyfield
2006-11-03 08:52 am (UTC)
Gotcha.
(Reply) (Parent) (Thread)
[User Picture]From: lazyman
2006-11-03 07:37 am (UTC)
Ok, I think I found the "source image" for the Time Magazine map. That is, a sensible display of the data which inspired the Time map.

This map makes a lot more sense to me because the spatial granularity is a lot more clear. Even better here. With these maps, the viewer can readily see the area of a high density region and so statements like "80% of the US population lives in a metropolitan area" have valid graphical support. I think Time Magazine tried to sex-up thier graphics and in doing so made a mess of the quantitative display of data.
(Reply) (Parent) (Thread)
From: natebunnyfield
2006-11-03 07:50 am (UTC)
Googling up the author's name could explain some stuff.
(Reply) (Parent) (Thread)
[User Picture]From: lazyman
2006-11-03 08:04 am (UTC)
Indeed. Wow.

He does do a nice job with the non-quantitative stuff.
(Reply) (Parent) (Thread)
[User Picture]From: megaman
2006-10-31 05:57 pm (UTC)
wow. that is a distractingly pretty map!
it conjures up all sorts of images of elementary school and growing crystals in a jar... so blissful...
(Reply) (Thread)