Mapping the intersection of education and money

January 14, 2011 9:32 PM

Does a better education really lead to a higher income? Take a map of the USA, overlay census data for high school graduation rates (red), college graduate rates (yellow) and median household income (blue). What do you get? A patchwork map of purples, blues, pinks and greens, that shows the relationship between education and income by county.

Some basic analysis explains that the closer to black a county is, the more prosperous it is. Closer to white, less education and less income. The spectrum of colors in between show you the interaction between education and income at a glance.

posted by Joh (60 comments total)

Don't get me wrong, this map has a lot of pretty colors, but I can't help thinking it would be more effective if there was some level of interactivity here.
posted by girih knot at 9:39 PM on January 14, 2011 [2 favorites]

This is something like When Graphic Designers Attack.
posted by sanko at 9:46 PM on January 14, 2011 [12 favorites]

Median household income includes children who work and live at home.
posted by Brian B. at 9:46 PM on January 14, 2011

And why by county? It's not granular enough. By zip code would be better.
posted by Mo Nickels at 9:52 PM on January 14, 2011

Ecological fallacy. When you aggregate, you don't actually know that the people with high incomes in high-education counties are the people who have the high education. It's consistent with the data that in highly educated counties, low-education people get paid more money. This is interesting, but if you want to know if people with higher education make more money, look at the data using individuals, not counties as the units of analysis. Aggregated data can't answer the question.
posted by If only I had a penguin... at 9:52 PM on January 14, 2011 [29 favorites]

Median household income includes children who work and live at home.

No, it's the income of the whole household (yes, including income generated by kids, presumably generally 0), and then take all the households and take the median household. What you're thinking of would be median income of individuals within a household. Again, it's all about keeping your units of analysis straight.
posted by If only I had a penguin... at 9:54 PM on January 14, 2011 [1 favorite]

I'm confused by one thing: college graduates would also be high school graduates, but these two things are treated as independent colors.

OK, I'm confused by another thing: what is that largely poor and uneducated state at the bottom, to the left, doing in a map of the USA? Did Puerto Rico become a state when I wasn't looking?
posted by twoleftfeet at 9:56 PM on January 14, 2011

Why is good.is trying to kill Edward Tufte?
posted by boo_radley at 9:58 PM on January 14, 2011 [6 favorites]

I'm confused by one thing: college graduates would also be high school graduates, but these two things are treated as independent colors.

You're right, but this allows you, to some extent, locate the places which have relatively high levels of college graduates, but with relatively low levels of high school graduates. (The two colors require different percentages per shade.) In practical terms, this can show places where extremely educated people live near areas with populated by the largely uneducated.
posted by Dee Xtrovert at 10:03 PM on January 14, 2011 [1 favorite]

If they are taking household income, I've often wondered if lower income households end up with their income inflated, because they've got extended family or kids working too. In upper middle class families, kids tend to be off at college or working hard at extra curriculars.
posted by acoutu at 10:06 PM on January 14, 2011

Also, though they both measure educational attainment, and "college-educated" is always a subset of "high school-educated," the two often correlate to different demographic groups. A place can have a lot of people with high school educations but a very low number of college-educated people, or vice versa.

The former seems to apply more to rural/suburban areas (everyone graduates high school, but it's not a place where people go to college or, alternatively, where people with college degrees are likely to stick around). The latter applies to urban centers (lots of inner-city kids don't finish high school, but a disproportionate part of the population is highly-educated urban professionals). See, e.g., Hennepin County, MN and New York County, NY.

This might be kind of obvious, but I think it's interesting to see it mapped out, and it wouldn't be as striking if educational level were plotted only on one single spectrum.

(Preview: what Dee Xtrovert said, basically.)

That said, I'm really having trouble deciphering this map, though I'm going to keep trying because it seems interesting.
posted by dixiecupdrinking at 10:13 PM on January 14, 2011 [1 favorite]

overlay census data for high school graduation rates (red) ... and median household income (blue).

I'm pretty sure those colors are Magenta and Cyan, not Red and Blue. It's pretty easy to understand if you understand subtractive color, but it would have made more sense to use RGB since more people are familiar with it.

I'm confused by one thing: college graduates would also be high school graduates, but these two things are treated as independent colors.

Interestingly, if you treat this a 3-dimensional vector space, the college graduation level would certainly be dependent on highschool, but there is obviously a third variable there: the college graduation rate minus the highschool graduation rate. (that is, the college rate projected onto the subspace generated by removing highschool)
posted by delmoi at 10:27 PM on January 14, 2011

college graduates would also be high school graduates

I'm a high school drop out with no undergraduate degree and two Masters, so not necessarily.
posted by obiwanwasabi at 10:28 PM on January 14, 2011 [4 favorites]

the two often correlate to different demographic groups.

