The Low Information Voter and the Strategic Voter are not the Same
Biden’s support, to a large degree, is driven by the “strategic”, or insincere voter, who votes for a person they don’t actually want, to prevent an outcome that they see as even worse.
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.
Recently, there has been a backlash, if you could call it that, against a claim that someone apparently made somewhere, that South Carolina voters voted for Joe Biden because they were somehow “low information” voters, who needed to be told what to do, most notably by Representative James Clyburn, and then just dutifully followed his instructions. The implication was that whoever said this and whoever repeats it is somehow promoting a racist narrative about a predominantly black electorate in the Deep South.
One NBC article pointed to a single tweet by some dude with about 6000 followers as evidence that Sanders supporters had unified around the narrative of the “low information Voter” in South Carolina electing Biden. Whether there were more or not, I am not sure — I can easily imagine that this term has caught on with a number of people who are low-information themselves, and do not actually know that “low-information voter” is an actual term with a real meaning. The same article went on to dismiss the other attempts to theorize Biden’s overwhelming victory in South Carolina.
This NBC article, and a LOT of commentary on the wins in both South Carolina and on Super Tuesday, has tried to downplay the idea that Democratic Party Elites were scrambling to find a single “…candidate [who] was best positioned to fend off Sanders’ leftward-push.” Never mind that the following quote is lifted from the same piece:
“It was reported that Democratic Party leaders were working behind the scenes to find an alternative to Sanders, a democratic socialist who has long labeled himself an independent. But black voters like the ones who pushed Biden to big victories in South Carolina, and then in Virginia, Alabama and North Carolina, aren’t The Establishment, and it would be a mistake to characterize them as moderate, or even, as certain pundits maintain, somewhat conservative.”
It may be so that the “only concern” for many South Carolina Democrats is defeating Trump. And it is certainly so that a milquetoast Biden is probably preferable for these voters to a “full reset” in the case of Sanders. But it is absolutely NOT true that Biden won his massive victories across the south without Conservative votes, or that massive constituencies are fully on board with Biden’s policy preferences, such as they are. We know, for example, that much of Biden’s support in northern counties of Virginia came from Republicans who no longer supported Trump. Gov Terry McAuliffe, himself closely associated with the Clintonite wing of the Democratic Party, said as much in the New York Times. This Daily Beast editorial essentially makes the same argument and says in black and white what many Dems already know in the subtitle of the article: “ No, I don’t vote for many Democrats, and I’ll probably sit out November. But on Super Tuesday, I’m going to help the Democrats nominate their best candidate against Donald Trump.”
There is enough evidence then to suggest that, at least in some degree, southern voters who have essentially forced Biden on the Democratic Party were helped by conservatives who do not like Donald Trump, but don’t like Sanders even more.
What it says to me, though, is that these people who voted for Biden across the American South are certainly NOT Low information voters at all. They are strategic voters, who vote based on who they think other people will vote for. A strategic voter is also an insincere voter, who votes not for what he or she may actually want, but based on a rational calculation of utility: They would rather vote for a person they believe will win than voting for what they actually want, because they feel that conformity is a safer bet.
Philosophy of Science and Theory of Causation
Let’s take a step back here.
We need to dip our toes into the philosophy of science for a second before we can continue. Specifically, I am thinking about the rules surrounding the positivist notion of causality. We will leave alone the common trope of the sophists that “Correlation is not causation.” That is true. We admit this. And so we will assume that that doesn’t even apply here. Also we will make the a priori assumption that this theory is built from some data set that imaginarily exists out there somewhere, and this is not simply ad hoc theory. I shouldn’t have to say that but I am convinced that if more than one person takes the time to read to this point, they will object just to try to start a dilatory argument.
So, for the record, I am asserting that causation exists, that we can know it, and that we can determine, even if only probabilistically, that x causes y, or x→y.
Now: Some of the rules of Humean (or, of Hume) causation that I recall from my research design course are things like there has to be a correct temporal ordering between x and y (x has to come before y to have caused it), x and y are theoretically linked (we can’t have spurious correlations, or causal variables which have nothing to do with the variable under investigation) and that when several things cause the same outcome, there is something common to all of those causes which is the actual cause of the outcome.
