INTRODUCTION
T 1 We granted certiorari to review whether the court of appeals applied the correct standard of review when it reversed a juvenile court determination that Z.D., the seven-month-old son of S.B.D. and LL.D., suffered a fractured femur while in S.B.D.'s care and that Z.D. was an abused and neglected child under Utah law. We remand this case to the court of appeals for application of the standard of review as discussed below.
*402FACTUAL AND PROCEDURAL BACKGROUND1
12 On Saturday, November 16, 2002, S.B.D. (father) cared for his infant son Z.D. while his wife, LD. (mother), was away from the home. Before the mother left the home, she gave Z.D. a dose of Tylenol for an earache, constipation, and teething pains that Z.D. had been experiencing. Later, when Z.D. awoke from a nap, his father observed that he was fussy and guarding his left leg. The father attributed Z.D.'s apparent leg pain to a flu shot that had been administered in ZD's left leg the day before. He wrapped Z.D. tightly in a blanket in what the parents called a "burrito wrap." After being wrapped and held, the symptoms of fussiness subsided.
T3 That evening, when Z.D.'s mother returned home, she also noticed that Z.D. was favoring his left leg. The parents contacted the Kids Care clinic and were told that there was no need to worry and that Z.D. did not need an examination. That night Z.D. slept normally and evidenced no signs of fussiness or pain.
14 On Sunday morning, the father noted that Z.D. continued to favor his left leg. He took Z.D. to Primary Children's Medical Center (Primary) in Salt Lake City for an examination of the leg. An x-ray revealed that Z.D. had a fractured left femur. The record reflects that Z.D. was in no apparent distress when he was examined, but cried when his leg was manipulated.
T5 Primary follows certain procedures when a non-ambulatory child is diagnosed with a fracture. First, the hospital admits the child for further examination and treatment-Primary admitted Z.D. to the hospital. Next, it notifies the Division of Child and Family Services (DCFS) for investigation of abuse-Primary did so. However, this case posed complications for DCFS because the father was employed by DCFS as an in-home child welfare investigator. Therefore, DCFS assigned independent investigator Paul Dean to the case. Third, Primary's emergency room staff alerts the hospital's in-house child abuse investigation team, the Center for Safe and Healthy Families (CSHF), to investigate-Primary immediately notified CSHF of Z.D.'s condition.
T6 Dr. Bruce Herman, a member of the CSHEF and a pediatrician, examined Z.D. and concluded that the child likely sustained the fracture on Saturday, November 16, and probably became acutely symptomatic on the same day. Dr. Herman determined that the fracture was most likely the result of excessive axial loading-direct application of force or smashing-of the femur.
17 On Tuesday, November 19, 2002, Dr. Herman visited Z.D.'s room while Z.D.'s father, mother, and grandmother were present. The grandmother asked Dr. Herman if it were possible that the fracture took place on Wednesday, November 18, 2002, when Z.D.'s leg became stuck in a baby walker and she released it by pulling the leg through the hole in the walker. Dr. Herman did not discuss this theory with the family because, according to his testimony at trial, he did not consider it a likely cause of this type of fracture.
118 Approximately a month later, the family requested a meeting with Dr. Herman and the members of the CSHF team to further discuss their "walker theory." At this time, the grandmother provided further details about the incident. She said that when Z.D.s left leg became stuck in the walker his knee was bent. She then placed her left hand and thumbs on that leg just above the knee, pushed, and pulled it through the walker. Z.D. let out a shrill, vigorous ery when this happened but calmed down quickly.
T9 After the grandmother gave this account and a demonstration of her movements in extricating Z.D. from the walker, a group of physicians from Primary met and agreed that the walker theory and demonstration did not change their opinion that the break was the result of some other type of axial load.
{10 Kari Cunningham, the hospital lHaison to DCFS, testified that the doctors agreed *403that it was possible for a fracture to occur by use of one's hand on Z.D.'s leg, but that the force used by the grandmother in the walker was insufficient to cause Z.D.'s fracture. According to Ms. Cunningham, the doctors formed their opinion without considering either the Tylenol that Z.D. had been given, his tight wrapping in a blanket, or the reduced level of activity that Z.D. exhibited between Wednesday and Saturday.
