Tags
2016 Election, 2020 Election, absentee ballot, alpha, At Large, beta, branches of government, California, Candidate, Checks and Balances, Civil War, Colorado, Congressional District, data points, Democratic Party, disenfranchisement, Donald Trump, Egalitarianism, Electoral College, Electoral Votes, exit polls, felons, Florida, formula, Hillary Clinton, House of Representatives, Joe Biden, larger states, Maine, Majority, Margin of Victory, minority, Mitt Romney, Nebraska, New York, non-citizens, past performance, Political Party, polls, Populism, prediction, Presidential Election, privilege, redistricting, Relative Influence, Republican Party, Senate, smaller states, South Carolina, stock broker, swing states, Texas, Third Party Candidate, Tipping Point, Transgender, U.S. Constitution, Union, United States, Urban Areas, US Presidents, Utah, volatility, Voter, Voting, voting rights, Winner Take All, Wyoming
Introduction
This should be used for academic purposes and discussion, not to discourage anyone from voting. I have left out party labels because they are not relevant to the discussion. This is not intended to encourage anyone to vote for any particular candidate.
I have been reading some discussions and have seen some charts discussing the Electoral College regarding its fairness and which states have the most power under the system. Leading up to the 2016 election, it was generally accepted that the larger states with their large block of electoral votes had the advantage. Since all states with the exception of smaller states Nebraska (5 electoral votes) and Maine (4 electoral votes) are winner take all in terms of the Electoral College, if a candidate in states like California (55 electoral votes), Texas (38 electoral votes), Florida or New York (29 electoral votes each) should win that state’s vote by just one vote, they would be awarded all of the electoral votes from that state.
Because most large urban areas tend to vote for Democratic Party candidates and most of those large urban areas are in the states with the most electoral votes, the prevailing opinion going into the 2016 Presidential election was that Secretary Clinton had a built-in advantage going into the election. Of those top four states, California and New York had voted for the Democrats in the last four Presidential elections, Texas had voted Republican and Florida was a swing state. Assuming that trend continued, the best Mr. Trump could hope for was to trail 67-84, and the worst case would be a deficit of 38-113 to overcome with the rest of the country. Adding in the remaining two states with at least 20 electoral votes (Illinois and Pennsylvania: 20 apiece), both of those states had voted for the Democrats in the previous four elections. This would increase Secretary Clinton’s advantage to a minimum of 57 electoral votes and a maximum of 115 electoral votes. With 270 electoral votes needed to win the election, it was seen as a huge starting advantage for Secretary Clinton, especially since a significant number of small and medium-sized states also voted for the Democrats during the previous four elections. Mr. Trump would have to hold Texas, swing Florida his way, and win a large number of small and mid-sized states to have a chance at winning.
But when Mr. Trump won the election, contrary to the polls conducted just before the election and even contrary to some of the exit polling, analysts and the Democratic Party leadership have been searching for answers. As a result, some of the attention has been focused on the Electoral College, whether it is fair and whether it should be replaced. In particular, because of the way that electors are assigned to each state (one for each Congressional seat; i.e. every state gets two electors for their Senate seats and then additional electors in accordance with the number of seats they have in the House of Representatives; and the District of Columbia gets assigned 3 electors), it was claimed by some that smaller states have an unfair advantage because the electors corresponding to Senate seats give small states a disproportionate number of electors per resident.
However, that claim assumes that the smaller states have a common interest to vote in a block against the larger states. This is not so. Just as there larger states are not unified, neither are the smaller states. There is a diversity of interests that have some solidly Democrat, some solidly Republican and some in the swing state column. Of the ten states with the lowest population per electoral vote, five tend to vote Republican and five tend to vote Democrat.
Furthermore, the Electoral College was not made part of the Constitution solely for the purpose of fairness, although it would be wrong to claim that it is totally unfair. One of the primary purposes of the Constitution was to create a system of checks and balances to minimize the possibility that one group could impose itself on the rest of the country and a majority could oppress the minority. Thus those in charge of the three branches of government are selected by different means.
