This is an admiralty case in which a key issue is an interpretation of state law. We took this ease en bane, at the request of the three-judge panel that initially heard this case, to decide whether we should accord special deference to a district judge’s interpretation of state law or whether we should review such determinations under the independent de novo standard that we apply to a district judge’s interpretation of federal law.
This case involves a personal injury action and a wrongful death action arising out of the navigation of three skiffs, two of which collided off Kodiak Island, Alaska. As a part of that action, the plaintiffs asserted an in personam liability claim against two of the defendants based upon an Alaska statute. The interpretation of this statute is the issue of state law that presently concerns us. There has been no definitive interpretation by the Alaska Supreme Court. The district judge held that the statute did not apply to the circumstances of this case. The three-judge appellate panel of this court unanimously requested en banc review because they found the standard of review to be controlling. The panel indicated that if the question of law were reviewed under the deferential standard that we have applied in the past, which permits reversal only for clear error, then they would affirm; but if they were to review the determination under an independent de novo standard, they would reverse.
Generally in the past we have applied a deferential standard of review to a district judge’s construction of the law of the state in which he or she sits, accepting that construction unless it is “clearly wrong.” Jablonski By Pauls v. United States, 712 F.2d 391, 397 (9th Cir.1983); Fleury v. Harper & Row, Publishers, Inc., 698 F.2d 1022, 1026 (9th Cir.), cert. denied, — U.S. -, 104 S.Ct. 149, 78 L.Ed.2d 139 (1983); Airlift International, Inc. v. McDonnell Douglas Corp., 685 F.2d 267, 269 (9th Cir. 1982); Camacho v. Civil Service Commission, 666 F.2d 1257, 1262 (9th Cir.1982); Gaines v. Haughton, 645 F.2d 761, 770 (9th Cir.1981), cert. denied, 454 U.S. 1145, 102 S.Ct. 1006, 71 L.Ed.2d 297 (1982).1 Today we adopt as the law of the circuit the rule that questions of state law are reviewable under the same independent de novo standard as are questions of federal law. In this case, the question of state law arose in an admiralty action; however, our standard of review of a district court’s interpretation of state law would be the same in a diversity case or any other case in which the district court’s decision involves, state law.
*1398In our view, a decision to give less than full independent de novo review to the state law determinations of the district courts would be an abdication of our appellate responsibility. Every party is entitled to a full, considered, and impartial review of the decision of the trial court. We review questions of fact under the clearly erroneous standard and we review conclusions of law de novo. There is no justification for being less thorough, for abdicating any portion of our appellate responsibility, or for curtailing the parties’ appellate rights simply because the law involved is state law. The parties are entitled to the same careful, independent consideration of the issues of law by the appellate court whether the case involves state law or federal law.
The parties are accorded independent de novo review of issues of law in the appellate courts, not because of any greater wisdom of appellate judges than of trial judges, but because of the structural differences between the two courts. This court, sitting en banc, has recently considered at length the structural relationship between the district courts and the court of appeals. United States v. McConney, 728 F.2d 1195, 1201-04 (9th Cir.1984) (en banc). We observed that the standards of review we apply to trial court decisions have been developed to protect the parties’ rights to appeal and to achieve sound judicial administration by recognizing the structural differences in the two types of courts. Thus, the application of Fed.R.Civ.P. 52(a)’s dearly erroneous standard to the district courts’ factual determinations “emphasizes ... the trial judge’s opportunity to judge the accuracy of witnesses' recollections and make credibility determinations ____” Id. at 1201. On the other hand, we observed in McConney that the application of the de novo standard to the trial courts’ conclusions of law reflects a different policy concern. As we stated in McConney:
Structurally, appellate courts have several advantages over trial courts in deciding questions of law. First, appellate judges are freer to concentrate on legal questions because they are not encumbered, as are trial judges, by the vital, but time-consuming, process of hearing evidence. Second, the judgment of at least three members of an appellate panel is brought to bear on every case. It stands to reason that the collaborative, deliberative process of appellate courts reduces the risk of judicial error on questions-of law. Thus, de novo review of questions of law, like clearly erroneous review of questions of fact, serves to minimize judicial error by assigning to the court best positioned to decide the issue the primary responsibility for doing so.
Id. (footnote omitted).
