The 2004 Presidential Election Maps of the Vote

There has been a lot of talk in the media, even from fairly trusted news sources, that Bush won by a landslide, or the vast majority of people voted for Bush, or that Bush has a mandate from the people. This is simply not true. But some maps seem to show that it is. How is this possible? This is a great opportunity to look at how one can simply misuse presentation techniques to your advantage. Here I present the same data in a variety of ways to illustrate this. (I appologize for not making the maps the same size but I just grabbed them pretty much as is from the sites. I leave it to the reader to rescale if they like.)

The map as presented by USA Today paints a county Red if Bush won and Blue if Kerry won. This discounts all votes by whichever side lost in a given county. Also counties are not of equal populations. It pretty much looks like a landslide for Bush. Another way to think about this kind of representation is if Bush won by 1 vote in each county in the US the whole country would be completely red and yet he could have won by less than a 5000 votes out of the whole country. (click on the image below if it does not load first time.)

USATodayMap04.gif 35K

While not perfect, here is the map given the same data as done by professor Robert Vanderbei at Princeton. It also has the problem that counties are not of equal populations, but it does shade the counties from blue to red with ratio of votes for Kerry and Bush. Suddenly there is a lot of purple and it doesn't look like the country was all for Bush. It seems more like the 51% to 48% win for Bush. A 51 to 48 victory in a football game is a close game in anyone's book. Don't be fooled that this was a mandate by the people as some would like you to believe. (click on the image below if it does not load first time.)

princetonMapSM.jpg 301K princetonMap04.gif 233K

Another problem with this map is it is hard to visually determine near equal votes. That is the scale could look tipped depending on how we perceive color. A broader color scheme that looked more like a false color map might be more appropriatel. However...

Both maps suffer the problem that the amount of color you see is more like the "vote by square mile" rather than "vote by ballot". The shaded version doesn't solve the area vs vote problem but at least the degree of victory in each county is at least accounted for. To get a true voter count we either have to distort the maps area protrayal or just do a raw area=vote sort of display.

Here is the vote as a bar chart of Bush vs Kerry. Note that if we change the question to Bush vs not Bush then the blue bar goes up from 48% to 49% (1/3 of the distance between the two bars as displayed in image 1 below.). That more represents the "mandate" by the voting public for or against Bush. Compare the red and blue areas in the bar chart below with the first map.

bushKerry.jpg bushNotBush.jpg

To show you how close the vote really was in the following two
Who won: Red? Blue? Who won: Red? Blue?
One of these two maps has a 49:51 ratio of red to blue and the other 51:49. Can you tell without actually counting? Want to know what the answer is to see if your right? If it was obvious you wouldn't need to know, would you?

If you want to still have that state map feel but get the vote by population and not by square mile then you have to reshape the map so the states have a size proportional to their population. Here is a state by state version that has some distortion because the granularity of measure is only by state.


This is a version that is ganularity by county and can be compared directly to the county versions above. Again it is an all or nothing coloring. Note that the finer the subdivisions the closer it will get to the bar chart proportions above.


These maps are from Mark Newman's site at Department of Physics and Center for the Study of Complex Systems, University of Michigan. Where you can also find things like nonlinear color maps and cartograms of the linear shading. Follow links from there to the wonderful world of data visualization via maps.