Three-Fifths of a Floridian:
A Brief Synopsis of The Electoral College Primer
Peter A. Taylor
November, 2000
A number of good articles have come out recently in anticipation of the mess that the Electoral College (EC) has made of the 2000 Presidential election. However, I have also heard some arguments in favor of the EC, and the best way I can think of to show how misguided these arguments are is to write a short synopsis of an academic work called The Electoral College Primer, by Lawrence Longley and Neal Peirce (1996 edition).
Here is my proposed fix.
One argument in favor of the EC is that it supposedly protects the small states.
Part of the confusion here is that some of the biases the EC introduces into Presidential politics act in conflicting directions. States get one elector for each Senator (two per state) plus one for each US Representative, the latter numbering in proportion to the state's population. The extra two votes introduce a bias in favor of the smaller states. But the "unit rule" (the winner of a plurality of votes in a state takes all of the electors) used by all states except Nebraska and Maine means that a block of swing voters in a large state can determine the selection of more electors than an equal number of swing voters in a small state (although a larger block of swing voters is also likely to be needed to tip the outcome in a large state). Which effect dominates? Answering this, unfortunately, requires some statistical analysis, but fortunately, Longley and Peirce have provided some (see their Table 20, pp. 143-4). The answer is that the bias towards larger states dominates. The weakest voters are those in Montana, which is just barely shy of having enough population to warrant a second US Representative. Swing voters in Wyoming, with the lowest population, have an easier time tipping the outcome within their state, but get the same number of electors. Thus a Wyoming voter is worth 1.327 of a Montana voter. On the other extreme, the most populous state is California, followed by New York, Texas, and Florida. A California voter is worth 2.663 Montana voters, a New Yorker 1.888, a Texan 1.891, and a Floridian 1.663 . Thus a Montana voter is worth 60.13% (just a hair over 3/5) of a Floridian, or 37.55% of a Californian. However, this assumes that the local race is relatively close, an issue to which I turn next.
Here is a simplified example that tries to approximately reproduce these relative probability numbers.
Let's also take a more extreme example. As a thought experiment, imagine if California grew to 62% of the total US population. Its block of Electoral College seats would then, despite the numerical overrepresentation of the smaller states, have a clear majority, and no voter in any of the smaller states would have any effect on the outcome of the election. How does this benefit the smaller states?
Another argument for the EC is that it supposedly helps keep the country from becoming polarized by giving politicians incentives to try to be popular in every geographic region, rather than just in strongholds, and writing off other states.
In order to make sense of this claim, we first have to assume that there is some reason why geographic polarization is more important than, say, racial, economic, or ideological polarization, a point which I doubt, but which may well have been true prior to the Civil War. Now let's take an example: suppose we ignore the different sizes of the various states, and consider only two candidates. Suppose that Smith has 60% of the vote in 60% of the states, and 10% in the remaining 40%, giving him 40% of the total vote. Jones has the remaining 60% of the total vote (40% in 60% of the states plus 90% in the remaining 40%). Jones wins the popular vote, and Smith the Electoral College, but whose support is the more widely distributed? If we try to measure this in terms of statistical "variance" (the sum of the squares of local support minus average support), this will always be the same for both candidates under the assumptions we have just made (two candidates, equal sized states). If we use a "maximin" criterion, Jones should have won--he had a minimum of 40% compared to Smith's minimum of 10%. Another way to bias the results in favor of candidates with a little support in a lot of places is to add the square roots of the local percentages instead of just adding the percentages. By this criterion also, Jones should have won ( 60 * SQRT(40) + 40 * SQRT(90) = 758.95 versus 60 * SQRT(60) + 40 * SQRT(10) = 591.25 ).
With the Electoral College, it's true that once Smith has a firm majority in a particular state he has little reason to seek more support there, but unfortunately, neither does Jones--it's all or nothing. Rather than motivating a candidate to court voters where he is weakest, or even to court them equally hard everywhere, the EC merely encourages him to court voters where his popularity is close to being even with his opponent's. We saw the effects of this in the Elian Gonzales case, where a fairly small block of Cuban-American swing voters in a large state that might go either way in a Presidential election year were able to thumb their noses at the entire US legal system for months on end.
The idea that the Electoral College helps prevent polarization is even harder to take seriously if one considers minor candidates. A regional candidate like George Wallace can realistically hope to win some electoral college votes. A national candidate like Ross Perot cannot.
Much of the sentiment in favor of the Electoral College seems to me to be a vague sort of political ancestor worship, a sense that the founding fathers knew something we don't. So just what were they thinking? And does the Electoral College actually do what they wanted it to do? According to Yale law professor Akhil Amar, "The biggest reason it was set up was to protect slavery." My title is a play on a book title, Three-Fifths of a Man, which is a reference to the formula used in the Constitution for counting slaves towards a state's population for purposes of determining the size of that state's Congressional delegation. Another claim (these are from the Center for Voting and Democracy website) is that direct popular vote would have given the states an incentive to enfranchise women and non-landowners in order to increase their influence over the Federal government. Longley and Peirce give four reasons for the Constitution's framers to hesitate to select the direct popular vote:
- the lack of awareness and knowledge of candidates by the people, with unforseen consequences resulting from the scattering of votes by the electorates in the various states among favorite sons they knew best; [For a better way of dealing with the "spoiler" effect, see my essay on Instant Runoff Voting--PAT.]