11 Two additional physicians, Dr. William G. Nixon, a pediatric radiologist, and Dr. John Smith, a pediatric orthopedist, did not participate in the walker demonstration or subsequent meeting, but did propose conflicting opinions. Dr. Nixon had earlier expressed his opinion that the fracture was not made by an axial load, but rather by angular leverage. Dr. Smith explained that he thought the fracture was likely the result of force on the leg over a fulcrum.
12 The State then filed a petition pursuant to Utah Code section 78-8a-103(1)(a)@), alleging that Z.D. was an abused child, that pursuant to Utah Code section 78-Sa-Z.D. was a neglected child subjected to mistreatment or abuse, and that pursuant to Utah Code section 78-82a-108(1)(g)(E), Z.D.s sister, A.D., was at risk of abuse and neglect because she resided in the same home as Z.D. At trial, both sides offered dueling experts and testimony.
113 During trial, the State offered the expert opinions of several physicians, relying principally on the testimony and opinion of Dr. Herman, that Z.D.'s injury was not sustained accidentally. Dr. Herman grounded his opinion concerning causation on three diagnostic observations: the type of Z.D.'s fracture, which provided clues to the mechanism and force associated with the cause of the fracture; the age of the fracture; and the symptoms that were associated with Z.D.'s fracture.
T14 Based on his observations, Dr. Herman expressed the opinion that the father deliberately fractured Z.D.'s left leg on Saturday, November 16, 2002, by direct and excessive force applied to the bent left knee. Dr. Herman testified this was the only reasonable conclusion supported by the relevant diagnostic facts. His confidence in his conclusion was not, however, without qualification. While Dr. Herman rejected the walker theory, he conceded that he was "51/49" percent sure that it was direct and excessive force that caused the fracture. He also testified that, while he disagreed with Dr. Nixon's view that Z.D.'s injury was caused by angular force, his disagreement was "not significant."
115 Dr. Herman's view that the walker explanation was implausible in his opinion because the femur fracture would have caused Z.D. so much pain that neither the Tylenol nor the "burrito wrap" would have been sufficient to mask Z.D.'s discomfort from a "vigilant caretaker" between the date of the walker incident on November 13 and November 16, the date Dr. Herman believed the father had inflicted the injury. He did not explain Z.D.'s apparent absence of pain when brought to the hospital.
{16 In their defense, the parents offered the testimony of various experts including Mr. David Ingebretsen, a biomechanical engineer; Debbie Hosseini, a nurse with expertise in premature baby development; and Dr. Steven Scott, a pediatric orthopedic expert.
{17 Mr. Ingebretsen testified that the walker theory was consistent with his interpretation of the pattern of Z.D.'s fracture. Ms. Hosseini testified that she had often worked with Z.D. and his family because Z.D. was born premature. She testified that she had never seen any bruising or swelling and that Z.D. was a very happy baby who was easily consoled.
1 18 Finally, Dr. Seott testified that he was initially skeptical of the parents' explanation of Z.D.'s injury because in his experience a fracture in a non-ambulatory infant is generally the result of abuse. However, upon reviewing the medical records, including x-rays, Dr. Scott became persuaded that Z.D.'s fracture was, in fact, caused by the type of lateral force explained by Z.D.'s grandmother. Dr. Scott also testified that the absence of bruising was a significant issue because it is much easier to bruise soft tissue than to break a bone. Finally, Dr. Scott testified that it was not rare for a child to be brought to the hospital days after a fracture because *404parents may attribute the symptoms to other causes.
T19 The juvenile court issued findings of fact and conclusions of law. The court acknowledged the difficulty in the case due to its length, complex legal and factual issues, and the emotions of all the parties involved. The court concluded that Z.D.'s leg was fractured on Saturday, November 16, 2002, while he was in the exclusive custody of his father. The court acknowledged that the "details of how the child received the femur fracture has [sic] not been provided to the [cJourt but sufficient and clear and convincing evidence has been established for the [clourt to conclude it was non-accidental trauma without a reasonable and acceptable explanation from either parent as to its causation." The juvenile court determined that it was satisfied that the State had proven each of the allegations of its petition by clear and convincing evidence.
