dissenting.
Because I believe that the majority’s analysis of the question of actual malice is fundamentally flawed in two ways, I dissent.
First, as I read the majority’s opinion, it purports not to decide whether the alleg*1234edly defamatory statements in the Fortune 1 article are substantially true or substantially false. Rather, the opinion focuses exclusively on whether Fortune made the statements with actual malice. However, the majority does not analyze whether Fortune made the statements as they appeared in the article with actual malice. Rather, the majority first homogenizes the article’s statements into a single sentence that it calls the “gist” of the article, and then asks and answers the question whether that “gist” of the article was published with actual malice. In revising the article’s statements in this manner, the majority purports to follow the Supreme Court’s decision in Masson v. New Yorker Magazine, Inc., — U.S. -, 111 S.Ct. 2419, 115 L.Ed.2d 447 (1991). However, nothing in Masson authorizes a reviewing court to revise an article’s allegedly defamatory statements in any way before the court analyzes whether the statements were false or whether they were made with actual malice.
Second, although I agree that as the appellate court we must make an independent review of the facts in this defamation case, I cannot agree with the manner in which the majority has performed this task. Two fairly recent Supreme Court cases, Bose Corp. v. Consumers Union of U.S., Inc., 466 U.S. 485, 104 S.Ct. 1949, 80 L.Ed.2d 502 (1984), and Harte-Hanks Communications, Inc. v. Connaughton, 491 U.S. 657, 109 S.Ct. 2678, 105 L.Ed.2d 562 (1989), establish the procedures for performing such an independent review. One aspect of this process that is absolutely central is the requirement that we accord great deference to the jury’s credibility determinations while we perform our independent review of the allegedly defamatory statements and the circumstances under which they were made. In the present case, however, my reading of the majority opinion convinces me that this requirement was not met. For these reasons, I must dissent.
Unlike some other defamation cases in which public figure plaintiffs have prevailed both in trial and on appeal, in this case plaintiff-appellee Marcus Bressler cannot simply rely upon one facet of the case as the “key” to demonstrating malice.2 Nevertheless, when the facts and circumstances surrounding the investigation are closely examined, it becomes clear that the reporters’ claims of an “extensive” investigation are negated by the overwhelming evidence which shows that Fortune’s reporters wrote an article with reckless disregard for the truth of many of its accusations. Because I would hold that the jury was entitled to find from the evidence presented that many of the article’s actual statements were false and were published with actual malice, I must also address the arguments by the defendant-appellant, Fortune Magazine, that the statements in the article were not false and that the trial court erred in instructing the jury as to the burden of proof on the issue of falsity. First, however, I shall set out a brief factual introduction, followed by a discussion of what I believe to be the most significant aspect of this case, the manner in which the majority has approached the actual malice determination.
1. Factual Introduction
Because the legal issues in this case are so heavily dependent on the facts, I shall present a more detailed discussion of the facts in the sections of this dissent where they are most pertinent, and offer only a brief factual background here at the outset.
In late August and early September of 1986, Fortune reporters Brian Dumaine and Brett Fromson were working on an article about mismanagement at the Tennessee Valley Authority (“TVA”), the focus of which was TVA’s chief, Admiral Steven White. In the course of their investigation, *1235the reporters received some information about a report (“NCR 5609”) that concerned certain welds at the Watts Bar Nuclear Power Plant in Tennessee. After performing an investigation of the circumstances surrounding that report, the reporters included in their article on the TVA a discussion of NCR 5609 and Marcus Bres-sler’s involvement with it.
NCR 5609 is what is known as a non-conformance report or a non-conforming condition report (“NCR”). Such reports document conditions during construction of a nuclear facility which do not meet the requirements of certain construction codes. The report is created after an Authorized Nuclear Inspector (“ANI”) indicates to TVA that there is a non-conformity. At Watts Bar, TVA contracted with Hartford Steam Boiler Inspection and Insurance Company (“Hartford”) to supply ANIs to inspect construction of the facility. The testimony presented at trial indicates that NCRs were commonplace (J.A. at 619), and were in TVA’s public files. (J.A. at 408-09, 412). The testimony further established that NCR 5609 had been marked “significant” by the very persons who prepared it, which ensured that it would be sent to the Nuclear Regulatory Commission for their approval (J.A. at 369 (Dumaine); J.A. at 1212), and would be reviewed by TVA’s Nuclear Licensing Staff. (J.A. at 718-22 (Bressler)).
ANI Howard Haston, an employee of Hartford, initially documented that certain welds on pipes leading to the containment area of the Unit 1 reactor at Watts Bar had not been visually inspected during hydrostatic testing. Since such viewing is required by the Code of the American Society of Mechanical Engineers (“ASME”), an NCR . had to be prepared to document the variance from the Code. NCR 5609 was prepared for that purpose and specifically recommended that even though the visual inspection had not been performed, the welds should be “used as is,” since the hydrostatic test had been performed and radiographic tests had also been performed. (J.A. at 1211-12). Among the persons to “sign off” on NCR 5609 was Bressler.
II. Discussion
A. Majority’s Rewriting of the Fortune Article
I believe that the approach adopted by the majority of rewriting the content of the Fortune article and then discussing whether that rewritten version was published with actual malice represents a significant change in the constitutional law of defamation. In my view, the majority has fashioned this approach without support from any case law. Since this issue is integral to the discussion of both the falsity and aetual malice issues, it requires discussion at the outset of this dissent.
The majority begins its analysis by citing Masson v. New Yorker Magazine, Inc., — U.S.-, 111 S.Ct. 2419, 115 L.Ed.2d 447 (1991), for the proposition that a public official plaintiff in a libel case must prove that the “gist of the article” is false. Opinion at 1228 (emphasis supplied). From its statement in a footnote, I shall assume that the majority means by this sentence that a plaintiff must prove that the “gist” of the statements about him were false.3 The majority looks next to the tests for establishing actual malice, noting that its conclusions in that regard obviate the need to address whether the article’s “gist” was false, and proceeds to analyze whether this gist of the article was published with actual malice. I do not believe this approach is supported by Masson or any other case of which I am aware.
The majority opinion states:
*1236The “gist” of the contested portion of the Fortune article was that plaintiff Bressler allegedly played a lead role in pressuring an independent inspector to certify, contrary to fact, that certain safety-related welds in the Watts Bar plant met the engineering code requirements, and that Bressler also attempted to cover up that safety violation. We now examine the defendant’s basis for reporting such allegations.
Opinion at 1229 (emphasis supplied; footnotes omitted). In my view, the most obvious flaw in this approach is that this Court cannot possibly examine Fortune’s basis for reporting this capsulized “gist” because it is not what the Fortune article actually said. Rather, the article contained numerous, specific falsehoods about Bressler.
The Fortune article did not state that Bressler “played a lead role” in pressuring Authorized Nuclear Inspector Howard Hasten; it stated that an ANI (Hasten) “was forced by TVA and superiors of his own company to approve a safety report against his will.” (J.A. at 10, 1687) (emphasis supplied). Significantly, the reporters conceded at trial that TVA here meant Bres-sler. (J.A. at 255-251).
The article did not merely say that the welds did not meet “engineering code requirements”; instead it alleged that “TVA failed to test welds on [56] crucial steam and water pipes” (J.A. at 14, 1691) (emphasis supplied). Furthermore, the article did not say that the welds were “safety-related”; it referred to them as welds on “crucial” pipes (J.A. at 14, 1691), and stated that “[a] burst pipe could set off a serious nuclear accident” (J.A. at 14, 1691).
The article did not simply state that Bressler was involved in “pressuring” an inspector to certify that the welds met code requirements; rather, the article reported, “Bressler, the head of TVA’s engineering codes office, ... warned Hartford Steam Boiler to get its inspectors in line or TVA would not renew its inspection contract, ...” (J.A. at 15, 1692) (emphasis supplied).
Finally, although the Fortune article did state that Bressler attempted to “cover up that safety violation,” as the majority states, it says much more. The article’s second paragraph explicitly stated, “FORTUNE has uncovered evidence that the TVA tried to suppress inspection records showing that one of the plants — in Watts Bar, Tennessee — could be unsafe.” (J.A. at 10, 1687) (emphasis supplied).
Bressler and his counsel spent almost two weeks in trial (and months, if not years, in pretrial preparation) trying to prove both that the article’s specific statements were false and that the Fortune reporters made them with actual malice. Having won the battle by proving these issues (and the others he was required to prove) to the satisfaction of a jury and a district court judge, Bressler nonetheless loses the war because the majority has changed the statements which he must prove false, by rewriting them into the “gist” of the article. It is also worth noting that Bressler has never had an opportunity to attempt to prove false this revised version of the article’s statements.
I can find no case that authorizes a court, for purposes of analyzing whether statements are defamatory, to revise the statements first — leaving only innocuous generalizations — and then determine whether the revised statements are defamatory. Masson, the only case on which the majority appears to rely in adopting this approach, certainly does not authorize such an approach. In Masson, the Court addressed the standards for proving falsity.
Minor inaccuracies do not amount to falsity so long as “the substance, the gist, the sting, of the libelous charge be justified.” Put another way, the statement is not considered false unless it “would have a different effect on the mind of the reader from that which the pleaded truth would have produced. ” Our definition of actual malice relies upon this historical understanding.
