Thursday, May 19, 2005

Should Courts Use Numbers to Describe Standards of Persuasion in Legal Proceedings?

This judge is Judge Jack B. Weinstein.

The opinion is United States v. Copeland, 2005 WL 1109441 (E.D. N.Y., May 4, 2005).

Defendant was indicted for illegally re-entering the United States. Defendant had been previously deported. The issue was whether the indictment was valid or should be dismissed.

Defendant had been denied due process in the deportation proceedings that resulted in his deportation. The Second Circuit instructed the trial judge (Weinstein) that the indictment against defendant for defendant’s subsequent illegal (re)entry into the United States should be dismissed only if the trial court found that defendant was prejudiced by the due process defect in the deportation proceedings.

Judge Weinstein held that the Second Circuit’s prejudice requirement requires a determination of the probability that the result in the deportation proceedings would have been different had there been no denial of due process in the deportation proceedings that actually resulted in deportation. Judge Weinstein held that defendant did not have to show that defendant was required to show by a preponderance of the evidence that the result would have been “no deportation order” in the absence of the due process error; defendant’s burden was less than that.

Judge Weinstein summarized the situation this way:

The rule of law adopted by the United States Court of Appeals for the Second Circuit requires the district court to undertake the highly speculative task of determining whether, absent the constitutional violation, the immigration judge would have ordered the defendant deported. In effect, this enterprise necessitates a present reconstruction of a hypothetical deportation hearing in 1996. It requires the court to accurately predict, in the deportation context, what would have happened at another time, in another place, with different lawyers, and before another unknown judge, using variable and subjective factors.

An indictment charging Richard Copeland with illegal reentry into the United States was dismissed by this court based on a finding that the underlying prior ordered deportation was not consistent with due process. See United States v. Copeland, 228 F.Supp.2d 267 (E.D.N.Y.2002). Though the United States Court of Appeals for the Second Circuit was in agreement that due process had been denied, it nevertheless vacated dismissal of the indictment. See United States v. Copeland, 376 F.3d 61, 75 (2d Cir.2004). Concluding that the law required both denial of due process and a showing of prejudice before the dismissal of an illegal re-entry indictment, the Court of Appeals remanded the case to the district court for an evidentiary hearing and findings on the question of whether the fundamental procedural error in the defendant's deportation hearing had been prejudicial.

As I have already noted, Judge Weinstein concluded that defendant did not have to show that it was more probable than not that he would not have been deported. But if defendant was not required to show by a preponderance of the evidence or to show that it was more probable than not that the result of the deportation proceedings would have been different in the absence of the due process defect, how much less of a burden did defendant have? And how is that burden to be described?

Judge Weinstein’s answer was nothing if not forthright. Judge Weinstein held defendant had to show that there was a probability of at least .2 (20%) that his deportation would not have been ordered (by another immigration judge?) if there had been no denial of due process.

The result under this liberal standard was that defendant still lost. Judge Weinstein wrote:

The [trial] court [i.e., Judge Weinstein] conducted a full evidentiary hearing. It finds that the defendant was not "prejudiced." The decision is based on a determination, by a standard of clear, unequivocal and convincing evidence--80% or more probability--that had the section 212(c) hearing and administrative appeal before deportation been properly conducted by an immigration judge, the defendant would have been ordered deported. Accordingly, the indictment is reinstated.
And how did Judge Weinstein explain his choice to use digits (80%; 20%) to describe the burden of persuasion on the issue of “prejudice”?

Judge Weinstein wrote:

Prejudice is shown where "defects in the deportation proceedings may well have resulted in a deportation that would not otherwise have occurred." Id. …

The [Second Circuit] Court of Appeals did not say what degree of "likelihood," i.e. what probability, must be established. …

It concluded that prejudice is shown where there is a "reasonable probability" that the deportation at issue would not have been ordered absent the error complained of.

2. Quantifying Prejudice

Typically, courts have not quantified burdens of proof. See J. MAGUIRE, J. CHADBOURN, J. MANSFIELD, ET AL., CASES AND MATERIALS ON EVIDENCE 871-73 (6th ed.1973) (collecting literature on quantification of burdens of proof); RICHARD H. GASKINS, BURDENS OF PROOF IN MODERN DISCOURSE 20 (1992) (discussing implications of burdens of proof); TERENCE ANDERSON & WILLIAM TWINING, ANALYSIS OF EVIDENCE 338 (1991) (correlating verbal and mathematical measures of certainty and doubt).

The term "reasonable probability" should be quantified to the extent possible, given the difficulty of assessing what another adjudicator would have done when applying complex and subjective criteria. One of the most astute observers of problems of proof has emphasized the importance of attempting to set standards of probability in deciding whether fact-finding burdens have been met. He wrote:

The possible frailty of the fact-finding process in adjudication is an important and complex problem. It threatens to corrode public confidence in a vital part of the legal system; it raises significant moral, ethical, and legal issues ...; it raises the possibility that legal rules cannot be efficiently or effectively administered and implemented; and it raises broader questions about the fairness, morality, acceptability, and efficiency of matters such as the legislative process and administrative rule-making proceedings. The problem of evidence and uncertainty is not simply a "technical" or "abstract" one. However one proposes to deal with the problem of uncertain inference--whether by "abstract" models, by "common sense" reasoning, by dismissing it as unmanageable by rational analysis, or in some other way--what one thinks about probability and inference in [adjudication] profoundly affects what one thinks about the "rule of law."
Peter Tillers, Introduction to the Boston University School of Law Symposium on Probability and Inference in the Law of Evidence, 66 B.U. L. REV. 381, 381-82 (1986). Agreement on quantification, while not a solution, does help move beyond the mere anecdotal to at least a rough consensus in application.