Due to populist and egalitarian movements over the years, especially as the country expanded westward, has evolved to more of the democracy elements and less of the republic elements of this democratic republic. Senators, previously elected by state legislatures, are now elected by popular vote. The same is true for the electors in the Electoral College. South Carolina was the last state to have their state legislature vote for the electors by vote of the state legislators rather than by popular vote. They switched to popular vote when they were readmitted to the Union after the Civil War. By a fluke, Colorado was the last state to choose their electors for a Presidential election by state legislature vote. In 1876, they were admitted to the Union too close to the time of the general election to organize a popular vote for President.
So for me, it is intuitive that the larger states have the greater amount of influence on the election, even if they have a larger ratio of residents per electoral vote. They have a much greater chance of being the tipping point in an election.
But my interest is in the individual registered voter. Each person has to decide who to vote for or if they should even vote. The focus of my study is the relative probability that any individual voter can influence the Presidential election.
Methodology
I chose to look back over the past five elections which would roughly equate to a generation worth of elections. There were no third party candidates nationwide in any of these elections who would have significantly skewed the results, although there were occasionally third party candidates who polled well in an individual state here and there.
I recorded the margin of victory in each election between the two major parties. I assigned the margin to be a negative number if it was in favor of one of the parties and a positive number if it was in favor of the other party. Because my data source assigned a positive number to the winning party and a negative number to the losing party, and because the same party did not win all five elections, I made this assignment so that I would have to change the sign for the fewest number of elections. The sign has to be consistent throughout so it will accurately show the potential for an election to be close.
For example let’s take two states. For state A, the party X wins all five elections by 10,000 votes each election. For state B, party X wins three elections by 10,000 votes and party Y wins two elections by 10,000 votes. For state A, the average Δ is 10,000 votes per election. If I take the absolute value of the vote differential for state B, the average Δ would also be 10,000. But obviously the chance for a close election is much greater in state B than it is in state A because the parties are alternating in winning the election. The average Δ for state B therefore would be 2,000: [((3*10,000)+(2*(-10,000)))/5].
Once I have an average over the 5 elections, then I take the absolute value because I only care about how close the elections are in those states. I don’t care which party is more likely to win. At this point, I take the inverse of this number. My reasoning is that this shows me the likelihood that any one voter will change the outcome of the election. If the average Δ for my state is 10,000, the odds that any voter will change the outcome are 1 in 10,000. At this point in the calculation, it does not matter if there are 200,000 total votes cast or 10,000,000 votes cast. What matters is the closeness of the election. (Later on, the total number of votes will matter indirectly. See below.)
Of course, this assumes that the vote is cast for one of the top two contenders for the election. I could calculate the odds for any third party that runs candidates with regularity each election cycle, but I would have to record the margin between that party’s candidate and the winning candidate in the state. If third parties are performing poorly, then the odds of a voting for a third party influencing the election decrease.
However, this does not take into account situations where the voter would have voted for one of the major party candidates if the third party candidate was not on the ballot. A more sophisticated model would be needed for that situation and far more detailed data would also be needed. Based on raw numbers, there is no way to know which candidate would have been voted for if the third party candidate was not on the ballot. It might have been one of the major party candidates, but it might not have; it might have been another third party candidate, it might have been a write-in or no candidate might have received a vote from that voter.
Earlier I mentioned Maine and Nebraska. These states have chosen to vote for two electors at large corresponding to the electors assigned based on their Senate seats and for their remaining electors to be chosen severally by congressional election district corresponding to their seats in the House of Representatives. Therefore, in addition to taking the statewide totals, I had to take the data for each congressional district. But there is a problem with that. Based on the census taken every ten years, the House of Representatives is reapportioned, although in this case, the number of House seats allotted to Maine and Nebraska did not change.
Even so, redistricting can occur, and if there was a change of the party in power between the 1990 and 2010 census, the more likely a significant change in redistricting would have occurred at least once. The districts in the 2000 election were based on the 1990 census; in the 2004 and 2008 elections based on the 2000 census; in the 2012 and 2016 census based on the 2010 census. Therefore, I took the statewide margins for all five elections for these states, but the congressional district margins for only the last two elections.
Does this make the data much less reliable for those congressional districts or only marginally less reliable? Based on my methodology, it would make it much less reliable since there are only two data points, not five. Some might propose I should have used weighted averages, giving higher weight to more recent elections and a valid argument could be used for that method. But that also might give undue weight to an anomaly (for example a much wider Republican margin in Utah in 2012 because of Mitt Romney as the Republican candidate).