I raise these rules of causation because these talk about the basis of determining, using data, that something causes something to happen. In Aristotelian terms, me causing something to happen is called “The efficient cause” (there are other types according to Aristotle, but I won’t talk about them.) For example: if I am looking for a theory of why the light came on, I can argue that the efficient cause of the light coming on was that I flicked a switch. In this sense, then, the theoretical statement would look like this:
Flicking a switch →light turned on
This theory is kind of boring. It just announces an observation about the causal relationship between two variables. The first variable, called “x” or the independent variable, is the thing that happens that is, at least for our purposes, not caused by anything else (we know, intuitively that it too was caused… What caused me to flick the switch? But for the sake of the model, we will say, we are not investigating the cause of me flicking the switch.) while the second variable is called the”y” or the dependent variable. The variation in the dependent variable is dependent upon the variation in the independent variable. When “Flicking switch” is “zero”, or not happening, then the value of “light turned on” is also zero; when “Flicking a switch” becomes “one”, as in, happening, then “light turned on” becomes “one”. In this case, we can say that flicking a switch is correlated to the light turning on, and if we can rule out all other theoretically relevant competing explanations, like “telekinesis” or “sudden shift of gravitational poles” or “air flow” or whatever, we can say, confidently, that x → y and that my flicking the switch caused the light to turn on.
The important thing here is that the independent variable MUST vary to produce variation in the dependent variable. Two corollaries to the Humean rules listed above are that non-variation cannot be used to explain variation, and variation cannot be used to explain non-variation. In other words, if I never flicked the switch, but the light came on, flicking the switch is not the cause of the light coming on; something else is. And if I flicked the switch on and off but the light never came on one way or another, flicking the switch cannot be used to explain the fact that the light is still off. Something else must be going on there.
We can also briefly touch on causal mechanism. This is where the theory becomes quite interesting. How does the mechanical action of flicking a switch on one side of the room cause the incandescent bulb to begin glowing on the other side of the room? There is a whole complex story involved with supply of power to a circuit, which is attached to a lamp, which causes a filament to glow brightly, involved in the simple x→y formulation. But the causal mechanism must be sequential and it must be consonant with the theory. You can’t have the filament glowing without first supplying power to it via a circuit, and you cannot supply the power without first flicking the switch. If your mechanism involves something like magic or trans-dimensional effects, we can say that it is not the flicking of the switch but the magic which is lighting the bulb.
The Myth of the “Low Information Voter” in Super Tuesday
With that background out of the way, lets move to “low information” voters.
The “Low Information voter” was satirized on Saturday Night Live very effectively in 2012. They called them “undecided” voters, but the gist was that these people were being asked to cast a vote in a significant national election, but knew almost nothing about the process, the candidates, or a host of other non-related issues such as how people get pregnant, and where one guy put his power cord for his laptop. The joke of course was, “In an age of media saturation, how can there possibly be a group of people who haven’t made up their minds yet? How can a low information voter even exist?”
Fact is: they can’t really. Even among fetishists who like to think of themselves as “independent” because they do not want the labels attached to them of being partisan, and all of whom think they are so much smarter than people who attach partisan labels, almost all of them vote for only one party or the other each election. These are the so-called “swing” voters, and as some data scientists suggest, they do not actually exist. “Low Information” voters, as thought of by political scientists, are typically completely disengaged from the process, view politicians with disdain and suspicion, and have very little information about a host of important issues and candidates, except for a ton of information about a single topic. You might better refer to them as “single issue” voters.
For example: A person who votes solely on gun rights will not have much information on trade or education policy, but will be able to cite laws and regulations of the Federal Government and several states, concerning the lawful ownership and use of firearms. The actual “low information voter” doesn’t really participate in the process, except when their chosen issue is on the ballot, so to say there are people who vote and are low information seems to be a contradiction in terms.
Voters are, instead, quite rational, and almost always partisan. Even if you brag about voting for “the person” and not for “the Party”, you still almost always only ever select from candidates of the Right or the Left, depending on where you stand. (This also kind of makes me suspect the mythical Obama-Trump voter, who claims to have voted for the first Black President and then turned around and backed an open xenophobe who could also be successfully be labeled a racist. I personally think those voters, when asked, all lied about supporting Obama due to beliefs about the social unacceptability of saying a person opposed the “first black President”.) Most people will vote based on who they think will win, and if they do not think their preferred candidate will win, they will often prefer to stay home, determining that the time and effort necessary to go vote for a loser is not worth the effort.
We don’t find many Low Information voters in primary elections campaigns which began more than a year ago. Fact is voters in South Carolina, and then on Super Tuesday were not “low information.” It is possible that they had already decided to vote for Biden before South Carolina, but didn’t want to waste their vote on some other “non Sanders” candidate unless they had no other choice. So they waited as long as they could before casting the vote, to get as much information and offset the risk they felt over who to back. There is some evidence of this, as people thought they might like to vote for Pete, but when he dropped out, voters were demanding new ballots so they could change their votes to Biden.
