Critical Ideological Miscalculation Could Spell Disaster for Democrats in November
A failure to understand voter preferences and the Party’s position in relation to those preferences will lead to the loss of the Left to a Third Party, and therefore an easy win for the Incumbent.
As Super Tuesday ended and pundits on MSNBC and CNN were falling all over themselves to declare Joe Biden the presumptive nominee of the Democratic Party, those people who understand American politics were likely split on the question.
Like all articles on The New Haberdasher, this story is presented to you for free. If you like what I do, consider supporting my work with a small monetary contribution at my Patreon and thank you.
Some political and/or data scientists probably agreed with the Media’s presentation of a completed primary process. It was obvious, after all: We had been told for months that South Carolina was reflective of the Democratic Party as a whole, and not only they, but many other different kinds of states had solidly backed Joe Biden in that short period of time, taking him from being on “life support” to being declared no less the “Presumptive Nominee” in just a few short, miraculous days before the virus set in and cancelled most of the remaining contests, as per the demands of the Ragin Cajun himself.
In addition, many of us, who mispredicted the 2016 contest have not yet regained our self confidence from such a catastrophic failure in our models of understanding the election. As I have heard many times, Trump needed a host of random events to fall his way for him to win, and he got enough of those lucky drops to put him over the line.
But it is possible that, like generals, people who forecast elections really do just refight the last war. We build our prediction models based on available data, though some of us have taken to simulating data as well, to account for strange occurrences which we haven’t yet seen, but which are not outside the realm of possibility. Given the fact that much of our work is done using existing data and theoretical models which explain why some one thing x happened this one time and not thing y as we might have expected, it is also possible that we have completely misinterpreted the cause behind thing x happening in the first place while we are parsing the past to glean insights into the future.
In short: perhaps we have applied the wrong theory. Or, worse still, we tried to make the world reflect our theory, rather than the other way around. (If you think this is bonkers, just know that both liberal economists and Marxists do it all the time.)
So: maybe we can bring in a different theory to help us going forward, since the ones we tried failed us so thoroughly.
The Median Voter Theory
I’ve written about Median Voter Theory of Anthony Downs in the past, when attempting to predict the 2012 election following the Conventions. In that article, I built a conceptual model based on both the Median Voter Theory and other work relating to US presidential elections. So I figured: Why not give this one another spin in an attempt to apply it to what is currently happening in the Presidential Elections.
For some temporal context: Today is April 28, 2020. We have been in a coronavirus pandemic situation for about a month and a half now. Just yesterday, they cancelled the New York Primary, much to the chagrin of Bernie Sanders’ supporters, but he officially suspended his campaign and endorsed his primary competitor, former Vice President Joe Biden weeks ago, also to much chagrin and disappointment of his supporters. As a result, the feeling that the primary elections were “stolen” by the DNC elite (whether they were is up for debate) is pervasive among most of Sanders’ base, who have, to my knowledge, for the most part sworn to not vote for Joe Biden, under any circumstances, and are currently looking for an excuse to join the Greens in protest.
Good. Now we’re caught up.
We need to cover the Median Voter theory though before we can go forward. What I am about to provide is a VERY basic summary of a very nuanced theory, popularized by the work of Anthony Downs in his groundbreaking and influential book An Economic Theory of Democracy. It is in no way complete as I am writing it. There are so many more parts to it than I am going to write about in this very basic overview. If you really want to read more of this when you complete this article, I’ve linked a paper written by Downs that then formed the basis of his book. And if you look hard enough at this article, you may be able to locate the full text of the book.
The theory is one that attempts to explain why a Party, composed of rational politicians, would adopt certain policy preferences. The notion here is that, in an electorate, which is split between people who hold one of two competing visions for the country, and which is governed by majoritarian,
“winner take all” and “first past the post” election rules as we are here in the US, the politician who obtains “half plus 1” of those votes cast, or the plurality, as the case may be, wins the election. So the politician, whose rational goal is to get re-elected, will look for the way to adopt and run on the political preferences of the “median voter”, or that voter who could go either way, equidistant between the two ends of the political spectrum of the whole electorate. If that one voter is attracted by the politician, then the election is won, and the politician stays in office.
This theory suggests a reason why, in a country with an electorate and a political structure like the US possesses, the voters might WANT our politicians to be as radical as possible, but the politicians will rush to the center in General Elections, fight for the votes of “undecided voters” and promote the most moderate policies they can get away with. If they can keep the median voter of their own Party AND gain the median voter of the general electorate, they will be guaranteed reelection.