Quantification requires analysis in terms of probability. Cf. Anne W. Martin & David A. Schum, Quantifying Burdens of Proof: A Likelihood Ratio Approach, 27 JURIMETRICS J. 383 (1987); Peter Tillers & David A. Schum, Charting New Territory in Judicial Proof: Beyond Wigmore, 9 CARDOZO L. REV. 907, 910 (1988) ("The primary motivation for the use of most mathematical models of proof is to facilitate consistent thinking about very complex problems.").

While "reasonable probability," the term of art selected by the Court of Appeals, seems deliberately designed to be fuzzy in concept and articulation, it is suggested that a probability of 20%--the approximate inverse of "clear, unequivocal and convincing evidence"--represents a sensible and enforceable standard, considering that deportation often has such serious consequences for the deportee and his or her family.

It has been said that,

[t]ime is irreversible, events unique, and any reconstruction of the past at best an approximation. As a result of this lack of certainty about what happened, it is inescapable that the trier's conclusions be based on probabilities.
J. MAGUIRE, J. CHADBOURN, J. MANSFIELD, ET AL., CASES AND MATERIALS ON EVIDENCE 1 (6th ed.1973). See also United States v. Fatico, 458 F.Supp. 388 (E.D.N.Y.1978). In Fatico, the court noted: "Quantified, the preponderance standard would be 50% probable.... [T]he probabilities might be in the order of above 70% under a clear and convincing evidence burden.... In terms of percentages, the probabilities for clear, unequivocal and convincing evidence might be in the order of 80% under this standard." 458 F.Supp. at 405. Id. at 411 (concluding that "clear, unequivocal and convincing evidence" means a probability of "about 80%"). Cf. United States v. Shonubi, 895 F.Supp. 460, 514 (E.D.N.Y.1995) ("[Non-statistical] evidence offers nothing more than a basis for conclusions about a perceived balance of probabilities."), rev'd, 103 F.3d 1085, 1092 (2d Cir.1997) ("Though [the district court's] comprehensive opinion is a valuable addition to the legal literature on the subject of evidence in particular and judicial decision-making in general, we conclude that he relied on evidence beyond the category of 'specific evidence' that our prior opinion ruled was required for determination of a "relevant conduct" drug quantity for purposes of imposing a criminal sentence ."); Peter Tillers, Introduction: Three Contributions to Three Important Problems in Evidence Scholarship, 18 CARDOZO L. REV. 1875, 1884 (1997) ("One possible way to make sense out of [the opinion of the Court of Appeals in Shonubi ] is to view it as a condemnation of statistical evidence in general.... But there are ... problems with the view that [the opinion] is, at bottom, a repudiation of statistical evidence and statistical methods. [For instance, the Court of Appeals for the Second Circuit] seems to accept the use of statistics and statistical methods for [related] purposes ....").

When, as here, a relatively "simple fact"--what happened in the real world of defendant's life--is combined with what an unknown administrative judge would have done in evaluating the evidence supporting that finding of "fact," and analyzing the "fact" in the context of a "legal rule," the problem of determining how the judge would have decided the "law-fact" issue is complex. It is compounded by many factors--among them the egocentricity of the judge. At most a band of probabilities is all that we can expect. Since the defendant's constitutional rights have been violated he is entitled, it is submitted, to the most favorable band border--here, it is proposed, 20%. An attempt to quantify in order to provide some uniformity in application of the rule is justified even though it must be conceded that the percentage chosen is based on public policy favoring enforcement of constitutional rights and somewhat arbitrary.


Do you think Judge Weinstein’s ruling will stand? Or do you think the Second Circuit Court of Appeals will again slap down Judge Weinstein, perhaps on the ground that Judge Weinstein is again improperly trying to expressly quantify, or grade numerically, things that cannot be or ought not be quantified, measured, or graded numerically?

Even though I see force in the argument that standards of persuasion such as the reasonable doubt standard should be couched in numerical terms, cf. Questions about Winship, I routinely tell my students there is not a snowball’s chance in %$^& that this will happen. But seeing a judge who has the great courage of his very determined convictions makes me wonder if I am being too timid. Perhaps the sort of “mathematical” approach that J. Weinstein favors will someday prevail, at least in some sectors of the law.

Monday, May 16, 2005

More about Biography of H.L.A. Hart

Update: As Nicola Lacey's biography of H.L.A. Hart progresses, there is increasing detail about the intellectual currents at Oxford and elsewhere. This largely answers one of the concerns I voiced earlier. See my post dated May 13, 2005.

Edinburgh Seminars on Law, Probability, and Risk

Professor Colin Aitken and the University of Edinburgh are hosting additional seminars on law, probability, and risk. See

One seminar is slated for October 14-15, 2005. The general topic topic is criminology and the organizer is Sarah Armstrong.

A second seminar is scheduled for December 2-3, 2005. The general topic is evidence evaluation and the organizer is Colin Aitken.

The seminars are free but participation is by invitation. For details about the seminars get in touch with either Sarah Armstrong or Colin Aitken at:

Sarah Armstrong: sarmstro at

C.G.G. Aitken
School of Mathematics
The King's Building
The University of Edinburgh, EH9 3JZ

E-mail: cgga at
Phone: (0)131 650 4877
Fax: (0)131 650 6553