Sorry, it’s another semi-political post!
We at the Common Data Project are definitely interested in more than politics, but this Amazon map of political book-buying state by state was too interesting not to blog about it. It illustrates so many things I believe in.
One: Information-sharing can be fun.
People love patterns, and even more, knowing where they fit into them. The Amazon customers who are most likely to be drawn to this map are those who have bought political books, books that fall into the red, blue, or purple categories. No one is likely to be outraged that his purchase of Thomas Friedman’s book in the last 60 days got counted in designing this map. Although there’s a lot of data collection that Amazon prefers to keep on the down-low, this kind of tracking is refreshingly open and explicit. We know it’s being collected, and most of all, we get something in return. We all get to enjoy the data as well.
Two: Data has limited value if there is limited context.
As pretty as this map is, it doesn’t really provide much information. Junk Charts lays out a lot of the deficiencies that limit our ability to draw any meaningful conclusions. Providing the map with just the states colored in, but without real sales numbers, doesn’t give you a real sense of which books are selling better, in the same way that the 2004 election red-blue maps with their wide swaths of red in the middle didn’t provide real information about population density and how close the election had actually been, nor how seemingly blue or red states actually contained significant pockets of people who had voted for the other guy. How many people in South Dakota bought a “red” book? Ten, twenty, or a hundred thousand?
The paucity of information on how books were rated red, blue or purple drove me crazy, too. Every place I clicked to “Learn more,” it took me to the same very short four paragraphs. It says that the categorization was based on the book’s own promotional materials and the tags readers added to them, but I still wonder who categorized these books and precisely how they did so. Would all the authors necessarily have labeled their books as blue or red?
And if they were categorizing books as purple, as neither obviously liberal or conservative, why didn’t they include them in the percentage calculations by state?
Three: Underlying data should always be available for alternative analyses.
A lot of people are wary of data; they’ve heard too many times how numbers can be twisted to serve any purpose. We at the Common Data Project make no promises that data = truth, only that when data is truly open and available, conclusions based on that data can then be prodded, tested, and possibly refuted.
In this case, I’m not quite sure if Amazon does have a conclusion to assert, but the decisions it made about which data to include and exclude have shaped the map presented. One conclusion you might draw from a cursory glance might be the same one drawn by one of the commenters to the Junk Charts post—that people only read books they’re likely to already agree with. Imagine now if we could test that conclusion, if we could count how many readers in each state bought both “red” and “blue” books, or if there were readers who would consider themselves “conservative” but bought “liberal” books. Maybe there’s a very active and large political book club in Wyoming buying books from across the spectrum!
It may very well be true that people who identify as conservative buy “red” books, while people who identify as liberal buy “blue” books, but the map as provided doesn’t provide enough information to truly test that conclusion or propose interesting hypotheses of why that’s happening.
Still, I had a good enough time playing around with the map that I was reminded me of a book I’ve been meaning to read, which is probably Amazon’s ultimate goal anyway!