Above, in Figure 1, we see a model of the Median Voter Theory. In this model, there is one “mode”, at the ideological center of the ideological spectrum. A number of voters on the extremes of the spectrum are guaranteed to vote for their preferred candidate, and the real fight is over the VAST majority of voters in the center, who could, in theory go either way. There is some probability that a voter will vote for one candidate over the other, reflected as a point on the blue or red outside arches corresponding to a given ideological preference. For example, a voter who expresses preference “-2", in this model, has about a 38% chance of voting for the candidate, and a much greater chance of not voting at all. A person whose preference is measured at “-1” is almost assured to vote for the candidate from the Left side. A voter whose preference is at -.5 is near 100 percent likely to vote for the Dem, and also has a pretty good probability of voting for the Republican too — Hence they are “undecided”, though in this model, they will more likely vote for the Democrat, because the policies of the Left conform more to their preferences.
This model is just that — a model. It is not supported by data of any kind. Nor is it useful for anything other than actually considering the theory in a visual form and teasing out some of the implications of the theory. But: it is a model which many political strategists, and strangely enough, voters this year have implicitly printed in mind, though they may never have heard of the Median Voter Theory. Democrats, especially, seem to think that the more centrist the candidate is, the more likely he will be to not only pick up ALL the Democratic voters, but also some of the Republican voters as well. This is one of the thought processes behind the “strategic voting” we saw this year in the South, among older conservative Democrats who are more concerned about defeating Trump than ANY other issue or group of issues.
And this concept was echoed in the words and warnings of countless Democratic-aligned pundits and officials who fretted over Sanders’ potential alienation of the middle. Unfortunately, this particular model is a relic of a electoral structure which no longer exists in an age of polarization and tribalism. So policy options and prescriptions based on this outdated model will be counterproductive.
A 2020 Median Voter Theory Model
The flaw in the logic, I believe is in the belief that the vast majority of voters are in the middle. Instead, I would like to present a different model, which is also covered under Median Voter Theory. In Structural Realism, we would call the Uni-modal distribution “uni-polar”. But I think what our country looks more like a “bipolar” world. I won’t talk more about structural realism, but those who know it will understand why I made the reference when they think about the viciousness that Biden and Sanders supporters fight with one another, or the constant “rank policing” that occurs daily on the Right.
In Figure 2, we see something which I think is more akin to our tribal, polarized reality. Here, voters are spread normally in a bimodal distribution, and the center of those distributions form the median ideological preference of each of the Parties. There is not one median preference, therefore, as there was above, but two.
For the sake of this model, we need to make some a priori assumptions which would likely not stand up to a strict rigorous test with data but I adopt now to make the illustration:
- Both sides have the same number of voters.
- The sum total of all the ideological scores of all the preferences of all the members of each Party added up to some number and becomes the center of the full range of ideological positions held within the Party, from -5 to 0 for Democrats, and 0 to +5 for Republicans. In Figure 2, that median ideological score for Dems and Republicans is -2.5 and 2.5 respectively.
- That median is merely a place holder for “some number which is equidistant away from the center and some distance from the fringes of the spectrum.”
- Each of these ideological preference “profiles” if you will, is anathema to the other side. There is nothing that the Dems who sit on the ideological median could want that Republicans can support, and vice versa.
- All Democrats and Republicans are normally distributed over their respective medians, with more being clustered right over the top of that median and fewer and fewer situated the further from median you get, with NO Democrats adopting positions of the Republican side, and vice versa.
This is an objective, or structural condition of the field. Against this structural distribution, candidates from either Party have laid out their platforms. These platforms cannot possibly conform exactly to both distributions, because the candidates themselves are MEMBERS of the distribution, and they still need to find a way to win a majority, not merely the 50 percent that happens to be on their side of the spectrum.
As you might guess, therefore, the individual preferences of the candidates themselves have a very high probability of being either to the right or the left of their Party’s Median, but Median Voter Theory will suggest that the candidates’ preferences are closer to the center of the ENTIRE structure, than further away from the center, toward the fringes, in order to attract the median voter of the whole electorate.
We happen to be able to observe this in the real world, though real world observation is not really all that necessary for this discussion of theory. Biden is widely considered to be a moderate — it was one of his selling points in the conservative South. Meanwhile, during the three and a half years of the Trump Administration, the President has been constantly surrounded by people whose sole mission was to pull him further to the Right, when in fact many of the positions he ran on were fairly standard GOP positions. He still, in fact holds most of those positions, even if SOME of his policies on immigrants and judges have been deeply offensive to the Left. Neither Biden nor Trump are radicals, even when compared to one such as Sanders, who himself has never called for nationalization, though he claims to be a socialist.
