Hacker News new | past | comments | ask | show | jobs | submit login
Reasons to consider a cartogram (medium.com/joshuatauberer)
85 points by danso on Feb 14, 2016 | hide | past | favorite | 22 comments

I shortened the headline because, well, it didn't fit...and also, I think "silences minorities" is a little strong. If the 538 map [1] in question aimed to argue something along the lines of, "Why do minorities have so much voting power when they represent so little of the country?"...then sure. But sometimes a standard choropleth is used when geopolitical boundaries and distinctions are important. I'm not sure that they're particularly important for "The Facebook Primary"...on the other hand, in the context of presidential elections, geopolitical boundaries are generally more relevant to the reader...because of the way presidents are decided via electoral college rather than popular vote.

That said, I absolutely agree that awareness and practice of cartograms is needed, as there are plenty of times when such geopolitical-agnostic granularity is needed. But it's not just a tradeoff in difficulty in production and design...cartograms are more abstract to the viewer, and the more geopolitical data that you fuzz over, the more you might as well just make a labeled bar chart...which also is an underused format these days, because very few people will want to decorate their articles and essays with bar charts when maps are just as easy to make.

[1] http://projects.fivethirtyeight.com/facebook-primary/

I think labeled bar charts are worse than maps and cartograms for conveying coarser grained information.

Bar charts beyond a single row are hard to compare because they use height rather than colour and they don't say anything about the location (e.g. I don't really know where wyoming is).

They also don't say anything about population size, which is what this article is about. You could do some height + colour thing to get the same effect, but then you're just building a crappy cartogram.

> You could do some height + colour thing to get the same effect, but then you're just building a crappy cartogram.

Yeah, that's what I meant. The spectrum for a cartogram has total geographical fidelity on one end, to total abstraction on the other. The latter would basically be a bar chart in which the geographical data has been basically ignored.

On the more abstract side, the Guardian did a radial chart for U.S. gay rights...the positions around the circle corresponded roughly to regions, i.e. "Northeast" is the top-right of the circle. Of course, it has all the problems inherent to radial charts...but it's a nice example of something that could be a bar chart (a horizontal array of blocks) that attempted to include some geo-positioning data, even at a very general level:


What about a horizontal bar chart (ordered by value)? Yes you'll have to scan the list to find Wyoming, but it would better relate how Wyoming compares to West Virginia (which seems to have almost the same size square in the cartogram).

> Maps of political sentiment give more space to the political right. Although Obama took just over 50% of the vote in the 2012 election, his voters account for only 38% of the space on a typical election map — making it look like his opponent should have won.

I don't recall encountering sour grapes from even the least educated Republicans over the amount of the map colored red. Most people are aware of population density. And they always show the total votes bar chart next to the map.

> why [do] we keep drawing maps of land when it's the people we're interested in

Well, because America has states and electoral votes, so we really are kind of interested in land when talking about elections.

So would you object to a map like this[1] because Alaska and Hawaii are not geographically to scale with the other states? (Or in the correct geographic position for that matter)

Clearly distorting those 2 states does not distort the information the map conveys. So too with the other 48.

>...they always show the total votes bar chart next to the map...

I'm afraid not. Indeed the map I cited does not. Anyone glancing at it could fairly likely think Red won when in fact Blue won by very large margin. Actually it's hard to imagine a more confusing graphic.


The full Wikipedia article you conveniently snipped the map from does in fact show the popular vote in a bar chart:


My point is the claim that these maps are being used to trick people, is exaggerated. Why? Because it's rare that the popular vote tally is not shown, and most people are aware of population density.

I think that having as many different views as possible on the data is a great thing, and the maps in this article are excellent. However, because of the Electoral College, the standard map is not useless, and in fact shows you something no other map can - where a state is located along with how it voted in the election, projection distortions notwithstanding.

>most people are aware of population density

Half of Americans can not find New York on a map. Half can not find Mississippi. Most believe Alaska is more populous than any New England state etc.

>...they always show the total votes bar chart next to the map...

That image is used in at least 4000 places on the web[1]. Very few include the the vote bar you claim is always there. Even MIT omits it [2]

>I don't recall encountering sour grapes from even the least educated Republicans over the amount of the map colored red

And I have encountered quite a bit. Seems the people we happen to bump into at random are different.

>My point is the claim that these maps are being used to trick people, is exaggerated

Claiming that it is intentionally tricking people would be exaggerated. Claiming that it is incidental misleading people is spot on.

>the standard map is not useless

It is not completely useless. But is so nearly so that it clearly should not be the default means of presenting the information to the public.

