How much does the Electoral College affect elections? And how much is your vote worth?

[#1]

Did you know that one vote for president in Wyoming is worth almost four votes for president in Texas?

How is that possible? Why would the residents of any state agree to have their votes count less than another?

I became interested in this topic a couple weeks ago when I was visiting my parents. My parents were very excited to discuss the election process. The conversation went a little like this:

Mom – “The US should switch to the popular vote to elect the president.”

Me – “Why do you think that?”

Mom (standing on a soapbox) – “Every vote should count, and we don’t want to disenfranchise voters by discounting their votes. Why would a conservative vote for president in California? Why would a liberal vote for president in Texas? Why should some people have more say than others when it comes to who is president?”

Dad – “Is dinner ready?”

I admit this is something I had not given much thought since my middle school social studies class. We continued the conversation during dinner:

Me – “How much difference can the value of a vote be?”

Mom – “I don’t know the exact numbers, but some states get twice the electors per vote than others.”

Me – “That seems broken. Can we do anything to fix it, so all votes count?”

Mom – “Yes, the National Popular Vote are a group of states that have agreed to cast all of their electors for the candidate that wins the popular vote. Right now, the states add up to 172 electoral votes, and they need another 98 votes to override the electoral college.”

Dad – “I need some wine. Anyone else need wine?”

My dad’s passion was awe-inspiring, so I had to dig into the data. Not that I didn’t believe my mom, but in the words of the famous statistician Dr. W. Edwards Deming, “In God we trust; all others bring data.” I knew the United States uses a system called the Electoral College to elect the president. I discovered that the Electoral College was established in the US Constitution as a compromise between a pure popular vote and a legislative selection. The Electoral College grants one elector per member the Congressional delegation. Each state has two Senators and a varying number of Representatives based on population. This means that every state, even those with very low population, will have at least three electors. I pulled the population and congressional delegation information from Wikipedia. Below is a map of the number of people per congressional seat (elector).

In 2017, the state with the fewest people per Congressional seat is Wyoming, where there are 193,105 people per Congressional seat. The state with the most people per Congressional seat is Texas, where there are 744,858 people per Congressional seat. That means one vote in Wyoming is worth approximately 3.86 votes in Texas when it comes to the presidential election. This conjures up thoughts of the book Animal Farm by George Orwell, who wrote: “All animals are equal, but some animals are more equal than others.” Below is a distribution of the population per Congressional seat for each state.

As you can see, the average population per congressional seat when you break it up by state (including the District of Columbia) is about 510,725 people. This is an example of a skewed distribution where most states have more people per congressional seats. The states with fewer people per Congressional seat have a lot fewer. Nineteen states and the District of Columbia have fewer people per congressional seat than the average. I applied a filter to the map to highlight those states.

If your state is highlighted in this graph, you live in a state where your presidential vote is worth more than it would be in an average state. If your state is not highlighted, your vote is not as valuable in a presidential election. The people in the non-highlighted states would be more equally represented if the popular vote were used to determine the president.

I guessed this difference in vote value must be a new thing or at least be getting worse over time due to the trend of urban migration. I pulled population and congressional representation data from Wikipedia dating back to 1970 to check it out. What I found surprised me: The discrepancy in vote value was much worse in 1970 than now, and since 1980 the trend has been flat, not getting worse on average. I made a table of summary statistics below. I also made a map of the trend and utilized the Column Switcher in JMP to show how the map changes over time.

At the state level, maximum votes to equal the most valuable vote is higher than it has been since the 1970s. This means Texans should be particularly motivated to dispense of the Electoral College. On the other hand, if you live in Wyoming or Vermont, you have a more valuable vote than every other state.  Seems unfair, but this begs the question: How much does the Electoral College really affect elections?

If the US used the national popular vote in place of the Electoral College, how many past elections would have had different results? I found that five presidents were elected despite losing the popular vote: John Quincy Adams, Rutherford B. Hayes, Benjamin Harrison, George W. Bush and Donald Trump. The 1824 election saw John Quincy Adams elected over Andrew Jackson even though Jackson received more than10% more of the popular vote. After that election, irritated supporters of Jackson created the Democratic party (Jackson ran as a Democratic-Republican in 1824). Ironically, every candidate since 1824 whom has lost a presidential election despite winning the popular vote has been a Democrat.

The United States was founded by the Declaration of Independence, which states, “All men are created equal.” Should all people have equal say in who becomes the next president? Is it time to the give the Electoral College the boot?