True, but in this map income level isn't distinguished for these demographic groups, so if the purpose is to understand the relationship between education and income isn't there some confusion here?

I for one am still confused. (And I'm a college graduate, darn it!)
posted by twoleftfeet at 10:29 PM on January 14, 2011

I'm confused by one thing: college graduates would also be high school graduates, but these two things are treated as independent colors.

I'm assuming that this is because the colors are a reflection of a question like "what is your highest level of education?", rather than "did you graduate high school?" and "did you graduate college?".

Delmoi - good point on Cyan/Magenta, that makes sense. I was just thinking primary colors as I looked at the map.
posted by Joh at 10:39 PM on January 14, 2011

Oh wait, scrub that last comment about the questions. I looked again and see that in the pop-out examples that doesn't fit.
posted by Joh at 10:42 PM on January 14, 2011

Goddammit, they went and made data that could be comprehensible into a completely incomprehensible visualization. This is information design at its worst, something produced by people who are either convinced that they are clever or thought of a bad idea and ran with it because it was too late to stop. Those of us without intimate, intuitive knowledge of color relationships—or god forbid, those of us who may be color blind—are screwed. grumble grumble
posted by dubitable at 10:56 PM on January 14, 2011 [11 favorites]

What you're thinking of would be median income of individuals within a household. Again, it's all about keeping your units of analysis straight.

No, I was thinking that median household income can't be useful to compare education level when it includes those children, and others, who work and live in the home, especially households that consist of many working people who share living arrangements out of necessity, which is common enough.

Definitions:

Household income is the sum of money income received in the calendar year by all household members 15 years old and over, including household members not related to the householder, people living alone, and other nonfamily household members. Included in the total are amounts reported separately for wage or salary income; net self-employment income; interest, dividends, or net rental or royalty income or income from estates and trusts; Social Security or Railroad Retirement income; Supplemental Security Income (SSI); public assistance or welfare payments; retirement, survivor, or disability pensions; and all other income.
posted by Brian B. at 11:24 PM on January 14, 2011

Which color is coded for being born to wealthy parents. Last I tried to get adopted that was the salient factor.
posted by vapidave at 11:24 PM on January 14, 2011 [1 favorite]

It's too much. I tried to make sense of it but I just couldn't. I can't tell the difference between blue-green and red-blue and everything in between. In fact, I can't tell what most of the data is supposed to be. California and New York are both kind of greenish. What's that, high income, high college, and low high-school? Doesn't make sense, but then what is it? It's a neat idea though. I think if you restricted it to two dimensions instead of three it could work -- income is white to black, and education is blue to red, or something like that.
posted by PercussivePaul at 11:35 PM on January 14, 2011

How is mobility taken into consideration? Is there a color for that? Perhaps there should be a cross hatch pattern for that. And then polka dots for high brith rates. And a wavy patterns for racial make up. Oh yeah... then we'd have a Jackson Pollack map. Sooo preeetttty.
posted by helmutdog at 11:48 PM on January 14, 2011 [1 favorite]

I rather enjoyed one of the example counties they chose to highlight: Holmes County, Ohio. The most Amish county in the United States. Of course it's going to have a low high school and college graduation rate. Over half the population are Amish!

I'm generally unimpressed with the map, it is colorful and doesn't say a hell of a lot.
posted by Mister Fabulous at 11:53 PM on January 14, 2011

I'm a high school drop out with no undergraduate degree and two Masters, so not necessarily.

Well, you need to stop that immediately. Can't you see you're screwing up the data?
posted by strangely stunted trees at 11:56 PM on January 14, 2011 [2 favorites]

Wyoming seems to have an impressive high school graduation rate.
posted by gottabefunky at 11:56 PM on January 14, 2011

I'm having no trouble reading the map or understanding what the creators intended. Don't take it too literal. It reveals a lot of cool things. Look at the colors like pixels and sort of stand back and blur your eyes, it's all about light and dark, part information and part art.

Kansas is more like Nebraska and Oklahoma is more like Texas.

The south-east is lighter, the north-east is darker.

Every state seems to have at least one darkest spot.

Kentucky and Mississippi are the epicenter of light.. Nashville sits right between them dark as a night.
posted by stbalbach at 12:12 AM on January 15, 2011

...geh...
posted by nj_subgenius at 12:53 AM on January 15, 2011

It's not that hard to read if you're familiar with the subtractive CMYK model. Green would be Lots of cyan and yellow, but low magenta: Lots of education but little income. Blue is Cyan and Magenta, but low yellow - so Lots of high-school education and money, but not too many college graduates.
posted by delmoi at 12:54 AM on January 15, 2011

The map is a faliure:

Look at one of the highlighted counties in the lower right portion of the map.

