Skip to content

An election just happened and I can’t stop talking about it

Some things I’ve posted elsewhere:

The Electoral College magnifies the power of white voters (with Pierre-Antoine Kremp)

I’m not impressed by this claim of vote rigging

And, in case you missed it:

Explanations for that shocking 2% shift

Coming soon:

What theories in political science got supported or shot down by the 2016 election? (with Bob Erikson)

A bunch of maps and graphs (with Rob Trangucci, Imad Ali, and Doug Rivers)


  1. Lewis says:

    >>>> “The Electoral College isn’t just a goofy way of cumulating votes…”

    “Goofy” perhaps, if one is unfamiliar with the history & concepts of the United States founding. The Electoral College is certainly much less goofy than Parliamentary systems of selecting a chief executive (e.g., Great Britain), which totally reject use of the popular vote.

    The American Electoral-College process is constantly criticized by those oblivious to the origin of the ‘United States’ nation and its “federal” structure. America is a federation of individual sovereign states; popularly elected ‘representatives’ from within each state… together elect the chief executive of their federation — the “President of the United States”. Complaints against the Electoral-College are basically complaints against the concept of representative-government and the federal-republic form of government.

    • Andrew says:


      Yeah, it’s all gone downhill since they abolished slavery, huh?

    • Andrew says:


      Just to explain a bit: The U.S. electoral college is a relic. If it really were an example of representative government, so that we voted for electors who then, on their own, chose a president and vice-president, that would be one thing. But that’s not what we have.

      • Miss Ingdata says:

        I wonder why you did not adjust for turnout rates. Shouldn’t our concern be that electoral vote power is equally distributed among eligible voters, not actual voters? Granted there may still be a problem after adjusting for that, however.

        • Andrew says:


          The voting power of a nonvoter who decides to vote is 1/2 times the voting power of a voter who decides to switch his or her vote. In either case, the voting power depends on the number of people who actually voted.

        • George says:

          “Shouldn’t our concern be that electoral vote power is equally distributed among eligible voters, not actual voters?”

          Why would that be our goal or concern? The best reason to switch to a “popular vote” is that it would (should) cause more people to actually vote since every vote WILL count, unlike now. If you’re in the minority (or majority for that matter) in a state that always votes for the same party by 60/40 (even 55/45), your vote really does not count.

          In that case, the odds of it ever counting are … well, I’ll let one of the local probability gurus define that calculation.

          • Miss Ingdata says:

            Based on the data I found, voter turnout rates are not equal across ethnicities. Why should we ‘bake that in’ to our analysis of the fairness of the electoral-college-based voting power distribution across ethnicities? If people choose not to vote, that is their choice; I don’t think we should blame that on the electoral college (assuming voter suppression is not a factor, which it might actually be).

            This is not an argument for or against the electoral college.

          • Miss Ingdata says:

            I’m sorry, what I meant to convey with the ‘bake that in’ phrasing is that I don’t think we should expect electoral college vote distribution to correct for differential voter turnout across ethnicity.

      • Ian Fellows says:

        I have such a hard time understanding arguments in favor of the EC. The idea that votes from rural individuals should count for more than city individuals doesn’t pass the basic “democracy” smell test. By that logic, maybe we should give African Americans two votes each so that whites don’t have “undue” influence, or perhaps two votes for all those with college degrees as they are “better able” to make informed decisions.

        I’ve yet to encounter an internally consistent argument in its favor. And yet… every comment thread is flooded with apologists. Its almost as if there is a coordinated PR campaign. The feeling is almost similar to what you get on articles critical of China (

        • anon says:

          I think the EC can only be justified on empirical grounds. No social choice theory perspective (which studies the properties of different systems in terms of how they aggregate individual preferences) will recommend it.

          “rural areas would be forgotten!” sounds good, until you realise that it’s totally ad hoc and needs a theory of Undueness in order to not be ad hoc.

          • Andrew says:


            Also it’s hard to come up with a good reason why rural residents of Pennsylvania or Virginia should count more than rural residents of Texas (or why urban or suburban residents of Pennsylvania or Virginia should could more than urban or suburban residents of Texas).

        • Wayne Folta says:

          We may as well do away with the US Senate as well. How totally unrepresentative that small states get two senators, the same as large states.

          I think we need to distinguish two aspects of the Electoral College: 1) the “elect wise men who will consider and select a wise president” aspect, and 2) the aspect of winner-take-all state votes that forces the election to be state-wise.

          The first part strikes me as a relic of the 18th century, a fantasy, and in the current election we’re reading about death threats against Trump electors, we’re hearing about “faithless electors”, etc, and so this part doesn’t make any sense in our times. There’s really no argument for this mechanism which certainly doesn’t operate as originally envisioned.