As noted above, we apply the de novo standard to the trial courts’ determinations of federal law questions. In re Bialac, 712 F.2d 426, 429 (9th Cir.1983); Hoptowit v. Ray, 682 F.2d 1237, 1245 (9th Cir.1982); United States v. One Twin Engine Beech Airplane, 533 F.2d 1106, 1108 (9th Cir.1976). When the trial courts are called upon to resolve questions of foreign law, we also review their conclusions under the de novo standard. Fed.R.Civ.P. 44.1. See also Kalmich v. Bruno, 553 F.2d 549, 552 (7th Cir.), cert. denied, 434 U.S. 940, 98 S.Ct. 432, 54 L.Ed.2d 300 (1977); 9 C. Wright & A. Miller, Federal Practice & Procedure § 2446 (1971). There is no sound reason why we have a lesser appellate duty to the parties to make a correct, independent determination when the question is one of state law. The policy concerns supporting the de novo standard apply as well to questions of state law as to questions of federal law. The appellate function is the same in each case and the same structural advantages encourage correct legal determinations.
The parties to a civil action may appeal “as a matter of right” under Fed.R.App.P. 3 from the final judgment of a district court to the circuit court of appeals except where direct review may be had in the Supreme Court. See 28 U.S.C. § 1291 (1982); 9 J. Moore, B. Ward & J. Lucas, Moore’s Federal Practice ¶ 203.03 (2d ed. 1980); 15 C. Wright, A. Miller & E. Cooper, Federal Practice and Procedure § 3901 at *1399398 (1976); see also United States v. De Bright, 730 F.2d 1255, 1259 (9th Cir.1984) (en banc) (statutory right of appeal of criminal defendant under 28 U.S.C. § 1291). The Supreme Court has direct appellate review of district court decisions in a few eases under 28 U.S.C. §§ 1252 and 1253.
It has been argued that Supreme Court precedent requires our deference to conclusions of state law by the district court because the Supreme Court has chosen not to review questions of state law in a number of cases and has deferred to the interpretation of the lower courts. In each instance, however, the parties had already received the mandatory appellate review in the circuit court. In none of these cases' was the Supreme Court exercising direct appellate review from the district court. For example, in Runyon v. McCrary, 427 U.S. 160, 181-82, 96 S.Ct. 2586, 2599-2600 (1976), it is apparent that the Supreme Court did not defer to the initial conclusion by the trial judge, but accepted the determination by the appellate court.
The petitioners’ contention is certainly a rational one, but we are not persuaded that the Court of Appeals was mistaken in applying the two-year state statute. The issue was not a new one for that court, for it had given careful consideration to the question of the appropriate Virginia statute of limitations to be applied in federal civil rights litigation on at least two previous occasions____ We are not disposed to displace the considered judgment of the Court of Appeals on an issue whose resolution is so heavily contingent upon an analysis of state law, particularly when the established rule has been relied upon and applied in numerous suits filed in Federal District Courts in Virginia. In other situations in which a federal right has depended upon the interpretation of state law, “the Court has accepted the interpretation of state law in which the District Court and the Court of Appeals have concurred even if an examination of the state-law issue without such guidance might have justified a different conclusion.” Bishop v. Wood, 426 U.S. 341, 346, and n. 10 [96 S.Ct. 2074, 2078, and n. 10, 48 L.Ed.2d 684], citing, inter alia, United States v. Durham Lumber Co., 363 U.S. 522 [80 S.Ct. 1282, 4 L.Ed.2d 1371]; Propper v. Clark, 337 U.S. 472 [69 S.Ct. 1333, 93 L.Ed. 1480]; Township of Hillsborough v. Cromwell, 326 U.S. 620 [66 S.Ct. 445, 90 L.Ed. 358],
(Citations and footnote omitted; emphasis added.)
In Runyon and the cases cited in that opinion, there had been no disagreement between the district court and the court of appeals. The approach of the Supreme Court becomes more apparent in the case where the district court and the court of appeals disagree. Butner v. United States, 440 U.S. 48, 99 S.Ct. 914, 59 L.Ed.2d 136 (1979) is such a case. There the court of appeals, in a two-to-one decision, had reversed the district court on an issue of state law. The Supreme Court affirmed, stating “We decline to review the state law question.” Id. at 57-58, 99 S.Ct. at 919. The Court thus upheld the determination of the two-judge majority on the court of appeals even though it was contrary to that of the district judge and the dissenting appellate judge. The words chosen by the Court are important in that they accurately indicate what the Supreme Court is doing: it is not giving weight or deference to the decision of the district judge or the court of appeals, rather it is simply not reviewing the state law question that has been fully reviewed and determined by the intermediate appellate court.
