When Are Future Predictions Clearly Erroneous? Matter of A-A-R-
In the early 1970s, the proposal for the film Star Wars was rejected by multiple Hollywood studios (including Disney), on the grounds that the concept was “too out there” to appeal to mainstream audiences. Eventually, Twentieth Century Fox decided to take a chance on the film. Based on the prevailing view of the project at the time, would Twentieth Century Fox’s decision have been deemed “clearly erroneous?”
In hindsight, I don’t believe that too many people expected the American Cardinal Robert Prevost to be selected as Pope. And according to a CBC Poll Tracker, Canada’s Liberal Party had a one percent chance of winning the country’s federal elections a little more than two months prior to the election. And we can all think of our most memorable sports upset. Would it have been “clearly erroneous” to predict in advance that Cardinal Prevost would be chosen to be the first American Pope, that Mark Carney would be the current Prime Minister of Canada, or that, say, the U.S.A. Hockey Team would upset the team from the Soviet Union in the 1980 Olympics?
These examples demonstrate the problem with judging the reasonableness of predictions that have yet to come to fruition. How should they be assessed?
Immigration Judges make regular predictions about the risk of persecution or torture to those facing deportation. The Board of Immigration Appeals (“BIA”) has to review these predictions speculatively, while the at-risk individuals are still in the U.S. But the Immigration Judge’s predictions don’t have to be found correct on appeal; they must be allowed to stand as long as they are not “clearly erroneous.” The meaning of that term will be explained in much detail below.
With all of the above in mind, in September 2024, an Immigration Judge found that if deported to El Salvador, a reformed former gang member, who all agreed would be detained upon arrival under the country’s state of exception and jailed in President Bukele’s infamously horrible prison system, would likely face torture. Was that decision clearly erroneous?
In a very recent precedent decision, Matter of A-A-R-, the BIA said yes. To understand why the Board was wrong, let’s review what clear error means in the context of appellate review.
We will begin by looking at the history of the Board’s clear error standard.
The 2002 Regulations
In 2002, during the George W. Bush administration, the Department of Justice, headed by then Attorney General John Ashcroft, published regulations barring the BIA from engaging in independent fact finding, and limiting the Board to “clear error” review of fact finding by Immigration Judges on appeal. In publishing the final regs, the Department stated that “the final rule continues to focus on the primacy of immigration judges as factfinders and determiners of the cases before them. The role of the Board is to identify clear errors of fact or errors of law in decisions under review, to provide guidance and direction to the immigration judges, and to issue precedential interpretations as an appellate body, not to serve as a second-tier trier of fact.”1
The preamble to the rule further explained:
A finding is “clearly erroneous” when, although there is evidence to support it, the reviewing Board member or panel is left with the definite and firm conviction that a mistake has been committed. A factfinding may not be overturned simply because the Board would have weighed the evidence differently or decided the facts differently had it been the factfinder. Anderson v. City of Bessemer, 470 U.S. 564, 573 (1985).2
There was more. The agency continued by explaining that:
The “clearly erroneous” standard reflects the major role of immigration judges under the Act and implementing regulations as determiners of facts. In removal proceedings, it is the immigration judges, not the Board, who have been given authority to “administer oaths, receive evidence, and interrogate, examine, and cross-examine the alien and any witnesses.” 8 U.S.C. 1229a(b)(1).3
And even more:
The rationale for changing to a “clearly erroneous” standard of review of fact findings is not limited to the consideration that immigration judges may be better positioned than the Board to decide factual issues, including issues of credibility. See generally Anderson, 470 U.S. at 574-75. As the Supreme Court has opined in another setting, the “clearly erroneous” standard rather than a de novo standard of review is appropriate for factfindings by trial courts because “[d]uplication of the trial judge's efforts [by an appellate body] would very likely contribute only negligibly to the accuracy of fact determination at a huge cost in diversion of judicial resources.” Id. “[T]he parties to a case on appeal have already been forced to concentrate their energies and resources on persuading the trial judge that their account of the facts is the correct one” and “requiring them to persuade three more judges at the appellate level is requiring too much.” Id. at 575. The “clearly erroneous” standard of review recognizes that an evidentiary hearing on the merits should be the “ ‘main event’ * * * rather than a ‘tryout on the road.’ ” Wainwright v. Sykes, 433 U.S. 72, 90 (1977).
Notice that the agency did not limit its view of the primacy of Immigration Judges to credibility findings, but rather, extended it to all factual issues, “including issues of credibility.” This is consistent with the language of the Supreme Court that the agency cited to in Anderson v. Bessemer City, which states in part that “[t]he rationale for deference to the original finder of fact is not limited to the superiority of the trial judge's position to make determinations of credibility. The trial judge's major role is the determination of fact, and with experience in fulfilling that role comes expertise.”4
A few years later, the U.S. Court of Appeals for the Second Circuit took a similar view regarding determinations of whether conditions in a particular country have changed so fundamentally that one who suffered persecution in the past would no longer have a well-founded fear of future harm. The court held that deference was due to the Immigration Judge regarding such findings, stating “We… assume that an IJ has a certain advantageous and expert working knowledge — this includes the salient historical events and conditions of countries that are the subject of an appreciable proportion of asylum claims.”5 The court added: “We cannot assume that an IJ comes to each case ignorant of the history, political status, and evolving conditions in countries from which many petitioners seek asylum.”6
So let’s recap: more than 20 years ago, a very conservative Attorney General under a Republican administration severely limited the BIA’s ability to reverse findings of fact made by Immigration Judges, and supported that change with very strong language about the primacy of Immigration Judges, finding much support from Supreme Court decisions. And circuit courts have further recognized that Immigration Judges accumulate considerable country condition expertise through the nature of their work.
So when is it appropriate for the BIA to intervene and reverse this type of finding by an Immigration Judge?
The Standard: What is Clear Error? The “Dead Fish” Test
The U.S. Court of Appeals for the 11th Circuit has explained the standard in particularly vivid terms. Saying that the legal error must be “beyond the scope of reasonable debate,” and further noting that “this is a high bar,” the court provided this analogy: "To be clearly erroneous, a decision must strike us as more than just maybe or probably wrong; it must ... strike us as wrong with the force of a five-week-old, unrefrigerated dead fish.7
Although not nearly as memorable, the Supreme Court (in addition to the language quoted above) has also provided this helpful guidance: “Where there are two permissible views of the evidence, the factfinder’s choice between them cannot be clearly erroneous.”8 So when is a prediction impermissible?
The Second Circuit provided examples of what clear error might look like in the immigration context. The court offered two hypotheticals which it stated were unlikely to arise: where no evidence at all supports a finding of fact, or where indisputable evidence shows that a claimant was at a location far from where he claimed to have been beaten. The court continued:
A more likely example might arise where an IJ has obviously misunderstood the testimony of a witness and based a finding of fact on that misunderstanding. Situations might also arise where the evidence opposed to the claimant's version, though not indisputable, has overwhelming persuasive force. How overwhelming the opposing evidence must be will often be a close question for the entity applying clear error review and for the court reviewing a clear error conclusion. What is not in doubt, however, is that the phrase “clear error” is to be taken literally: the error must be clear.9
So to summarize, clear error is not a difference of opinion, and not even “we think the judge was wrong;” or even “the judge was probably wrong.” Rather, it’s reserved for situations where the judge screwed up royally in a way that is clear to all, and reached a conclusion that is implausible based on the record.
Matter of A-A-R-: The BIA Ignored the “Clear Error” Standard
Matter of A-A-R- involved a tattooed former MS-13 gang member from El Salvador who was granted protection under the Convention Against Torture by an Immigration Judge.10 The judge found that the respondent would be detained upon arrival in El Salvador under that country’s state of exception, and relied on a written statement by a country expert, as well as reports from organizations including Amnesty International and Human Rights Watch, to conclude that the respondent would more likely than not be tortured, as that term is defined by regulation and case law.
ICE appealed. As explained above, the BIA was barred from engaging in its own factfinding, and under its narrow clear error scope of review, it was limited to determining if the Immigration Judge’s prediction of the possibility of future harm was so obviously, overwhelmingly wrong as to hit the Board with the force of a five week old dead, unrefrigerated fish.
Instead, the Board did exactly what the department said it should not - it relitigated the issue, treating the IJ’s analysis as if it was not the “main event,” but rather merely a” tryout on the road.”
The Third Circuit’s Prior Statement of the “Clear Error” Standard
It bears noting that A-A-R- arose within the jurisdiction of the U.S. Court of Appeals for the Third Circuit. And not long before, in 2022, that court had rejected a decision of the BIA arising in similar circumstances in which it reversed a factual finding of an Immigration Judge that a Guatemalan woman would likely be subjected to torture if removed to her country. The Immigration Judge in that case had relied on country condition evidence, and “noted that while the Guatemalan government has passed a law to combat violence against women, the law is not fully prosecuted, and did not preclude the finding that the government would still acquiesce to torture.”11
On appeal in that case, “The BIA explained that, while the country conditions report evidenced violence against women in Guatemala generally, it was not persuaded that Arreaga-Bravo faced a particularized risk of harm. As such, the BIA held it was “unable to uphold” the IJ's decision granting the application for protection under CAT.”12
The Third Circuit found that the BIA had not applied the “clearly erroneous” standard. According to the court:
It was not the BIA's role to determine whether it agreed with the IJ's weighing of the evidence in Arreaga-Bravo's favor. Its role was to point to findings by the IJ that were “an obvious, plain, gross, significant, or manifest error or miscalculation.”...Rather than do this, the BIA substituted its view of the evidence for that of the IJ, rather than reviewing for clear error, and concluded that it was “not sufficiently persuade[d]” with the IJ's conclusion… But the clear error “standard plainly does not entitle a reviewing court to reverse the finding of the trier of fact simply because it is convinced that it would have decided the case differently.”13
Later in the decision, the court found that the BIA had once again applied the wrong standard of review in concluding that “the overall evidence is insufficient to show that the police would specifically fail to act, or that their inability to provide assistance would constitute ‘consent or acquiescence’ under the regulations.’ ”14
The Third Circuit again called the Board out, stating “the BIA seemed to review de novo the IJ's factual findings as to how the government is likely to respond to Arreaga-Bravo's harm. This was error. The BIA stepped out of the bounds of its permissible role.”15
In A-A-R-, The BIA Repeats Its Prior Errors
And yet, in spite of this recent guidance from the Third Circuit, the BIA in A-A-R- committed the exact same error, again reviewing the factual findings de novo and reversing solely because it reached a different conclusion than the Immigration Judge, which is the exact definition of what clear error can never be.
First, note the language used by the Board in A-A-R-: “Although there is no clear error in the Immigration Judge’s finding that the applicant will likely be identified as a former gang member and detained in El Salvador, we disagree with the Immigration Judge that the applicant has satisfied his burden of proving that it is more likely than not that he will be tortured in detention…” (emphasis added). For the reasons explained above,“we disagree” is absolutely the wrong standard for clear error.
Nowhere in its decision does the Board claim that the judge’s factual finding was based on no evidence at all. Nor does it point to either indisputable or powerfully persuasive evidence of the impossibility of the factual finding. Not does it claim that the judge misunderstood facts of record and relied on that misunderstanding in calculating the risk of torture.
What the Board did claim was that the Immigration Judge relied on mere “anecdotal reports of some instances of people having been severely harmed or killed in detention in El Salvador.” First of all, these anecdotal reports were contained in reports of the U.S. Department of State, and of other highly respected non-governmental human rights organizations, the type of reports that are often relied on by adjudicators in calculating the risk of harm.
Secondly, the Immigration Judge also relied on the written report of a witness deemed to be a country conditions expert. The expert had access to the same country reports that the BIA mentioned. And although I haven’t read the statement myself, I have it on good authority that the expert concluded that the respondent would likely experience torture in prison under the state of exception, relying on country reports and the expert’s own field research. Somehow, the Board omitted the expert’s conclusion from its published decision.
Instead, the Board reinterpreted reports of the U.S. Department of State, Amnesty International, and Human Rights Watch. The Board also added “Although the country conditions evidence reflects that hundreds of prisoners have died, that a substantial number showed signs of physical violence, and that there have been reports of severe physical violence committed by prison officials, the applicant must do more than show that some individuals in detention suffer harm rising to the level of torture.” (Emphasis added).
The BIA’s approach constituted impermissible factfinding. Furthermore, in stating that in light of over 70,000 being imprisoned under the state of exception, evidence “that hundreds of prisoners have died, that a substantial number showed signs of physical violence, and that there have been reports of severe physical violence committed by prison officials” isn’t enough to support the Immigration Judge’s finding, it was unclear what the Board would have found to be enough. Would a respondent have to show more than 35,000 cases of torture to satisfy the “more likely than not” requirement in the Board’s view?
If we apply the Board’s apparent approach to the examples at the beginning of this post, how many prior futuristic space sagas would have had to be blockbusters for Star Wars to pass the BIA’s test? How many American popes would have had to be previously selected for the Board to have allowed a correct prediction of Pope Leo XIV to be permissible?
In addition to brushing aside the opinion of the country expert, the Board also dismissed the Immigration Judge’s reliance on the statements of government officials in El Salvador who claimed that the gang members deserved the terrible prison conditions to which they were being subjected. The Board reduced these public pronouncements as merely reflecting “the public officials’ moral judgment about gang members” from which no inference of intent could be drawn.
An Aside: A Variation to Be Aware Of
Since we’re discussing the BIA and clear error, I thought it worth mentioning another way the BIA has sought to circumvent its limited scope of review. In Francois v. Garland,16 an Immigration Judge granted withholding of removal to Haiti to a respondent from that country who suffers from severe mental illness. When ICE appealed the grant, the BIA (which clearly disagreed with the decision) found no clear error in the Immigration Judge’s prediction of persecution. The Board therefore should have dismissed the appeal, allowing the grant of withholding to stand.
Instead, the BIA claimed that further factfinding was required by the judge below, and returned the record to the judge with instructions to make those findings. The Board’s action was clearly a signal to the Immigration Judge that it sought a different result, and the Immigration Judge obliged, denying relief on remand.
The case eventually made its way to the U.S. Court of Appeals for the Fifth Circuit. In a decision issued just this past October, the court criticized the BIA’s actions, noting that the Immigration Judge “had already made these exact findings” that the BIA requested in its remand order. According to the Fifth Circuit, the first Immigration Judge’s decision “already answered every question the BIA remanded for the IJ to address,” and that “the BIA shirked its duty by ignoring the IJ's relevant findings, in violation of 8 C.F.R. § 1003.1(d)(3)(i) and (iv).”17 Those regulatory provisions limit the BIA to reviewing factual findings for clear error only, and prohibit the Board from engaging in factfinding in the course of deciding cases.
So in summary, the BIA demonstrated in Francois that it does understand the “clear error” standard (as it found no clear error in the decision below in that case), but then tried to accomplish an end-around to achieve a reversal and denial anyway, in clear circumvention of the binding regulations, which was stopped by the Fifth Circuit.
Conclusion
In summary, the Board in Matter of A-A-R- impermissibly reinterpreted the evidence in a very selective way in order to reach a different conclusion than that reached by the trier of fact. As the Fourth Circuit recently held in reversing the Board, “In conducting clear error review, the BIA may not reweigh evidence or ‘substitute[ ] its own judgment for that of the IJ.’”18
For all of the above reasons, a prediction that A-A-R- will not withstand circuit court scrutiny would not be clearly erroneous.
Copyright 2025 Jeffrey S. Chase. All rights reserved.
Notes:
67 FR 54878, 5480 (8/26/2002).
Id. at 5489.
Id.
Anderson v. Bessemer City, 470 U.S. 564, 574-75 (1985).
Hoxhallari v. Gonzales, 468 F.3d 179, 186 (2d Cir. 2006).
Id. at 187.
Cox Enters., Inc. v. News-Journal Corp., 794 F.3d 1259, 1272 n.92 (11th Cir. 2015) (quoting Parts & Elec. Motors, Inc. v. Sterling Elec. Inc., 866 F.2d 228, 233 (7th Cir. 1988)).
Anderson v. Bessemer City, supra at 573-74.
Wu Lin v. Lynch, 813 F.3d 122, 127 (2d Cir. 2016).
Matter of A-A-R-, 29 I&N Dec. 38 (BIA 2025).
Arreaga Bravo v. Att’y Gen. of U.S., 27 F.4th 182, 185 (3d Cir. 2022).
Id.
Id. at 187 (quoting Mendoza-Ordonez v. Att'y Gen., 869 F.3d 164, 169 (3d Cir. 2017)).
Id. at 188.
Id. at 189.
120 F.4th 459 (5th Cir. 2024).
Id. at 465-66.
Funez-Ortiz v. McHenry, 127 F.4th 498, 505 (4th Cir. 2025) (quoting Crespin-Valladares v. Holder, 632 F.3d 117, 127–28 (4th Cir. 2011)).