"Breathitt County, KY", the numbers tell us it's

Magenta 60% Yellow 9% Cyan 19%

And yet, anyone looking at the map will swear that the resultant color is 'light yellow', a color suggesting a light level of college education and perhaps just a little bit of something else. Certainly, one wouldn't imagine a 60% value for Red (high school education).

It's because we're not accustomed to thinking of color values in terms of different levels of cyan, yellow, and magenta. In fact, who the hell can accurately read this map???

What's 100% yellow, 100% magenta, and 0% cyan?
Orange?
No, Red
Wakka wakka!

This is just terrible.
posted by lemuring at 1:10 AM on January 15, 2011 [4 favorites]

I would prefer to distil this into two pieces of information:

1. Some regions of the US are more prosperous than others; in more prosperous regions people tend to be better educated.

2. Here is a picture of a candy jar. The picture is taken from something in Scientific American so you can feel intellectually OK about enjoying the look of it. But there is no requirement to make sense of it.
posted by rongorongo at 1:12 AM on January 15, 2011

Correction: "60% value for Magenta"

P.S. the athore comments are TOTOL LIES
posted by lemuring at 1:18 AM on January 15, 2011 [1 favorite]

lemuring: The map is a faliure:

Look at one of the highlighted counties in the lower right portion of the map.

"Breathitt County, KY", the numbers tell us it's
Magenta 60% Yellow 9% Cyan 19%

Well, the numbers don't say it's 60% magenta, but 60% highschool graduates. The color level is not linear with the percentage of graduates, as you can see on the legend (for highschool graduation rate, it starts at around 50%, that's why 60% is such a light shade of magenta). The yellowish colour you see is because there are few high school graduates above what the map authors consider their zero.

None of the colours are linear, by the way. The three variables are divided in five stretches, and each stretch represents a slice of the population, but they are not evenly distributed. Check the cyan legend (for median household income):

Lightest: under 25k
Lighter: between 25k and 40k
Medium: between 40k and 50k
Darker: between 50k and 65k
Darkest: above 65k

I agree the map could be easier to read, but the mapping of the data over the colours is very well done, and it is quite clear how to read it, at least when you know what you are looking for:

Do you want to find places with many well-off people, with few college graduates and relatively few highschool graduates? Look for cyan patches on the map. I can see a number of counties in Texas (oil money?), a big patch in the middle of Nevada (Las Vegas?), Holmes County, OH isn't bad (especially compared with the neighbouring counties). Even disregarding the distribution within the county, there are places in the US where it seems easier to make money even without a high school diploma. But they are few, and it may not be the same people being very rich and without a high school diploma. The fact that the measure for income is the median is more difficult to skew than an average.

Dark blue tells you places with many high school graduates (relative to the 50% baseline, see above comment about the legend) and high median income. Again, might be skewed by the fact that it's per-household and not per-capita income, but we do see a lot of dark blue in the map. I don't know enough geography of the US, do these correspond with industrial or agricultural areas where economic life goes on like in the 60s?

Comparatively, there are many more counties of unmitigated magenta: high school graduates with not many college grads or many folks who make more money. There are some red areas: places where there is a high number of college graduates, but not much money to be made. There are some areas of almost white like the one lemuring pointed out, some of which seem to be in poor appalachia, and the rest mostly in the traditional South.

I do agree that it would be cool to have an interactive map where you can see the numbers per county just by hovering your cursor on the map (and then use the colour to match other counties with similar numbers).

The map may not be great for the web, but it's as good as you can make it for a print magazine. It is, however, crying for an interactive remake.
posted by kandinski at 2:38 AM on January 15, 2011 [3 favorites]

And why by county? It's not granular enough. By zip code would be better. --- You don't think it's illegible enough as it is? By zip code, it would look like an old television that wasn't tuned to any broadcast.
posted by crunchland at 3:08 AM on January 15, 2011

Correlation != cause, etc. And doesn't this assume that people don't move from their county? For instance, Boulder County CO is colored black. There are good jobs there (govt scientific agencies), that pay well, that require PhDs. There are (some) good local schools. The university is okay. However no-one really goes along the path school - university - career - income just in Boulder county.

If anything, the presence of good jobs has led to better schools, not the other way round.
posted by carter at 4:31 AM on January 15, 2011 [1 favorite]

kandinski: "I agree the map could be easier to read, but the mapping of the data over the colours is very well done."