          The second part still seems relevant — at least as long as states, the senate, etc are relevant. And I think it also points out three aspects of democracy that we could examine:

          1. Numerics: majority rule.

          2. Neighbors: the spatial aspects of our actual living arrangements.

          3. Niceness: preserving certain rights for the non-majority.

          In a modern, electronic nation, states (and state governments and winner-take-all by states) might also be outdated, so perhaps we should also eliminate them? Holdovers from the partisan, bad old pre-revolutionary days, perhaps?

          Those who want Chicago, New York, LA, and Boston to rule the nation tend to minimize points #2 and #3. The Electoral College at least indirectly addresses #2, though it could also do so as a theoretical construct that simply means winner-take-all in each state.

          There are advantages of a direct election, of course. Hopefully, more people would vote since their vote would count. Though I wonder if the elections would turn out as some of the (current) most vocal opponents of the Electoral College hope? For example, Hillary got 2 million more votes than Trump nationwide, but in California alone she got 4 million more votes. My amateur feeling is that in California — where all branches of the state government are in Blue hands — more Red voters have no reason to vote (local or national) than Blue voters, so perhaps in a a world without statewide winner-take-all Hillary still may not have won?

          Perhaps Andrew could comment on whether more Red or Blue votes are suppressed in California by the “my vote doesn’t count” factor?

          And of course a direct election is simple and understandable as well.

          It’s interesting that some feel the need to compare those who disagree with him as “apologists”, shills, and Chinese-style censors and propagandists. There’s no argument quite as classical and 18th century as a good old ad hominem argument, eh?

          • Ian Fellows says:

            I would certainly agree with you that you can’t separate the election from the win conditions. If the rule were majority rule, then Trump may have won the majority. Since the rule was the EC, campaign strategies and voter behavior reflected that.

            Regarding #2, spatial relations are only one way in which we are tied together as individuals. We also have race, gender, religion, etc. Among these aspects, I don’t see the spatial location of a voter as having any particular privilege.

            From my perspective the fundamental question isn’t about who should have won this particular election, but rather the ideal that we are all created equal and should have equal rights (and say) under the law. I wouldn’t want Chicago, New York, LA or Boston to rule the nation, and I wouldn’t want Helen, Paia or Camden to rule us either. I’d like each person to have equal say and protection under the law.

            Regardless, you definitely appear to be a real person who is (almost?) arguing in favor of the EC ;).

        • Jayson Virissimo says:

          The states united on the condition that these rules would become the law of the land. Arguably, they would not have agreed to do so without something like the electoral college, equal number of senators, etc… It is perfectly consistent and valid to affirm that one should always keep one’s promises, and that therefore we should not renege on this particular contract (between the states).

          Not that I think the electoral college is an optimal institution, but to change it without also letting the states opt out of their side of the bargain (essentially, the liberty to secede instead) would seem to be required by conventional justice norms. I’d wager that the bad arguments defending the EC you’ve encountered are simply rationalizations for these kinds of moral sentiments.

          • Jayson Virissimo says:

            Whoops, I meant to say that to change it without also letting the states opt out of their side of the bargain would seem to violate conventional justice norms, as I understand them.

            • Ian Fellows says:

              Depends how it is done. One current strategy is to put laws in place in states around the country so that each states delegation is committed to the winner of the national popular vote. If enough states join in on this, then the electoral college becomes functionally equivalent to the national popular vote.

              Secondly, we have come a long way from being a collection of united states. We are THE United States, and there have been a couple events in our history that I can think of establishing the primacy the whole USA over individual states.

              • Jayson Virissimo says:

                That compact you mention would be a way to move to a straight popular vote without violating conventional justice norms, so I predict that there would be less of these poor arguments (rationalizations) if it were this specifically that was being argued for, rather than a general “we need to end the electoral college” claim, since the latter seems much more like simply reneging on a prior agreement.

                The idea that previous agreements between the states are void because the civil war just seems like a might makes right type of argument. I apologize if you mean something else and I’m not being charitable enough.

    • George says:

      Lewis gets it. Andrew? He just likes nonsensical, snarky rejoinders. That seems to be the only life-skill of the Andrews around here.

  2. Miss Ingdata says:

    I agree that a rigged election seems unlikely, but there were a couple of points in your Slate article that I thought could use further elaboration.

    First, you claim that we should only take stories seriously if they are being covered by the mainstream media — You really trust the mainstream media to be completely unbiased in their choice of stories to cover? Given the slant of the election coverage (largely pro-Clinton bias), however, that claim may hold some weight, but in general that is not a principle that I would advocate.

    You give one counterexample (Wisconsin) to Palmer’s observation that “In any given hotly contested primary state, Donald Trump tended to perform the same as, or worse than, his polling averages.[…] In contrast, Hillary Clinton tended to perform about the same as, or better than, her polling averages in most states.” Did you actually look at the data on all the states and find that it was actually not the case in most states, as Palmer notes?