Earlier I said that the total number of votes matter indirectly. That’s because a state with 10,000,000 total votes (approximately the total in California which has 55 electoral votes) will have more electoral votes assigned to it than a state with 200,000 total votes (approximately the total in Wyoming which has 3 electoral votes). So to calculate the potential influence an individual voter has in the election, I multiplied the inverse of the absolute value of the average margin (i.e. the odds) times the number of electoral votes. Since this is a unitless number, I arbitrarily multiplied this result for each state (and congressional district where applicable) by one million. This was because the original result was a tiny fraction with a lot of leading zeroes. Most people would find it easier to compare numbers greater than zero, even if taken to two decimal places to the right of decimal point. I have labeled this result as “Relative Influence”.
What matters most is the comparison of the numbers for different states to get a relative value, not the raw number. For example, if state A has a relative influence of 100 and state B has a relative influence of 20, a voter in state A voting for a major party candidate has 5 times as much chance affecting the result of the national election as a voter in state B voting for a major party candidate. In reality, the amount of influence any one voter has during a national election is much smaller. Remember that we multiplied by one million to get a number that is easier to relate to.
Each voter in Maine and Nebraska affects 1 electoral vote in their congressional district and 2 electoral votes statewide. So to calculate the Relative Influence of any major party voter in these states, their statewide Relative Influence based on 2 electoral votes was added to their district Relative Influence based on 1 electoral vote.
Chart: Probable potential influence of an individual voter by state, ranked from highest to lowest
Jurisdiction | EV | Relative Influence |
FL | 29 | 789.47 |
VA | 13 | 573.00 |
IA | 6 | 351.87 |
OH | 18 | 298.05 |
ME – CD 2 | 1 | 291.91 |
CO | 9 | 184.27 |
NH | 4 | 177.41 |
NV | 6 | 173.59 |
NM | 5 | 94.31 |
NE – CD 2 | 1 | 88.43 |
PA | 20 | 80.96 |
WI | 10 | 80.36 |
NC | 15 | 70.76 |
MN | 10 | 68.91 |
AZ | 11 | 68.57 |
GA | 16 | 50.88 |
DE | 3 | 49.68 |
MI | 16 | 48.62 |
MO | 10 | 47.11 |
AK | 3 | 45.81 |
OR | 7 | 42.79 |
ME – CD 1 | 1 | 42.39 |
SD | 3 | 41.30 |
MT | 3 | 40.09 |
ND | 3 | 39.59 |
SC | 9 | 38.30 |
RI | 4 | 37.36 |
VT | 3 | 37.30 |
WV | 5 | 35.12 |
IN | 11 | 34.00 |
WA | 12 | 32.41 |
AR | 6 | 32.38 |
MS | 6 | 32.31 |
HI | 4 | 31.55 |
TX | 38 | 31.25 |
WY | 3 | 30.83 |
NE – CD 1 | 1 | 29.89 |
TN | 11 | 27.87 |
NJ | 14 | 27.54 |
CT | 7 | 27.31 |
LA | 8 | 26.24 |
KS | 6 | 24.89 |
IL | 20 | 23.08 |
KY | 8 | 21.40 |
CA | 55 | 21.03 |
AL | 9 | 20.15 |
ID | 4 | 19.60 |
NE – CD 3 | 1 | 18.17 |
MD | 10 | 18.17 |
UT | 6 | 17.69 |
NY | 29 | 16.40 |
OK | 7 | 16.20 |
MA | 11 | 14.09 |
DC | 3 | 13.89 |
ME – At Large | 2 | |
NE – At Large | 2 | |
Total | 538 |
Summary
As I stated in the introduction, this study is focused on the probable potential influence an individual voter can have on the presidential election, not how much influence one state has compared to another. This has been determined by two factors: the likelihood that one vote can swing the results of an election and the likelihood that one state (or in the case of Nebraska and Maine, one congressional district) can swing the results of an election. The former is based on how likely the election will be close within a state. This is more important than the number of voters per electoral vote. The latter is determined by the number of electoral votes. The more electoral votes a state has, the greater the chance that it could tip the election from one candidate to the other.