20 Z.D.'s parents appealed, claiming that the evidence was insufficient to support the juvenile court's result. The court of appeals reversed. The court of appeals reasoned that " TwJhile it rests primarily with the trial court to determine whether the evidence is clear and convincing, its finding is not necessarily conclusive'" S.B.D. v. State (In re Z.D.), 2004 UT App 261, ¶ 19, 98 P.3d 40 (quoting Lovett v. Cont'l Bank & Trust Co., 4 Utah 2d 76, 286 P.2d 1065, 1068 (1955)). The court went on to detail that
it is the duty of the appellate court in reviewing the evidence to determine, not whether the trier of facts could reasonably conclude that it is more probable that the fact to be proved exists than that it does not, ... but whether the trier of facts could reasonably conclude that it is highly probable that the fact exists.
Id. (emphasis in original) (internal quotation marks omitted).
1 21 The court of appeals also pared away the deference it afforded the juvenile court findings based on our holding in Alta Industries v. Hurst, 846 P.2d 1282, 1284 n. 2 (Utah 1998), that appellate courts do not give factual findings made by a trial judge the same deference as those made by a jury. Id. As a result, it explained that "an appellate court does not, as a matter of course, resolve all conflicts in the evidence in favor of the appel-lee." Id.
122 The court of appeals then put these misunderstood principles of review into practice. It concluded that the evidence, particularly the testimony of Dr. Herman, did not clearly and convincingly demonstrate that the fracture was caused by an axial load, or that it took place on Saturday, when Z.D. was in his father's care. In re Z.D., 2004 UT App 261, ¶ 26, 98 P.3d 40. We granted the State's petition for certiorari to review the standard of review adopted by the court of appeals.
" 23 We hold that the court of appeals used the incorrect standard of review and should, on remand, apply the "clearly erroneous" standard discussed below.
ANALYSIS
1 24 Appellate courts are removed temporally and geographically from trial courts. They do not see juries impaneled or oaths administered to witnesses. They do not view first-hand witnesses' "tells" of posture, infleetion, or mood that strengthen or erode credibility. It is the lot of appellate judges to take their sustenance from the printed page; to peer into the facts as deeply as the flat plane of paper will permit. By the time the trial transcript reaches the hands of the appellate judge, the universal adjective describing its condition is "cold." Thus, appellate courts have ample cause to defer to the judgment of trial judges on matters that cannot be reliably extracted and examined from such a two-dimensional record. The doctrine that shapes and guides judicial review is that it is not within the province of an appellate court to substitute its judgment for that of a front line fact-finder except when exceptional cireumstances warrant more rigorous serutiny.2
*405( 25 On this occasion, a unanimous panel of the court of appeals overcame its reluctance to disturb factual findings and reversed a juvenile court judge's determination that the father was responsible for fracturing the left femur of his seven-month-old son Z.D. The court of appeals evaluated the factual record for its sufficiency with the aid of a two-part test. In re Z.D., 2004 UT App 261, ¶ 19, 98 P.3d 40. First, the court looked to the burden of proof-clear and convincing evidence in this case-required to sustain the juvenile court's conclusion that Z.D. was an abused and neglected child. Id. (citing our opinion in Lovett v. Cont'l Bank & Trust Co., 4 Utah 2d 76, 286 P.2d 1065, 1068 (1955)). The court of appeals concluded that, where the degree of proof required to support a result exceeds a mere preponderance of the evidence, an appellate court may set aside an outcome based on the insufficiency of its factual basis even if one could reasonably conclude that it was more probable than not the fact to be proved existed. Id. The court of appeals read Lovett to require an appellate court to conduct the more searching inquiry into "whether the trier of facts could reasonably conclude that it is highly probable that the fact exists." Id. (emphasis in original) (internal quotation marks omitted).
126 The court of appeals coupled its view that appellate deference to factual findings was linked to burden of proof with the principle that lesser deference was due factual determinations made by a trial judge than factual findings incorporated within a jury's verdict. Id. (citing Alta Indus. v. Hurst, 846 P.2d 1282, 1284 n. 2 (Utah 1993)). With these implements, the court of appeals cleared the way to reverse the juvenile court decision.