These people are not “low information”. They are acting strategically, trying to prognosticate, using all available data, who will stand the best chance of doing the only thing they care about, which is defeating Trump in the fall. This is the strategic goal of the Biden voter. They literally don’t care about anything else, if they believe it will prevent a Trump defeat by preventing a Sanders’ win. In fact, they are the epitome of the “Strategic“ or insincere voter.
Voting and the Parimutuel
The Strategic voter votes not for what he really wants, but what he thinks everyone else is going to vote for. It is a little bit like playing the parimutuel at the horse track. You have to decide where to place your money based on a couple of things: The daily handicaps printed in the bulletin are only about historical data, based on how the horse has done in past races. You also need the lines printed on the board to help you make your decisions. As more money floods into one horse, the line goes down. Note: these are not the probability of the horse winning, but the expected payouts based on where the money in the pool is. If the line is really low, that means that lots of people expect that horse to do well.
But the payouts will go up if there is little money behind a horse. This is an inducement to get people to put money into that horse’s line so the books can square, and the bookmakers can cover any longshot bets. When the race is run, and completed, the house pays out its bets based on how the money was divided. You get a share of the winning pool, which is augmented by the losses of the other horses.
Voters in the South on Super Tuesday were making those sort of bets, on Biden. They might not like Biden, but they were attracted to the line on Biden. In this graph, we see that on the morning of South Carolina, Sanders (blue) had been leading for over a month. Biden (green) was in 4th through most of January behind Bloomberg and Buttigieg. What happened is that the “bookmakers” in the Democratic Party dramatically increased the potential payout by upping the stakes with the Clyburn endorsement, which was made strategically to provide maximum effect in a key moment. Doing so was specifically aimed at pulling more votes to Biden. The result was stunning as you can see in the above chart. Biden went, overnight from 4th in the pool to tied effectively with Sanders. By Tuesday morning, with the additional endorsements cooked in, Biden was far and away the favorite among bettors. And all the voters followed what they now saw as a “sure thing”, that being a Biden nomination over the person who now seemed like a “long shot”.
This dramatic change lends some credence, if you were looking at it from a strictly parimutuel perspective, to the argument that the process is, if not rigged outright, at least bears undue influence which encourages strategic, instead of sincere voting. The punters, in this case the voters, were simply following the bookmakers, or in our case, the DNC elites who were pressuring candidates to coalesce behind Biden, and they were getting their information about what the handicappers were doing from sources running constant earned media coverage aimed specifically at aiding Joe Biden, like MSNBC and CNN.
Theory application
So how does this relate to South Carolina and Super Tuesday Voters and the idea of the strategic voter?
Well, I think the data I posted in this article was key to understanding the causal mechanism which led from Clyburn’s endorsement to Biden’s rout of Sanders during Super Tuesday. I’ll post the specific chart once more.
As you can see here — assuming that these poll results were accurate, and given modern polling methodology, we have to assume that they are something like the “best information that is available,” for what that is worth — late voters broke hard for Biden, in all post-South Carolina contests. Similarly, they walked away from Sanders. Why? We need a theory to understand that question. Why is it that “late voters” were significantly more likely to support Biden than “early voters” were? Did something significant happen between Clyburn’s endorsement, when Biden’s campaign was already being lamented over for its failure to stop Bernie from “destroying” the Democratic Party and guaranteeing Trump’s re-election? What is the “x” that caused the “y”, being in this case the dramatic move toward Biden on same day voting?
The thing that happened was South Carolina’s primary, which then led to a very rapid series of endorsements, by Party elites, including surprise withdrawals of Biden’s chief moderate opponents Mayor Pete Buttigieg and Senator Amy Klobuchar, as well as the unexpected endorsement by Texas Representative And El Paso native Beto O’Rourke, along with dozens of other very public endorsements. As I noted in the above article, the nearly 80 hours of solid pro-Biden and anti-Sanders media coverage following the South Carolina Primary gave Biden an inestimable earned media advantage. He got basically a free 80 hour infomercial from MSNBC and from CNN ahead of the Super Tuesday Primaries, and was also given significant help from other major news networks, on their Sunday programming. There are a number of stories on the topic besides my own, written the day after Super Tuesday, including this very explicitly outright one from CNN.
This series of articles completely contradicts the later, and somewhat apocryphal narrative that “the establishment” didn’t tip the scales in Biden’s favor in an effort to stop Sanders. Countless tweets from defensive Biden supporters have been written making some variation of the following claim: “I’m a voter in South Carolina. I am working class. I voted Democratic all my life. I have finally found someone to trust in Joe Biden. #Iamtheestablishment.” These Biden apologists are aided, whenever possible by Democratic Party elites, like, for example, Senator Klobuchar, who strangely insists that she is “not the face of the establishment.” The goal is to invalidate the critique that Democratic Party elites ALSO coallesced around Biden ahead of these elections and their endorsements, as well as the massive earned media blitz helped reassure Voters that Biden was the most likely to prevail against Trump in the fall.
