After a while, all State of the Union addresses start to sound alike. But how alike are they really? To find out, data analysis company Civis Analytics looked at every State of the Union address starting in 1790 to create visualizations that show what the key eras of political rhetoric are in U.S. history, as well as what topic each president made his pet.
To chart similarity across all States of the Union, Civis developed a pair of matrices that show how similar every SOTU was to one another: the darker the square, the more alike the addresses that meet on X- and Y-axis are to one another. Not surprisingly, addresses that are closest in time tend to be the most similar. When you pull back, though, the matrices reveal various turning points in U.S. history where the political rhetoric suddenly shifted: the Great Depression, or 9/11, for example. It also allows you to see how similar the speeches of different presidents are to one another, revealing similarities between both Democratic and Republican candidates.
In the first matrix, Civis provides an overview of every State of the Union since George Washington's first address in 1790. Each cell on the matrix represents a single year, which is why each box is so small: there's 225 speeches to fit in. When you pull back, though, you see some important takeaways. First of all, before 1815, State of the Union addresses tended to be very different from one another, even in consecutive years: a sign, perhaps, that the nascent SOTU's unspoken rules were slowly being formalized. By 1815, though, the art of writing State of the Union addresses stabilizes, at which point, the matrix reveals that, at least in terms of addresses, American history can be split into three important eras: 1815 to 1912 (pre-World War I), 1923 to 1932 (Coolidge and Hoover), 1946 to 2016 (post-World War II). Each of these eras is represented in the matrix by a dark, self-contained block.
The second matrix takes a deeper focus on the post-World War II era to show how similar the text of U.S. presidents' speeches are to one another. What's surprising here is that while we tend to think of presidents like George W. Bush and Barack Obama as being wildly different from one another, they were actually very similar, at least in terms of what they talk about in their State of the Union addresses. Same with George Bush and Bill Clinton. In fact, these last four presidents are all more alike each other than any other president since Truman, from a data science perspective.
Drilling down, though, there are differences, which help pinpoint the different presidential eras. Bill Clinton tended to talk more about families and family values than other presidents. Not so shockingly, George W. Bush mentioned terrorism far more than any other president. Carter was big on talking about energy. Every candidate since Carter has talked about jobs during the State of the Union, but no one has beat the drum harder about them than Obama. Surprisingly? Ronald Reagan didn't talk about jobs all too much, despite being Mr. Free Market Economy President himself.
So what can we expect the State of the Union to look like in 2017? Civis Analytics doesn't provide any predictions, but I think we can all assume that if either Donald Trump or Bernie Sanders is elected, the similarity matrix of the State of the Union is likely to shift pretty dramatically. Cruz, Rubio, or Hilary? Probably not so much. Read Civis's full data study here.