The personal representative of Lonnie Bell’s estate sued the United States under the Federal Tort Claims Act, 28 U.S.C. §§ 1346(b), 2671-2680. The plaintiff claimed that Bell’s death was caused by the failure of physicians at the Veterans Administration Medical Center, where Bell was hospitalized, to diagnose timely Bell’s abdominal aortic aneurysm. After a bench trial, the district court found the plaintiff had established that the Veterans Administration doctors breached their duty of care toward Bell. However, the court also found that the plaintiff failed to prove that Bell had a better than 50% chance of surviving an operation to correct the aneurysm on and after the date that the duty of care was breached. Thus, the court concluded that even if Bell’s condition had been diagnosed when a careful doctor would have diagnosed it, the evidence indicated that Bell’s life would not “have been spared dr prolonged....” Accordingly, the court entered judgment for the defendant.
We reverse. Our review of the record has left us with the “ ‘definite and firm conviction’ ” 1 that the key finding of fact relied upon by the district court in entering judgment for the defendant is clearly erroneous. The district court found that “significant leaking” of the aneurysm had occurred by December 24, 1983, the date by which the court also determined that the duty of care was breached. Given such leaking, the district court concluded that the probability Bell would have survived an operation to correct the aneurysm was less than 50%.
We accept the district court’s finding that the' duty of care was breached by December 24, 1983, as that finding is sup*883ported by the record. However, while there is evidence that some leaking had occurred by December 24, and that leakage is a significant event in the development of an aneurysm, we find no evidence which supports the conclusion that the leakage of Bell’s aneurysm had risen to such a level that there was not a “reasonable probability” that Bell would have survived an operation to correct the condition in the days immediately after the duty of care was breached.
Moreover, we do not interpret Michigan law as requiring a plaintiff to prove that the deceased had a better than 50% chance of survival when the duty of care was breached in order to establish that the negligent failure to diagnose a medical condition was the proximate cause of death or injury. All Michigan requires is proof that the deceased had a “reasonable probability” of recovery if the medical condition had been discovered and treated within the appropriate time. Although the evidence concerning Bell’s chance for recovery after December 24, 1983, was confusing, the evidence clearly placed his prognosis near 50%. Accordingly, the district court erred when it entered judgment for the defendant on the ground that Bell did not have a better than 50% chance of surviving an operation to correct the aneurysm once the duty of care was breached.
I
Bell was admitted to the Saginaw Veterans Administration Medical Center on November 21, 1983, for surgical treatment of a prostate gland condition. Except for a period of approximately twenty-four hours, Bell remained hospitalized at the Center until his death on January 16, 1984.
Dr. John A. Feemster, a board-certified doctor in general and thoracic cardiovascular surgery, testified extensively at the trial on behalf of the plaintiff. He stated that he is familiar with the diagnosis and surgical treatment of abdominal aortic aneurysms. Dr. Feemster’s testimony focused on when the standard of care was breached and whether the breach of that standard was the proximate cause of Bell’s death.
According to Dr. Feemster, “[a]neu-rysms are sacs formed by weakness in the wall of a retaining blood vessel. This weakness can occur by several different ways but it results in a ballooning out of the blood vessel....” The pressure an aneurysm causes in a blood vessel can make the blood vessel rupture which in turn can cause a person to bleed to death.
According to Dr. Feemster, abdominal aortic aneurysms are especially common in black males between the ages of 50 and 70. Bell was a 55-year-old black male. Dr. Feemster testified that in light of these facts, and the facts that Bell had a history of peripheral vascular disease and persistently complained of symptoms indicating an abdominal aneurysm while hospitalized at the Center, his doctors should have diagnosed his aneurysm “within a period of a week or two weeks.... [from the time of] the first development of symptoms.”