In our model, we see two arches that rise above the distributions. These, like the Figure 1, show the probability of a person holding a particular ideological position to vote either “blue” or “red” as the case were. The two anchor points of these arches along the X-axis are set by the candidates themselves, and the curve of the arch is meant to intersect both the measured “enthusiasm” and the apex of the distribution over the Median preference.
I calculated “Enthusiasm” by adding “Definitely” to “Somewhat Likely” to vote for said candidate and throwing in 2/3rds of the “Unlikely” respondents, because I know voters habitually lie when they claim they are independent. My source is an ABC poll from March 28, which I used just to have some number. The reason I intersected those two points with the “probability” arch is because I needed a way to represent that those over the median are some high probability likely to vote for a candidate closer to the center than they are, and also, people closer to the center, and therefore closer to their own preferences are MORE likely to vote for that same guy.
We’ll call these two arches the Democratic Probability Arch (DPA) colored blue and the Republican Probability Arch (RPA) colored red, though they are more like index lines than structural features. As you see, where the DPA and the RPA intersect, THAT voter has a 50–50 chance of voting for the Dem or the Republican. As you can see by this model, that voter sits over a negative value, meaning that not even all moderate Democrats have a greater than 50/50 chance of voting for Joe Biden. According to this model, he’s lost the Median Voter. More about that later.
In our model, the left and the right policy boundaries are set by the candidates and the Party. Here, these boundaries, the anchor points of the DPA and RPA, represent educated guesses based on each of the candidates’ stated platform positions.
For example: The GOP, since 2010, has adopted a hardlined approach to several major economic issues, such as taxes and immigration. Trump is running this year as a person who gives his entire base what they want. My placement of the RPA anchors is therefore reflected in the fact that even after 3 and a half years, more than 90 percent of all Republicans still support everything he does, and the ones that do not either do not belong to the Republican Party (and therefore make up their own distribution which I haven’t included here) or are on Trump’s Left.
Meanwhile, the policies adopted by Biden and the Democratic Party this year have significantly alienated the Democratic Party’s Left Wing, especially those under the age of 45. It seems that this was an attempt to secure “Never Trump” voters and older, conservative Democrats, at the expense of a group of people who “never vote anyway” and therefore cannot be counted on to vote in the Fall. Biden’s Left anchor, especially therefore must be placed to reflect that he and the DNC have shorn off the ideological preferences of about 31% of his Party, while his right anchor has to be placed about 2 SD into the Republican distribution to show that he can at least attract the attention of about 10 percent of the GOP.
We can see in this figure that the area under the distributions are colored different colors. The first color I wish to draw your attention to is “purple”. The area under the DPA on the GOP side are people who we call “Never Trumpers” — These are about 10% of the GOP. Since his base is almost completely with Trump, we can assume that the only folks who have any probability of voting for Biden are the center-right rump of the old Republican Party, who find Trump repulsive. They would in theory still vote for Trump, and are probably more likely to do so, but Biden also offers them an alternative.
Similarly, on the Democratic Side, you will see some Democrats under the RPA who are called “Obama Trump Voters” who voted for Obama in 2012 and then switched to Trump in 2016. These are more likely to vote for Biden this year than Trump, because many of their votes cast for Trump in 2016 were cast because they were voting against Clinton.
You will also see blocks of blue and blocks of red. These are “Guaranteed” Dems and Republicans respectively, though they might be thought of as “Reliable” populations. They are the number of people distributed over various ideological positions on the chart. The closer they are to the center of their Party’s ideology, the more of them there are — Hence they are normally distributed over the Party’s Median position, whatever that happens to be in real life.
You can use these red and blue colors to determine the relative strength of those voting blocks which will reliably vote for Biden or Trump. In this model, Trump has reliably about 90 percent of all those who will vote for a Republican based on preference this year, while Biden has something more like 65 percent. This is NOT a prediction of the final vote totals of course; but based on the model, we can tell that, at this moment there are a LOT of competitors for the Democratic Party voter, even still, while comparatively few for Republican voters.
Finally, you will note the share of the Democratic voters beyond the DPA and a tiny sliver of Republicans beyond the RPA. The Left is colored Green, while the Right is colored Yellow. These represent opportunities for Third Parties to come in and easily poach voters from the respective candidates. As you will note, due to the fact that Biden has anchored the left side of the DPA to a position well inside the full spectrum of Democratic voter preferences, there is AMPLE opportunity for a third Party to rise on his Left and either pose a serious challenge to him in the general election or pull him to the Left in an attempt to get him to support policies which neither he nor the DNC, and certainly not his conservative base want to support this year. Meanwhile, Trump does not face such a threat. Any opposition he has on his Right will not take away from his own base of support.