— U.S. at-, 111 S.Ct. at 2433 (citations omitted; emphasis supplied). I am nonplussed that the panel majority in the present case can read that direction from the Supreme Court as a license to rewrite allegedly defamatory statements before *1237subjecting them to the test of actual malice.
Of course, it is unfortunate that the Supreme Court in Masson used the word “gist” at all. Even as the Court recognized the confusion that has resulted from the use of the term “actual malice” in the law of defamation, — U.S. at-, 111 S.Ct. at 2429-30, the Court appears to have placed its imprimatur on another term — “gist”— that is almost certain to generate similar confusion and difficulty in application. In my view, the Court added nothing beneficial to its opinion by stating that a plaintiff must show that the “gist” of an article is false. It is enough to say that a “statement is not considered false unless it would have a different effect on the mind of the reader from that which the pleaded truth would have produced.” — U.S. at-, 111 S.Ct. at 2433 (citation omitted).
The present case is a good example of how the term “gist” can cause confusion. The majority here has seized upon the term “gist,” which was used by the Court, and extrapolated from it that the reviewing court should set forth its own version of the article’s content and subject that version to the falsity and actual malice tests. However, the actual language of Masson is,
Minor inaccuracies do not amount to falsity so long as “the substance, the gist, the sting, of the libelous charge be justified.” Put another way, the statement is not considered false unless it “would have a different effect on the mind of the reader from that which the pleaded truth would have produced.”
— U.S. at-, 111 S.Ct. at 2433 (citations omitted; emphasis supplied). These standards do not permit the reviewing court to forego an analysis of whether each allegedly defamatory statement was false by revising the statements into a generalized “gist,” and subjecting the revised version to the requisite analysis. Rather, the reviewing court is to determine whether those actual statements are “minor inaccuracies” within the context of the relevant portions of the entire article such that they do not render the actual substance of those portions of the article false. Put another way, the reviewing court must determine whether the article’s actual statements would have a different effect on the mind of the reader than the pleaded truth would produce. But these tests for determining falsity must be applied to the actual words used in the article to determine whether they are minor inaccuracies or something more than that. If those statements are more than merely minor inaccuracies, or if they would produce a different effect than would the truth, the court must then determine whether the actual statements were made with actual malice. Thus, the Supreme Court’s standards guide the reviewing court in performing this task. While the Supreme Court’s use of the term “gist” is unfortunate, it is even more unfortunate that the majority in the present case has chosen to focus exclusively on that term, without considering the context in which it was used.
My difficulty with the majority’s treatment of Masson is perhaps most clearly pointed up by the footnote in the majority opinion which observes:
The dissent relies on Masson in criticizing our purported rewriting of the article’s defamatory statements. Masson, however, in addition to confirming that it is the “gist” of the statements which must be examined, — U.S. at-, 111 S.Ct. at 2433, considered the significance of rewriting, or misquoting, by a defendant-publisher, and not by a reviewing court. Further, there has been no allegation that the Fortune reporters altered or fabricated any quotations.
Opinion at 1229 n. 2. First of all, in Mas-son, the Supreme Court made it clear that it was “rejectpng] any special test of falsity for quotations,” choosing instead to use the case before it to “illuminate a broader principle” of falsity. — U.S. at-, 111 S.Ct. at 2432. Secondly, it is not to be wondered at that the Supreme Court did not consider the significance of rewriting or misquoting by the reviewing court. Neither the trial court nor the appellate court in Masson undertook to rewrite the allegedly libelous words published by the defendants. Unless the lower courts had en*1238gaged in such a rewriting, the Supreme Court would have had no occasion even to contemplate the idea, since the issue in that case, (as in any libel case,) was the extent to which the expression of the defendant-publisher, not the expression of the reviewing court, is protected by the First Amendment.
I believe that by revising the article’s accusations into the two simple statements which the majority characterizes as the article’s “gist,” the majority has simply eliminated the stinging accusations contained in the article, the very statements that Bres-sler alleges are defamatory. The statements in the article are detailed and specific. It is impossible to determine whether those specific statements are false and, if so, whether they were made with actual malice, by analyzing only the generalized gist of the article as summarized by the majority instead of analyzing the statements themselves. I find nothing in Mas-son which even implies that a court may determine the merits of a libel action by doing so.
My continual emphasis in this dissent on what the Fortune article actually says, in contrast to what both Fortune and the majority say it says, is not an overly restrictive approach. Minor misstatements must not be treated as materially false, as the Supreme Court made clear in Masson. However, there is a world of difference between inquiring whether the “substance, the gist, the sting” of a statement is false and was published with actual malice, and rewriting allegedly defamatory statements into some other, watered-down version and asking if that version is false and was published with actual malice. A hypothetical will help to illuminate this point.
Assume a newspaper published an article stating that John Doe, a bank manager and public figure, was discharged by his employer for having embezzled $250,000. If, in fact, Doe had embezzled only $220,000, we would not treat that discrepancy, as rendering false the entire article under the law of defamation. This article would have the same effect on the reader as would the truth — both advise the reader that John Doe was discharged for embezzling a very substantial sum of money from his employer.
In contrast, assume again that the newspaper published the story that John Doe was discharged for embezzling $250,000. In fact, Doe was discharged because he authorized $250,000 in loans which immediately went into default, and the newspaper published the embezzlement story with knowledge only of Doe’s authorizing the bad loans. Were Doe to sue the newspaper for falsely reporting that he had embezzled money, would it be appropriate for the reviewing court to summarize the gist of the story as being that “John Doe was discharged by his employer because he was responsible for the employer’s incurring large losses of money,” and to determine the presence or absence of malice by examining the newspaper’s basis for reporting that statement? Surely not. If that were what the newspaper had reported, it probably could not be found to be false or malicious since the paper had information about Doe’s responsibility for the bad loans. But, of course, the newspaper did not report that “John Doe was discharged by his employer because he was responsible for the employer’s incurring large losses of money.” The newspaper published the statement that Doe was fired for embezzling money. If the court examines the newspaper’s basis for reporting that Doe was fired for embezzling money, very likely it would find malice, since the newspaper had no information to support that statement, and certainly that statement would have a very different effect on the mind of the reader than the truth would have had. It is one thing to be called incompetent. It is quite another to be labelled a thief.
By rewriting or recharacterizing allegedly defamatory statements into a “gist,” a court may end up with a statement (the “gist”) that lies somewhere between the truth and what was actually published. Subjecting this “gist” to the tests of falsity and actual malice will often produce different results from those which would be produced if the actual statements at issue in the case were subjected to those same tests. The victim of such an approach will *1239always be the defamed party, for the more generalized and innocuous the “gist” becomes, the greater the likelihood that there will be evidence justifying the publication of those more innocuous statements. In my view, the majority’s approach produces just such results. It has reviewed its own rewritten version of the article, what it calls the “gist,” and concluded that there was justification for the Fortune reporters to have reported the gist, and thus there was no malice. From that', the majority concludes, but does not hold, that the “gist of the article was substantially true.” Without belaboring the further watering down of the test for falsity which would result from looking to see whether the rewritten “gist”' is “substantially true,” I shall demonstrate in some detail below that if the article’s actual statements are examined, it becomes clear that they were both false and published with actual malice.
B. Falsity
1. Burden of Proof on Falsity
Fortune contends that the district court “erred in instructing the jury that the plaintiff was obliged to prove falsity by a mere preponderance of the evidence.” (Brief of Appellant, at 43). Instead, Fortune asserts that a proper charge would have instructed the jury that Bressler was required to prove falsity by “clear and convincing evidence.” Fortune’s argument is as follows:
The “clear and convincing” standard follows ineluctably from the requirement, imposed by the Supreme Court in New York Times Co. v. Sullivan [376 U.S. 254, 84 S.Ct. 710, 11 L.Ed.2d 686], that actual malice be established by “convincing clarity.” 376 U.S. at 285-86 [84 S.Ct. at 729]; Gertz v. Robert Welch, Inc., supra, 418 U.S. [323] at 342 [94 S.Ct. 2997 at 3008, 41 L.Ed.2d 789], If knowledge of falsity must be established by clear and convincing evidence to satisfy the actual malice standard, so too must the falsity. If the underlying falsity could be established by a mere preponderance, proof of the knowledge of such falsity could rise no higher, i.e., the proof of knowledge could, by definition, be no better than the proof of the underlying alleged falsity. This would gut the Supreme Court’s requirement that actual malice be established by clear and convincing evidence.
(Brief of Appellant, at 43-44). Essentially, Fortune argues that if a jury, applying a preponderance of the evidence standard, errs by concluding that a statement was false, their determination that the statement was published with actual malice would be undermined, thereby compromising the First Amendment protection that the actual malice standard was meant to guarantee.4 To begin with, the logic of Fortune’s argument suffers from a fatal flaw. Nothing in the Supreme Court’s decisions indicates that a public figure libel plaintiff must prove knowledge of falsity. In fact, many of the Court’s decisions recognize that actual malice may be proved by showing that the publisher acted with reckless disregard for or in purposeful avoidance of the truth. See generally New York Times Co. v. Sullivan, 376 U.S. 254, 279-80, 84 S.Ct. 710, 726, 11 L.Ed.2d 686 (1964); Harte-Hanks Communications, Inc. v. Connaughton, 491 U.S. 657, 666, 109 S.Ct. 2678, 2685, 105 L.Ed.2d 562 (1989).