Implicit in the Supreme Court’s practice of not reviewing the state law question is the assumption that it need not exercise its discretionary jurisdiction to do so because the appellate panel has exercised its mandatory appellate jurisdiction by giving full and independent review to the decision of the trial judge. This practice by the Supreme Court is a jurisprudential decision that recognizes the structure of the federal court system; the Supreme Court is free to conserve its discretionary powers because the parties have already been accorded full appellate review. To fulfill our role within the federal court *1400structure, we should review all legal issues, whether of state or federal law, under the independent de novo standard.
It is most appropriate, of course, that we respect the views of the district court judges and review with great care their determinations of both state and federal law. Indeed, this respect is inherent in the adversary system, which assigns to the appellant the duty to establish the errors in the trial court’s decision. The district judge is an expert in determining and applying the state law, but is no less an expert in determining and applying the federal law. There is no sound reason for according greater weight to the former than to the latter.
Our independent determination of state law should be based upon recognized sources that are available to the parties and that may be argued and contested before the district court as well as before the appellate court. These sources include the relevant statutes, legislative history, treatises, restatements, and published opinions. Our determination should not be based upon some undefined special knowledge or feeling for the state law that the district judge may be presumed to have, but that cannot be articulated by the judge, argued by the parties, or reviewed by the appellate court.
The deferential standard, whether the appellate court affirms in the absence of “clear error” or gives the “great weight” that the dissent would accord to all state law determinations by the district judge, is not based upon the reasoning and persuasiveness . of the judge’s decision, which is always entitled to careful consideration. Instead, it is based on an assumption that the district judge has some particular knowledge or experience in the field of law in issue that is to be given great weight apart from the authorities presented- by the parties or articulated by the district judge. This is an unsound ground on which to base our decisions. First, it depends upon ad hominem factors not part of the record. Second, it invites exploration of each judge’s actual experience. See, e.g., In re Big River Grain, Inc., 718 F.2d 968, 970 (9th Cir.1983) (per curiam) (comparing state court experience of district court judge and bankruptcy judge); Metropolitan Life Insurance Co. v. Ease, 718 F.2d 306, 307 (9th Cir.1983) (according less deference where panel members were former state court judges); Yamaguchi v. State Farm Mutual Automobile Ins. Co., 706 F.2d 940, 946 n. 5 (9th Cir.1983) (declining to accord deference where trial judge sitting by designation). Bringing such considerations into the decisional process is neither proper nor efficient. It shifts the focus from the appropriate legal authorities to the biography of the judge.
' The present case illustrates the point. The district judge is a respected jurist sitting in the state of Alaska, a former practicing attorney in Alaska, and a former state trial judge in Alaska. One of the members of the appellate panel is also a respected jurist, a resident of the state of Alaska, a former practicing attorney there, and a former Justice of the Alaska Supreme Court. It is not sensible to give “great weight” to the decision of one of these judges and not the other. More importantly, it misdirects the focus of the inquiry.
The application of the “clear error” standard or the dissent’s “great weight” standard certainly cannot be justified by any desire to conserve judicial resources or to promote administrative efficiency. The parties have a statutory right to appeal state law questions and we must undertake the same full and careful review of the pertinent legal authorities whether or not deference is to be accorded. We are the first level of appellate review of the trial court’s conclusions of law. There would be no justification for us to use deference to the trial judge’s determination as an excuse for cursory or more limited inquiry into the state law question. There is an entirely different situation when the Supreme Court reviews cases involving state law when there has been a thorough appellate review of the state law issue at a lower level.
*1401It is worth noting that if the parties were to proceed in a state court to litigate a state law issue, they would have the right to an independent de novo review of the trial judge’s determination by a multi-judge appellate panel with the structural advantages heretofore noted. The parties should be entitled to the same appellate consideration in the federal court system. There is no reason why an appellant should have a greater burden in seeking review of the trial court’s conclusions of law because he is in a federal forum.