Implications of this Model
I have stressed, a number of times that this is merely a graphic representation of a concept. It is not based on data that I have collected and analyzed. It merely is designed to demonstrate the interaction between the candidates’ more moderate preferences and their bases’ more radical ones.
As a candidate courts votes from the center, as Biden has done this year, his Left wing becomes less and less likely to vote for him. Ultimately, he will move the Left anchor of the DPA so far to the right, that Left wing voters will not vote for him under any circumstances. “Needing to defeat Trump” is not nearly enough to placate voters who believe that Single Payer Healthcare, Climate Change, and sexual assault charges are more important. They will be far more likely to become fodder for a Third Party to come and offer them what they want, and Biden, who has staked out a position on his Party’s Right Wing, in an effort to appeal to Republicans, will lose them and their votes in the General Election.
But if Biden attempts to move left to pick these voters up, the probability of a Never Trump voter coming to vote for Biden decreases, and the appeal of either Trump or abstention becomes more attractive for them. He also slightly decreases his probability of Rightist Democrats, like those who put him over in the South, voting for him, because as they and even Barrack Obama have said repeatedly, they do not want a “revolution” or any significant change: they just want to get rid of Trump.
Meanwhile, Trump, who enjoys a higher enthusiasm among his supporters, and near complete hegemony over his entire base, can afford to move his RPA anchors out toward his fringes on the Right without alienating a large portion of his main base. He can also continue to infringe on Democratic issue areas like “prison reform” and “working class concerns.” He can easily jump to Biden’s Left on a host of issues such as the War and Trade, without losing his right wing, who are deeply and personally committed to him and his agenda. And he can continue to throw red meat to his Right wing without worrying about losing his Left wing, who are for the most part gone anyway as Never Trumpers, because he has no serious challengers to his Right.
This model talks about tradeoffs. How much do you lose on one side by moving the other way? Will you retain enough votes to help you win, if you trade votes over one ideological position for votes over another? What will happen if a challenge comes in from your wing which forces you off the position you think you have carefully staked out for yourself, either by pulling you toward the challenge, or crowding you out toward the middle, where there are almost no actual votes?
We see that of the two candidates left, Biden is in a far more precarious position than Trump is. He and his DNC strategists have misread the structure of the electorate, in a very crucial way, and the adoption of positions in the Center while sacrificing the Left causes him to trade a large swath of numbers AND energy, for a few conservative votes in states which will not be voting for Democrats in the fall anyway. Yes, he will keep votes in the Midwest and in the Suburbs, but he had those votes anyway. What he and his Party lost by their sacrifice of the Left will not be made up for at all when the Electoral College votes are cast and all those “landslides” he got in South Carolina, Mississippi, and Alabama aren’t much more than a bit of gravel compared to the avalanche that the GOP is going to turn in in those states.
In short: Where he believed all the votes are, there is almost nobody there anymore.
Can Trump lose this Election?
Sure he can. I want to stress, that while Downs asserted that this model was designed to explain individual, rational behavior, my take on it is a structural model of the electorate. Therefore, it seems more deterministic than the events on the ground actually bear out. I am saying, based on the structure of our electorate, Biden has made a number of critical errors, which make it far easier for Trump to not only win the Electoral College Vote, but to also reliably expect to win the popular vote this Fall.
The misreading of the structure opens up opportunities for Trump which Biden has foreclosed to himself.
But none of that means that Trump will win. There could be a number of things that happen within the structure which could give Biden the advantage. My numbers could be, and probably are, completely off. Like a plastic Snaptite model of an F-14 is never meant to fly, but instead is just designed to show various characteristics of the physical fighter jet, this model is not meant to be used to do the heavy lifting of prediction. If we actually had data to put in and create a more accurate and realistic model, it might show the same thing as I showed here, but it might show something entirely different.
Trump himself could do something so far out that even his most diehard supporters can no long back him. He may lose issue positions to Biden’s left, or he may not be able to hold his right wing together as he has so far. What it would take to accomplish these feats seems to be more than the effort and random luck it took to get him elected in the first place, but there is no accounting for the unexpected, and can be none in a structural model.