{27 We acknowledge that both burden of proof and identity of the finder of fact may under appropriate cireumstances affect the degree of deference extended by an appellate court to findings of fact. These are, however, tools that are subordinate to the standard of review set out in rule 52(a) of the Utah Rules of Civil Procedure.
{28 Despite attempts by the father to characterize the question under review as a mixed question of law and fact, his challenge to the juvenile court's ruling was one of pure fact: was there sufficient evidence to establish the father's culpability for his son's injury?
{29 An appellate court must launch any review of factual findings from rule 52(a) of the Utah Rules of Civil Procedure and its "clearly erroneous" test, yet rule 52(a) goes unacknowledged by the court of appeals.
130 The absence of an attempt to explain its reasoning within the context of the "clearly erroneous" standard of rule 52(a) exposed the court of appeals to the charge advanced by the State in its brief that it conducted an appellate trial de novo, affording no deference to the juvenile court's factual findings or to its advantaged position with regard to determining credibility of expert and fact witnesses.
131 We reiterate the observation made above that the considerations which persuaded the court of appeals that it was justified in examining with more rigor the juvenile court's ruling-a heightened standard of proof and findings made by a judge instead of a jury-are relevant to a sufficiency of the evidence review conducted under rule 52(a). These are, however, subsidiary principles that exist only in the service of an accurate evaluation of whether findings were "clearly erroneous." They cannot be applied independently from a "clearly erroneous" inquiry.
' 32 We provided sound guidance to appellate courts regarding the seope of their authority to review factual findings under the clearly erroneous standard of rule 52(a) in State v. Walker, 748 P.2d 191 (Utah 1987). There we stated that
the content of [rJule 52(a)'s clearly erroneous standard, imported from the federal rule, requires that if the findings (or the trial court's verdict in a criminal case) are against the clear weight of the evidence, or if the appellate court otherwise reaches a definite and firm conviction that a mistake *406has been made, the findings (or verdict) will be set aside.
Id. at 198.
133 Several interpretative observations may be fairly drawn from the Walker commentary on rule 52(a). These include the recognition that it is simply unrealistic to expect an appellate court to conduct a review of a sufficiency of the evidence challenge without weighing evidence. The appellate court must, however, go about weighing the evidence with one eye on the scales and the other fixed firmly on its duty of deference to findings of fact. Thus, it may only disturb findings that offend the "clear weight" of the evidence. An appellate court must be capable of discriminating between discomfort over a trial court's findings of fact-which it must tolerate-and those that require the court's intercession. It must forebear disturbing the "close call."
134 Judge Richard Posner made an able attempt to give practical meaning to the rule 52(a) standard when he wrote:
It requires us appellate judges to distinguish between the situation in which we think that if we had been the trier of fact we would have decided the case differently and the situation in which we are firmly convinced that we would have done so. Our serutiny of the district judge's findings of fact thus is deferential, but it is not abject. As the Supreme Court pointed out in the Concrete Pipe case, we need not, to overturn a finding under the clear-error standard, adjudge the finding "so unlikely that no reasonable person would find it to be true."
Carr v. Allison Gas Turbine Div. Gen. Motors, 32 F.3d 1007, 1008 (7th Cir.1994) (emphasis in original) (quoting Conerete Pipe & Prod. of Cal., Inc. v. Constr. Laborers Pension Trust, 508 U.S. 602, 623, 113 S.Ct. 2264, 124 L.Ed.2d 589 (1993)).
1 35 It is the role of subsidiary principles, like the ones employed by the court of appeals here, to help make principled distinctions between a belief that a trial court was mistaken and a firm conviction of error. Thus, an appellate court may consider whether the findings were made by a judge or by a jury. This distinction matters. An appellate court must indulge findings of fact made by a jury that support the verdiet. No such indulgence is required of findings made by a judge. We made this point in Walker in the context of interpreting rule 52(a) when we looked favorably to the position adopted by Wright & Miller concerning federal rule 52(a) that
"Lt is not accurate to say that the appellate court takes that view of the evidence that is most favorable to the appellee, that it assumes that all conflicts in the evidence were resolved in his favor, and that he must be given the benefit of all favorable inferences. All of this is true in reviewing a jury verdict. It is not true when it is findings of the court that are being reviewed."