It feels like 30% of the map isn't readable. We're only able to look for combinations of colors that are relatively simple to decipher in a CMY color space. It would have made a world of a difference, I feel, if they went with a CMYK color space with a constant value for K; Breathitt County, KY would appear as this shade of purple as opposed light-yellowish gray. One would clearly be able to see that meant a certain degree of magenta and cyan, with very little yellow.

I guess part of my outrage is the website presents itself as a source of good infographics, and yet it seems as though they've learned very little from ideas Edward Tufte has discussed in his books.

I can only excuse them for so much, especially with such important data. Treated properly, the infographic would be a lucid presentation of the current socio-economic condition in the US. As it stands, it's just another colorful infographic.

dammit I love design, but some designers give the field a bad name...
posted by lemuring at 4:38 AM on January 15, 2011 [2 favorites]

If anything, the presence of good jobs has led to better schools, not the other way round.

This comment runs completely counter to the generally accepted thought about the reason for Massachusetts' relatively good job market. Many people come here to attend one of the many colleges, then stay after graduation. This produces an educated workforce, which attracts and generates companies that pay well.

It would be interesting to see a map that showed density of colleges.
posted by Kirth Gerson at 5:13 AM on January 15, 2011

None of the colours are linear, by the way. The three variables are divided in five stretches, and each stretch represents a slice of the population, but they are not evenly distributed.

There's the real problem. When I saw the map colored in CMY, I instantly thought I could take it into Photoshop and use the Info palette to read the color values. Then I realized it was an RGB jpeg image, the RGB>CMYK conversion is nonlinear. I could figure out the transfer function, since they put a legend under the individual C M and Y maps, showing the color density for each level. And then I'd have to consider whether they used GCR or UCR in the conversion, although I suspect the designer used the default conversion settings (nobody ever changes them). And about this point, I lost interest in the whole idea of trying to get accurate data out of this map.

Curse you, foul graphic designer!
posted by charlie don't surf at 5:21 AM on January 15, 2011 [1 favorite]

Green means lots of college graduates (relative to the national average) in an area with relatively fewer high school graduates (rtna) making pretty good money (rtna). Am I reading this right? I really wanted to like this map...but numbers are much more my thing.
posted by Shutter at 5:34 AM on January 15, 2011

Lemuring: if they went with a CMYK color space with a constant value for K

Given that the K in the CMYK colour model specifies a black key, I don't know that "a constant level for K" makes any sense. A constant value of 100% for K would give a completely black map. A constant value in between 100% and 0% would only serve to murkify the map with a shade of gray.

A constant value of K equal to 0 would give us the map we already have. I think the one we have is as legible or more than any map with any other constant value for K.

That aside, I have found another easy pattern: different shades of green means "lots of college graduates and high median income with under 88% percentage of high school graduates".
posted by kandinski at 5:40 AM on January 15, 2011 [1 favorite]

As a resident of Alachua county (the county in Florida that is singled out for combining high education and low income), it seems to me that this map would benefit from having the median income combined with cost of living.

It's true that the median income here is lower, but it also costs a lot less to live here. So, relatively speaking, the well educated are well off.

For those who don't know Alachua is the home of the University of Florida and Santa Fe College. So we have tons of highly educated people in the middle of a mostly rural, deep-south area of Florida.
posted by oddman at 5:43 AM on January 15, 2011

Also, infographic != information, unfortunately. This is more of a datagraphic.
posted by carter at 5:49 AM on January 15, 2011

On wishing I had previewed: what Shutter said about the green.
posted by kandinski at 5:52 AM on January 15, 2011

Brian B: Ok, my bad, sorry. I still think it doesn't matter, though, since neither variable uses households or individuals as the units of analysis. The unit if analysis (counties) is consistent for both variables.
posted by If only I had a penguin... at 6:06 AM on January 15, 2011

What a beautifully executed failure. I love the data, I love the effort to combine it. But people just can't intuit colour mixing. I fully understand CMY and subtractive colour and all, but when I look at the patchwork map I just see chaos, no information. It's beautifully done, just a terrible idea for a visualization.

It'd work better if they let you toggle each of the 3 layers; then you could view each independently, and maybe get some better intuition for the combined views.
posted by Nelson at 7:28 AM on January 15, 2011

I agree that the map might not have followed through on its execution, but it's still a nifty concept. If anything, it might be an aberration in terms of how well thought-out the infographics usually are from Good (they're a print magazine but the infographics are primarily for the web). Their print mag is gorgeous (they've been finalists for the National Magazine Award in design as well as for their infographics section). They generally have a new designer come on every month or so and design a few new infographics. See their infographics section.
posted by jng at 7:39 AM on January 15, 2011

Does a better education really lead to a higher income?