The overall argument suggests that it is offensive for Sanders to suggest that the elites of the Democratic Party put their finger on the scale, as if the voters themselves were so low information that they just did whatever they were told. Nobody ever argued that seriously, except for either people who are themselves low information, and do not understand the concept of the “low information voter”, which is, itself, something of a myth, or are looking for a cheap, superficial target to blame for a deep structural problem within the Democratic Party nomination process that favors Biden and disfavors Sanders.
At the same time, we can also rule out competing causes for wins. Theoretically, we know that spending in states on media, and ground organization which is primarily focused on Get Out the Vote efforts (or GOTV) are important factors in whether or not someone will do well in a state. Usually, an increase in spending and organization will lead to an increase of voter turnout for a particular candidate. But we also know that Biden didn’t have much of that, well, anywhere. In a state as important as Texas he apparently spent less than 100,000 dollars. He didn’t spend any money in Minnesota, but was put over by his recent rival turned support Amy Klubuchar, who apparently acted as a “Jim Clyburn of the North Woods.” Biden spent almost nothing, because he had almost nothing going into super Tuesday. His ground organization was similarly weak.
Using our causation discussion above, then, we can make a theory of Super Tuesday and say, pretty convincingly that the the causes for at least some of Biden’s overwhelming success was key Elite endorsements and massive earned media. We can also discount money and spending as being influential. It clearly wasn’t, especially given that Bloombug dwarfed all of his rivals combined, and he got delegates from Guam on Super Tuesday, but nowhere else.
We can argue that the Clyburn endorsement and the SC win came first in the mechanism, which led eventually to Democratic Party elites, who were just looking for someone, anyone not Sanders to back, “coalescing” around Biden, which led to pressure and then finally cave-in from Biden’s rivals, who conceded that they were preventing the Democratic Party from stopping Sanders by splitting the anti-Sanders’ vote in the primary. This led them to endorse Biden in prime time on Monday Night, which in turn signaled to jittery voters who believed that Sanders was too radical to beat Trump in the fall that the Democratic Party was unifying around Biden as its selected nominee. Voters responded to THAT fact when they went to vote on Tuesday. You can see the residual evidence of that mechanism in the variation of numbers from early and same day voting in the chart above.
Conclusion
This result is not the result of “low information” at all: it is equivalent to placing your bet right before the window closes, instead of beforehand, because you spent your time watching where everyone else was putting their money. In terms of theory, then, the variation that caused the change in outcome for Joe Biden was the variation in the robustness of his support from the Democratic Party, and its associated elements on Opinion Cable. Contrary to the expectations of the “low information” voter, voters across the South were EXTREMELY well informed on Super Tuesday: they knew not only what the Democratic Party wanted to happen, but they also devoured polling data and talked with all their friends and family before making their vote. They may not have voted for who they actually wanted, but they voted based on who they thought everyone else was voting for.
There might be a sizeable number of “sincere” voters out there. A large bloc might actually like Biden. We can take seriously the words of those black voters who voted for Biden because they knew him from the Obama years, and they others who actually listened to the words of Clyburn, who called Biden a decent person. But we can also bet that Biden’s margins were not driven by sincere voters: Given his poll numbers going into South Carolina across Super Tuesday states, and the change in his support between early voters and same day voters, at least a large portion of his margins were these insincere voters who were voting based on what they thought others were doing.
The people we must take a bit more skeptically are those in the Democratic Party leadership who claim that there was no attempt to prevent Bernie Sanders from obtaining the nomination, that there was no effect on voters from all these endorsements, and that this process is being decided entirely based on votes of voters. There is ample evidence of a coordinated effort to “stop Bernie” or to prevent him from succeeding at the Convention, including a discussion about changing the rules to allow superdelegates, which are presumably all lined up behind Biden already, to vote on the first ballot, thereby effectively negating the results of a close plurality for Sanders in won delegates. There was also pressure on other candidates to get out of Biden’s way and let him take on Sanders one on one, dramatically improving his chances to defeat Bernie state by state. The betting pool data above also supports this conclusion. There are also claims of voter suppression in Texas and California, which might make the difference for Sanders in terms of delegates.
Whether the process is rigged is something for readers to decide for themselves. At any rate, we should conclude that voters themselves, who were largely moderates and conservatives in terms of the Democratic Party, acted strategically, and though they might not actually like Biden, in the Primary, Sanders is worse to them than Trump, and they would rather vote for someone they do not want, then get stuck with someone they fear.
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.