The dissent seems to agree that independent de novo review of the district court’s conclusions of law is required, and yet it advocates a standard of deference to the decision of the trial judge. We cannot have independent de novo review and still defer to the decision of the district judge. The concepts are inconsistent. Either we defer, as we do under Fed.R.Civ.P. 52(a), or we reach an independent judgment, as we do on the district court’s conclusions of federal law. The dissent would shift from a “clear error” standard of deference to a new expression of that deference that gives “great weight” to the decision of the district court. Whether this would create more or less deference is difficult to say, but it is clear that in deferring to the decision of the district court we would not be according the parties an independent determination of the law. There is a very real distinction between deferring to the conclusions of the district judge, as opposed to considering the reasoning of the district judge with the respect that is certainly due. If we give “great weight” to the conclusion of the district judge simply because it is his or her conclusion, then we have not made an independent determination of the law.
The functions of the appellate court have traditionally been described as two-fold. The first is to review for correctness, which is the prime consideration of the parties. The second is the institutional function of announcing, clarifying, and harmonizing the rules of decision employed by the legal system in which they serve. See P. Carrington, D. Meador & M. Rosenberg, Justice on Appeal, 2-3 (1976). As noted in McConney, de novo review of questions of law is dictated in part because of the precedential effect of those questions on future litigants. While the trial courts’ factual determinations bind only the parties, the determination of legal issues affects the rights of future litigants. McConney, 628 F.2d at 1201. Although this aspect is not as important when we are concerned with state law as when we are concerned with federal law, because our opinions are not binding precedent, yet our determinations of state law are also of significant precedential importance. When we interpret state law under the doctrine of Erie v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed. 1188 (1938), we are bound to strive for uniform application of state law in state and federal courts. Fidelity Trust Co. v. Field, 311 U.S. 169, 179-80, 61 S.Ct. 176, 179, 85 L.Ed. 109 (1940); see also Note, Deference to Federal Circuit Court Interpretations of Unsettled State Law, 1982 Duke L.J. 704, 706, 732. “Diversity jurisdiction, especially in its post-Erie incarnation, should not create needless diversity in the exposition of state substantive law.” Factors etc., Inc. v. Pro Arts, Inc., 652 F.2d 278, 283 (2d Cir.1981), cert. denied, 456 U.S. 927, 102 S.Ct. 1973, 72 L.Ed.2d 442 (1982). Uniformity among federal interpretations of state law tends to create state-federal uniformity. See id. at 282. For this reason, federal court interpretations of undecided state law issues do have precedential value, see 1A J. Moore, Moore’s Federal Practice, § 0.309(2) at 3122, 3123 n. 19 (2d ed. 1978), and therefore deserve the attention we generally grant to the creation of precedent.
The precedential importance that our appellate determinations of state law can have can be illustrated by two Ninth Circuit cases. In 1969, the Ninth Circuit in Monolith Portland Midwest Co. v. Kaiser Aluminum & Chemical Corp., 407 F.2d 288, 293 (9th Cir.1969), decided the issue of when a particular cause of action arose under California law for purposes of applying the statute of limitations. This holding was cited and relied upon by the California Supreme Court in Davies v. Krasna, 14 *1402Cal.3d 502, 121 Cal.Rptr. 705, 711, 535 P.2d 1161, 1167 (1975). Both cases were then cited and relied upon in Whittaker Corporation v. Execuair Corporation, 736 F.2d 1341 (9th Cir.1984).
A second illustrative case is Scandinavian Airlines System v. United Aircraft Corp., 601 F.2d 425, 429 (9th Cir.1979), which involved an application of California’s strict liability law to two large corporations contracting with each other from positions of relatively equal strength. This holding was in turn cited and relied upon by the Second Circuit in construing California law in Tokio Marine & Fire Insurance Co. v. McDonnell Douglas Corp., 617 F.2d 936, 939 (2d Cir.1980). One or both of these two circuit cases have been cited and relied upon in numerous other cases construing California law.2 One recent Ninth Circuit case upheld the dismissal of a strict liability claim solely on the basis of the Scandanavian Airlines holding on this point of California law. S.A. Empresa de Viacao Aerea Rio Grandense (Varig Airlines) v. Walter Kidde & Company, Inc., 690 F.2d 1235, 1239-40 (9th Cir.1982). These cases are merely illustrative of the many cases in which we construe state law that are subsequently relied upon by other courts.