    The rest of your points are good, but could have benefited from more data and examples (e.g. how election day voters are different from early voters, past incidences where upsets in many states involved small winning margins). Anyway, I appreciate you sharing your perspective; I personally don’t have a lot of time to research the details and perhaps you don’t either. Also, I wouldn’t necessarily want to discourage further investigation or recount either.

    Enjoying the cat photos.

    • Andrew says:


      No, I only looked at Wisconsin. This was enough to make me not bother reading the rest of Palmer’s article, since he seemed to have no grasp of what he was talking about.

      And, yes, I agree that these differences could be studied further, most naturally by looking at earlier elections I did in my Florida 2004 analysis.

    • George says:

      “you claim that we should only take stories seriously if they are being covered by the mainstream media”

      Good one. Welcome to the Bubble. Make yourself at home. Tea? Cookies? Coloring book?

  3. Anonymous says:

    My understanding is that the electoral college was put in because in the 18th century, the citizens didn’t have the means to get to Washington to have their votes counted so the electorate would ride in on horseback and deliver their district’s popular vote. So now it is obsolete. Not everything in the Constitution works with today’s technology and couldn’t have predicted it’s obsolescence today.

  4. Julien Couvreur says:

    There are many instances of the electoral college pattern. For instance, in primaries, in Congress and Senate, in the EU, NATO and any kind of group of groups.
    Regardless of merits, I find it interesting how the electoral college became such a hot topic after the election. If there had been a long, slow and growing effort to lawfully adjust those rules, it might seem a bit less self-serving and opportunistic to question how the system is organized.

    Even with a popular vote, the way to deal with third parties will still create situations that can be argued about. After contested elections, I have no doubt such questions would come about (should we ask for each person’s second and third choice? should votes be pondered by the stake/wealth you own? how about education or age?).

    As many have studied (going back to Condorcet, and later famously with Arrow and public choice theorists), groups don’t have preferences in the way individuals have. But we have a tendency to anthropomorphize groups as if they did and assume some rational and objective method of aggregation despite its impossibility. The choice of rules and control of the agenda becomes the hard problem to “decide collectively”.

    Personally, I see this as an inherent trap of democracy. The permanent bureaucracy and the rulers are little affected by such trifles given the amount of power they still have a good chance to get their hands on. In some ways, they even gain when the ruled are distracted arguing against each other.

  5. Andrew [not Gelman] says:

    Funny how this only became a problem when you didn’t get the result you wanted.

    • Andrew says:

      Andrew [not Gelman]:

      I’m not sure which topic you’re talking about here, as there were five items discussed in my post above. So, here goes:

      1 . No, people have been talking about problems with the electoral college for a long time. The discussion just gets more intense after an election such as 2016 where the popular vote winner loses in the electoral vote.

      2. Same with vote rigging: people have been talking about it for awhile. I wrote my post expressing skepticism in response to a vote-rigging claim that appeared after the election.

      3. In my post about the 2% shift, the problem was that the polls had non-sampling error, or maybe the problem was that naive poll aggregators took their models too seriously. This has nothing to do with people getting the result they did or didn’t want. Had Trump won the popular vote, then we’d still be talking about a shocking 3% shift, maybe. Or had Clinton won by a landslide, we’d be talking about a shocking shift the other way.

      4, 5. Of course these are post-election items too. As political scientists, we can consider any unexpected outcome as a possibility for learning.

  6. John Goodwin says:

    If we were to adopt popular vote on a national scale (either a majority or plurality required) the effect would be tantamount to giving each HR district 1 electoral vote. You in effect give the cities a permanent lock on the executive branch, and marginalize their hinterlands. Disenfranchised groups tend to get peckish, in a democracy. What could possibly go wrong?

    The Founding Fathers were wise to balance blood with soil — the more soil you have, the more sunlight (solar power) and water you have. Cities need energy and water.

    A system of government that was allocated based solely on the number of buckets of blood the wetware has, with no reference to the interests of the light gathering power of the soil or rain gathering power of the watershed, would be doomed to failure.

    The Electoral College isn’t Red or Blue — it’s Green.

    • Ian Fellows says:

      No, it is tantamount to giving each citizen in our country one vote and an equal say. Giving sub-groups or Americans different voting privileges based on their location, property ownership status, race or gender is something that we have been moving away from in this country over the last couple hundred years. Each time we have acted to enfranchise our citizens equally, it was proven to be not simply the right move, but the Just one.

    • anon says:

      > You in effect give the cities a permanent lock on the executive branch, and marginalize their hinterlands

      They aren’t marginalized. Their collective voting power is proportional to how many people people are living there.

      If you worry about giving each individual the same power to determine the outcome of the election by voting marginalizes some groups based on clusters of correlations of voter preferences, we should aim for a principled fix, and not some ad hoc move that looks only at city vs hinterlands clusters.