- the loss in relative influence of the South because of its large nonvoting slave population;
- the dislike, on the part of the small, less populous states, of too open an admission of an inferior role in the choice of the president; and
- the fear of many that direct election of the president would consolidate too much power and influence in one person.
Whatever their misgivings, the delegates to the Constitutional Convention were tired when they were negotiating over the presidential election. They had been arguing for months over the composition of Congress. The Virginia Plan would have given representation in direct proportion to population. The New Jersey Plan would have given equal representation to each state. Finally (July 16th, 1787) they settled on the Connecticut Plan, the "Great Compromise." But then a similar struggle ensued between supporters of a direct popular election and supporters of having the President chosen by Congress. Eager to avoid another protracted battle that might derail the Convention, on September 7th, they adopted James Wilson's proposed compromise, and the Electoral College was born. There was little debate. As Alexander Hamilton wrote,
The mode of appointment of the chief magistrate of the United States is almost the only part of the system, of any consequence, which has escaped without severe censure....
It was, as John Roche put it,
...a jerry-rigged improvisation which has subsequently been endowed with a high theoretical content.
I will not dwell on the question of whether the Electoral College actually does what the framers wanted it to do. The short answer is, "no." According to John P. Feerick,
The system which emerged in practice is not the system contemplated by the founding fathers.
One factor is that the framers were not successful in discouraging the development of political parties. Another is that the states all eventually came to choose their electors by popular vote. Finally, all but two of the states adopted winner-take-all, the "unit rule," in deciding how the electors should reflect popular opinion. For the full story, read the book.
We pay for this "jerry-rigged improvisation" in a number of ways. The most obvious disadvantage is the occasional event of a candidate winning the election while losing the popular vote. This is an especially serious problem in the presence of spoilers (ie. Ralph Nader in 2000). A more serious problem, in my opinion, is that a candidate with a very small margin of victory, but whose support is distributed more or less uniformly throughout the nation (ie. Reagan in 1980), gets a false "mandate," a "landslide" victory that in some cases may not even represent a plurality of the voters. This undermines the intent of our system of checks and balances, which is to force political factions to compromise and work out their differences. Finally, and most seriously, is the problem illustrated earlier this year by the Elian Gonzales case, and then later by the famously geographically skewed distributions of money and effort that went into the Bush and Gore campaigns. I heard a report on the radio that the air quality in Houston under Bush's tenure as governor was being raised as a campaign issue in some of Gore's political advertisements in Michigan. I live in Houston, but the presidential race wasn't a tight one in Texas, and so I never heard the ads. This is despite Texas having the third largest number of electors, at 32. The President is supposed to represent all the people, not just the ones in closely contested large states like Michigan and Florida. As Longley and Peirce concluded in early 1996,
It is all too likely that the 1996 presidential election, or one in the future, finally will provide the American public with indisputable evidence of the failings of the electoral college as a means of electing the people's president.
I hope that the 2000 election has now done so.
Appendix A: Numbered summary points in the last chapter of the Primer (direct quotes)
The Electoral College at Its Best
The Electoral College at Its Worst
Appendix B: The Electoral College and the "discretization" of Voting Fraud (5-8-2001)
The electoral college has complicated effects on how sensitive presidential elections are to voting fraud. The EC makes small amounts of fraud irrelevant in many states, but increases the likelihood that a small number of fraudulent votes in one or a few critical states will determine the overall outcome of the election. Which of these effects dominates depends on how uniformly the politicians' support is spread across the various states. It also depends on the nature of the fraud. If fraud acts in opposite directions in different parts of the country, the effects would tend to cancel in a direct election, but are less likely to cancel in the electoral college. And how does fraud vary as a function of the major parties' relative strength? Is there more fraud when the parties are closely matched, or when one has a dominant position?
If we assume that all voters decide how to vote on election day by tossing coins (binomial distribution), and that fraud is proportional to population and acts in the same direction everywhere, the conclusion that I draw is that the defenders of the electoral college are correct: the electoral college reduces the likelihood of a presidential election being determined by fraud. But I have two caveats: One is that this result is tentative. I don't think this is a very realistic model of how electoral fraud works. With a different model, the results could easily go the other way. The other caveat is that even if this argument in defense of the electoral college is correct, I don't think it carries very much weight. I don't think voting fraud is all that hard a problem to deal with tolerably well. Also, the point of combatting voting fraud is to have fair elections. Embracing the electoral college out of fear that elections will be unfair without it is too much like committing suicide for fear of dying.
Appendix C: A proposed Constitutional Amendment (9-12-2006)
"Electors shall be chosen using the same districts and methods as the corresponding members of Congress."
This Amendment would cause the Electoral College to work the way most people think it works now, bringing a small benefit to the smaller states because of their disproportionate number of Senators. I believe it would also satisfy concerns about voting fraud. I would also like to couple this with another Amendment bringing the number of Senators closer to being in proportion to the States' populations, but the latter is not essential.
Here are some quick examples of how minor parties can try to cope with the tendency of our election system to force their candidates into the role of being spoilers [14k].
Here is an advocacy piece on Instant Runoff Voting [55k], the system used to elect Australia's lower house. It is suitable for either Congressional or Presidential elections, and eliminates problems with spoilers.