Yes.

Whatcha could do, and might be interesting, would be to run a multilevel model of how individual education links to individual income, and then generate a map showing high intense or weak the connection is in different areas.

You couldn't do it for counties, b/c the Census doesn't release the microdata by county, but you could do it for states or for other areas within states (called "PUMA"s).
posted by ROU_Xenophobe at 8:10 AM on January 15, 2011

Look for cyan patches on the map. I can see a number of counties in Texas (oil money?), a big patch in the middle of Nevada (Las Vegas?), Holmes County, OH isn't bad (especially compared with the neighbouring counties). Even disregarding the distribution within the county, there are places in the US where it seems easier to make money even without a high school diploma.

Vegas is in Clark County, the muddy-blue looking shape at the bottom of the state (in the Y with California and Arizona). The area you're looking at (Lander Co.?) is very sparsely populated and bound to have erratic data. My home county in California is roughly the same color as Vegas and is known for having very few college graduates and a low cost of living; from the color of the counties to the south, it appears they only appear blue because educational attainment is so low that even subsistence-level income is going to dominate color perception. The areas of West Texas you're looking at could represent oil wealth or could just be similarly dire (Texas has one of the nation's highest poverty rates).
posted by kittyprecious at 9:30 AM on January 15, 2011

I have a top notch education and I'm poor. Do I win something?
posted by Lutoslawski at 11:10 AM on January 15, 2011

The color scheme is a failure because an area full of hypereducated people who make extremely little money would be bright red.

It should be blue, because those people are called Democrats.
posted by gurple at 11:42 AM on January 15, 2011 [1 favorite]

The color scheme is a failure because an area full of hypereducated people who make extremely little money would be bright red.

It should be blue, because those people are called Democrats.
posted by gurple at 11:42 AM on January 15 [+] [!]

There's a joke in there somewhere, but I can't figure it out; it's obscured by confusing color references.
posted by dubitable at 11:48 AM on January 15, 2011

an area full of hypereducated people who make extremely little money would be bright red.

Is that what's going on with Benton County, OR? It's more of a brick red, so maybe it's not as low-income as your example. I couldn't figure it out because I'm not very fluent in CMYK, but "hypereducated & middle-class" totally checks out.

I would love this map if only I could just click on the county to see the percentages they're using. Sigh.
posted by dialetheia at 12:08 PM on January 15, 2011

This is just bad.

Good infographic maps draw you into the data and make it easy to understand. I had to force myself to figure this one out.
posted by Defenestrator at 12:25 PM on January 15, 2011

Nashville sits right between them dark as a night.

Actually, that black spot in middle Tennessee is Williamson County, where I grew up; these days, that's where all the rich folk who work in Nashville (just to the north in Davidson County) actually live. Not surprisingly, Davidson is a lot less cyan, and Williamson is a lot more cyan, than any of their neighbors.
posted by erniepan at 12:58 PM on January 15, 2011 [1 favorite]

including income generated by kids, presumably generally 0

Lot of stage moms in NY and LA
posted by IndigoJones at 4:44 PM on January 15, 2011

There's over 3 million households in LA county. A few hundred, or even many thousand, households with high-earning kids isn't going to skew median household income.
posted by plastic_animals at 5:02 PM on January 15, 2011

You can pretty much tell where Kentucky is. Haha.
posted by XhaustedProphet at 10:30 PM on January 15, 2011

[e-mail]
posted by humanfont at 9:22 AM on January 16, 2011

So according to the article, dark blue indicates high education and relatively high income? I live in Clay County, Texas which is the dark blue county along the border between Texas and Oklahoma. Clay County is entirely rural without much oil money, and a median income of slightly more than $35K according to the 2000 census. This doesn't make sense really.
posted by tamitang at 9:59 AM on January 16, 2011

Median household income, that is.
posted by tamitang at 10:00 AM on January 16, 2011

Yes, now I get it.

I'm looking for places to NOT move to, i.e., places that have high education and low salaries. As oddman pointed out, red indeed indicates that statistic. There is also a very red patch in Eastern Washinton, which turns out to be Whitman County, the home of Washington State University and lots of low wage rural agricultural jobs.

Now I gotta look for very dark spots, or at least dark cyan to find someplace were I could get a decent high-paying job.
posted by f5seth at 4:10 PM on January 16, 2011

You can pretty much tell where Kentucky is. Haha.

Duuuuude, low blow!
posted by pecanpies at 7:20 PM on January 16, 2011