To highlight the importance of close elections on the potential influence a single voter can have, we can compare Florida and New York. Both have the same number of electoral votes this year, 29. But the average margin of victory in Florida in 36,733.4. In New York, the average is 1,768,022. According to my formula, the probable potential impact of a Florida voter is ~48 times greater than that of a New York voter.
Does this mean that the New York voter need not bother to vote? Absolutely not (I live in New York and I vote). Voting patterns change over time. Trends change over time. And if I used a weighted average of the margins, giving greater weight to the more recent elections, the 537 vote margin in the 2000 election in Florida would have had much less influence on the results. Also, if I do this analysis again for the 2024 election, the 2000 election results will be deleted and replaced by the 2020 election results. Also the electoral votes will be reallocated based on the 2020 census.
When I was a stockbroker, one thing we had to tell clients (and the prospectus usually included this statement as well) is that past performance is no guarantee of future results. (Of course elsewhere in the prospectus, the fund company shows you how much your $10,000 would be worth today if you had invested it ten years ago.) The same is true of margin of victory in presidential elections. Based on what has happened in prior elections, there is no way to predict what the margin will be in the next election.
If I had wanted to make my formula more sophisticated (which I might do in the future), I would have included a measure of volatility in the margin. Let’s look at two hypothetical states. For State A, the margins were 300, 100, 125, 75 and 400. For State B, the margins were 60,000, -20,000, 1,000, -80,000 and 40,000. The average margin is the same for both states: 200. But the volatility is much greater for State B than State A. Therefore, the likelihood of a close election is much higher for State A. To continue the investing analogy, I have factored in the alpha component (or more accurately inverse alpha, since we want low numbers not high performance) but not the beta component. But this is just a first draft to elicit comment and get it posted by the end of Election Day 2020.
There are additional factors that should be remembered when deciding whether to vote. The first is that voting is a right that many people around the world do not have. Furthermore, it was a right denied to different groups of people and in most cases it took hard fought battles to win those groups the right to vote. Taking that right lightly could be looked upon as disrespectful of those who fought hard to win the right to vote and even of those who fought hard to win independence and the right to self-determination for this country.
There is also an element of privilege to voting as well as it being a right. People under a certain age do not have the right to vote. While it isn’t a perfect standard as some people gain these qualities faster than others, the idea is that a person should have a certain level of maturity and experience before being allowed to vote. Non-citizens do not have the right to vote in federal, state and most local elections. In most cases, a person must have lived in this country under a certain visa or work status for at least 5 years and also pass a citizenship test (which sadly I suspect that many native-born citizens could not pass). And some people who have lived here longer than 5 years simply do not choose to apply for citizenship, forgoing that right to vote.
There are some other categories where the right to vote varies from state to state. The right of felons to vote varies from total enfranchisement to total disenfranchisement (except for the right of petition to have the right restored), with various levels or conditions for reinstatement in between.
While there is no law expressly denying transgender individuals the right to vote, in some places, transgender individuals have been blocked from voting because their gender presentation does not match the gender marker on their ID. This is why I voted by absentee ballot in 2012. Because of my profession as a professional tax preparer, I had a relatively tight window between October 16 and mid-January of the following year to legally change my name, change it on my legal ID, change it with Social Security and the IRS, and change it with the State of New York Department of Taxation (which heavily regulates paid tax preparers compared to most other states). There was no way that I could coordinate all that and also change my voter registration between October 16 and the first Tuesday after the first Monday in November. So before the deadline passed, I requested my absentee ballot, checking the most applicable reason on the application. Since then, I have always voted in person, whether in general elections or primaries. And it has never been a problem where I live. But it has been a problem for other transgender people. https://en.wikipedia.org/wiki/Transgender_disenfranchisement_in_the_United_States
Wherefore, brethren, look ye out among you seven men of honest report, full of the Holy Ghost and wisdom, whom we may appoint over this business. And the saying pleased the whole multitude: and they chose Stephen, a man full of faith and of the Holy Ghost, and Philip, and Prochorus, and Nicanor, and Timon, and Parmenas, and Nicolas a proselyte of Antioch: – Acts 6:3,5