Dr. Feemster testified, based on Bell’s medical records, that Bell first complained of low abdominal and low back pain on December 9, 1983. Dr. Feemster identified these complaints as the first signs that Bell had an aneurysm. From December 9 through mid-December, 1983, Bell continued to complain of low back pain and also of “pain radiating out to his flanks.” According to Dr. Feemster, pain in the back and lower abdomen are “pretty classic” symptoms of an aneurysm. He explained that an aneurysm
can cause pain just from [the] pressure [it creates] alone. Pain can also, of course, be caused by the [blood] leakage or what is called dissection, and dissection is the process by which the blood does not leak directly out of the blood vessel into the abdominal cavity, but leaks between the walls of the blood vessel.
Dr. Feemster stated that according to the medical records, on December 23, 1983, Bell complained that the pain in his back was starting to radiate down into his sacrum. The following questions and answers *884from the trial transcript explain the significance of this event:
Q. What significance, if any, does radiating pain into the sacrum have?
A. Well, radiating pain ... would lead you to believe that the patient is starting to leak....
******
Well, when we say leak, we're talking about leakage versus open rupture.
Let’s say you have a cup of water and you can have a pinpoint hole in that cup of water and the water leaks, drips through that small hole.
As that hole enlarges, then the water gushes out.
And that would be considered a rupture.
******
THE COURT: Like a slow flat tire; is that it?
A. That’s a good analogy.
******
.... When the hole gets big enough, it’s a blowout, and then it collapses right away.
Where with the slow leak, the tire can go for days, and then you notice it as you start to lose air.
THE COURT: As this leaks out, that means the blood that’s down below there is actually coming into the abdominal cavity?
A. Usually what happens in a slow leak — now this is why the patient lives ... [is] that it gives you warning and gives you enough time to get something done.
Dr. Feemster testified that the records indicate Bell’s pain was worsening between December 28, 1983, and January 2, 1984. On January 4, it was noted that Bell had lost 251/2 pounds since his admission to the Center. Dr. Feemster stated that weight loss is consistent with an aneurysm because the abdominal pain it causes also causes loss of appetite.
On January 12, 1984, Bell was noted as having a fever and had suffered what Dr. Feemster characterized as “a significant loss of blood” based on the hemoglobin level recorded that day. Dr. Feemster testified that with the addition of this blood loss, “your diagnostic possibility is almost 100 percent that you’re dealing with an abdominal aortic aneurysm which is leaking.” Dr. Feemster testified that the hemoglobin level recorded on January 12 indicates that the aneurysm ruptured on that date. At some time during the next few days, Bell suffered a cardiac arrest and, on January 16, 1984, at 5:58 a.m. he was pronounced dead.
Dr. Feemster explained that an aneurysm need not be a fatal condition because it can be treated surgically. However, his testimony was somewhat confusing regarding the probabilities for a successful operation dependent upon the stage of the aneurysm’s development when the operation is performed. His clearest testimony concerning this issue was that the probability for success of an operation performed before a rupture occurs is, depending on where the aneurysm is located, “anywhere from 50 to 75 percent, and even higher in some skilled hands.” Once a rupture occurs, however, a person’s chance of surviving the surgery is “less than 30 percent, probably even lower than that.”
Dr. Feemster also testified that the probability a person will survive an operation decreases as leakage from the aneurysm increases. He never testified, however, as to what the probability of success is after “significant leaking” has occurred.
When asked the question, “Based upon a reasonable degree of medical probability, would Mr. Bell have survived the surgery to repair the aneurysm if it had been performed prior to rupture?” (emphasis added), Dr. Feemster answered, “Yes.” He later added that Bell’s medical records indicate that if his aneurysm had been repaired, Bell would have lived “a normal life span.”
However, during Dr. Feemster’s cross-examination, the following exchange took place:
Q. [By the Assistant United States Attorney]: Now, once the aneurysm begins to leak, then the likelihood of sue-*885cess at surgery goes way down; is that correct?