Walker, 743 P.2d 191, 193 (quoting Wright & Miller, Federal Practice and Procedure § 2585 (1971) (citations omitted)).
T 36 The degree of skepticism to which an appellate court subjects a factual record can significantly affect the firmness of a court's conviction that a trial court was mistaken and influence its reading of the scale that weighs those facts. Yet, neither this principle nor any of its associated subsidiary rules of factual review can substitute for rule 52(a) analy-sig.
137 It is with some trepidation that we attempt to elaborate further on the dimensions of the Walker operational guide to determining what is "clearly erroncous." We will mention, however, two issues of importance that bear on this case: the duty to conduct a review of the "whole record" where a sufficiency challenge is made, and the pro-pricty of considering heightened standards of proof as part of a "clear error" review.
{38 Like rule 52(a) itself, the interpretive tool that we announced in Walker was borrowed from federal jurisprudence. Id. (citing Wright & Miller, Federal Practice and Procedure § 2585 (1971)). Walker acknowledged as much when it adopted "the definition of 'clearly erroneous' in the federal rule" as defined in United States v. United States Gypsum Co., 333 U.S. 364, 68 S.Ct. 525, 92 L.Ed. 746 (1948). The Gypsum definition explained that "[a] finding is 'clearly errone*407ous' when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed." Id.
T 39 The Gypsum definition, which has lost none of its vitality over the decades, includes the important recognition that the seope of review contemplated by rule 52(a) encompasses the entire factual record.3 It is, therefore, the responsibility of the appellate court to provide some indication that it performed its sufficiency of the evidence review in the context of the whole record, or at least that portion of the record to which its attention was drawn by the appellant's marshaling obligation or the appellee's response to the appellant's marshaled evidence.
'I 40 It is also appropriate when evaluating whether a result was "clearly erroneous" for the reviewing court to consider the standard of proof the prevailing party below was required to meet. See Lovett, 286 P.2d at 1067. The Walker standard requires reversal under rule 52(a) when a result is "against the clear weight of the evidence." Walker, 748 P.2d at 193. The amount and quality of evidence required to sustain a result based on a preponderance of evidence is, of course, less than that required to meet a clear and convincing standard. Accordingly, it is more difficult to demonstrate that the evidence supporting a "preponderance" based outcome is so wanting as to be "clearly erroneous" than it is to demonstrate the required dearth of evidence under a more exacting evidentia-ry standard. Still, it is not the role of the appellate court to reverse a trial court merely because it is convinced that the evidence is inadequate to sustain the result under the standard of proof applied below. Walker demands more. The result must be against the clear weight of the evidence or leave the appellate court with a firm and definite conviction that a mistake has been made.
1{ 41 We stop short of applying the proper standard to the facts in this case, electing instead to remand the matter to the court of appeals. By finding the standard of review applied by the court of appeals to be flawed, we do not intend to suggest that its result is likewise in error. To the contrary the record as a whole may well leave the court of appeals with a firm conviction that the juvenile court was mistaken in concluding that the State had established each of the allegations of its petition by clear and convincing evidence. '
4 42 For example, the juvenile court's conclusion that Z.D.'s fractured femur was the result of abuse by his father on Saturday was based primarily on its finding that Z.D. was asymptomatic from Wednesday, the day of the walker incident, until Saturday.4 But this finding does not account for an abundance of evidence that explains the absence of symptoms, including the painkillers Z.D. had been taking at the time and the "burrito wrap," which stabilized Z.D.'s leg. The finding also leaves unexplained the uncontradiet-ed testimony that Z.D. was in no apparent distress when he was first examined at Primary Children's Hospital. He was described by hospital workers as cheerful, interactive, alert and slightly fussy, but consolable. In fact, the first physician at Primary observed 7.D. independently move his leg. Even after moving Z.D.'s leg around, the physician could not find anything wrong. Far from being the "sudden and demonstrative appearance" of symptoms that "dumbfounded" the juvenile court, the record reveals that Z.D.'s distress was transient and muted.