Furthermore: People are random. Rationality is assumed as a part of the theory, but it is rarely ever observed, at least in the terms that the observer is expecting. I assume that Leftists will be so alienated by Biden’s centrism that they will not vote for him under any circumstances. But, many of them are expected to “come home” to Biden. Just as in 2016, when Sanders supporters claimed to despise Clinton, still, 90 percent of them voted for her anyway. Or: they could be so disgusted that they vote for Trump, which is likely for a few at least, but which is not reflected by this model as even being remotely possible.
Also, we do not know the effect that a Third Party will have. I expect it, based on this model, to eat into a significant chunk of Democratic Voters who would otherwise not vote for Biden but would probably stay home. But a lame Third Party option on the Left will likely also serve as a push factor to push those voters toward a pragmatic vote, rather than an ideological one. Depending on how a Third Party candidate positions him or herself, that Third Party could ALSO completely undermine Biden’s chances, and either assure Trump a win, or win themselves.
One thing which could significantly alter this calculation would be a Third Party challenge from the Center. This model suggests that there really is no room at the center, and no votes anyway, but the recent news of Michigan Congressman Justin Amash’s launch of an exploratory committee for a Libertarian Bid could threaten potential Never Trumper support for Biden. Surely Democrats will know this is a possibility, but I imagine they think Americans aren’t paying attention and hoping nobody will notice. On the other hand a Libertarian challenger would be doomed if it came from Trump’s Right, because while there are few voters in the center, there are almost no voters to Trump’s right this year.
The only way a challenge from the Center could help Biden is if it forced him to move his left anchor to the left in a dramatic and credible way. He would lose the Never Trump Republicans most certainly, but he probably didn’t have them anyway. But he would gain a vast swath of his voters which he is leaving on the table. This is not unprecedented: Robert Kennedy did it in 1968. But it is VERY rare in American politics, because doing so means turning your back on your erstwhile supporters in an attempt to gain new ones. It is a gamble: such a move could very easily appear cynical. But doing so could assure that a Third Party in the center, aligned with the Right as the Libertarians are, hurts Trump rather than helping him.
The Miscalculation is a major problem for Joe Biden
The thing I want to point out, however, which I think this model correctly demonstrates, is that Biden’s position, especially his anchor point to the Left, is the problem for him. By refusing to adopt Leftist policies, especially those favored by his lately primary competitor, he runs a significant risk of completely alienating those who disagree with him ideologically. He seems to believe that a rightist, moderate, “change nothing” platform will attract enough Republicans to make up for this, but I think he does so because he does not understand the structure of the electorate, and those who are offering him advise do not either.
We got this way because he was declared the winner after the states which voted in person did so. The characteristics of Democratic voters in those states fit nicely with the profile that is represented in this model. There are fewer Democratic voters in those states relative to the population as a whole (i.e. States in the Deep South are overwhelmingly Republican, so those who voted for the Democratic candidate tend to be more conservative than the Party’s Median), and those states which voted for him are not even the majority of the Party’s voters.
Additionally, Biden has only won 41 percent of the popular Democratic vote thus far, most of it coming from Southern States and states won by Donald Trump in 2016. Of the 30 contests (not including Samoa, Marianas, and Dems Abroad) which have been held so far, only 13 of them were in states that Dems won in 2016, and Sanders won half of them, some of those by spectacular margins, while Biden’s biggest margins have come in Southern States and states which have held their contests since Biden was declared the defacto winner by the Dem Party aligned Press following South Carolina and Super Tuesday. In only two of the states where Dems won in 2016 that Biden took this year was his margin greater than 9%. And now, given New York’s desire to cancel its primary election and voting occurring during the Pandemic, there will not be any more opportunity for Leftists to express their preference in the Primaries. Those avenues of participation in the Democratic Party are now entirly closed to anyone in the Left third of the Party.
So: there is some support for my model’s claim that Biden’s ideological support comes from centrist and conservative voters, especially in the South. I do not believe that there will be any pressure from official sources for Biden to move left between now and November. The more his current position is confirmed by Party endorsements and what are essentially “rubber stamp” ratifications instead of competitive elections, he will feel no pressure from below to move left either.
In this case, I suppose we will see in the fall what happens to the Left of the Democratic Party. My model predicts that it is ripe for 3rd Party picking. And since they will not be helping Biden win, while Trump’s supporters are almost entirely on board, this critical miscalculation with regard to ideological positioning due to a failure to correctly read the structure of this year’s electorate can be expected, with some reasonable degree of certainty, to Joe Biden’s demise and an easy victory for Donald Trump.
Like all articles on The New Haberdasher, this story is presented to you for free. If you like what I do, consider supporting my work with a small monetary contribution at my Patreon and thank you.