Minutes before a New York jury convicted Donald Trump on thirty-four counts of falsifying business records, I had posted the following on social media:

Many trial watchers were curious whether the law would prevail over mathematics. Here, I explain why mathematics predicted a hung jury. I also explain how legal realism theories that the Trump team advocated for trashing the trial do not fully explain the jury verdict. The hardcore legal formalism we teach in law schools triumphed over mathematics and legal realism. The triumph of law over mathematics is no less intriguing than the triumph of legal formalism over legal realism, the terms I explain below.

Probability Model

The probability model, not law, predicted that the jurors drawn from New York voters would be divided just like the polls showing Trump with a slight lead over Biden. Throughout the Trump trial, Biden trailed Trump by a narrow margin in nationwide polls. The jurors heard the evidence, and the people watched television and read newspapers to follow the Trump trial. Since the jury stands for the community, a broad legal premise, pro-Trumpers on the jury should be roughly equal to pro-Trumpers among voters. The notion that voters are politically split for and against Trump, but the jurors are not, runs counter to the probability model. The following analysis shows how the public polls did not change much as the New York trial went ahead. Did the pro-Trump jurors, if any, change their minds after hearing the evidence? The timeline tells the probability story.

On April 19, 2024, a jury of 12 jurors, five men and seven women, and six alternates is seated. Among the jurors are an oncologist, a nurse, an IT specialist, and a lawyer. The alternate jurors are five women and one man. (Thus, out of eighteen jurors, there were 12 women and six men). On April 20, polls showed Biden-Trump at 45%-47%. Since the jurors are a random sample of New York citizens, they should likely represent a partisan split: nearly half of the jurors should be pro-Trump. However, jurors are screened through voir dire, a selection process to recruit a fair and impartial jury. The screening process affects the randomness of the sample, thus generating a disconnect between the community and the jury.

New polls kept coming as the trial proceeded. On May 7, Stormy Daniels took the stand to tell the jurors about the sexual encounter she had with Trump. On May 9, Trump’s lawyers grilled Daniels for making up a false story to mint money out of the scandal. The public heard the evidence, though not directly and not in as much fine detail as did the jurors and alternates. On May 10, polls showed that Biden and Trump were evenly split at 43% each. If jurors were processing the evidence like the larger community they represented, six jurors and three alternates would probably remain pro-Trump after Daniels’ testimony.

From May 13 to May 20, former Trump lawyer Michael Cohen testified that Trump instructed him to “take care” of the payments to hush Daniels and that Trump approved a payment of $130,000. The defense team painted Cohen as a pathological and prolific liar. On May 20, the prosecution rested its case. A day later, the defense team rested its case. The May 13-23 polls showed that Trump was in the lead over Biden. People have not changed their minds since hearing Cohen. Could we assume that the pro-Trump jurors also have not changed their minds?

One could argue that the nationwide polls do not represent the views of New York citizens or those of the Trump jurors from New York. On April 17, New York polls showed that Biden led Trump by 10% with a 47%:37% split. If the New York polls were more representative of the jurors, four or five jurors would probably be pro-Trump. On May 29, at the conclusion of the trial, New York polls showed Biden leading Trump by 10% again, thus keeping the partisan split among voters after the evidence of Daniels and Cohen. Could we assume that the pro-Trump jurors did not change their minds either?

Suppose some jurors were unaffiliated, as many polls did not add up to 100%. Even in that case, no more than two jurors out of twelve would be unaffiliated. However, only one dissenting juror was needed to cause a mistrial, known as a hung jury, since the New York laws require a unanimous jury verdict to convict.

That there was not a single pro-Trump juror who held on to their bias is questionable under probability theory, which relying on the persistent split polls predicted a hung jury. Note, however, that probability is not certainty. What is most probable may not occur, and what is least probable might, like a favorite horse losing the Belmont race. Unlikely outcomes are compatible with probability theory. However, legal formalism explains the Trump conviction much better as it contends that after hearing the evidence, all jurors, including the pro-Trump jurors, if any, decided to go for the guilty verdict.

Legal Formalism

Under legal formalism, evidence prevails over biases.

Legal formalism is the mainstream model that lawyers and judges trust as the most reliable method to prosecute a crime before a jury. Under this model, the prosecutors and defense team select twelve jurors from a panel of citizens, dropping those they find unreliable, biased, or otherwise unfit to be jurors. This screening is critical for finding the twelve people who would hear the evidence and apply the facts to the law. The judge instructs the jury to convict only if there is evidence beyond reasonable doubt that the defendant has committed the crimes charged.

Under the formalistic model, the defense team cross-examines and creates doubts about the testimony of witnesses like Stormy Daniels and Michael Cohen. The state makes its case, proving each element of the crime with credible evidence. Near the end of the trial, prosecutors and defense summarize the evidence and the legal arguments to persuade the jury of their conflicting positions.

The judge instructs the jurors to be neutral, focus only on the evidence they hear in the court, and give a verdict accordingly without bias of race, religion, or any other factor. After hearing the evidence, jurors spend hours discussing what they heard and deciding whether the defendant committed the crimes beyond a reasonable doubt.