Thus it is clear that many of our decisions construing state law have significant precedential effect, even though they are not binding precedent in the state. Yet when closely analyzed, if those holdings are based on the clear error standard, they are not really holdings by appellate panels that construe state law. Instead, they are holdings that the district judges’ construction of state law is not clear error.3 We owe a greater appellate duty than this when we establish precedent to be relied upon by other trial and appellate courts. This deficiency in our present standard of review would not be alleviated by the standard proposed in the dissent. When an appellate panel is reviewing the decision of the district court under a deferential standard which accords “great weight” to the conclusions of a district judge, the holding of that panel is not an independent appellate construction of state law, but one that is greatly weighted toward the construction given by the district judge. With the significant precedential effect that our opinions construing state law can have, this affords another reason why our standard of review should be the same as the standard of review we utilize in determining questions of federal law and, indeed, the same standard utilized by the state appellate courts in reviewing the trial courts’ construction of their state law — that is, an independent de novo appellate review.
There is more agreement between the majority opinion and the dissenting opinion on this standard than at first might appear. The dissent agrees that the proper exercise of our appellate function is “de novo review over all questions of law, whether state or federal.” The dissent also agrees that “it is our duty to exercise our independent judgment” and that our appellate court review of state law questions should not be “narrower in character than its review of other legal questions.” The dissent concedes that we cannot properly perform this appellate function under our circuit’s prior “clear error” standard. Where the dissent parts company with the majority opinion is in the assertion that if we simply express this deference or weight given to the district court’s conclusion in *1403different terminology, we somehow meet this appellate obligation. In our view, if we give great weight or defer to the conclusion of the district judge, it is not an independent judgment, but a weighted or deferential judgment and the review is clearly under a narrower standard than questions of federal law.
Both the majority and the dissenting opinions agree that careful consideration should be given to the reasoned explanations of state law by the district court. The dissent completely mischara'cterizes the majority opinion when it states that we adopt a view that “the district court is entitled to no consideration.” It is difficult to understand how the dissent could arrive at this conclusion, when we have emphasized that the district court’s reasoned explanation for a holding on a question of state law will be given full, thorough, and respectful consideration, just as it is on questions of federal law. It is even more difficult to understand how the dissent could possibly conclude that the time district courts spend “thoroughly explaining a state law is time wasted” and that “we therefore will be deprived of their expertise.” This is certainly not the case with the district court’s determinations of federal law, to which the same standard applies. It is inconceivable that the highly competent and thorough district judges in this circuit would in any way be dissuaded from considering just as fully and fairly questions of state law as they do questions of federal law. In fact, because it is the reasoned explanation of state law that we will look to, rather than simply the district court’s conclusion, it would seem that there would be a strong incentive for thorough and well-reasoned explanations.
In conclusion, in order to perform properly the appellate function in the review of state law questions, we must apply the same standard of review as we do in determining all other questions of law. We see no justifiable reason to distinguish between the standard of review for determinations of federal law and of state law. We hold that our appellate review of conclusions of state law should be under the same independent de novo standard as conclusions of federal law.
The case is remanded to the original panel for disposition consistent with this opinion.
. Two of our recent cases indicate some departure from the "clear error” standard. See Insurance Company of North America v. Howard, 679 F.2d 147, 150 (9th Cir.1982) and Bank of California, N.A. v. Opie, 663 F.2d 977, 979 (9th Cir. 1981) (according little or no special deference where the district court relies on state law that offers only general guidance).
. Some of the cases relying upon Scandinavian Airlines are Airlift International, Inc. v. McDonnell Douglas Corp., 685 F.2d 267, 269 (9th Cir. 1982); Aeronaves De Mexico, S.A. v. McDonnell Douglas Corp., 677 F.2d 771, 772 (9th Cir.1982); S.A. Empresa de Viacao Aerea Rio Grandense (Varig Airlines) v. Boeing Company, 641 F.2d 746, 754 (9th Cir.1981); Purvis v. Consolidated Energy Products Co., 674 F.2d 217, 220, 224 (4th Cir.1982); International Knights of Wine, Inc. v. Nave Pierson Winery, Inc., 110 Cal.App.3d 1001, 1006, 168 Cal.Rptr. 301, 303 (Cal.Ct.App.1980).
. Theoretically, if a second district judge disagreed with the first district judge and the appellate panel has only held that the decision of the first judge was not "clear error,” the second judge could hold to the contrary and the appellate panel should affirm if the decision of the second district judge also was “not clear error.” This illustrates the fallacy of our present standard of review. As a practical matter, the circuit court opinion would probably be relied upon as though it were an independent de novo construction of state law, which it is not.