[lee_creighton]

I like your article, and especially like the topic. The Electoral College gets a lot of attention every four years, most of it bad. Your analysis is just a bit off if you used the state population from the cited Wikipedia site. The whole population is used for the apportionment of the house, but when computing vote power you need to use voting population size. The calculations with the complete state’s population does show the relative power per vote in legislative matters. States are Constitutionally guaranteed to have three electoral votes each. Those are the five smallest states, and one *could* make a case that they shouldn’t be included since they don’t have a large enough population to participate in the apportionment process. The others, though, have varying per-person power in everything that comes before the House Representatives. The way we apportion the house (which has changed over our history, but that’s another story) essentially figures out *exactly* how May representatives each state should have. By computing the exact number, we’ll, the math doesn’t work out “nicely” and there are decimals in every state’s number. Say, for example, that dividing the US population by NC’a population results in 13.08 representatives and PA has 17.81 representatives. If we could give 13.08 and 17.81 representatives, all voters would have the same power. Well, we can’t have fractional representatives, so each state gets the number of representatives when you round down the calculated number. Doing that leaves us with fewer that 335 representatives apportioned. So, there has to be a method for doling them out that is somehow fair to the states. One way is to see how close a state was to being rounded u there than down. PA in our example is really close to 18, so they get one more representative. That process is repeated for all the states until all 335 are assigned. (It’s slightly more complicated—only slightly—because they don’t use our typical rounding rules. The method we use is called the Huntington-Hill method which uses the geometric mean of your representative possibilities—-13 and 14 for NC, 17 and 18 for PA in our example). Our current apportionment method is optimized to make sure that states that most deserve the extra representative get them with fewest possible anomalies. ANYWAY, the apportionment is as fair as can be, but for elections, it really doesn’t come into play when computing voting power. States that cast primary votes earlier in the process certainly have huge amounts of power, far more than those who vote later. Party division comes into play too. One other thing your analysis didn’t grab is the fact that all fifty states aren’t winner-take-all. Nebraska and Maine give out individual electoral votes based on district voting, so they need to be excluded from your data set, or included per district rather than as complete states. There are a lot of reasons that the Electoral College is the right way to conduct our one-and-only Federal election. One of them relates to your topic, power per vote. If the candidate only needs to get 50% + 1 vote, then they can be far more targeted in their campaigning. I took a list of all counties in the US and ranked them by population. I started coloring in counties as I accumulated a total, and as soon as I crossed 50%, I stopped. I can’t include a picture here, but you can see the map at https://www.dropbox.com/s/tw16cxc2hoe8p1j/NoElectoralCollege.JPG?dl=0 What does this show? That is, if only popular vote is used, where should the campaign resources go? States like Montana, Idaho, both Dakotas, Arkansas, Louisiana, and others would get NO CONSIDERATION. There’s no county in there that is “worth” winning. Some states (Nebraska et al.) have only one county that the candidate would need. Point is that using strictly the popular vote would mean that huge swaths of the country would have zero power in the election. That probably means that Federal money wouldn’t trickle down to them either—after all, the colored counties are enough to win. Sure, one candidate wouldn’t win them all in real life, but the fact that man many people would have no voice in the election stands. So our apportioned method that uses the electoral college, while not perfect, gives people in every state a voice. That’s a preferable outcome. [If I was a nitpicker, and writing JMP docs for all those years kinda made me one, I’d balk at the longitude and latitude measurements with Alaska and Hawaii dropped into the map without boxes around them to indicate that they’re not part of the axis measurements. Luckily, I’m the most forgiving person you’ll ever meet, so I say BRAVO for using JMP’s capabilities like you did. Clearly you’re thinking the JMP way, and it’s serving you well.

[lee_creighton]

What happened to my paragraph breaks‽ Removinf then makes this look like a long manifesto!

[Peter_Hersh]

Lee great point on the voting age population.  I should of taken that into account although I don't think it would change the overall message.  A vote for president in Wyoming (~241,000 registered voters ~80,000 voters per congressional delegation) is still worth roughly 3.2 times a vote for president in Texas (~9,946,000 voters, ~ 262,000 voters per congressional delegation member) .  I am glad you are clearly passionate about this topic.  I love to debate the merits of the systems.

You have a good point about states with low amounts of population being ignored without the electoral college.  I would argue that is already happening and even to a worse degree as the only states that get any attention are battleground states and the area of those states with large population densities.  Here is a map of presidential campaign stops for the 2016 election (again no boxes for Alaska and Hawaii that is one for the developers).  Twenty four states received no visits from either candidate including every state you mentioned.  

As you can see the targeting marketing for campaigning is already happening and low population states are not part of the campaigning plan to get to 270 electoral votes.  I would argue the popular vote would make targeted marketing tougher as you could not ignore states that are firmly for one party or another.

Your point on huge swaths of the population have no say in the election is very valid.  I agree completely we should not disenfranchise voters.  Like the over 4.7 Million registered republicans in California, do they get much of a say in a presidential election?  Why should their votes not matter?  If you add the total number of registered voters in the Dakotas, Wyoming, Montana, and Arkansas you have less than 3.5 Million.  All of those votes should count and in a pure popular vote every vote would count not just the battleground states.  

[louv]

Nice blog Peter.

I'm with your Dad....Go with the whine