A. [By Dr. Feemster] Well, it starts to diminish. The most calamitous event is rupture.
Once you have a free rupture, of course, that is the worst scenario that you can get, but you can have a leaking aneurysm, you operate on that aneurysm within a reasonable period of time before there is a rupture. Even though you’re diagnosing it in an urgent manner, it is not an emergency situation.
So it is considered an elective situation.
Q. Before it leaks, an aneurysm is easier to operate on; correct?
A. It is the easiest.
Q. Those are the figures that you relied on when you said 50 percent, 50 percent are successful?
A. Yes.
Q. After it begins to leak, then the percentage drops off after that?
A. Starts to drop off, yes.
Q. After it ruptures, then it’s real low? (emphasis added)
A. Very low.
Q. Less than 50 percent?
A. Yes.
Q. So after it begins to leak, then the percentage is below 50 percent; is that correct? (emphasis added)
A. Yes.
Later, during redirect examination, Dr. Feemster again stated that he believed Bell could have survived an operation to correct the condition if it had been performed before the aneurysm ruptured:
Q. [By plaintiff’s attorney] If Mr. Bell had been operated on prior to the frank rupture, would there be a reasonable degree of medical probability he would have survived?
A. Yes.
However, during the recross examination which immediately followed this answer, another series of questions and answers complicated the doctor’s testimony:
Q. [By the Assistant United States Attorney] But the probability, Dr. Feem-ster, depends on whether the aneurysm is leaking or not; is that a variable factor?
A. That’s one variable factor.
Q. Before it leaks, it’s 50 percent or better. After it leaks, it’s less than 50 percent? (emphasis added)
A. Yes.
Relying particularly on this last exchange during the recross examination, the district court concluded that the date of significant leakage of the aneurysm was the date that Bell’s chance for survival dropped below 50%. The court acknowledged that Dr. Feemster had testified that Bell would have survived an operation so long as it was performed before the aneurysm ruptured. However, the court concluded that the doctor’s testimony regarding the effect of leakage meant that “[although Dr. Feemster believed that plaintiff’s decedent had a ‘reasonable degree of medical probability’ of survival, he clearly defined this chance as being less than 50 percent” after “significant” leakage had occurred. The district court determined that significant leakage had occurred “by December 24th.” Since the court already had found that the duty of care had been breached by that same date, the court, concluding that the plaintiff had not met the burden of proof, entered judgment for the defendant.
II
The alleged negligence in this action occurred in Michigan. Consequently, the Federal Tort Claims Act mandates that Michigan law governs the substantive aspects of this case. See 28 U.S.C. § 1346(b). Federal law governs the procedural issues involved, however, and thus, under Fed.R. Civ.P. 52(a), the district court’s findings of fact, “shall not be set aside unless clearly erroneous_” As this court recently noted,
[a] district court’s finding of fact may not be set aside lightly, nor because of a mere difference of opinion as to credibili-ty_ However, ‘findings that are unsupported or arbitrary’ should be set aside. “ ‘[A] finding is “clearly errone*886ous” 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.’ ” Under such circumstances a reviewing court has the power, and the duty, to overturn findings of a district court.
Batts v. NLT Corp., 844 F.2d 331, 336 (6th Cir.1988) (citations omitted).
This court adopts the position that “determinations of negligence and causation present mixed questions of fact and law ... [and] has long held the view that a district judge’s findings regarding negligence and causation are subject to the clearly erroneous standard.” Hasler v. United States, 718 F.2d 202, 204 (6th Cir.1983), cert. denied, 469 U.S. 817, 105 S.Ct. 84, 83 L.Ed.2d 31 (1984) (collecting cases).
Thus, acknowledging the high level of deference we must afford the district court upon review of this case, we nevertheless have been left with the “definite and firm conviction” that a factual mistake has been committed. In addition, we conclude that the district court erred in its interpretation of Michigan law.