{43 Moreover, there is secant evidence regarding the causation of Z.D.'s injury. The State's medical expert testified that he was only "51/49" percent certain that the injury was caused by axial force, which was consistent with the State's theory of the case, rather than angular leverage, which was con*408sistent with the parents' explanation of the walker incident. And while the defense's expert admitted the possibility that Z.D.'s injury could have been caused by axial force, he testified that the absence of any bruising, swelling, and contusions surrounding the fracture rendered this mechanism of injury highly unlikely.
1 44 Finally, the juvenile court was unable to reach any conclusion about what actually caused Z.D.'s injury. Instead, the court relied entirely on Utah Code section 78-Sa-305.1, which created a presumption that any abuse or neglect suffered by a minor was caused by the person having the minor "under his direct and exclusive control" at the time of the abuse.5 Inasmuch as the State presented no evidence of how the abuse occurred, the juvenile court's conclusion that the broken femur was the result of abuse levied by Z.D.'s father, rather than an accidental injury, was entirely dependent upon this presumption. We note that subsequent to the court of appeals' opinion, the Utah Legislature repealed this presumption. See 2005 Utah Laws 1889 (H.B.89).6 Thus, we ask the court of appeals to determine, under the standard announced today, whether the complete lack of evidence-let alone clear and convincing evidence-regarding the mechanism of Z.D.'s injury, coupled with the juvenile court's failure to account for the evidence that Z.D. was in no apparent distress upon his arrival at Primary supports a reversal of the juvenile court.
CONCLUSION
4 45 Because it is beyond the scope of the issue we agreed to take up on certiorari, we decline to comment on whether we believe that the juvenile court's findings and result satisfy rule 52(a). We therefore remand this matter to the court of appeals for the purpose of applying the "clearly erroneous" standard as illustrated in Walker.
146 Chief Justice Durham, Justice Dur-rant, and Justice Parrish concur in Justice Nehring's opinion.
. The sole issue before us concerns the standard of review adopted by the court of appeals. Our recitation of the facts should not, therefore, be taken to reflect this court's view on the sufficiency of those facts under the standard of review we articulate in this opinion.
. Examples of circumstances in which we extend less deferential review to findings of fact include our review of challenges to searches and seizures conducted without a warrant, issues of "constitutional fact' implicating First Amendment guarantees, and punitive damage awards, which by *405their nature carry the risk of infringing due process protections. Jensen v. Sawyers, 2005 UT 81, ¶ 61, 130 P.3d 325; State v. Brake, 2004 UT 95, 103 P.3d 699; Campbell v. State Farm Mut. Auto. Ins. Co., 2004 UT 34, 98 P.3d 409.
. The Gypsum language "has become standard fare in subsequent [U.S.] Supreme Court opinions. [Moreover,] [mJlost of the cases in the circuits that define clearly erroneous also routinely follow the Gypsum formula." Childress & Davis, Federal Standards of Review § 2.05 (3d ed.1999).
. The juvenile court's finding reads: "The [cJourt is both astonished and dumbfounded why the symptoms would be absent on Wednesday [through Saturday morning], but yet make such a sudden and demonstrative appearance on the afternoon ... after being in the sole care of the father all day."
. It provided:
In determining whether a minor is an abused child or neglected child it may be presumed that the person having the minor under his direct and exclusive care and control at the time of the abuse is responsible for the abuse or neglect.
Uiah Code Ann. § 78-3a-305.1
. As the presumption was simply an evidentiary rule that did not "enlarge, eliminate, or destroy" a substantive right of the parties, its repeal is to be applied retroactively. Goebel v. Salt Lake City S. R.R. Co., 2004 UT 80, ¶ 39, 104 P.3d 1185. Even when existing procedural rules have been followed at trial, a party whose case has not yet reached final judgment has the right to take advantage of changes in those procedural rules. See B.A.M. Dev. v. Salt Lake County, 2006 UT 2, ¶¶ 17-21, 128 P.3d 1161 (applying a procedural statute retroactively to an appeal).