*1240Fortune cites several cases that it contends support its position. In Buckley v. Littell, 539 F.2d 882 (2d Cir.1976), cert. denied, 429 U.S. 1062, 97 S.Ct. 785, 50 L.Ed.2d 777 (1977) — the only appellate case that supports Fortune’s position — the court stated, “The appellee, a public figure, must rather have demonstrated with convincing clarity not only that appellant’s statements were false, but that appellant knew that they were false or made them with reckless disregard of their truth or falsity.” 539 F.2d at 889-890. The court cited Gertz v. Robert Welch, Inc., 418 U.S. 323, 342, 94 S.Ct. 2997, 3008, 41 L.Ed.2d 789 (1974), and New York Times, 376 U.S. at 279-80, 285-86, 84 S.Ct. 710, 725-26, 728-29, in support. The Supreme Court, however, has taken no position on the standard of proof for falsity. See Harte-Hanks, — U.S. at-n. 2, 109 S.Ct. at 2682 n. 2. Thus, the Buckley court’s conclusion, made without further analysis or explanation, provides little guidance or persuasive force on the question.
Fortune also cites Judge Griffin Bell’s concurring opinion in Firestone v. Time, Inc., 460 F.2d 712, 721-24 (5th Cir.), cert. denied, 409 U.S. 875, 93 S.Ct. 120, 34 L.Ed.2d 127 (1972), in which Judge Bell stated,
The Supreme Court has not expressly added the requirement of clear and convincing proof of falsity to the plaintiff’s burden of proof. As stated, the burden of showing falsity has been imposed upon the plaintiff in First Amendment cases. Garrison [v. Louisiana, 379 U.S. 64, 85 S.Ct. 209, 13 L.Ed.2d 125 (1964)]; Rosenblatt [v. Baer, 383 U.S. 75, 86 S.Ct. 669, 15 L.Ed.2d 597 (1966)]. Such a standard of proof seems implicit however, in the stated requirement in New York Times that plaintiff has the burden of showing by clear and convincing proof that publication was with knowledge of falsity or with reckless disregard as to falsity vel non. I conclude for the same constitutional reasons giving rise to this stringent proof requirement that the clear and convincing proof standard would also apply to proving that the statement was false in the first instance.
460 F.2d at 722-23 (Bell, J., specially concurring). Judge Bell, however, went no further in analyzing the question, relying only upon the Supreme Court’s reasoning in New York Times, Garrison, and Rosen-blatt. As noted above, however, the Supreme Court has not ruled on the extent of the burden of proof on falsity in those cases or any others.
A somewhat more persuasive argument that the proper standard of proof is that of clear and convincing evidence is found in Robertson v. McCloskey, 666 F.Supp. 241, 248-50 (D.D.C.1987), the third case cited by Fortune. In Robertson, the court noted that in Philadelphia Newspapers, Inc. v. Hepps, 475 U.S. 767, 106 S.Ct. 1558, 89 L.Ed.2d 783 (1986), the Supreme Court con-stitutionalized the falsity element of defamation in cases involving public figure plaintiffs by striking down the common law presumption that defamatory statements are also false. 666 F.Supp. at 248; see Hepps, 475 U.S. at 776, 106 S.Ct. at 1563. The district court in Robertson accepted the defendants’ argument that the Court’s holding in Hepps supported a conclusion that falsity should be proved by clear and convincing evidence, reasoning as follows.
[A] clear and convincing standard of proof for falsity would resolve doubts in favor of speech when the truth of a statement is difficult to ascertain conclusively. Indeed, as a practical matter, public-figure plaintiffs already bear such a burden, for in order to prove malice they must, of necessity, show by clear and convincing evidence that the defendant knew the statement was false or acted in reckless disregard of its truth. Finally, defendants’ argument has more than merely a logical or symmetrical appeal. To instruct a jury that a plaintiff must prove falsity by a preponderance of the evidence, but must also prove actual malice, which to a large extent subsumes the issue of falsity, by a different and more demanding standard is to invite confusion and error.
666 F.Supp. at 248. The court thus supported its holding with three separate justifications: (1) the need to resolve doubts in *1241favor of speech; (2) the fact that actual malice practically subsumes falsity; and (3) potential jury confusion. But these reasons are not sufficient to justify elevating the burden of proving falsity to the level of clear and convincing evidence. First, as noted above, I do not believe that actual malice subsumes falsity. Public figure plaintiffs need not show knowledge of falsity to prove malice, as Fortune contends they must; it is enough that they show reckless disregard for the truth.
Second, merely requiring a public figure plaintiff to prove by a preponderance of the evidence that the statements were false is to resolve doubts in favor of speech. The Supreme Court’s discussion of this issue in Hepps makes this clear.
There will always be instances when the factfinding process will be unable to resolve conclusively whether the speech is true or false; it is in those cases that the burden of proof is dispositive. Under a rule forcing the plaintiff to bear the burden of showing falsity, there will be some cases in which plaintiffs cannot meet their burden despite the fact that the speech is in fact false. The plaintiffs suit will fail despite the fact that, in some abstract sense, the suit is meritorious. Similarly, under an alternative rule placing the burden of showing truth on defendants, there would be some cases in which defendants could not bear their burden despite the fact that the speech is in fact true. Those suits would succeed despite the fact that, in some abstract sense, those suits are unmeritorious. Under either rule, then, the outcome of the suit will sometimes be at variance with the outcome that we would desire if all speech were either demonstrably true or demonstrably false.
This dilemma stems from the fact that the allocation of the burden of proof will determine liability for some speech that is true and some that is false, but [for] all of such speech [that] is unknowable true or false. Because the burden of proof is the deciding factor only when the evidence is ambiguous, we cannot know how much of the speech affected by the allocation of the burden of proof is true and how much is false. In a case presenting a configuration of speech and plaintiff like the one we face here, and where the scales are in such an uncertain balance, we believe that the Constitution requires us to tip them in favor of protecting true speech.
475 U.S. at 776, 106 S.Ct. at 1563-64 (emphasis supplied). This discussion, couched in the traditional language of the preponderance of the evidence standard, i.e., “scales,” “balance” and “tip [the scales] in favor,” and containing no reference to the clear and convincing standard, indicates that the Court has already made a judgment to resolve doubts in favor of speech by assigning the burden of proof to plaintiffs on the falsity issue.
In addition, when the Court in Hepps assigned public figure plaintiffs the burden of proving falsity, it indicated that it was not making any significant shift in the allocations of burdens between plaintiffs and defendants.
We note that our decision adds only marginally to the burdens that the plaintiff must already bear as a result of our earlier decisions in the law of defamation. The plaintiff must show fault. A jury is obviously more likely to accept a plaintiff’s contention that the defendant was at fault in publishing the statements at issue if convinced that the relevant statements were false. As a practical matter, then, evidence offered by plaintiffs on the publisher’s fault in adequately investigating the truth of the published statements will generally encompass evidence of the falsity of the matters asserted.
475 U.S. at 778, 106 S.Ct. at 1565. In contrast, requiring that plaintiffs prove falsity by clear and convincing evidence, as Fortune and amici urge this Court to do, certainly would result in a significant change in the burdens on plaintiffs, a shift I view as unwise because it would upset the balance between the competing interests involved in defamation litigation.
There are two sets of interests at stake in constitutional decisions on the law of defamation, the states’ interest which un-
*1242derlies the law of libel in “compensation of individuals for the harm inflicted on them by defamatory falsehood,” Hepps, 475 U.S. at 773, 106 S.Ct. at 1562, and the First Amendment interest in providing “breathing space” for freedom of expression in order to avoid self-censorship. Id. at 772-73, 106 S.Ct. at 1561-62. Requiring plaintiffs to prove falsity by clear and convincing evidence would result in the unnecessary sacrifice of state law protection against defamation, while securing extremely marginal gains in constitutional protection. See, e.g., Time, Inc. v. Firestone, 424 U.S. 448, 456, 96 S.Ct. 958, 966, 47 L.Ed.2d 154 (1976) (discussing the depreciation of the individual’s interest in protection from the harm from defamatory falsehoods that some additional constitutional requirements would cause). The Supreme Court has already resolved doubts in favor of speech by assigning the burden of proof on falsity to public figure plaintiffs. I would not add to that burden by increasing the level of proof.
Since the Supreme Court has struck a balance between protecting citizens from defamation and ensuring adequate breathing room under the First Amendment, it is not for this Court to upset that balance. Certainly, such factors as jury confusion and the interrelatedness of malice and falsity would not justify such a reordering of priorities. Therefore, I conclude that the district court did not err by instructing the jury that falsity may be proved by a preponderance of the evidence.
2. Standard of Review
Having discussed the appropriate standard of proof for falsity, I must next address the issue of the proper standard of review of the jury’s finding of falsity. In Connaughton v. Harte-Hanks Communications, Inc., 842 F.2d 825 (6th Cir.1988), affd, 491 U.S. 657, 109 S.Ct. 2678, 105 L.Ed.2d 562 (1989), this Court implicitly sanctioned the clearly erroneous standard of review for the jury’s finding of falsity. In that case, we stated, “Equally apparent from the jury’s answer to the second special interrogatory is that it considered the published Thompson charges to be false.... Thus, upon reviewing the record in its entirety, this court concludes that the jury’s determinations of the operational facts bearing upon the falsity of the article were not clearly erroneous.” 842 F.2d at 841. See also Harte-Hanks Communications, Inc. v. Connaughton, 491 U.S. 657, 109 S.Ct. 2678, 2683 & n. 4,105 L.Ed.2d 562 (1989) (recognizing that the court of appeals concluded that the jury’s determination of falsity was not clearly erroneous).
3. Analysis
Fortune devotes a considerable portion of its brief to arguing that the statements in the article are not false. However, this argument is premised on Fortune’s characterization of the article's content, and this characterization is not accurate. Fortune summarizes the “gist” of the article in the following sentence: “Marcus Bressler was a key participant in a process in which a purportedly independent inspector was compelled to certify that certain safety-related welds in a nuclear power plant were in conformity with the applicable engineering code provisions when they were not and in a subsequent effort to cover up that process.” (Brief of Appellant, at 28). Fortune breaks this statement down into these statements/headings: “The ASME Code Was Violated,” “The Nonconforming Condition Report Affected Safety Issues,” “Bressler Pressured Hartford To Have the Report Signed,” and “Bressler Participated in a Cover-Up.” (Id. at 28-31). In reviewing the evidence, the majority essentially adopts Fortune’s approach, although not Fortune’s specific characterization of the article. In so doing, the majority relies upon Masson v. New Yorker Magazine, Inc., — U.S. -, 111 S.Ct. 2419, 115 L.Ed.2d 447 (1991), as support.
In Masson, the plaintiff, a public figure, sought to recover from the New Yorker Magazine for publishing allegedly defamatory statements. The plaintiff’s principal complaint was that the author, “with full knowledge of the inaccuracy, used quotation marks to attribute to him comments he had not made.” — U.S. at-, 111 S.Ct. at 2424. After a thorough review of the *1243facts and a brief discussion of constitutional principles applicable in libel cases, the Court then set out to analyze the inaccuracies in the context of those principles. As a starting point, the Court rejected the idea that “any alteration beyond correction of grammar or syntax by itself proves falsity in the sense relevant to determining actual malice under the First Amendment.” Id. — U.S. at-, 111 S.Ct. at 2431-32. The Court recognized the reality of reporting an interview.
Even if a journalist has tape recorded the spoken statement of a public figure, the full and exact statement will be reported in only rare circumstances. The existence of both a speaker and a reporter; the translation between two media, speech and the printed word; the addition of punctuation; and the practical necessity to edit and make intelligible a speaker’s perhaps rambling comments, all make it misleading to suggest that a quotation will be reconstructed with complete accuracy.
— U.S. at-, 111 S.Ct. at 2432. This served as the background for the Court’s discussion of falsity in the broader context.
The Court rejected the idea of a special test for falsity for quotations. Instead, the Court used the case before it involving misquotation to illuminate broader principles of falsity. The common law'of libel, it noted, “overlooks minor inaccuracies and concentrates on substantial truth.” Id. — U.S. at-, 111 S.Ct. at 2433. The standards for proving falsity are as follows.
Minor inaccuracies do not amount to falsity so long as “the substance, the gist, the sting, of the libelous charge be justified.” Put another way, the statement is not considered false unless it “would have a different effect on the mind of the reader from that which the pleaded truth would have produced. ” Our definition of actual malice relies upon this historical understanding.
Id. (citations omitted; emphasis supplied). An examination of the evidence relative to each of the following actual statements contained in the Fortune article demonstrates that the jury’s finding of falsity is not clearly erroneous, but is supported by the great weight of the evidence.
a. “TVA [Bressler] tried to suppress inspection records showing that one of the plants — in Watts Bar, Tennessee— could be unsafe”
The Fortune reporter who drafted the article, Brian Dumaine, agreed that the inspection records referred to in the article were NCR 5609 and its attachments. (J.A. at 254). “To suppress” means “to keep from public knowledge, to refrain from divulging, leave undisclosed.” Webster's Third New International Dictionary, at 2298. Dumaine admitted that when he wrote the article he knew that the TVA “already had [NCR] 5609 and all its attachments clearly in their public files.” (J.A. at 412). Dumaine also admitted that there was no physical suppression of the documents. (J.A. at 373). These facts alone provide a sufficient basis from which the jury could conclude that the statement was false. But, in addition, it is undisputed that an NCR exists solely to disclose nonconforming conditions. Thus, by its very nature, it serves to reveal, not conceal.
b. Authorized Nuclear Inspector Haston “was forced by TVA to approve a safety report against his will,” and, “A campaign then developed to force Ha-ston to sign [TVA’s safety report] recommending that the welds be used as is”
Dumaine admitted that these statements refer to Bressler and his alleged involvement in Haston’s signing of NCR 5609. (J.A. at 255-257). At trial, Haston testified that on April 20, 1984, he had a conversation with Bressler regarding the welds covered by NCR 5609. (J.A. at 621). Haston stated that he did not feel pressured by Bressler during that conversation and that he had no other conversations with Bres-sler regarding NCR 5609. (J.A. at 621, 623, 635-36). He further testified that he did not recall telling Fortune that Bressler or anyone at TVA tried to get him to sign the NCR; “there would be no reason to say that.” (J.A. at 641). On cross-examination he also testified that he was not forced by *1244anyone to sign the NCR, that it was an issue between himself and his employer, that TVA was not involved, and that the NCR was not a safety report but rather a non-conformance report. (J.A. at 660).
Bressler testified that he spoke to Ha-ston only once regarding NCR 5609 to explain that he felt it was acceptable to fore-go visual inspection. (J.A. at 710). In addition, Dorwin Etzler, who worked with Bressler at TVA and prepared the documentation for NCR 5609, testified that if Haston was not willing to sign the report, TVA was willing to go to the Nuclear Regulatory Commission for its approval. (J.A. at 544).
Fortune offered no evidence that Bres-sler and TVA, as opposed to Haston’s company, Hartford Steam Boiler, even had any means of “forcing” Haston to sign the report, let alone that he was actually forced by TVA. Instead, Fortune relied upon the allegations, made by TVA Investigator Mansour Guity, of pressure being exerted by Bressler on managers at Hartford. Guity did not testify at trial and the reporters’ notes of their interviews with him did not reveal specifics about the sources of his information. It certainly was not clearly erroneous for the jury to find falsity in the article’s statements that Bressler forced Haston to sign.5
c. “TVA failed to test welds on [56] crucial steam and water pipes”
Fortune argues that the “ASME Code was violated.” (Brief of Appellant, at 28-29). That there was non-conformity with the ASME Code is not disputed, nor is it an issue in this case. What is at issue is what the article stated. The majority characterizes this portion of the . article as stating that the safety-related welds did not meet engineering requirements. Opinion at ■1229. The article, however, did not merely state that provisions of the ASME Code were violated; it falsely stated that “TVA failed to test welds.’’ (J.A. at 14,1691). In fact, it is undisputed that the welds were tested, both radiographically6 and hydro-statically. (E.g., J.A. at 497 (Robison); 529-32 (Etzler)). The welds were not, however, visually inspected by an ANI or TVA personnel during the hydrostatic test. It is clear that the evidence supports Bressler’s contention that the statement that the welds were not tested is false.7
d. “A burst pipe could set off a serious nuclear accident”
Once again, both Fortune and the majority opinion understate the gravity of the statement in the article by contending that the welds covered by the NCR were “safety-related.” (Opinion at 1229; Brief of Appellant, at 29). This simply does not convey the alarmist impact of this portion of the article. The substance of this statement — what it would mean to the average reader — is not that the welds were somehow related to safety, but rather that these welds were not tested at all and the failure to test them makes it possible that they would cause a nuclear accident.
To demonstrate the falsity of the statement, and its implication that these pipes were a danger to public safety, Bressler offered the testimony of Howard Haston, Dorwin Etzler, Gary Pitzl, and Harold Ro-bison. Each of these witnesses testified *1245that although the welds did not entirely conform to the ASME Code because they had not been visually inspected during hydrostatic testing, there was not a safety problem with the welds. (J.A. at 628, 631 (Haston); at 499 (Robison); at 539 (Etzler); at 600 (Pitzl)). Pitzl, a managing engineer in TVA’s nuclear plants, opined that “this disposition does not pose any safety concern.” (J.A. at 600). Etzler stated, “We had a very high confidence that these welds were perfectly adequate for their intended function.” (J.A. at 539). In a deposition that was presented at trial, Robi-son, a manager for Hartford, testified that even after he found out that the welds in unit one had not been visually inspected by anyone during hydrostatic testing, the disposition of the nonconforming condition did not change and it was not a safety problem. (J.A. at 498-99).8
Fortune presented no evidence to substantiate the article’s statement. Instead, Dumaine testified that it was “common knowledge in the nuclear power industry.” (J.A. at 283). Dumaine admitted that he would not be qualified to make the statement in the article on his own knowledge, (J.A. at 284), and that if he had done so without someone trained in the nuclear industry giving him the information, it would have been reckless. (J.A. at 284-85). He also admitted that nothing in the editorial reference file supported the article’s statement. (J.A. at 398).
e. “Bressler, the head of TVA’s engineering codes office, tried to cover up the breach of safety standards,” and “The pressure Bressler had exerted to cover up the safety violation”
In its brief, Fortune attempts to show that Bressler was involved in a coverup during 1985. The article, however, clearly refers to an alleged coverup involving NCR 5609, the events surrounding which took place early in 1984. In addition, an earlier draft of the article specified that the pressure referred to took place in 1984. (J.A. at 1665; see also J.A. at 365-66).
The key words in the article’s statements are “cover up” of “safety violations” and “breach of safety standards.” Bressler proved the statements to be false. Brian Dumaine, Fortune’s associate editor who wrote the article, opined that there was a coverup because Bressler recommended that the welds be used as is, instead of submitting the non-conformance to NRC for their opinion.
Q. So you believe that the making this [NCR 5609] significant and sending it up to the Nuclear Regulatory Commission for approval was a coverup of this NCR 5609?
A. That’s right, the way it was done. That’s right.
(J.A. at 370). He later testified that the substance of the coverup in 1984 was that even though NCR 5609 said the welds did not meet the Code, “it said ‘use as is’ when it should have said ‘rejected,’ and the Code should have been debated within the Nuclear Regulatory Commission.” (J.A. at 406). NCRs, including NCR 5609, are public documents located in TVA’s public files, (J.A. at 408-09, 412); the purpose of NCRs is to disclose conditions which do not conform to code standards; and people at TVA and Hartford had signed off on NCR 5609. There is no evidence in the record to support Dumaine’s unfounded opinion that the “use as is” directive on the NCR or the fact that the directive was not debated within the Nuclear Regulatory Commission constituted a coverup, and clearly, Bressler demonstrated that these statements in the article were false.
In sum, many of the article’s statements about Bressler are materially false; they are not minor inaccuracies, and each of them would have a different effect on the reader than the truth would have. Thus, the jury’s determination that Bressler had proved the falsity of the actual statements *1246contained in the Fortune article cannot be found to be clearly erroneous.
C. Actual Malice
1. Standards of Law for Determining “Actual Malice”
In New York Times Co. v. Sullivan, 376 U.S. 264, 279-80, 84 S.Ct. 710, 726, 11 L.Ed.2d 686 (1964), the Supreme Court held that for a public official to recover in an action for libel he must prove that the defamatory statement “was made with ‘actual malice’ — that is, with knowledge that it was false or with reckless disregard of whether it was false or not,” and that this constitutional standard demands proof of malice with “convincing clarity,” id. at 285-86, 84 S.Ct. at 729.
Since New York Times, the Court has applied and refined the standards set out in that case. For example, in St. Amant v. Thompson, 390 U.S. 727, 88 S.Ct. 1323, 20 L.Ed.2d 262 (1968), the Court noted that while “ ‘[rjeckless disregard’ ... cannot be fully encompassed in one infallible definition,” id. at 730, 88 S.Ct. at 1325, to satisfy the standard “[t]here must be sufficient evidence to permit the conclusion that the defendant in fact entertained serious doubts as to the truth of his publication,” id. at 731, 88 S.Ct. at 1325. The standard is a subjective one; “[TJhere must be sufficient evidence to permit the conclusion that the defendant actually had a ‘high degree of awareness of probable falsity.’ ” Harte-Hanks Communications, Inc. v. Connaughton, 491 U.S. 657, 687, 109 S.Ct. 2678, 2696, 105 L.Ed.2d 562 (1989) (quoting Garrison v. Louisiana, 379 U.S. at 74, 85 S.Ct. at 215 (1964)) (internal quotation marks omitted).
In St. Amant, the Court stated,
The defendant in a defamation action brought by a public official cannot, however, automatically insure a favorable verdict by testifying that he published with a belief that the statements were true. The finder of fact must determine whether the publication was indeed made in good faith. Professions of good faith will be unlikely to prove persuasive, for example, where a story is fabricated by the defendant, is the product of his imagination, or is based wholly on an unverified anonymous telephone call. Nor will they be likely to prevail when the publisher’s allegations are so inherently improbable that only a reckless man would have put them in circulation. Likewise, recklessness may be found where there áre obvious reasons to doubt the veracity of the informant or the accuracy of his reports.
390 U.S. at 732, 88 S.Ct. at 1326. The Court has also noted that “[fjailure to investigate does not in itself establish bad faith.” Id. at 733, 88 S.Ct. at 1326. More recently, however, the Court has made it clear that “the purposeful avoidance of the truth” is fundamentally different from negligence in investigating. Harte-Hanks, 491 U.S. at 691, 109 S.Ct. at 2698.
2. Standard of Review
In defamation cases, the question whether the evidence in the record is sufficient to support a finding of actual malice is a question of law. Harte-Hanks Communications, Inc. v. Connaughton, 491 U.S. 657, 683-84, 109 S.Ct. 2678, 2694, 105 L.Ed.2d 562 (1989). In New York Times Co. v. Sullivan, 376 U.S. 254, 279-80, 84 S.Ct. 710, 726, 11 L.Ed.2d 686 (1964), after the Court held that “actual malice” must be proved by evidence of “convincing clarity,” the Court further held that appellate judges “must make an independent examination of the whole record, so as to assure ourselves that the judgment does not constitute a forbidden intrusion on the field of free expression.” Id. at 285, 84 S.Ct. at 729. With respect to appellate review, the Supreme Court recently stated,
The question whether the evidence in the record in a defamation case is of the convincing clarity required to strip the utterance of First Amendment protection is not merely a question for the trier of fact. Judges, as expositors of the Constitution, must independently decide whether the evidence in the record is sufficient to cross the constitutional threshold that bars the entry of any judgment that is not supported by clear and convincing proof of “actual malice.”
*1247Bose Corp. v. Consumers Union of U.S., Inc., 466 U.S. 485, 511, 104 S.Ct. 1949, 1965, 80 L.Ed.2d 502 (1984). This review is not the same as de novo review.
There are, of course, many findings of fact in a defamation case that are irrelevant to the constitutional standard of New York Times Co. v. Sullivan and to which the clearly-erroneous standard of Rule 52(a) is fully applicable. Indeed, it is not actually necessary to review the “entire” record to fulfill the function of independent appellate review on the actual-malice question; rather, only those portions of the record which relate to the actual-malice determination must be independently assessed. The independent review function is not equivalent to a “de novo” review of the ultimate judgment itself, in which a reviewing court makes an original appraisal of all the evidence to decide whether or not it believes that judgment should be entered for plaintiff. If the reviewing Court determines that actual malice has been established with convincing clarity, the judgment of the trial court may only be reversed on the ground of some other error of law or clearly erroneous finding of fact.
466 U.S. at 514 n. 31, 104 S.Ct. at 1967 n. Si.
In conducting this independent review, however, the appellate court must have due regard for the ability of the finder of fact to view the witnesses’ demeanor and to make credibility judgments. In Bose, the Court held, “When the testimony of a witness is not believed, the trier of fact may simply disregard it. Normally the discredited testimony is not considered a sufficient basis for drawing a contrary conclusion.” Id. at 512, 104 S.Ct. at 1966. In Harte-Hanks Communications, Inc., v. Connaughton, 491 U.S. 657, 109 S.Ct. 2678, 105 L.Ed.2d 562 (1989), the Court stated,
In determining whether the constitutional standard has been satisfied, the reviewing court must consider the factual record in full. Although credibility determinations are reviewed under the clearly erroneous standard because the trier of fact has had the “opportunity to observe the demeanor of the witnesses,” the reviewing court must “ ‘examine for [itself] the statements in issue and the circumstances under which they were made to see ... whether they are of a character which the principles of the First Amendment ... protect.’ ”
109 S.Ct. at 2696 (citations omitted). The Court then approached this review of the record as follows.
Given the trial court’s instructions, the jury’s answers to the three special interrogatories, and an understanding of those facts not in dispute, it is evident that the jury must have rejected (1) the testimony of petitioner’s witnesses that Stephens was not contacted simply because Connaughton failed to place her in touch with the newspaper; (2) the testimony of Blount that he did not listen to the tapes simply because he thought they would provide him with no new information; and (3) the testimony of those Journal News employees who asserted that they believed Thompson’s allegations were substantially true. When these findings are considered alongside the undisputed evidence, the conclusion that the newspaper acted with actual malice inextricably follows.
491 U.S. at 689, 109 S.Ct at 2697. It is significant that in Harte-Hanks there was specific evidence presented which, if believed, refuted the respondent’s claims of actual malice. The Supreme Court found that the jury “must” have rejected that testimony and without it the jury’s finding of actual malice “inextricably” followed from the remaining evidence. Thus, the Court reviewed de novo the substance of the rejected testimony, because it involved the statements at issue in the case and the circumstances surrounding the making of those statements, but it did not attempt to substitute its own judgment regarding whether that testimony was credible. The jury’s rejection of the testimony, which was crucial to the outcome of that case, was reviewed for clear error and was permitted to stand.
Recently, the Ninth Circuit analyzed and applied these Supreme Court decisions in Newton v. National Broadcasting Co., 930 *1248F.2d 662 (9th Cir.1990), cert. denied, — U.S. -, 112 S.Ct. 192, 116 L.Ed.2d 152 (1991). With respect to credibility determinations, the court stated, “In sum, we read Bose and Harte-Hanks as creating a ‘credibility exception’ to the New York Times rule of independent review. However, a determination of actual malice cannot be predicated on the factfinder’s negative assessment of the speaker’s credibility at trial.” 930 F.2d at 671. Citing the Supreme Court’s decision in Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255, 106 S.Ct. 2505, 2513, 91 L.Ed.2d 202 (1986), the court in Newton noted that the factfinding function consists of “credibility determinations, the weighing of the evidence, and the drawing of legitimate inferences from the facts.” 930 F.2d at 671. The court noted that there is “a thin line drawn between highly deferential review of credibility determinations and less deferential review of the factfinder’s evaluation of other evidence relevant to the malice issue.”9
3. Analysis
The discussions in preceding sections of this dissent dealing with the majority’s recasting the Fortune article into broad generalizations which remove the sting of the article’s allegations, and analyzing those generalizations, are necessary as well to my consideration of the malice issue because of the interrelationship between actual malice and falsity. See Masson, — U.S. at-, 111 S.Ct. at 2443. In particular, I would refer again to the majority’s statement on page 1229 of the opinion that its examination in this regard is of the defendant’s basis for making the allegations as rewritten and condensed by the majority into the article’s “gist.” When the actual allegations of the article and the evidence in regard to them are considered, I believe Bressler presented “sufficient evidence to permit the conclusion that the defendant actually had a high degree of awareness of probable falsity.” Harte-Hanks, 491 U.S. at 687, 109 S.Ct. at 2696 (internal quotations omitted).
The reporters’ interview notes indicate that Haston explained to Fortune reporter Fromson in detail the circumstances surrounding NCR 5609. (J.A. at 1285). Ha-ston told Fromson that he thought the NCR disposition of “use as is” was a funny one, and that he did not want to sign it. (J.A. at 1286). • In his interview notes, Fromson paraphrased Haston’s statement as, “They tried to get me to sign even though I disagreed with the recommended solution. Higginbotham was getting calls from Bressler, the guy in charge of ASME codes and standards division of construction. I dont [sic] I don’t think Bressler ever directly threatened Higginbotham to pull the contract, but the fear that it might be pulled was always there for Hartford. Higginbotham had a saying, ‘Woe to the inspector who costs us a contract.’ ” (J.A. at 1286). Haston said the ANIs were constantly overridden by superiors in Atlanta, (J.A. at 1286-87), but he also recognized that this was a common occurrence throughout the nuclear industry. (J.A. at 1287). Haston also told the reporter that the penetration welds had undergone radio-graphic and hydrostatic testing, “so perhaps not a huge problem but they were never inspected [during the hydrostatic *1249test] and so did not meet code.” (J.A. at 1287).
At trial, Haston testified that on April 20, 1984, he had a conversation with Bres-sler regarding the welds covered by NCR 5609. (J.A. at 621). He also stated that he did not feel pressured by Bressler during that conversation and that he had no other conversations with Bressler regarding NCR 5609. (J.A. at 621, 623, 635-36). He testified that there was not a safety problem with the welds, but rather “a failure to meet minimum code requirements.” (J.A. at 628). He also testified that “[t]here’s nothing wrong with the welds.” (J.A. at 631). According to Haston, it was a strictly a matter of “procedure,” and he felt that as an ANI he was there to sign to show that code requirements had been met. (J.A. at 631). More significantly with respect to the malice issue, Haston testified that those facts are basically what he told the Fortune reporter, although he elaborated when speaking to the reporter. (J.A. at 637). Haston also testified that he told the Fortune reporters that it “wasn’t a matter of being forced or coerced.” (J.A. at 638). Haston testified that he told the Fortune reporter, Fromson, that he thought they (Fortune) were “wasting their time with this article, that there wasn’t a story there,” in response to which Fromson said (according to Haston’s paraphrase), “Well, that may be, but I have a deadline to meet, and this is the story I’m going to write.” (J.A. at 638).
I recognize that this last statement by Fromson is susceptible of differing interpretations, and that it fairly may be viewed as one aspect of the circumstances under which the defamatory statements were made. Accordingly, it is our duty as the reviewing court to examine for ourselves this comment of Fromson’s as it relates to the issue of actual malice. However, any reading of Harte-Hanks requires that we accord great deference to the jury’s right to determine the credibility of the witnesses — in this case Haston, who testified that Fromson had made the comment, and Fromson, who also testified at the trial— and to determine whether or not the comment was made at all, and if so, what, in light of those credibility determinations, was meant by it. Our duty to make an independent assessment of Fromson’s comment does not entitle us simply to disregard an entirely reasonable interpretation of that comment arrived at by the jury. If the jury believed that Fromson made the statement as Haston testified, and interpreted the statement as meaning that Fromson, and therefore Fortune, had no regard for the truth of the article’s statements, the jury was entitled to view all of the article’s accusations in that context. Since evidence concerning motive may bear some relation to the actual malice inquiry, Harte-Hanks, 491 U.S. at 667, 109 S.Ct. at 2686, and since I believe that the jury was permitted to interpret the statement as showing a reckless attitude on the part of Fortune, I consider Fromson’s statement to be important evidence of actual malice.
Haston further testified that when he first read the article, his impression was that it “was not based upon the conversation we had directly.” (J.A. at 646). He testified at some length that what was written in the article was not based on what he had told Fortune. (J.A. at 647-53).
In addition to Haston’s testimony and the reporters’ notes of his interview, plaintiff Bressler also relies heavily on the trial testimony of the reporters and the deposition of Fromson, parts of which were also read at trial. On cross-examination, Brian Dumaine admitted that the part of the article which discussed the suppression of inspection records referred to Bressler and NCR 5609. (J.A. at 254). Dumaine also admitted that the portion that indicated Haston was forced by TVA to sign the report also referred to Bressler. (J.A. at 255). Dumaine further admitted that the “coverup of safety questions” referred to Bressler (J.A. at 256), as did the reference to the campaign to force Haston to sign (J.A. at 257), along with the portion about complaints to Hartford from managers at TVA, (J.A. at 257). Also, Dumaine testified that the efforts to “cover up breach of safety standards” referred to NCR 5609. (J.A. at 260).
*1250As noted above, the facts established that the allegation that Bressler had tried to “suppress” inspection records is false. Dumaine knew when he wrote the article that he had no facts to support this statement. Dumaine admitted on cross-examination that the term “suppress” referred to NCR 5609 (J.A. at 254, 374), and that at the time he wrote the article he knew that there had been no physical suppression of NCR 5609 (J.A. at 373), that the document actually went to the Nuclear Regulatory Commission for review and that it was in TVA’s public files. (J.A. at 370, 412).
The statement that Bressler had forced Haston to sign the NCR was also shown to be false. Fortune argued at length about the “pressure” Bressler had put on Hartford managers, but Dumaine admitted that none of his sources was present during any conversations that Bressler had with Higginbotham or any other Hartford manager. (J.A. at 296-97). Haston said to the reporters that Higginbotham would probably perceive phone calls as threatening to Hartford’s contract (J.A. at 639, 662), but Haston also told them that he would have no basis for saying that there were any threats. (J.A. at 641).
The evidence also showed that the assertion that TVA “failed to test” the welds is false. The welds in Unit 1 (which were the only ones covered by NCR 5609 at the time the article was published) were both radio-graphically and hydrostatically tested. Du-maine knew this at the time the article was prepared. (J.A. at 282-83, 292-93). Although he testified at trial that the Unit 2 welds had not been tested (J.A. at 283), at the time the article was written Dumaine knew that the Unit 2 welds had been separated from the Unit 1. welds, and that the Unit 2 welds were covered by a separate report, NCR 6420. (J.A. at 399).
The evidence showed that the statement that Bressler tried to “cover up” safety violations is false. The testimony presented at trial indicates that NCRs were commonplace (J.A. at 619), and were in TVA’s public files. (J.A. at 408-09, 412). Also, the fact that NCR 5609 had been marked “significant” by the very persons who prepared it ensured that it would be sent to the Nuclear Regulatory Commission, and be reviewed by TVA’s Nuclear Licensing Staff. (J.A. at 1212, see J.A. at 718-22 (Bressler); at 369 (Dumaine)). Dumaine testified that he knew that the report had been designated “significant” by Bressler and TVA and that it had been forwarded to the Nuclear Regulatory Commission for approval. (J.A. at 369). It was Dumaine’s theory, however, that because TVA’s recommendation was to use the welds as is, they were covering up the “safety violation.”
Q. So you believe that the making this [NCR 5609] significant and sending it up to the Nuclear Regulatory Commission for approval was a coverup of this NCR 5609?
A. That’s right, the way it was done. That’s right.
(J.A. at 370). Dumaine wrote about a coverup despite the fact that the Fortune investigation revealed that there was no suppression of any of the records documenting the procedures and the code violation. And he characterized this perceived coverup as fact, not as his opinion that these procedures constituted .a coverup.10
While Fortune presented no evidence to substantiate the article’s statement about the welds’ possibly causing a serious nuclear accident, Dumaine testified that this was “common knowledge in the nuclear power industry.” (J.A. at 283). Nevertheless, Dumaine admitted that he would not be qualified to make the statement about a nuclear accident by himself, (J.A. at 284), and that if he had done so without someone trained in the nuclear industry giving him the information, it would have been reckless. (J.A. at 284-85). He also admitted that nothing in the editorial reference file supported that statement in the article. (J.A. at 398).
*1251The reporters’ own testimony and the other evidence relied upon by Bressler constitutes clear and convincing evidence that at the time the reporters wrote the Fortune article, they knew the facts did not support many of the allegations about Bressler. Thus, there was sufficient evidence before the jury from which it properly could conclude that the reporters included those allegations with reckless disregard for their truth.
On appeal, Fortune points to several items in the record to support its contention that there is insufficient evidence to support a finding that the reporters wrote the statements about Bressler with knowledge of any falsity or with reckless disregard for the truth of the statements. First, Fortune cites the testimony of From-son and Dumaine that they entertained no doubts about the truth of what they reported. (J.A. at 1013, 398, 405). Both of these men testified before the jury, and it is quite apparent that the jury must have rejected their claims about their own state of mind. While the jury would not have been entitled to draw the contrary conclusion merely because it discredited this testimony, it was entitled to determine the reporters’ actual state of mind from circumstantial evidence, Harte-Hanks, 491 U.S. at 667, 109 S.Ct. at 2686, and therefore, all of Bressler’s evidence, discussed above, is relevant to the malice determination.
Besides the testimony of its reporters, Fortune relies on the notes taken in interviews of Mansour Guity, Owen Thero, Bill Kemp, and Henry Myers. The interview notes of these sources are summarized and discussed below.
Owen Thero had been an investigator with Quality Technology Corporation (“QTC”), a firm that had experienced non-renewal of its contract with TVA. Du-maine’s interview notes paraphrased Thero as saying, “If you don’t play ball — Hartford could lose contract — TVA said not to investigate Hartford they would take care of it. This all happened in mid 1985 ...” (J.A. at 1188). Dumaine testified that The-ro also told him that Bressler was involved with the non-renewal warning, and that Thero’s statements were relevant to the entire situation with Bressler and Hartford, but not relevant to Haston’s signing the NCR, (J.A. at 297-300), though his interview notes do not reflect this. Of course, the jury was entitled to disbelieve Dumaine’s testimony to the extent it supplemented the notes he wrote at the time of the interview. In addition, since the article clearly refers to an alleged coverup in 1984, it is doubtful that any of Thero’s statements are relevant at all.
Henry Myers, scientific advisor to the U.S. House of Representatives, was also interviewed by Dumaine. His statement appears in paraphrase in Dumaine’s notes: “We received a briefing fro[m] Hayes about the progress of the report on the ANI investigation. They have found evidence supporting most of the NSRS allegations. That there was a violation of the ASME code. That Bressler at TVA pressured Higginbotham. That Higg told Robi-son to pressure Haston. That Robison did so. Bressler is the guy who pressured Hartford in Atlanta. What is new is that the OI at NRC is finding evidence to support the Guidy [sic] allegations.” (J.A. at 1278).11 This “evidence” was not identified.
Bill Kemp had also been an investigator with QTC. Kemp summarized for the reporter the circumstances surrounding NCR 5609, and also spoke about the alleged contact that Bressler had with Hartford’s Atlanta office, specifically with Harold Robi-son’s boss, William Higginbotham. (J.A. at 1279-83).12 The reporters’ interview notes indicate that Kemp told them that Bressler called Higginbotham to pressure him to have Haston sign the NCR. (J.A. at 1282). As Dumaine admitted, however, Kemp had no direct knowledge that such a phone call had taken place, nor did Kemp indicate specifically from whom he had discovered *1252this alleged fact. (J.A. at 302). In addition, Kemp suggested that Fromson contact Harry Jackson of ASME, whom Kemp termed a “straight shooter.” (J.A. at 1283). ■
Mansour Guity was a TVA investigator. The Fortune article reported that he had filed a lawsuit against TVA alleging that TVA retaliated against him for raising safety issues by harassing him and eventually taking him off an investigation that dealt in part with the events that were the subject of the Fortune article. The suit was decided by the Department of Labor in Guity’s favor. Dumaine interviewed him once, and Fromson interviewed him three times. Fortune’s characterization of Guity as one of the reporters’ “major sources” is quite accurate. (See Brief of Appellant, at 36).13 The notes of his interviews contain statements about Hartford managers being “forced” to sign off on the welds covered by 5609 (J.A. at 1207), that there was “pressure” exerted by TVA on Hartford’s home office (J.A. at 1253), that there was “apparent collusion between the Hartford regional managers [Higginbotham and Ro-bison] and the TVA managers [Bressler] to persuade the ANI to approve penetration welds which were not hydrostatically tested.” (J.A. at 1255). The notes also indicate that Guity had investigated the matter and “had collected a substantial amount of evidence to support the collusion allegation.” (J.A. at 1255). This “evidence” was not identified. The notes reflect that Guity spoke to the reporters in detail about the penetration welds (J.A. at 1256-57, 1258-59), and Haston’s problems with their being used “as is” and Haston’s being pressured into signing NCR 5609, (J.A. at 1257-59), and Bressler’s alleged involvement. (J.A. at 1260-62).14
To support its position on the malice issue, Fortune also refers to some of Ha-ston’s trial testimony, specifically parts in which he made limited confirmations of some of Guity’s conclusions. As noted above, however, most of Haston’s testimony and even Fromson’s interview notes of Haston clearly support Bressler’s position on the question of actual malice.
Guity was clearly the reporters’ key source. Although several persons vouched for Guity’s credibility, even his statements to the reporters did not support the extent of the article’s statements. While Guity spoke about pressure, the article specifically accused the TVA and Bressler of orchestrating a coverup and said that TVA and specifically Bressler had forced Haston to sign the NCR. In addition, to the extent that Guity told the reporters about what Haston said and what had happened to Haston, when Haston told them otherwise (as both the interview notes and his testimony demonstrate), they exhibited reckless disregard for the truth by reporting only Guity’s version, which was secondhand, to the exclusion of Haston’s firsthand account, and by failing to utilize Haston during fact-checking. Certainly reporters must make credibility determinations when dealing with differing versions of stories, and reporting the version of a person who is hostile to the plaintiff is not evidence of malice. Here, however, the reporters made a deliberate decision to report as fact only the secondhand version given to them by Guity, and they did so without independent corroboration of that version, and without indicating in any way that Haston himself, the person who was indisputably at the center of the controversy, had reported to them a very different version of the facts. This constitutes more than mere carelessness. See St. Amant v. Thompson, 390 *1253U.S. 727, 732, 88 S.Ct. 1323, 1326, 20 L.Ed.2d 262 (1968) (recklessness may be found where there are obvious reasons to doubt the accuracy of reports).
As for the other sources, none undermines the jury’s finding of actual malice. One very significant factor in this regard is that none of the reporters’ sources testified at trial. Thus, to the extent that the reporters attempted through their testimony to supplement the sources’ statements contained in the interview notes with other statements allegedly made by those sources, the jury was of course entitled to reject such supplements as a belated attempt to justify the article’s unfounded accusations. For example, once Dumaine’s testimony about what Thero said is separated from the notes of the Thero interview, all that is left is a general statement about TVA’s saying not to investigate Hartford, with no mention of Bressler at all. In addition, the testimony demonstrated that Bill Kemp had no direct knowledge of any pressure by Bressler.
What is left then, is the Myers interview notes. Although Myers’ statement is fairly specific about Bressler’s alleged involvement in pressure on Hartford, several factors make this evidence a weak reed on which to rely in contending that the jury was not entitled to find that the article was published with actual malice. First, Myers admittedly had no firsthand knowledge of any of Bressler’s actions or statements; all his information came from a man named Hayes, who had briefed the House Committee on Commerce and Energy, to which Myers was an advisor. Second, all the allegations Myers made are conclusory: no piece of evidence is mentioned to support the allegations and no person is identified as a source of information for any of the accusations against Bressler. Thus, while Myers stated that Hayes said that they have found “evidence” to support the NSRS allegations, Myers gave the reporter no basis to believe the information was reliable, and no basis on which to follow up and verify the allegations.15 Under these circumstances, and considering the fact that the person most directly involved in these events, ANI Haston, had given the reporters firsthand information that contradicted Myers’ statements, the reporters would be more than merely negligent to report as fact such unsubstantiated allegations. See St Amant, 390 U.S. at 732, 88 S.Ct. at 1326.
Fortune also contends that the finding of actual malice cannot stand because of Bres-sler’s and TVA’s alleged refusal to speak with the reporters about the article’s content. (Testimony of Marc Bressler, J.A. at 755-58; Testimony of Brett Fromson, J.A. at 1004). Fortune cites Bichler v. Union Bank & Trust Co., 745 F.2d 1006, 1016 (6th Cir.1984) (en banc); Bell v. Associated Press, 584 F.Supp. 128, 129 n. 4, 132 (D.D.C.1984); Trapp v. Southeastern Newspapers Corp., 10 Media L.Rep 1985, 1995-96 (S.D.Ga.1984); and Moorhead v. Millin, 542 F.Supp. 614, 619 (D.V.I.1982), to support this contention. However, these cases are clearly distinguishable from the present one.16 For example, in Bichler, the plaintiff’s “silent partner” in business knew that the information about the plaintiff that the defendant WZZM-TV was go*1254ing to broadcast was false, but when he spoke to the defendant’s news anchorperson prior to the broadcast he made no mention that it was false. 745 F.2d at 1008, 1016. In Trapp, the defendant’s reporter had contacted the plaintiff, who said “no comment” to the information that the reporter had told her about. 10 Media L.Rep. at 1995. Similarly, in Moorhead, plaintiff was confronted with the defendants’ information prior to publication, but had no comment. 542 F.Supp. at 619.