Before discussing these errors, however, we note that, unlike appellant, we do not find the district court’s finding that the duty of care was breached “by December 24th” to be clearly erroneous. Dr. Feemster testified that Bell’s condition should have been diagnosed, at the most, two weeks from the date that the first symptoms of the aneurysm appeared. Dr. Feemster identified those symptoms as first appearing on December 9, 1983. Two weeks from that date is December 23,1983. The district court found that the duty of care was breached “by December 24th.” (emphasis added) We interpret this statement as merely another way of finding that as of December 23,1983, doctors exercising a reasonable standard of care, would have diagnosed Bell’s aneurysm. Bell’s doctors had not diagnosed his condition by that date and, thus, the district court correctly found that they breached their duty of care toward Bell “by December 24th.”
The district court incorrectly found, however, that leakage from Bell’s aneurysm was so significant by December 24, 1983, that his chance of surviving an operation to correct the condition was less than 50%. The district court correctly noted that during the cross and recross examinations of Dr. Feemster by the Assistant United States Attorney, Dr. Feemster agreed with the proposition that the chance of surviving an operation after the aneurysm begins to leak is less than 50%. However, on both occasions, the questions asked and answered immediately before the pertinent exchanges indicate that Dr. Feemster meant to imply that after the aneurysm ruptures, the probability for a successful operation is less than 50%. For example, during the cross-examination, the questions and answers, quoted in the order they were put to Dr. Feemster, were:
Q. After it [the aneurysm] ruptures, then it’s [the percentage is] real low?
A. Very low.
Q. Less than 50 percent?
A. Yes.
Q. So after it begins to leak, then the percentage is below 50 percent; is that correct?
A. Yes.
Similarly, during the redirect examination, Dr. Feemster answered “Yes” to the question of whether Bell had a “reasonable degree of medical probability of survival” if he had been operated on “prior to the frank rupture.” Then, during the recross examination which immediately followed this exchange, the Assistant United States Attorney asked, and Dr. Feemster gave, the following questions and answers:
Q. But the probability, Dr. Feemster, depends on whether the aneurysm is leaking or not; is that a variable factor?
A. That’s one variable factor.
Q. Before it leaks, it’s 50 percent or better. After it leaks, it’s less than 50 percent?
A. Yes.
We find these exchanges troubling because they are unclear as to whether Dr. Feemster’s testimony was that after leak*887age occurs, or after rupture occurs, the probability for a successful operation drops below 50%. Thus, we conclude that it was reversible error for the district court to have relied on these exchanges for its finding that once “significant” leakage occurred, Bell’s chance for recovery was less than 50%.
Still, if these exchanges were our only concern, we might be inclined to defer to the trial court’s role as the recipient and weigher of the evidence. However, we conclude that a reversal is warranted because of our additional conviction that no evidence supports the district court’s determination that “significant” leakage had occurred by December 23, 1983, the time the duty of care was breached. Indeed, when considered in its entirety, Dr. Feemster’s testimony indicates that, due to the fact Bell’s aneurysm appears to have been leaking relatively slowly, there may well have been a reasonable probability that Bell could have survived an operation during the period immediately after December 23, 1983.
Dr. Feemster stated that the spread of pain throughout the body is an indication that the aneurysm is leaking. He cited the notation entered on December 23, 1983, that pain was radiating into Bell’s sacrum, as a significant event in the course of the aneurysm’s development. While discussing the significance of the pain’s spread into the lower part of Bell’s body, Dr. Feemster emphasized that the slow spread of the pain “gives you a warning and gives you enough time to get something done.” Moreover, during cross-examination Dr. Feemster testified that, “you can have a leaking aneurysm, you can operate within a reasonable period of time before there is a rupture. Even though you’re diagnosing it in an urgent manner, it is not an emergency situation.”
Although there was some evidence that the aneurysm may have started to leak approximately 9 days earlier due to Bell’s complaints of “radiating pain,” Dr. Feem-ster never characterized the spread of pain by December 23 as being indicative of leakage so “significant” that Bell’s chance of surviving an operation after that date was less than 50%. Indeed, his testimony on both direct and cross examination indicates that after leakage occurred, and possibly up until January 12, 1984, the day Dr. Feemster testified that the aneurysm ruptured, the probability for a successful operation may well have been, if not greater than 50%, at least within the realm of “reasonableness” required under Michigan law.
Ill
The foregoing discussion leads us to the legal error in this case, the district court’s interpretation of Michigan law. At the end of the trial, before the district court adjourned to consider the evidence, the court stated,
[Rjecoverability depends upon whether or not [the failure to diagnose the aneurysm] was the proximate cause of his death. And under Michigan law, the proximate cause of his death does not— one cannot recover for the loss of the valuable right of a possibility of living through surgery.
If, for example, somebody told me or you tomorrow that we had a fatal disease but we had a 45 percent chance of successful surgery, we wouldn’t be very thrilled by it; but we would prefer the 45 percent to the alternative. And I would rather have that 45 percent than I would to win the Michigan Lottery this week; and yet, under the Michigan Law that loss of that recovery — that loss is not recoverable.
The district court did not cite any Michigan law at the time it made this statement. Later, in its written memorandum and opinion it cited two Michigan cases for the proposition that “[i]t is the plaintiff’s burden under Michigan law to show that the decedent would have had a 51%[2] or better chance of survival from such an opera*888tion,” Harvey v. Silber, 300 Mich. 510, 2 N.W.2d 483 (1942), and Vitale v. Reddy, 150 Mich.App. 492, 389 N.W.2d 456 (1986).
We have examined both of these cases, as well as other Michigan precedent, and have failed to find any mention of the rigid 50% or 51% standard that the district court relied upon. Harvey is cited in Vitale and appears to remain the Michigan Supreme Court’s most definitive statement concerning the burden of proof in an action for a negligent failure to diagnose. In Harvey, a physician misdiagnosed the course of a bullet in a patient and, therefore, failed to determine that the intestine had been pierced and that an operation was necessary to stop hemorrhaging. An expert witness testified that the “probability” that the patient’s life would have been saved by such an operation “was fairly good.” Harvey, 2 N.W.2d at 486. The Michigan Supreme Court noted that this testimony supported the jury’s determination that the negligent diagnosis was the proximate cause of the patient’s death and upheld a jury instruction which stated that the plaintiff did not have “to show that to a certainty surgical intervention would have saved [the deceased’s life]. It is sufficient if the plaintiff by a preponderance of the evidence has satisfied you that surgical intervention would with reasonable probability have saved his life.” Id. at 488. At no point in its opinion did the Michigan Supreme Court state that a “reasonable probability” requires proof of a 50% or greater chance of survival, nor did it discuss any numerical probabilities or indicate that evidence was received on that issue.
Vitale v. Reddy, 389 N.W.2d 456, also did not specifically address what degree of certainty constitutes proof that surgical intervention would, with “reasonable probability,” have saved a person’s life or explain how the testimony assessed the deceased’s chance for survival. However, in a footnote, the state appellate court acknowledged that “[p]roof of probability” continues to be the standard in Michigan3, citing Harvey and an even older Michigan Supreme Court case, Rogers v. Kee, 171 Mich. 551, 137 N.W. 260 (Mich.1912). Vitale, 389 N.W.2d at 460 n. 3.
Rogers also involved the negligent failure of a physician to diagnose a condition, which, in that case, was a fracture of the neck and left femur. The Michigan Supreme Court affirmed the jury’s judgment for the plaintiff. In its opinion, the state court rejected requiring a rigid quantitative standard of proof in these cases, stating,
the testimony given by the various physicians ... imports clearly that a patient suffering from such an injury on calling a physician is entitled to approved methods of treatment from which experience of the profession indicates beneficial results are probable and to be anticipated; and if not an entire recovery, a better ultimate condition than if left to chance.... We think such testimony presents an issue of fact for the jury — on probability, it is true. The issues of sickness and healing, life and death, are too uncertain to be otherwise forecast, but negligence which deprives a man of such probability is more than injuria sine dam-no.[4]
Rogers, 137 N.W. at 265.
In yet another later case, Stewart v. Rudner, 349 Mich. 459, 84 N.W.2d 816 (1957), the Michigan Supreme Court stated even more clearly that the plaintiff’s burden of proof in an action for negligent failure to diagnose does not rest upon proof of “mathematical” certainties. In affirming the jury’s determination that the doctor’s failure to perform a Caesarean section delivery “caused” the plaintiff to deliver a stillborn child, the court stated,
[the jury was] justified in finding ... that if plaintiff had been sent to surgery ... her baby would have been delivered alive. There is, of course, no absolute certainty thereof. The knife might have slipped or the child might have died at any instant because of forces beyond our *889control, beyond, possibly, even our comprehension. But absolute certainty we do not demand, particularly where there can be no certainty from the very nature of the case. The jury’s calculus does not demand certainties but only probabilities. As we observed, the defendant misconstrues the legal issue. The burden was not, as he asserts, upon the plaintiff to show that the failure to operate “caused” the loss of the baby. The plaintiff’s burden was lesser. It was her burden to show to the jury with reasonable {not mathematical) certainty that if the defendant had caused the operation to be performed, in accordance with his agreement, the surgeon would then have delivered [the baby] alive.
Stewart, 84 N.W.2d at 825-26 (emphasis supplied).
These cases illustrate that the Michigan Supreme Court long ago rejected the notion that a plaintiff must prove with mathematical certainty that had a physician diagnosed and treated a medical condition within the time period that a doctor exercising a reasonable standard of care would have done so, the probability that the patient would have recovered was better than 50%. Rather, under Michigan law, a “reasonable probability” is proven if a finder of fact legally could conclude from the evidence presented that the patient had, in the words of the expert witness in Harvey, a “fairly good” chance of surviving the operation. Although we acknowledge the imprecision which nearly always is the necessary companion of any legal standard that rejects quantitative analysis, Michigan law leaves no doubt that it has rejected such analysis in favor of a qualitative one.
Apart from indicating that some phrases are synonymous, the Michigan Supreme Court has never expressly defined what a “reasonable probability” is. Admittedly, an ambiguity of language affects our reading of the various cases and statements on this issue. To a statistician, the term “probability” means simply how likely something is, and can vary anywhere from zero (meaning, in truth, no probability at all, or impossibility) to one (meaning certainty). This is also the usage in common speech, where we may speak of a “small” probability (meaning much less than half), a “pretty good,” “reasonable,” or “fair” probability (meaning something in the middle range, perhaps a little more or less than one-half), or a “strong,” “overwhelming,” or “very great” probability (meaning something that will happen most of the time, well over half).
The other usage, derived from the definition of probability as “the state of being probable,” implies that rather than a continuum of states there are only two: a probability or not a probability. This is the 50+% test. Thus, if we attempt to forecast a baseball game between the Yankees and the Red Sox, and we believe it is more likely than not that the Red Sox will win, in this usage we will say that there is a “probability” that the Red Sox will win and we must then say that there is “not a probability” that the Yankees will win. On the other hand, the statistician employing the first definition might say that there is a 40 or 45 or 48% “probability” that the Yankees will win and a corresponding “probability,” slightly greater than half, that the Red Sox will win. The common sense usage might be that there is a “reasonable probability” that either team could win.
We recognize that when making a choice between a “bright line” cut-off of 50%, and a more amorphous standard of “reasonable probability,” we only have shifted the problem slightly, not eliminated it. In either case, a plaintiff falling below the threshold will obtain no recovery, even though something of value has certainly been lost. On the other hand, recovery for the full loss, when only some lesser chance for it has been taken away, may appear unjust as well.
In the current circumstance, however, we are not called upon to determine the ultimate measure of damages in a case such as this. We decide only that the Michigan standard, as evidenced in Harvey, does not require mathematically exact testimony that a patient had a better than 50% chance *890of a successful outcome. Instead, evidence that there was a reasonable probability of such outcome, that the probability was, in the words of Harvey, “fairly good,” is sufficient.
IY
Accordingly, the district court erred when it interpreted Michigan law as mandating proof of a probability of recovery which was greater than 50%5 and then concluded that, due to the .“significant” amount of leakage by December 24, 1983, Bell had a less than 50% chance of surviving an operation. As we stated above, we conclude that the district court’s conclusions that the leakage was so “significant” by December 24, 1983, that Bell did not have a reasonable probability of surviving an operation on and after that date to be factually and legally in error. Certainly, the amount of leakage which had occurred by December 24, 1983, is important to the determination of Bell’s prognosis. However, the evidence at trial was inconclusive regarding how much Bell’s aneurysm had leaked by December 24, 1983, and, more importantly, by how much the leakage had affected the probability that Bell would have survived an operation to correct the aneurysm.
We find it significant that Dr. Feemster indicated that a slowly leaking aneurysm provides time for successful surgery. Dr. Feemster did not indicate exactly how slowly Bell’s aneurysm was leaking and upon remand, this gap in the evidence may need to be filled before the district court can issue new findings. Another fact which may be relevant to the determination of whether Bell had a reasonable probability of recovery after December 23, 1983, is the fact that Bell lived for approximately three weeks after that date.
We note these facts only because they could be, but need not necessarily be, indications that the plaintiff has met the burden of proof in this case. We express no opinion as to whether Bell had a reasonable probability of recovery after the duty of care was breached. That determination is for the district court to make, applying the standard explicated in this opinion.
In summary, we affirm the district court’s conclusion that the duty of care was breached “by December 24th.” We reverse and remand this case in order for the district court to determine whether Bell had a reasonable probability of recovery if an operation had been performed to correct the aneurysm after December 23, 1983. Whether further formal proceedings are necessary in order to make this finding is left for the district court to decide.
For the foregoing reasons, the judgment of the district court in favor of the defendant is REVERSED and this case is REMANDED for proceedings in accordance with this opinion.
. Anderson v. City of Bessemer City, 470 U.S. 564, 573, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985) (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948)).
. In the district court's subsequent "Opinion and Order Denying Motion for New Trial,” the court characterized the plaintiff's burden of proof as having to show "a better than 50% chance of recovery.” For the reasons stated in the text, we find either characterization of the requisite burden of proof erroneous under Michigan law.
. Accord Berwald v. Kasai, 102 Mich.App. 269, 301 N.W.2d 499, 503 (1980).
. An injury or wrong done, "but from which no loss or damage results, and which, therefore, will not sustain an action.” Black's Law Dictionary 706, 1242 (5th ed. 1979).
. We note that the district court was not alone in its view that a plaintiff must prove a better than 50% chance of recovery in order to sustain the burden of proof. See Annotation, "Medical Malpractice: ‘Loss of Chance’ Causality," 54 A.L. R.4th 10 (1987); accord King, "Causation, Valuation, and Chance in Personal Injury Torts Involving Preexisting Conditions and Future Consequences,” 90 Yale LJ. 1353 (1981). We also note that, to our knowledge, the only federal appellate court to have defined "probable” stated that it means "more than 50 per cent [sic] of actual.” Price v. Neyland, 320 F.2d 674, 678 (D.C.Cir.1963). However, the Price case did not involve Michigan law and, therefore, is inapplicable to this action. Moreover, we note that Michigan law comports more closely with a statement made by the Ninth Circuit in a case involving workers' compensation: “because the conceptions of probability that have arisen in jurisprudence and in other branches of learning have far from achieved a perfect congruence, we prefer to describe ‘probability’ qualitatively, as requiring a very significant possibility, and not quantitatively, as mandating a ‘greater than fifty percent’ likelihood.” Sample v. Johnson, 771 F.2d 1335, 1343 (9th Cir.1985) (citations omitted).