[1]http://tineye.com/search/cdf0747ebb6fe87c0cc0f1dc1f21a35502b... [2]http://news.mit.edu/2012/presidential-election-conference-10...

I don't necessarily disagree with you.

Burning karma here but, imo, if someone can't find New York on a map they probably shouldn't even be allowed to vote for the President.

Quit possibly. But pushing the hypothetical further: simply depriving ignorant people of the vote, aside from possible justice issues, would certainly result in social instability. (not to mention the undoubtable drag on the economy that ignorance represents)

Better would be to see to it that everyone knows where New York is. And better still that everyone were indeed well informed enough not to be deluded by the map style we're talking about.

Given, unfortunately, that we do live in a world with a fair level of ignorance, we should probably take that into account when presenting information. Certainly there are cynical people who are all too aware of how to do so for negative ends. Or, as I think in this case, designers quickly putting together a pretty graphic for TV simply by following a graphic formula that isn't remotely the best.

Does anyone have any nice examples, other than the UK parliamentary seats map? It's widely used in the UK, but it's also perhaps the only decent cartogram I've seen.

Edit: Here's a reasonable one for the USA: https://upload.wikimedia.org/wikipedia/commons/5/5a/Cartogra...

And an interactive version of the UK one: http://www.theguardian.com/politics/ng-interactive/2015/may/...

Hey. I'm the author of the article.

If folks want to toss links to good examples of cartograms in replies here, I'll add them to the article!

The best I've seen is from the NPR article at http://www.npr.org/sections/itsallpolitics/2012/11/01/163632... .

One of these cartograms is a electoral vote cartogram of the US 2008 presidential election. Unlike the above examples, it uses a continuous color spectrum from red to blue (going through purple at 50/50) to represent the popular vote - which does a much better job of conveying voter proportions than the standard color every state red or blue. On the other hand, the map is geographically coarse grained on the state level - every state is shown in one averaged color. I have yet to find one that has both continuous color spectrum and is continous (ish) in space.

Why would you choose the 538 map to pick on when it's one of the few maps of this kind that actually allows you to zoom in on each area? That's as good a solution to the problem you're referring to as cartograms are. If not better.

Maps will always have inherent demographic bias, that's just how it is. And, sure, anyone who makes maps should be always looking for good solutions to these issues. But a smug Medium article proposing the use of cartograms, as if that's a new idea, doesn't help anyone. Except perhaps you, the author, if you're looking for some kind of attention or recognition.

> Why would you choose the 538 map to pick on

It was a convenient example. And I hadn't noticed the interactivity before writing the article. That said, I don't think the zoom function makes it much better.

If you think it's OK to routinely portray America as being only rural white America, and if you're not interested in knowing in a quantified way how different a map is from reality, then we live in different worlds. As I said at the top of the article, the point wasn't that there is a bias (because that's no surprise to anyone) but to find out how large that bias is.

It would be cool if someone added these to one of the d3-based mapping libraries that exist.

In your first example, is that a humongous San Francisco Bay?

This is a thesis without any supporting arguments.

tl;dr the thesis is Maps could be a reason policymakers don’t focus on minority issues

The American system of government is based on a bicameral system. That is, half of the legislative function (making new laws) is handled by a representation of the people, the other half by a representation of the land area.

This is by design. It is not a bug. The system is not a true democracy and while paying attention to minorities is important in any republic, it is not the only consideration. The reason land mass was called out was because the states establish and form the U.S., but I guess if you had to do it over again you could assign power to people with lots of money. The point being a legislative body representative of the existing system (and responsible for the architecture, not just the performance of government) was desired.

But none of that was in there. Neither was there much proof that maps cause harm to minorities. Both of these things should have been addressed for the essay to be worth much, sadly. Nice graphs though.

While I agree that cartograms beat standard maps for showing people-centric data, picking on FiveThirtyEight (which has done so much to de-obfuscate the American political system) seems a little silly.

> Maps could be a reason policymakers don’t focus on minority issues.

Well it could, but other factors (such racism as gerrymandering) play a definitive role in influencing policymakers and the constituents they answer to.

The other thing that goes unmentioned here is that congress works on both a representation by population (house of representatives) and a representation by state irrespective of population (Senate). In the case of issues the senate has a say in, every state has an equal say regardless of population so in that case it wouldn't be appropriate to size a visual by population.

It's a good rule of thumb that someone making a "could be" claim without even a little evidence or the scantest rationale is probably making a claim for which no evidence at all exists.

The US has a long history of racism, and minorities are, well, minorities in a majority-rules system. I don't think you have to look for maps as an excuse for the consequences of that.

Applications are open for YC Winter 2023

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact