Davis v. United States, 564 A.2d 31

Davis v. United States

564 A.2d 31

D.C.

1989

Check Treatment

NEWMAN, Associate Judge:

Milton Davis was convicted of carrying a pistol without a license, D.C.Code § 22-3204 (1981). He appealed, contending that the trial court had violated his fifth and sixth amendment rights to due process and to present witnesses by permitting a defense witness to refuse to testify on the basis of a claimed fifth amendment privilege against self-incrimination. A division of this court agreed with Davis that the trial court had committed constitutional error by failing to conduct the inquiry required to sustain a claim of the privilege. Davis v. United States, 482 A.2d 783 (D.C.1984). Because the nature of the trial court’s error rendered the record inadequate for the appellate panel to determine whether that error was harmless under the standard enunciated in Chapman v. California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967), the panel remanded for further proceedings and directed the trial court to determine, on the basis of the supplemented record, whether the error was harmless beyond a reasonable doubt. The trial court subsequently concluded that the error was harmless.

We review Davis’ conviction for a third time, this time en banc, to reconsider what role the trial court’s assessment of the harmlessness of its own constitutional error should play in our resolution of that question. 1 Stated another way, we must decide what standard of review we should apply in those unusual cases where the trial court has had an opportunity to consider, in the first instance, whether an error of constitutional magnitude may be excused as harmless under the Chapman test.

After considering the briefs and arguments of counsel, we are not satisfied that either the “clearly erroneous” standard, ordinarily applied to review a trial court’s factual findings, or the “de novo” standard, typically associated with the review of a question of law, adequately serves the interests — either analytical or functional— which underlie a determination of harmlessness. We reject the “clearly erroneous” standard because we believe, as a matter of analysis, that the appellate court must have a meaningful last word as to *34 whether or not to sustain a conviction that follows a trial marred by constitutional error. At the same time, we reject a pure “de novo” standard because we believe that the trial court sometimes in its ruling may make findings of historical or subsidiary facts (as well as reasonable inferences deduced therefrom) to which we should give deference. Accordingly, we choose an intermediate standard, a standard that we are persuaded is more finely calibrated to address the concerns presented by this case. In light of the standard we choose, we conclude that there is no longer any reason to elicit from the trial judge, as we did in this case, an opinion on the ultimate issue of harmlessness. That court’s proper function will have been served when it makes the necessary evidentiary record and returns the matter to us.

In Part I we derive and define our intermediate standard, applying general principles governing standards of review to the unique characteristics of an assessment of harmlessness. In Part II, we consider the evidence presented at Davis’ trial and at the proceedings on remand to determine whether the constitutional error committed by the trial court in this case was harmless beyond a reasonable doubt. Employing the standard of review developed in Part I, we conclude that the constitutional error was indeed harmless. Accordingly, we affirm Davis’ conviction for carrying a pistol without a license.

We are presented in this case with the novel problem of deciding what standard of review to apply to a trial court’s determination that its own constitutional error is harmless beyond a reasonable doubt. The question is one that arises infrequently; ordinarily, the determination of harmlessness rests with the appellate court in the first instance. Occasionally, however, the nature of the error and the inadequacy of the record on appeal prevent us, in the same appeal in which we make the decision that an error has occurred, from addressing the possible harmlessness of that error under Chapman, supra, 386 U.S. at 18, 87 S.Ct. at 824. When, for example, as in this case, the constitutional error resulted in the erroneous exclusion of evidence at trial, and that evidence is not- available to us by proffer or otherwise, we must remand to the trial court to supplement the record. The standard of review question arises because, in remanding this case, we also directed the trial court to evaluate the error for harmlessness in light of the supplemented record.

In formulating an appropriate standard of review, we seek a solution that will mediate between two competing concerns inherent in the problem posed by our remand order. On the one hand, we wish to avail ourselves of the unique operational advantage of the trial judge in making a determination requiring intimate acquaintance with the facts of the particular case as they evolved at trial. On the other hand, we seek to maintain our own role as primary expositor of law by applying a sufficiently penetrating measure of review to a trial court decision that, in effect, construes a legal right by denying its remedy.

A division of this court recently grappled with the problem of determining the appropriate standard of review, of a trial court’s ruling that the doctrine of collateral estop-pel prevented the government from relit-igating certain facts • at a second trial. United States v. Felder, 548 A.2d 57 (1988) (adopting “de novo” review). The panel set out at length the theoretical and practical underpinnings of standards of review. Specifically,- the court found that both “functional” and “analytical” concerns should command a.reviewing court’s attention when it is required to determine the appropriate standard of review. We- adopt that approach for the purpose of deciding our review powers here.

The most readily-made observation about standards of review is that they generally appear to be dictated by whether the court characterizes the matter under review as a question of law, a question..of fact, or a mixed question of law and fact. Courts *35 routinely engage in this classification process despite the well-recognized truth that the issue under review frequently defies rote categorization and that procrustean-like efforts often must be expended in order to fit the question into one of the fact, law or mixed fact and law categories. 2 Once the question is cast as fact, law or mixed fact and law, the court need only apply the designated standard of review for that category of question.

Hence, where the matter under review requires invocation or declaration of a fact-free general principle of law, the court will designate the issue as a question of law, and review the matter “de novo.” “De novo” review, otherwise referred to as “independent” or “plenary” review, empowers the appellate court, based on an original appraisal of the record, to reach a different result from the trial court without deference to that court’s findings. 1 S. Childress & M. Davis, Standards of Review: Federal Cases and Review Process § 2.14, at 76 (1986) (hereinafter Standards of Review); see Bose Corp. v. Consumers Union of the United States, Inc., 466 U.S. 485, 514 n. 31, 104 S.Ct. 1949, 1967 n. 31, 80 L.Ed.2d 502 (1984).

An issue is designated a question of fact if it involves the who, what, where, when and how details of the case; the case specific “historical,” “basic,” “primary,” “discrete,” “subsidiary” or “operative” facts. See generally 2 S. Childress & M. Davis, Standards of Review, supra, § 2.21, at 100-07. When the trial court sits as the factfinder, its factual findings are accorded considerable deference and are reviewed under a “clearly erroneous” standard. D.C.Code § 17-305(a) (1981); Super.Ct.Civ.R. 52(a) (1987); see, e.g., Auxier v. Kraisel, 466 A.2d 416, 418 (D.C.1983); Hummel v. Koehler, 458 A.2d 1187, 1191 (D.C.1983). The “clearly erroneous” standard precludes the appellate court from setting aside a trial court’s finding of fact unless the “judgment is plainly wrong or without evidence to support it.” D.C.Code § 17-305(a) (1981). Thus, where the facts admit of more than one interpretation, the appellate court must defer to the trial court’s judgment. Anderson v. Bessemer City, 470 U.S. 564, 574, 105 S.Ct. 1504, 1511-12, 84 L.Ed.2d 518 (1985); see Lee Washington, Inc. v. Washington Motor Truck Transportation Employees Health & Welfare Trust, 310 A.2d 604, 606 (D.C.1973).

A “mixed question” of law and fact exists where “the historical facts are admitted or established, the rule of law is undisputed, and the issue is whether the facts satisfy the statutory standard ... [or] whether the rule of law as applied to the established facts is or is not violated.” Pullman-Standard v. Swint, 456 U.S. 273, 289 n. 19, 102 S.Ct. 1781, 1790 n. 19, 72 L.Ed.2d 66 (1982); Cuyler v. Sullivan, 446 U.S. 335, 342, 100 S.Ct. 1708, 1715, 64 L.Ed.2d 333 (1980) (analysis applied to mixed questions of law and fact require the “application of legal principles to the historical facts.”); Townsend v. Sain, 372 U.S. 293, 309 n. 6, 83 S.Ct. 745, 755 n. 6, 9 L.Ed.2d 770 (1963) (mixed questions of fact and law “require the application of a legal standard to the historical-fact determinations”). There is no one standard of review that is uniformly applied to “mixed ques *36 tions.” Rather, mixed questions are “assigned, sometimes clumsily, either to the ‘clearly erroneous’ or to the ‘de novo’ category, depending, ostensibly, on whether the reviewing court regards the matter as more closely resembling a question of fact or a question of law.” Felder, supra, 548 A.2d at 61. Alternatively, some courts attempt to unmix the “mixed question,” reviewing the factual issue under the “clearly erroneous” standard and the legal issue under the “de novo” standard. See Miller v. Fenton, 474 U.S. 104, 112, 117, 106 S.Ct. 445, 450-51, 453, 88 L.Ed.2d 405 (1985); Strickland v. Washington, 466 U.S. 668, 698, 104 S.Ct. 2052, 2070, 80 L.Ed.2d 674 (1984); Felder, supra, 548 A.2d at 62; Curry v. United States, 498 A.2d 534, 540 (D.C.1985). As a matter of practice, it appears that many courts elect to review de novo in the mixed question context. Felder, supra, 548 A.2d at 61-62; see United States v. McConney, 728 F.2d 1195, 1202 (9th Cir.) (en banc), cert. denied, 469 U.S. 824, 105 S.Ct. 101, 83 L.Ed.2d 46 (1984).

It is apparent then, that the initial fact versus law determination plays a central role in an appellate court’s ultimate decision concerning which standard of review it shall employ. The fact/law determination, however, provides only a partial explanation of the process which culminates in the selection of a standard of review; indeed, frequently it does little more than describe that process.

A more thorough explanation of the process through which a standard of review is chosen comes from the unsurprising notion that courts are responding to common law “conceptions about how authority should be divided between the court in which a matter is originally heard and the reviewing court or courts.” Felder, supra, 548 A.2d at 63. As the Supreme Court recognized in Miller, the “practical truth” is that the decision to label a matter as fact, law or mixed fact and law “is sometimes as much a matter of allocation as it is of analysis.” Supra, 474 U.S. at 113-14, 106 S.Ct. at 451. “Allocation” is an operative factor in the selection of standard of review process that is most visible “in those instances in which the issue falls somewhere between a pristine legal standard and a simple historical fact, the fact/law distinction at times has turned on a determination that, as a matter of the sound administration of justice, one judicial actor is better positioned than another to decide the issue in question.” Id. at 114, 106 S.Ct. at 451.

In Felder, we adopted this approach and held that both the analytical factor and the allocation factor, which we renamed the functional factor, should be processed into the decision of selecting a proper standard of review. Supra, 548 A.2d at 61-64. We concluded that

a standard of review embodies either or both the following considerations: (1) which judicial actor logically should be entrusted to be the primary decisionmaker about a particular issue, in light of conceptions about the nature and significance of that issue in our legal system, and (2) which forum is, as a purely practical matter, more likely to render the decision with the higher degree of accuracy. The first approach, which we call “analytical,” depends on an a priori categorization of the nature of the matter under review. The second approach, which we term “functional,” focuses instead on which court, original or reviewing, has the best access to the tools and materials necessary to deciding the matter in question.

To be sure, the analytical and functional approaches are not unrelated. Indeed, what separates law and fact as analytical categories may, in actuality, reflect a recognition of divergent roles between an appellate court and trial court. In other words, questions that analytically would most likely be regarded as questions of fact are ordinarily those that can be answered by hearing testimony — a function in which only trial courts engage. On the other hand, questions that are regarded quintessentially as questions of law are precisely those to be resolved through reasoning, comparison with like cases, and review of a trial court record — tasks that the appellate *37 court is principally designed to undertake.

Id. at 63-64. 3

There is a “ ‘search-for-the-truth view’ of a criminal trial [that] is usually employed to mean maximum correspondence between the judgment of guilt and the defendant’s prior performance of the culpable conduct alleged. Yet we have developed a process which is designed over the long haul, to preclude conviction of an innocent person at the social expense of acquitting some guilty defendants.” Uviller, The Advocate, The Truth, and Judicial Hackles: A Reaction to Judge Frankel’s Idea, 123 U.Pa.L. Rev. 1067, 1078 (1975) (hereinafter The Ad vocate). That is why constitutional violations which involve rights that are so basic to a fair trial mandate per se “reversal without regard to the evidence in the particular case.” Rose v. Clark, 478 U.S. 570, 577, 106 S.Ct. 3101, 3105, 92 L.Ed.2d 460 (1986); see Gray v. Mississippi, 481 U.S. 648, 668, 107 S.Ct. 2045, 2057, 95 L.Ed.2d 622 (1987) (plurality opinion); Chapman, supra, 386 U.S. at 23, 87 S.Ct. at 827. 4 In contrast, we countenance some “constitutional errors which in the setting of a particular case are so unimportant and insignificant that they may, consistent with the Federal Constitution, be deemed harmless, not requiring the automatic reversal of the conviction.” Chapman, supra, 386 U.S. at 22, 87 S.Ct. at 827. (emphasis added). 5

*38 The harmless error doctrine serves the twin goals of the criminal trial: fairness of process and reliability of result. Rose, supra, 478 U.S. at 577-79, 106 S.Ct. at 3105-07; see United States v. Lane, 474 U.S. 438, 478, 106 S.Ct. 725, 747, 88 L.Ed.2d 814 (1986) (“A determination that an error was harmless is an extremely weighty conclusion; it implicates profound notions of fairness and justice.”). The doctrine measures the importance of enforcing a constitutional guarantee against the practical wisdom that “ ‘[wjrong directions which do not put the traveler out of his way, furnish no reasons for repeating the journey.’ ” R. Leflah, Appellate Judicial Opinions 117 (1974) (quoting Cherry v. Davis, 59 Ga. 454, 456 (1877)). Hence, reversal is not required to vindicate the right if “the beneficiary of a constitutional error [can] prove beyond a reasonable doubt that the error complained of did not contribute to the verdict obtained.” Chapman, supra, 386 U.S. at 24, 87 S.Ct. at 828; see United States v. Hasting, 461 U.S. 499, 510-11, 103 S.Ct. 1974, 1981, 76 L.Ed.2d 96 (1983) (reversal not mandated if absent the error it is “clear beyond a reasonable doubt that the jury would have returned a verdict of guilty.”).

Although the legal test for determining whether a" constitutional error is harmless is well-settled, the standard for reviewing such determinations is less clear, particularly for state appellate courts. The uncertainty concerning the scope of an appellate court’s review powers in the harmless error context is captured in part by such intrinsically dissonant expressions as “an appellate-level piece of fact-finding,” Bartram v. Maryland, 33 Md.App. 115, 152, 364 A.2d 1119, 1141, n. 1, aff'd, 280 Md. 616, 374 A.2d 1144 (1976), and “de novo factual review.” 2 S. Childress & M. Davis, Standards of Review, supra, § 7-3, at 9. The confusion persists despite the fact that the evaluation of harmlessness is “one of the most significant tasks of an appellate court, as well as one of the most complex.” R. TRAynor, The Riddle of Harmless Error 80 (1970) (hereinafter The Riddle). ' Indeed, the harmless error doctrine disposes of more appellate issues than any other single doctrine of review. 2 Childress & M. Davis, Standards of Review, supra, § 7-1, at 3-4.

The difficulty in discerning the scope of our review in the present circumstances derives in part, from a variety of factors which make comparison with existing case law a very trying matter, for example: 1) a reviewing court may have to engage in fundamentally different tasks, depending upon whether the doctrine is invoked and the issue resolved, in the first instance, at the trial court level, see, e.g., United States v. Buege, 578 F.2d 187, 188 (7th Cir.), cert. denied, 439 U.S. 871, 99 S.Ct. 203, 58 L.Ed.2d 183 (1978), or the appellate court level, see, e.g., Bartram, supra, 364 A.2d at 1141 & n. 1, and this may cause shifts in the dépth and breadth of the courts’ evaluation of whether the error is harmless; 2) obscure analysis or language may render it impossible to determine which standard the reviewing?court.is applying; 3) application of federal precedent at the state level is complicated by the constraints under which federal courts must operate, see, e.g., 28 U.S.C. § 2254(d) (1982); 6 and 4) the unavailability or inaccessibility of detailed pri- *39 or case history prevents a clear understanding of the often fine line distinctions between fact, law, and law application determinations, as well as which courts along the way engaged in these decisions.

Be that as it may, we start with the following established harmless error principles. First, since Chapman, the Supreme Court “has consistently made clear that it is the duty of a reviewing court to consider the trial record as a whole and to ignore errors that are harmless, including most constitutional violations.” Hasting, supra, 461 U.S. at 509, 103 S.Ct. at 1980; see Rose, supra, 478 U.S. at 583, 106 S.Ct. at 3109; Lane, supra, 474 U.S. at 445, 106 S.Ct. at 729-30; Harrington v. California, 395 U.S. 250, 254, 89 S.Ct. 1726, 1728, 23 L.Ed.2d 284 (1969) (court must rely on its “own reading of the record and on what seems to have been the probable impact ... on the minds of an average jury.”); accord Brooks v. United States, 367 A.2d 1297, 1310 (1976). This duty to make an “independent examination of the record when federal constitutional deprivations are alleged is clear, resting, as it does, on our solemn responsibility for maintaining the Constitution inviolate.” Napue v. Illinois, 360 U.S. 264, 271, 79 S.Ct. 1173, 1178, 3 L.Ed.2d 1217 (1959) (citations omitted).

That what the Court has in mind is “de novo” review, which is coterminous with “independent” review, is made plain by the Court’s classification of the determination whether a constitutional error is harmless as a question of federal law. Rushen v. Spain, 464 U.S. 114, 120, 104 S.Ct. 453, 456, 78 L.Ed.2d 267 (1983) (per curiam) (emphasis added); Chapman, supra, 386 U.S. at 20-21, 87 S.Ct. at 826. Although “de novo” review is required as to the legal determination of harmlessness, the Supreme Court nevertheless has held that federal courts engaged in the evaluation of harmlessness must defer to both the state trial and appellate court determinations of historical facts which were made en route to deciding the ultimate legal question of harmlessness under 28 U.S.C. § 2254(d) (1982). Rushen, supra, 464 U.S. at 120, 104 S.Ct. at 456; see Sumner v. Mata, 449 U.S. 539, 101 S.Ct. 764, 66 L.Ed.2d 722 (1981) (habeas case). 7

As to the legal determination of harmlessness then, we have a “constitutional *40 responsibility that cannot be delegated to the trier of fact.” Bose Corp., supra, 466 U.S. at 501, 104 S.Ct. at 1959. That the Supreme Court itself rarely engages in the independent review of the harmless error before it, does not alter in any way our responsibility to do so. The Court’s stated preference for remanding to lower appellate courts on the question of harmlessness, see Rose, supra, 478 U.S. at 584, 106 S.Ct. at 3109 (“Although we ‘plainly have the authority’ to decide whether, on the facts of a particular ease, a constitutional error was harmless ... we do so sparingly.”) (quoting Hasting, supra, 461 U.S. at 510, 103 S.Ct. at 1981); Parker v. Randolph, 442 U.S. 62, 78-81, 99 S.Ct. 2132, 2141-43, 60 L.Ed.2d 713 (1979) (Blackmun, J., concurring in part); Milton v. Wainwright, 407 U.S. 371, 377, 92 S.Ct. 2174, 2177-78, 33 L.Ed.2d 1 (1972); Harrington v. California, 395 U.S. 250, 254, 89 S.Ct. 1726, 1728, 23 L.Ed.2d 284 (1969); Chapman, supra, 386 U.S. at 24-26, 87 S.Ct. at 828-29; accord Lane, supra, 474 U.S. at 450, 106 S.Ct. at 732; Francis v. Franklin, 471 U.S. 307, 326 n. 10, 105 S.Ct. 1965, 1977 n. 10, 85 L.Ed.2d 344 (1985), appears to arise in part, from the view that such decisions are “best left to the” state appellate courts or the federal courts of appeal, in the first instance. Delaware v. Van Arsdall, 475 U.S. 673, 684, 106 S.Ct. 1431, 1438, 89 L.Ed.2d 674 (1986); Connecticut v. Johnson, 460 U.S. 73, 102, 103 S.Ct. 969, 985, 74 L.Ed.2d 823 (1983) (Powell,. J., dissenting) (harmless error “is a question more appropriately left to the courts below”).

Departing from the viéw that the decision whether a constitutional error was harmless is a question of law, e.g., Rushen, supra, 464 U.S. at 120, 104 S.Ct. at 456, the Ninth and Eleventh Circuits take the position that this determination is a “mixed question” of law and fact that should be reviewed “de novo,” save for basic, primary or historical fact findings to which deference will be given absent clear error. Adamson v. Ricketts, 789 F.2d 722, 727 n. 5 (9th Cir.1986) (en banc), rev’d on other grounds, 483 U.S. 1, 107 S.Ct. 2680, 97 L.Ed.2d 1 (1987); Fendler v. Goldsmith, 728 F.2d 1181, 1190-91, n. 21 (9th Cir.1983) (as amended 1984); Proffitt v. Wainwright, 685 F.2d 1227, 1259, n. 48 (11th Cir.1982), modified on reh’g on other grounds, 706 F.2d 311; cert. denied, 464 U.S. 1002, 104 S.Ct. 508, 78 L.Ed.2d 697 (1983). This departure may be more semantic than real; as for all practical purposes, the resulting standard of review is identical, at least for the federal courts.

For federal courts, “de novo” review of the legal conclusion of harmlessness means independent, plenary review of the record save for the section 2254(d) requirement that they defer to state court historical fact findings, a constraint that is grounded upon the principles of federalism and comity, see Sumner, supra, 449 U.S. at 547, 101 S.Ct. at 769, as opposed to the more general principle that appellate courts should defer to the trier of fact because, as a functional matter, the trier of fact is better positioned to make such findings. See supra, section 1A. Characterization of the harmless error decision as a “mixed question,” on the other hand, permits unmixing of the fact and law elements so that we may, under general principles which are familiar to and regularly relied upon by both state and federal courts alike, review the law portion “de novo,” and the fact portion for “clear error.” See Adamson, supra, 789 F.2d at 727 n. 5; Fendler, supra, 728 F.2d at 1190-91 n. 21; Proffitt, supra, 685 F.2d at 1259 n. 48.

In our view, the “mixed question” designation provides state courts a better-suited framework for working through the complicated and interdependent factual and legal conclusions typically encountered in harmless error cases: Although the Supreme Court has unambiguously classified harmless error determinations as questions of law, we nevertheless do not believe that we are precluded from reviewing a trial court’s purely historical or subsidiary fact findings under a clearly erroneous standard.

Our view that the primary role of state appellate courts engaged in the independent review of a trial court’s determination of. harmlessness is to .evaluate the legal *41 conclusion, finds support in the harmless error case law as well as in other cases reviewing for determinations of constitutional violations. In Delaware v. Van Arsdall, supra, 475 U.S. at 684, 106 S.Ct. at 1438, for example, the Supreme Court, upon concluding that the defendant had been improperly denied the right to impeach a witness for bias, remanded the case to the Delaware Supreme Court to determine, in the first instance — as neither the trial court nor the appellate courts had previously undertaken this analysis, Van Arsdall v. State, 486 A.2d 1, 6-7 (Del.1984) —whether that constitutional error was harmless beyond a reasonable doubt. In remanding, the Court explained:

Whether such an error is harmless depends upon a host of factors, all readily accessible to reviewing courts. These factors include the importance of the witness’ testimony in the prosecution’s case, whether the testimony was cumulative, the presence or absence of evidence corroborating or contradicting the testimony of the witness on material points, the extent of cross-examination otherwise permitted, and, of course, the overall strength of the prosecution’s case.

Id. (citations omitted) (emphasis added). We think this list of factors, while not exhaustive, suggests that the Court perceived the appellate court’s task as primarily one of resolving the remaining “law” and “mixed law and fact” issues.

“What clues are in the cold print to indicate where the truth lies? What clues are there to indicate where the half-truth lies?” R. TraynoR, The Riddle, supra, at 21.

We believe that functionally, the trial court, having the opportunity to observe the trial as it unfolds, is better positioned to weigh and reconcile the evidence; a far more difficult task for the appellate court, which “is limited to the mute record made below.” Id. at 20; see, e.g., Petrillo, supra, 821 F.2d at 88. Harmless error, however, entails an ultimate conclusion, that is, the court must determine “the legal effects of the findings” on the verdict. R. Tray-nor, The Riddle, supra, at 40. That is why, as an analytical matter, the appellate court must have the last meaningful word on the consequences of the denial of a constitutional right. We again caution that one must not readily assume that the functional and analytical approaches are at odds. Rather, we visualize the functional and analytical approaches operating in concert. See, e.g., Lacy, supra, 791 F.2d 980; Adamson, supra, 758 F.2d 441. As a result, we reject a pure “clearly erroneous” standard because it does not adequately recognize our responsibility to decide the effects of a constitutional violation. At the same time, we reject a pure “de novo" standard because it does not permit deference to the findings of fact mandated by D.C.Code § 17-305(a) (1981).

The United States urges us to rely on United States v. Agurs, 427 U.S. 97, 96 S.Ct. 2392, 49 L.Ed.2d 342 (1976), and hold that our measure of review of the trial court’s harmlessness conclusion is that of “clearly erroneous.” Agurs dealt with the issues presented by Brady v. Maryland, 373 U.S. 83, 87, 83 S.Ct. 1194, 1196-97, 10 L.Ed.2d 215 (1963) (“the suppression by the prosecution oi evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or punishment”). In Agurs, the Court held that suppressed evidence is “material” based on its likely impact on the outcome of the trial. Subsequently, in United States v. Bagley, the Court held that the suppressed evidence is "material only if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different. A ‘reasonable probability’ is a probability sufficient to undermine confidence in the outcome.” 473 U.S. 667, 687, 105 S.Ct. 3375, 3386, 87 L.Ed.2d 481 (1985). As Bagley makes clear, the issue in Brady cases is whether the evidence was “material” in the constitutional (due process) sense. It is only when it was constitutionally material that the suppression thereof by the prosecution is error in the first place. While perhaps conceptually similar, we are of the view that the task of *42 the trial court in determining constitutional materiality, i.e., whether there was error by the prosecutor suppressing evidence, is sufficiently different from the trial court’s evaluation of the impact of its own error, constitutional or otherwise, that we decline to apply the Bagley approach to our harmless analysis of trial court error. To do otherwise would be to abrogate “our solemn responsibility for maintaining the Constitution inviolate.” Napue, supra, 360 U.S. at 271, 79 S.Ct. at 1178. 8

We therefore hold that in reviewing a trial judge’s decision as to the harmlessness of a constitutional violation, we will ask ourselves, after reviewing the record: (1) whether the evidentiary and subsidiary facts recited by the trial court and those inferences deduced therefrom are fairly supported by the record; and (2) if so, whether in our judgment given the whole record, the- error was “harmless beyond a reasonable doubt.” Chapman, supra, 386 U.S. at 24, 87 S.Ct. at 828. The first step requires application of the “clearly erroneous” standard; whereas, the second step incorporates “de novo” review. We deem such a test particularly appropriate where, as here, the trial court is assessing the impact of its own constitutional error given the human tendency toward self-justification. 9

Having concluded that we owe no deference to the trial court’s views concerning harmlessness, we note that when because of an incomplete record upon which to assess harmlessness, it is necessary to remand for further proceedings in the trial court (as we were required to do both in Jaggers v. United States, 482 A.2d 786 (D.C.1984) and in this case), no useful purpose is served by requesting the trial court to opine on this question. The only proper role of the trial court, after resolving any legal, see id. at 794, and/or factual issues required by the remand, is to make the proper evidentiary record and return the matter to this court.

II.

Having explained our standard of review, our only remaining task is to apply it in this case. We direct the reader to Davis I, supra, 482 A.2d at 784-85, where the factual circumstances leading to Davis’ indictment are set forth in detail. At the remand hearing, Theodore Garvin testified that at the time of the arrest he was in the rear passenger seat, Perry was in the driver’s seat, Kemper was seated directly behind the driver, and Davis was in the front passenger seat. Garvin also testified that he never saw a gun. The trial court heard argument from counsel on whether the exclusion of this testimony at trial was harmless error beyond a reasonable doubt. After taking the matter under advisement to review the transcript and assess the impact of Garvin’s testimony, the trial court did an analysis which ultimately led it to conclude that the exclusion of Garvin’s testimony was constitutional harmless error.

*43 The trial court noted that there was trial testimony which placed Garvin in the front passenger seat and Davis behind him in the backseat, as well as testimony which showed that Kemper was observed in actual possession of the weapon. The trial court also recalled that Smith had positively identified Davis as the man who had shot Harris earlier in the day. Additionally, the trial court pointed to the testimony which inferentially identified Perry in the shooting, for example: (1) Smith testified that he had observed a woman give Davis a weapon, which Davis then used to shoot Harris and which weapon resembled the pistol recovered at the time of the arrest; (2) Smith also testified that the woman who gave Davis the gun was the same woman driving the vehicle, a yellow Chevette; (3) Smith’s description of the vehicle matched the description of the vehicle Perry, Davis, Kemper and Garvin were in at the time of their arrest; and (4) Perry was driving the vehicle when she was apprehended. The trial court also noted that Harris, the victim of the shooting, gave exculpatory testimony in favor of Davis and Perry, specifically, he testified that neither man was involved in the earlier shooting incident. We agree with the trial court’s foregoing analysis of the factual record. We further note that Davis does not challenge this analysis.

The trial court made an additional determination relating to the government’s theory of the case. The court found that the government had attempted to prove, primarily through Smith’s testimony, that Davis and Perry had joint constructive possession of the weapon recovered from the vehicle. Davis does not contend that the government did not present the case to the jury on this theory. Rather, he contends that the government also presented a second, alternative theory of constructive possession; a theory that hinged upon Davis’ sitting in the rear passenger seat. To support his argument, Davis references the following language from the prosecutor’s closing argument:

With respect to Ms. Perry and with respect to Mr. Davis, the evidence there is that one, without any reference to the fight at all, they are in a car. Two, it’s a very small car ... [A]sk yourself whether Cassandra Perry and Milton Davis knew there was a gun in the car. Remember then that Milton Davis was sitting in the back seat with Mr. Kemper, a very small car, and ask yourselves whether Mr. Milton Davis, in dependent [sic] of the fight, without any reference to the fight at all, whether Milton Davis had knowledge that the gun was in the car.

After making these remarks, however, the prosecutor also told the jury that the government’s theory of constructive possession was based upon the earlier shooting incident. The prosecutor further stated that the government was not claiming that Perry and Davis had possession of the weapon when they were stopped by the police. 10 More importantly, the trial court’s instructions to the jury on the constructive possession charge made clear the government’s theory:

In this regard, let me remind you that evidence was introduced about events alleged to have taken place at about 4 o’clock in the afternoon and the events about which you heard testimony from *44 Milton Smith and also from Enoch Harris concerning the shooting of Mr. Harris in his hand.

Now that evidence is admitted only for your consideration of whether it tends to show whether Mr. Davis and/or Ms. Perry was knowing[ly] in constructive possession of the pistol at the time the ear was stopped.

It is well-settled that absent “extraordinary situations,” which do not exist in this case, “the crucial assumption underlying our constitutional system [of trial by jury] is that juries will follow instructions.” Francis, supra, 471 U.S. at 324-25 n. 9, 105 S.Ct. at 1976-77 n. 9 (citing Parker v. Randolph, 442 U.S. 62, 99 S.Ct. 2132, 60 L.Ed.2d 713 (1979)). We agree with this analysis by the trial court. We likewise disagree with Davis that the jury considered an alternative argument, of prosecution. 11

In order for one to be guilty of carrying a pistol without a license, the government must prove, by direct or circumstantial evidence, that a defendant had “knowledge of the presence of the pistol or ammunition and the existence of dominion and control over them.” Logan v. United States, 489 A.2d 485, 491 (D.C.1985) (quoting Easley v. United States, 482 A.2d 779, 781 (D.C.1984) (emphasis in original)). “The right to exercise dominion and control may be jointly shared.” Wheeler v. United States, 494 A.2d 170, 172 (D.C.1985) (citations omitted). Further, although “an accused’s mere presence at the scene of the crime, his proximity to the ... [gun] or his association with one in possession do not in themselves substantiate a finding of constructive possession; nevertheless, ‘presence, proximity or association may establish a prima facie case of ... possession when colored by evidence linking the accused to an ongoing criminal operation of which that possession is a part.’ ” Easley, supra, 482 A.2d at 782 (quoting United States v. Hubbard, 429 A.2d 1334, 1338 (D.C.) (possession of narcotics) (citation omitted), cert. denied, 454 U.S. 857, 102 S.Ct. 308, 70 L.Ed.2d 153 (1981)).

As the trial court correctly found, Gar-vin’s testimony concerning where he and Davis were sitting in the vehicle is inconsistent with the trial testimony and thus, squarely presents the question of whether the constitutional error may have contributed to the verdict. This case, however, presents one of those rare instances where the facts provide a truly revealing gauge of how- the jury credited the trial evidence— facts which can assist us in determining whether the constitutional error was harmless. Specifically, we can extrapolate from what the jury concluded as to Perry to assess how it credited the evidence as to Davis.

The trial court recognized this avenue of analysis and we, applying our independent review, also recognize this opportunity. Davis argues that the jury could have concluded that Perry had constructive possession of the weapon on the basis that the weapon was recovered from behind her seat. We disagree. Indeed, in order for Kemper to have been convicted on a theory of actual possession, the jury would, have had to believe the. trial testimony which showed that he was observed holding, the weapon on his lap prior to placing it behind the seat within seconds after being stopped by police.. Accordingly, there is no basis for asserting that Perry was convicted as a result of the weapon being recovered from behind her seat. Thus, in order for _ the jury to have convicted Perry, it would have had to credit the evidence inferentially identifying Perry as a participant in the earlier shooting over Harris’ exculpatory testimony. We further note that the jury was so instructed and the government never argued otherwise.

As a result, there is no reasonable possibility that the jury would have credited the *45 government’s evidence with respect to Perry but not as to Davis. This is particularly true with respect to Davis, whom Smith positively identified, whereas Smith only inferentially identified Perry. Even in crediting Garvin’s testimony, we note that his testimony places Davis in the car. Moreover, Garvin’s testimony indicating that he did not see a gun is not inconsistent with the trial testimony because Garvin did not testify as to whether he was observing Kemper; nor did he testify about the earlier shooting incident.

Davis, however, contends that reliance upon Perry’s conviction is improper because the jury may only have convicted her because they “believed her to be Mr. Davis’ accomplice in the shooting and that if he were sitting next to the gun, she had constructive control over it.” We have great difficulty understanding how this argument assists Davis due to its implicit admission that for the jury to convict Perry it would have first had to conclude that Davis had possession of the weapon at the earlier shooting incident.

It is true that in prior cases this court has reversed a conviction when the weapon was not recovered in close proximity to where the defendant was sitting in a vehicle. See Easley, supra, 482 A.2d 779 (reversing conviction where weapon and ammunition were recovered under front seat of car and defendants were seated in back, and where the evidence was insufficient to demonstrate a joint criminal venture; cf . Logan, supra, 489 A.2d at 491-92 (upholding conviction despite recovery of weapon a few car lengths from where vehicle was stopped because driver and passengers were observed acting in concert to drop weapon from car).

Absent from these cases, however, is evidence that the defendant was in prior possession of the weapon. For example, in Porter v. United States, we upheld Porter’s conviction even though he was the driver and the weapon was recovered from underneath the front passenger seat of the vehicle. 282 A.2d 559 (D.C.1971). We found that the government had sustained its burden to support a theory of constructive possession because the evidence demonstrated that Porter had control of the vehicle, and that either Porter or his code-fendant was seen carrying a weapon at an earlier disturbance. Cf. Easley, supra, 482 A.2d 779; United States v. Bethea, 143 U.S.App.D.C. 68, 71, 442 F.2d 790, 793 (1971) (government failed to show connection between guns seized from car and the charged crime of possession of narcotics).

Granted, Davis was not the driver of the vehicle. Nonetheless, we find the facts of this case even more dispositive of Davis’ guilt, as they should be since we are reviewing a harmless error claim, than in Porter, supra, 282 A.2d 559. We make this finding based on the fact that Davis was positively identified in possession of the weapon within a relatively short time span from when he was apprehended. Stated otherwise, the evidence linking Davis to the earlier shooting, along with his presence in the vehicle and his association with Kemper and Perry, overwhelmingly demonstrates that he had constructive possession of the weapon. Thus, we conclude beyond a reasonable doubt that the excluded evidence did not contribute to the verdict. For the foregoing reasons, Davis’ conviction is

Affirmed.

Notes

. In granting Davis’ petition for rehearing en banc, Davis v. United States, 518 A.2d 712 (D.C.1986), we vacated a second division opinion, Davis v. United States, 509 A.2d 105 (D.C.1986), that had reviewed the trial court’s determination of harmlessness under the "clearly erroneous" standard, affirming the conviction.

. Of course, the analytical distinction between fact and law is necessarily an artificial construct. See, e.g., Damaska, Presentation of Evidence and Factfinding Precision, 123 U.Pa.L.Rev. 1083 (1975) (arguing that determination of relevant facts is impossible unless applicable deci-sional standards have been chosen, but that most of the time these legal standards cannot be understood without relating them to the factual situation of the case); see also Bose Corp. v. Consumers Union of the United States, Inc., 466 U.S. 485, 501 n. 17, 104 S.Ct. 1949, 1960 n. 17, 80 L.Ed.2d 502 (1984) (“A finding of fact in some cases is inseparable from the principles through which it was deduced. At some point, the reasoning by which a fact is ‘found’ crosses the line between application of those ordinary principles of logic and common experience which are ordinarily entrusted to the finder of fact into the realm of a legal rule upon which the reviewing court must exercise its own independent judgment.”); Pullman-Standard v. Swint, 456 U.S. 273, 288, 102 S.Ct. 1781, 1790, 72 L.Ed.2d 66 (1982) (the fact/law distinction has yet to yield to any "rule or principle that will unerringly distinguish a factual finding from a legal conclusion.”); Baumgartner v. United States, 322 U.S. 665, 671, 64 S.Ct. 1240, 1244, 88 L.Ed. 1525 (1944) (the fact/law distinction "is often not an illuminating test and is never self-executing.”).

.Federal Rule 52(a) illustrates well the compatible coexistence of the functional and analytical approaches within a single standard of review. In Felder, we explained:

The functional basis for the “clearly erroneous” standard has long been recognized in the proposition that "the trial court’s customary opportunity to evaluate the demeanor and thus the credibility of witnesses ... is the rationale behind Rule 52(a), United States v. General Motors Corp., 384 U.S. 127, 141 n. 16 [, 86 S.Ct. 1321, 1328 n. 16, 16 L.Ed.2d 415] (1966), a rule that by its terms adverts to the "opportunity of the trial court to judge of the credibility of the witnesses,” Fed.R.Civ.P. 52(a). The functional underpinnings of the Rule are also suggested by the principle that not all factual findings are entitled to equal weight upon review: "The conclusiveness of a ‘finding of fact’ depends on the nature of the materials on which the finding is based,” Baumgartner, supra, 322 U.S. at 670-71 [64 S.Ct. at 1243,] and has less significance in what is "essentially a ‘paper case,' ” General Motors, supra, 384 U.S. at 141 n. 16 [, 86 S.Ct. at 1328 n. 16] (being accessible to resolution by a reviewing court).

Nevertheless, the Rule superintending review of findings of fact is not wholly delimited by a functional perspective. For ‘‘[t]he same ‘clearly erroneous' standard applies to findings based on documentary evidence as to those based entirely on oral testimony, ... [even though] the presumption has lesser force in the former situation than in the latter, Bose, supra, 466 U.S. at 500 [, 104 S.Ct. at 1959] (emphasis supplied). Therefore, we know 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.” Anderson, supra, 470 U.S. at 574 [, 105 S.Ct. at 1512]. Instead there are some essential analytical differences between the finding of facts and the expositing of law. The finding of fact is "[t]he trial judge’s major role ... and with experience in fulfilling that role comes expertise” warranting appellate court deference, id. This analytical distinction between fact and law is what may be ultimately decisive in generating standards of review, for "[o]ne might therefore assume that the cases in which the appellate courts have a duty to exercise independent review are merely those in which the presumption that the trial court's ruling is correct is particularly weak [Le., where it does not depend principally on credibility judgments]. The difference between the two roles, however, is much more than a matter of mere degree,” Bose, supra, 466 U.S. at 500-01 [, 104 S.Ct. at 1959].

Supra, 548 A.2d at 64.

. Automatic reversal for a constitutional error was required, for example, upon the exclusion of a qualified prospective juror from a capital jury. Gray, supra, 481 U.S. 648, 107 S.Ct. 2045; complete denial of the right to counsel, Gideon v. Wainwright, 372 U.S. 335, 83 S.Ct. 792, 9 L.Ed.2d 799 (1963); introduction of a coerced confession, Payne v. Arkansas, 356 U.S. 560, 78 S.Ct. 844, 2 L.Ed.2d 975 (1958); and trial by a biased judge, Tumey v. Ohio, 273 U.S. 510, 47 S.Ct. 437, 71 L.Ed. 749 (1927).

. Among others, the following constitutional errors have been deemed harmless: restricted cross-examination concerning witness bias, Delaware v. Van Arsdall, 475 U.S. 673, 106 S.Ct. 1431, 89 L.Ed.2d 674 (1986); denial of the right to be present at trial, Rushen v. Spain, 464 U.S. 114, 104 S.Ct. 453, 78 L.Ed.2d 267 (1983) (per curiam); improper comment on a defendant’s failure to testify, United States v. Hasting, 461 U.S. 499, 103 S.Ct. 1974, 76 L.Ed.2d 96 (1983); admission of witness identification obtained in *38 violation of the right to counsel, Moore v. Illinois, 434 U.S. 220, 98 S.Ct. 458, 54 L.Ed.2d 424 (1977); admission of confession obtained in violation of the right to counsel, Milton v. Wainwright, 407 U.S. 371, 92 S.Ct. 2174, 33 L.Ed.2d 1 (1972); and admission of evidence obtained in violation of the fourth amendment, Chambers v. Maroney, 399 U.S. 42, 90 S.Ct. 1975, 26 L.Ed.2d 419 (1970).

. Section 2254(d) provides, in relevant part, as follows:

In any proceeding instituted in Federal court by an application for a writ of habeas corpus by a person in custody pursuant to a judgment of a State court, a determination after a hearing on the merits of a factual issue made by a State court of competent jurisdiction ... evidenced by a written finding, written, opinion, or other reliable and adequate written indicia, shall be presumed to be correct....

Id. (emphasis added). In harmless error cases, as in all cases brought to the federal courts via the writ of habeas corpus, section 2254(d) requires federal judges to presume the correctness of the state courts' historical or subsidiary fact findings. E.g., Rushen, supra, 464 U.S. at 120, 104 S.Ct. at 456; Cuyler, supra, 446 U.S. at 341, 100 S.Ct. at 1714.

. See also Savory v. Lane, 832 F.2d 1011, 1018—19 (7th Cir.1987) ("While the law/fact dichotomy is far from self-explanatory, we believe that questions such as harmless error, that are always determined by the court fall safely on the ‘law’ side of the line_ Like many ‘legal’ determinations ... a decision whether an error was harmless or not requires a full examination of the facts and circumstances of the case. In the course of this undertaking, state courts frequently will make subsidiary factual determinations en route to the ultimate conclusion. Such subsidiary findings are entitled to the presumption of correctness" under section 2254(d)) (emphasis in original) (citations omitted); Burns v. Clusen, 798 F.2d 931, 941 (7th Cir.1986) (On the question of harmlessness, the court must conduct its own examination of the factors the state court relied on because "although historical fact findings normally must be presumed correct, the legal conclusions drawn by state courts from those historical facts on the federal constitutional question raised are subject to independent review.”); Lacy v. Gardino, 791 F.2d 980, 986 (1st Cir.) (to determine whether error was harmless, court “must assess record as a whole,” and although it must defer under section 2254(d) to state court historical fact findings, the presumption of correctness does not extend to court’s conclusion that improperly admitted evidence was cumulative because this determination "went beyond a finding of historical fact ... and included the issue of the harmlessness of the error.’’) (citations omitted), cert. denied, 479 U.S. 888, 107 S.Ct. 284, 93 L.Ed.2d 259 (1986); Holland v. Attorney General of New Jersey, 777 F.2d 150, 156 (3d Cir.1985) (whether "constitutional error was harmless is a question of federal law ... [t]his frequently requires that the federal habeas court review the untainted evidence, a review that the district court undertook and that falls equally upon us since the issue is one of law”; section 2254(d) held inapplicable here because trial court made predictions, not historical fact findings on relevant issues) (citations omitted); cf. Cook v. Foltz, 814 F.2d 1109, 1113 (6th Cir.) (application of harmless error rule ” 'is not a question of mechanical application of sterile legal principles, but a sensitive judicial determination made upon a review of the record’”) (citation omitted), cert. denied, 484 U.S. 837, 108 S.Ct. 119, 98 L.Ed.2d 77 (1987); Burton v. Foltz, 810 F.2d 118, 121 (6th Cir.) (a finding of harmless error is appropriate only upon "review of the record as a whole,” citing Van Arsdall, supra), cert. denied, 484 U.S. 942, 108 S.Ct. 327, 98 L.Ed.2d 354 (1987); Merlo v. Bolden, 801 F.2d 252, 256 (6th Cir.l986)(same), cert. denied, 480 U.S. 909, 107 S.Ct. 1358, 94 L.Ed.2d 527 (1987).

. We note that while the U.S. Courts of Appeal are sharply divided on their measure of review of a trial court’s finding that suppressed evidence is not constitutionally material in the Bagley sense, compare, Carter v. Rafferty, 826 F.2d 1299, 1305-06 (3d Cir.1987), cert. denied, 484 U.S. 1011, 108 S.Ct. 711, 98 L.Ed.2d 661 (1988), and cases cited therein with United States v.Jenrette, 240 U.S.App.D.C. 193, 201, 744 F.2d 817, 825 (1984), cert. denied, 471 U.S. 1099, 105 S.Ct. 2321, 85 L.Ed.2d 840 (1985), and United States v. Mackey, 571 F.2d 376, 389 (7th Cir.1978), we have opted for a deferential review on that issue. See Derrington v. United States, 488 A.2d 1314, 1339 (D.C.1985); Davies v. United States, 476 A.2d 658, 661 (D.C.1984).

. Cf. Rose v. Mitchell, 443 U.S. 545, 561, 99 S.Ct. 2993, 3003, 61 L.Ed.2d 739 (1979) ("this same trial court will be the court that initially must decide the merits of such a claim, findings of facts and applying the law to those facts. This leads us to doubt that claims that the operation of the grand jury system violates the Fourteenth Amendment in general will receive the type of full and fair hearing [that is] deemed essential"); Chaffin v. Stynchcombe, 412 U.S. 17, 27, 93 S.Ct. 1977, 1983, 36 L.Ed.2d 714 (1973) ("the second sentence is not meted out by the same judicial authority whose handling of the prior trial was sufficiently unacceptable to have required a reversal of the conviction. Thus, the jury, unlike the judge who has been reversed, will have no personal stake in the prior conviction and no motivation to engage in self-vindication.”).

. Specifically, the prosecutor stated:

The Government does not charge that as of that moment in time [when the car was stopped] the defendants had possession of the gun, but rather that at that point in time, that incident, the occurrence of the shooting incident at 9th and O demonstrates that Cassandra Perry and Milton Davis had guilty knowledge about that gun. They knew the gun was there. They intended to possess it. They intended to have it in the car. And that they, therefore, knowingly possessed that gun.

The Government did not intend, nor did it attempt to prove or to prosecute that shooting incident that occurred on 9th and O, and that was not the Government's purpose. The Government’s purpose was simply to show that incident, as best it could, in order to demonstrate the very simple proposition that Milton Davis and Cassandra Perry had guilty, knowing possession of the gun when they were stopped at 9th and O; that Harrison Kemper had guilty possession of the gun by the fact that when police attempted to hide it [sic]. He had it directly in his possession and he attempted to hide it when police approached the car.

. Davis also argues that the trial court relied upon the testimony which showed that he was sitting next to Kemper, along with evidence pertaining to the earlier shooting incident, when it denied his motion for judgment of acquittal. ' Thus; he contends that it is ironic that the trial court should find the seating arrangement unimportant when determining the, error, was harmless. Be that as it may, this argument, does not negaje the fact that the trial court’s instruction on constructive possession was based upon the-earlier shooting incident.