      Alternatively, one could just stick with the renowned, age old idea of one person one vote, and devolve powers to states and local communities to address clusters of preferences, which is how most of the Western world does it.

  7. Wayne Folta says:

    Andrew: you have a point, but it also applies to the US Senate as well. And perhaps to Federal judges who are not directly elected by the people, but rather appointed by a president who is elected by states (unequal representation) and confirmed by the Senate (unequal representation).

    It seems to me that there is a geographical aspect to democracy that may not rise to the level of the numerical aspect but shouldn’t be ignored in practice.

    • Andrew says:


      There are three things going on here:
      1. The U.S. Senate massively favors voters from small states.
      2. The electoral college slightly favors voters from small states while massively favoring voters from swing states.
      3. Questions of political legitimacy.

      I don’t have much to say about item #3 right here, but I think items #1 and #2 are much bigger deals than would’ve been imagined by the authors of the Constitution, 200+ years ago. Item #1 is a bigger deal because the ratio of population of largest to smallest states is much much higher than it used to be. Item #2 is a bigger deal because, well, for one thing, there wasn’t much of a concept of swing state back then. The whole swing-state thing is a bigger deal now than it used to be, now that national elections are so close and state votes don’t vary so much from election to election.

  8. Jonathan (another one) says:

    Two things:
    (a) The Electoral College preserves an type of diversity: geographic diversity. It guarantees that an unusually geography-centered huge plurality can’t win an election. In any given state, it downweights the unanimity of opinion in that state. (Note: so would an electoral college in which the votes were precisely proportionate to population, which would eliminate one oddity without fundamentally interrupting the opportunity to have the EC not follow the popular vote.) Now you might find that a bug, and it certainly contradicts a different criterion, namely one person one vote, but it isn’t a tupid consideration ab initio. And it’s not accidental either… the EC came out of an elaborate compromise between the small states and the large states similar to the Connecticut Compromise (and indeed the makeup of the EC exactly mirrors the Connecticut Compromise.) Slavery, while a minor consideration in 1789 was somewhat more important in 1803 when the direct election amendment was passed. So, yes, an anachronism, but not without at least some virtues.
    (b) The observation above (downweighting more unanimous opinion in any particular state) explains why White voters have higher weight. They have higher weight because they are less monolithic than the African American community or the Latino community. If African Americans were as evenly split as Whites in voting, their votes would have higher weights. The swing states are swing states *because* they have fairly high White percentages *and* the White vote is more likely to move that the African American vote. So Black votes don’t count less simply because they happen to be in larger agglomerations — they count less (partly) because they are too predictable.

    • Andrew says:


      Interesting points, but I’m not sure it has to be that way, even given how different groups of people vote. After all, a lot of African Americans live in deep-South states which in recent elections have not been competitive—not because of the uniformity of the black vote, but because of the uniformity in the white vote in those states.

      • Jonathan (another one) says:

        I definitely agree that the lower weight for the black vote is a historically contingent fact and depends on the racial balance of the state and the typical split of the white vote in the state. But I think the Vox piece made it sound as if appealing to white voters was somehow baked into the electoral college system, and I don’t think that’s true at all. What makes a swing state a swing state is that it can swing. And what makes it swing is some largish bloc of voters who vote one way some times and one way another (where critically, the “largish” is relative to whatever the size of the nonswingers is.) If black votes rarely swing, they’ll end up downweighted or tend not to count very much at all. As it is, black votes count for little in the South (as you correctly state) as well as in California and New York, but for very different reasons — in one case because they don’t help, and in the other because they overwhelm the mean.

        Note by the way that I’m assuming elections aren’t about “get out the vote” but are about actually getting people to change their minds. I have no idea how much one matters over the other.

        • artkqtarks says:

          “And what makes it swing is some largish bloc of voters who vote one way some times and one way another (where critically, the “largish” is relative to whatever the size of the nonswingers is.)”

          Is this correct? Isn’t the reason why a swing state swings is because it is close to 50:50 and a little shift in votes can switch it from blue to red or vice versa, whereas even a bigger shift in votes in California is not likely to make California swing? In other words, isn’t it just the result of the winner-take-all system rather than the actual size of fluctuations in the voting?

          • Jonathan (another one) says:

            Yes, my wording was a little terse. By “size,” I meant the size of the gap between nonswingers for candidate A and candidate B.

            • artkqtarks says:

              Jonathan (another one): “By “size,” I meant the size of the gap between nonswingers for candidate A and candidate B.”

              I know that’s what really matters, but then it is not simply a matter of “you count less if you are too predictable.” The “swingers” don’t have a big influence simply because they are the ones who change their minds. Only those who are in competitive states matter. Even if a solidly blue or a solidly red state has a higher percentage of the “swingers” than some so-called swing state, their votes don’t matter.

Leave a Reply