If a juror disagrees with the other eleven jurors, they are free to dissent and cannot be pressured to join a unanimous verdict. Thus, a single pro-Trump juror could have generated a hung jury; the more, the merrier.

Since the New York jury convicted Trump on all thirty-four counts, there is a formalistic presumption that evidence was beyond reasonable doubt to prove to the jurors that Trump wished to influence the 2016 presidential election by paying Daniels to stay quiet. If there were any pro-Trump jurors, they valued evidence over their bias for Trump as a presidential candidate.

The New York criminal justice system worked as it should have. However, for legal realists, the formalistic model of evidence and jury neutrality is a fairy tale that reads fantastic in law books. For them, the social and political prejudice and partisanship in the air in which a case is tried is polluted, unpredictable, and toxic, particularly for famous defendants like Trump.

Legal Realism

Under legal realism, bias prevails over evidence.

Legal realism, a quintessentially American invention, doubts the assumptions of legal formalism. Legal realism builds on the epistemological nihilism that truth is unknowable. Legal realism, which gave birth to deconstructionism and critical race theory, has little faith in objectivity and argues that every human is a product of prejudices and preferences. They selectively pick evidence to confirm their preexisting mental states.

So, witnesses like Stormy Daniels and Michael Cohen tell lies that survive cross-examination; they reformulate their memories and begin to believe in the fabrications their minds invent. Human memories are plastic, not words written in stone. Furthermore, jury neutrality is a fable that common law invented centuries ago. Very few legal systems worldwide use juries to prosecute crimes for various reasons, including some derived from legal realism.

From day one, Trump’s criticism of the trial was anchored in legal realism. Trump highlighted the fact that the New York prosecutor, Alvin Bragg, is a Democrat, implying that a Democrat prosecutor cannot be fair to a Republican presidential candidate, a classical legal realism argument that prosecutors are not neutral when dealing with a defendant of the rival political party. “Make no mistake about it, I’m here because of crooked Joe Biden,” Trump claimed on May 28, seeing politics in prosecution.

Trump also criticized Judge Juan Merchan, who conducted the trial, for his court rulings and extra-legal facts presumptively influencing decision-making. Trump mentioned that Merchan’s $15 contribution to the Biden 2020 presidential campaign showed the judge’s political bias. Trump also highlighted that Merchan’s daughter makes money from a website supporting Trump’s political opponents. The trial provides fuel for the daughter’s website ideas.

Again, the legal realism argument is that the judge in the Trump trial is a biased political operative. Republican Senator JD Vance wrote a letter to Attorney General Merrick Garland to open a criminal investigation of Judge Merchan for his anti-Trump bias. Legal formalism argues that a judge who gives money to Biden can be neutral in the Trump trial. (Ironically, some noted Republicans propose to ban teaching critical legal theories in law schools while they employ the same theory to trash the Trump trial).

To legal realists, the broader critique that the Trump trial is too close to the 2024 presidential elections smells bad faith legalism. Trump’s sexual relationship with Stormy Daniels occurred in 2006, and the hush money was disbursed before the 2016 presidential elections. Why is an eight-year-old case built on an 18-year-old sex story being tried in 2024? This prosecution timeline fits more with legal realism than a desire to enforce the law.

Legal realism faces difficulty in explaining how twelve jurors agreed on Trump’s conviction. Legal realism must advocate the following thesis. The jurors were anti-Trump and just listened to the witnesses to confirm their animus against Trump, thus rejecting the formalistic postulate that jurors were neutral and decided based on evidence. But how could the defense team not see the anti-Trump jurors? Legal realism would argue that the jurors hid their prejudice and outsmarted the defense team. An unlucky defendant with no sympathetic juror is part of legal realism. Worse, legal realism may cogitate whether the pro-Trump jurors were intimidated or bribed to endorse the conviction verdict. Formalistically, it is much easier to reject legal realism and the probability model, arguing that jurors heard the evidence from both sides and believed the prosecution, not the defense.

Conclusion

The probability model predicts the partisan distribution of the Trump jury, reflecting the broader community, without concluding how each juror will vote. Yet the notion that there was not a single pro-Trump juror is baffling. There would have been a hung jury if jurors reflected the sentiments of the partisan community expressed in the polls. According to legal formalism, the partisan distribution of the jury is irrelevant to the final verdict since the only deciding factor for each juror is the evidence presented in the trial. Thus, legal formalism is bias-independent and places the conviction probability solely on the evidence presented in the trial. Therefore, a pro-Trump jury could convict Trump based on the evidence. By contrast, legal realism is bias-dependent; it respects probability theory and predicts a conflicted jury since trial evidence should confirm jurors’ preexisting biases for or against Trump. If the jury is partisan, no conviction or acquittal will occur, resulting in a hung jury. Since the jury convicted Trump, legal realism would conclude that the jury was anti-Trump from the get-go or worse, that pro-Trump jurors faced strongarming. The believers in legal formalism are content that the jury honored the evidence regardless of their biases and that the prosecutors presented evidence beyond reasonable doubt that Trump had committed thirty-four felony counts. I wonder whether a jury in a red state, like Kansas, would have reached a similar verdict.