In the last month, there has been a lot of discussion in the news about how Hillary Clinton lost the presidential election to Donald Trump, despite winning the popular vote. This quirk of the Electoral College has now occurred five times in U.S. history and twice in the last twenty years. Instead of rehashing other explanations for how this happened, I wanted to take a different look at the Electoral College, and by extension, the U.S. Congress. I’ll try to keep my personal politics out of this post and focus on the data, but discerning readers should keep in mind that I chose my examples carefully to illustrate a point.
Historical look: representation over time
The U.S. congress is a compromise to the different ways a legislative body can represent a collection of states. The Senate has two members per state regardless of size, while the House of Representatives assigns its members based on each state’s population. The intention is to ensure states with a small population are still fairly represented compared to bigger states. The Electoral College assigns its members in the same way, with Washington D.C. being the sole exception. It has three votes in the Electoral College, but only non-voting representation in the U.S. Congress. My graphs below focus on just congressional representation, but the same numbers apply for the Electoral College.
To begin, below I have a plot of the U.S. population over time, specifically in states with congressional representation. The counts are from the U.S. census (according to Wikipedia), which is collected every ten years.
The main thing to point out here is simply that the population keeps growing. While the pre-Civil War population grew at an approximately exponential rate, it has now slowed to pretty much a linear rate since World War II.
Next I have the sizes for the two houses of Congress over this same period. By law, the House of Representatives has to re-assign its members after each census, but as you can see, the size of the House has not changed in the last 100 years. Instead, the same number of members are re-apportioned according to relative population changes in each state
This shows how the U.S. Congress grew quickly in early years as new states were admitted. After the 1920 census, though, Congress couldn’t agree whether to to add new members, supposedly due to size constraints (and political reasons). This led to the Reappointment Act of 1929, which established a way for reappointment to occur automatically. Importantly, it did not increase the size of the House, a tradition which continues to this day.
As a result, a growing number of people are represented by a single member of congress, or put another way, representation per capita is shrinking. Below is a graph that illustrates this, obtained by dividing the values from the two graphs above. Note that this uses the total population, not the number of elligible voters.
Despite the increases in the size of Congress, the representation per capita has continuously fell in the U.S., with the largest decreases occurring before the Civil War. These are fairly consistent relative changes, so when plotted on a log scale (not shown), the rate of decrease is fairly constant.
I think the most important takeaway here is simply how low these numbers have fallen in recent years. Since the year 2000, there are fewer than 2 members of Congress per million people (and by extension, members of the Electoral College). This means each member of the Senate represents, on average, ~3 million people, while each member of the House represents ~700,000 people. These values have more than tripled since the last time the House increased in size 100 years ago. And of course, as the population continues to grow, these numbers will not get any better.
Representation by state
While the decrease in representation has a negative effect on every state in the union, it certainly affects some states more than other. Specifically, states with larger population will have fewer congress members (and electoral college members) per capita. As I said earlier, this is intentional by the U.S. constitution, but I still think it’s interesting to look at how it breaks down by state. Below is a chloropleth map below using data from the 2010 census.
Smaller states like Wyoming, Vermont, North Dakota, and Alaska have the highest representation, while many of the larger states have lower representation. Thirty states have fewer than 2 members of Congress per million people. For another look, I have the same values in a bar graph below.
Since the states are arranged here by population, the mostly decreasing trend implies smaller states have better representation than larger states. The only thing that keeps this trend from strictly decreasing is the discrete nature of assigning House members. Rhode Island is the smallest state to receive two House members, making its representation appear higher than Montana and some others with only one member.
So what? Warning: opinions ahead
I hope these figures provide some insight into whether the current system of the U.S. Congress and Electoral College provides a “fair” representation of the American people. The logic that led to the the compromise between states and individuals certainly still applies, i.e. smaller states still have reasons to fear that they would be “ignored” without this system. However, I want to argue that most small states are already being “ignored,” along with many of the larger states.
Specifically, I want to repeat what CGP Grey pointed out in his excellent video on the electoral college: the system does not guarantee that small states receive attention during a national campaign, since it’s still possible to win a Presidential election with just the 11 largest states. But even those states don’t receive most of the attention. Instead, a huge majority of campaign resources are delegated towards the “Battleground” states, i.e. states with close races that have a chance in tipping the Electoral College vote one way or another.
To illustrate this, I made maps of the states that had scheduled events with Hillary Clinton and Donald Trump during the official campaign, meaning after their respective party conventions but before the election. The counts for Donald Trump came from this list of rallies on Wikipedia, while the counts for Hillary Clinton came from this archive of speeches. I don’t pretend either source is 100% correct or complete, but they both demonstrate my point.
Both candidates spent most of their time in an extremely small number of states: Florida, Ohio, Pennsylvannia, and North Carolina. Also, it’s not as if the white states in these maps all received 1 or 2 visits. According to the sources, Donald Trump only visited 19 states, while Hillary Clinton only visited 16 states.
One could argue that this was time well spent, since states like Florida and Pennsylvania did end up having close races and results that were very different from poll predictions. However, I argue that these maps are not the intention of the Electoral College. It isn’t big states or small states that are benefiting from this system, it’s only the states with contentious and often bitter races. This system encourages polarization. And while I don’t know what would happen if the presidential election became decided by a popular vote, I think this is an opportunity to have an honest discussion about what “fair” representation really means.