Here, Fromson at least17 called and left a message on Bressler’s answering machine. (J.A. at 755-58). However, Bres-sler testified that he handled this media contact according to a memorandum circulated at TVA which required employees to inform the media relations office. (J.A. at 757). Thus, unlike the plaintiffs in Moor-head and Trapp, Bressler was not apprised of the allegations prior to publication and he did not respond with a “no comment” to the allegations as did the plaintiffs in those cases. Nor did Bressler, as did the plaintiff in Bichler, have a person with whom he was closely associated and who knew the underlying facts fail to deny the allegations on his behalf.18
In addition, although both Fromson and Dumaine testified about the process they went through in checking the article’s content (J.A. at 984-1009; 1095), it is clear that the jury must have rejected their testimony. There is nothing in the record which would render the jury’s rejecting of that testimony clearly erroneous, and there are several factors which support the jury’s doing so: (1) Fromson could produce no written notes from this checking process, and testified that some may have been lost (J.A. at 1027-28); (2) the reporters were pressed for time;19 and (3) none of the sources with whom the reporters said they checked testified at trial to corroborate the reporters’ testimony in that regard. These credibility determinations were for the jury, and their decision may not be disturbed by this Court unless it is clearly erroneous. Thus, there is effectively no evidence that source checking was in fact performed. Under Harte-Hanks, while this negation of the reporters’ testimony may not be used as evidence to support the malice finding, its absence certainly undermines Fortune’s argument that the investigation was extensive.
Two other pieces of evidence also tend to support the malice finding. First, Kemp told the reporters to contact Harry Jackson, whom he identified as “Mr. Codes and Standards,” and a “straight shooter.” (J.A. at 1283). Fromson testified that he tried to reach Jackson, but was unable to. (J.A. at 1048). The jury was entitled to disbelieve this testimony. Haston testified that he told Fromson there was no story here, to which Fromson replied that he had *1255a deadline and that this was his story. (J.A. at 638). In addition, Fromson told Dumaine about Kemp’s suggesting a call to Jackson, and Dumaine made no effort to contact Jackson. (J.A. at 401-02). In response to counsel’s question of whether they “[j]ust left that area of understanding the ASME Codes and understanding from an expert construction nuclear power plant, just left that undone?” Dumaine responded, “That’s right.” (J.A. at 402). From this evidence the jury reasonably could have concluded that Fromson purposefully avoided calling Jackson so late in the investigation process because he might offer contrary evidence. As the Supreme Court has noted, “[A] deliberate decision not to acquire knowledge of facts that might confirm the probable falsity” of statements is support for a finding of actual malice. Harte-Hanks, 491 U.S. at 691-95, 109 S.Ct. at 2698-99.
Second, Fromson admitted that he did not even call Hasten as part of his fact-checking. (J.A. at 1029).20 Since Hasten clearly was at the center of the entire incident at issue here, and was the one who had by far the most firsthand knowledge about what occurred with NCR 5609, the jury reasonably could view the reporters’ failure to check with him as purposeful avoidance of the truth. See Harte-Hanks, 491 U.S. at 691, 109 S.Ct. at 2698.
Finally, although Fortune and the majority discuss the Nuclear Regulatory Commission’s Report of Investigation (see J.A. at 1694-1735) as supporting their conclusions, it is important to recognize that the Report was not issued until November 20, 1986, a month after the Fortune article was published. Obviously, then, the Report itself can have no bearing on the malice determination, since the issue is what the reporters knew at the time the article was published. However, even that Report does not support the specific allegations made by the Fortune article against Bressler. See supra note 5.
4. Summary
The jury obviously rejected the “good faith” testimony of Dumaine arid Fromson and their testimony about source checking. This they were absolutely entitled to do, as they are the judges of the witnesses’ credibility. We, as appellate judges, must defer to the jury’s determinations in this regard unless they are clearly erroneous, even in cases involving claims of defamation. See Harte-Hanks, 491 U.S. at 687, 109 S.Ct. at 2696. In performing our review of the record, once we consider what testimony the jury must have rejected — here it clearly rejected the testimony of Dumaine and Fromson — the next step is to examine what evidence remains that either supports or undermines the finding of malice. In the present case, there is scant evidence to undermine that finding and ample evidence te support it. In sum, there is evidence of convincing clarity that the reporters maintained serious doubts as to the truth of the statements they published about Bressler.
The interview materials that the reporters had before them at the time they wrote the article simply do not support the article’s numerous very specific, defamatory statements about Bressler. Thus, while the reporters may have believed in the truth of some aspects of the very general characterizations of the article which Fortune and the majority present here, the article contained very specific, defamatory allegations, which the evidence shows they could not have believed were true. Du-maine essentially admitted that he had no support for some of the article’s statements. In addition, the evidence relative to the absence of source checking and the failure to talk to Jackson provides further support for the malice finding. The jury was entitled to consider all of this evidence, together with Fromson’s statement to Hasten that he had a deadline and this was the story he was going to write, and conclude that Fortune published the article’s statements in reckless disregard of the truth.
*1256The majority here, however, has failed to accord the deference required by Bose and Harte-Hanks to the jury’s credibility determinations. Instead, the majority has, in my view, applied a standard more akin to that required in reviewing a summary judgment or directed verdict. Because there is in the record evidence that creates a genuine factual issue as to what the reporters’ subjective beliefs were, the majority finds that the judgment cannot stand. But the test is not whether there is evidence in the record which puts the facts regarding the reporters’ state of mind into genuine issue, nor is the test whether there is evidence which, if believed, would show that the reporters acted in good faith. The test is whether the jury, in believing some of the testimony and in rejecting other testimony was clearly erroneous, and if it was not, whether there is sufficient evidence in the record clearly and convincingly to establish actual malice. This Court is required to perform an independent review of the record in regard to actual malice to determine whether the plaintiff has proved actual malice by clear and convincing evidence. Where the record contains such evidence, to overturn the jury’s finding of actual malice because there is also evidence to the contrary, without regard to the jury’s right to disbelieve that contrary evidence unless it was clearly erroneous in doing so, is, in effect, to hold that the jury may not find actual malice unless there is no evidence in the record which, if believed, would support a finding of good faith. That is not the law.
III. Conclusion
As I noted at the outset, this is not a case in which Bressler can point to the presence or absence of one “key” fact as demonstrative of actual malice. Rather, in this case it is necessary to examine carefully all of the evidence, giving due deference to the jury’s ability to view the witnesses and make credibility determinations. With these factors in mind, and in the exercise of independent review of the record, I believe there is clear and convincing evidence that the reporters subjectively entertained serious doubts as to the truth, or had a high degree of awareness of the probable falsity of what was written in the article.
The majority arrives at the opposite conclusion, I believe, because of two fundamental problems with its analysis. The first is its basing its malice determination on its revised version of the article’s content. The second is its failure to apply the appropriate standard in independently reviewing the record to determine the issue of actual malice. In my view, the only way to justify a conclusion that Bressler did not show malice by clear and convincing evidence is to characterize or rewrite the article’s accusations and review the evidence entirely de novo as the majority has done. For, if one analyzes whether Bressler presented sufficient evidence from which the jury properly could conclude that he clearly proved that the reporters entertained serious doubts about the truth of the accusations as they appear in the article, it is apparent that he did. By revising the article’s content in the manner that it does, determining whether the revised version rather than the actual content is defamatory, and doing so without regard to the credibility determinations that the jury made, the majority engages in an unprecedented approach to the law of defamation, and, I believe, an insupportable analysis of the facts in this defamation case.
For the reasons set forth above, I would affirm the jury verdict and the judgment for plaintiff-appellee, Marcus Bressler, and the district court’s denial of Fortune’s motions for JNOV or a new trial. Therefore, I must respectfully dissent.
. In this opinion, I shall use "Fortune" to denote the corporate entity of Fortune Magazine, and “Fortune" to denote the publication itself.
. For example, in Harte-Hanks, the Court upheld a jury verdict for the plaintiff principally based on the editors’ failure to talk to a key person who would have discredited their story. 491 U.S. at 692, 109 S.Ct. at 2698.
. To the extent that the majority means by this statement that a plaintiff who is a minor figure in an article must prove false the gist of the entire article, the vast majority of which does not concern him, in order to prevail on a defamation claim for demonstrably false statements about him contained in that article, I cannot agree. No such situation was present in Masson, and the holding in Masson neither requires nor logically leads to that result. Indeed, were this so, a publisher or writer could avoid ' liability for defamation merely by inserting the defamatory remarks into a larger article, the vast majority of which is apparently truthful. Obviously, the law could not tolerate such a result.
. Citing Masson v. New Yorker Magazine, Inc., — U.S. -, -, 111 S.Ct. 2419, 2431, 115 L.Ed.2d 447 (1991), Fortune contends that the Supreme Court has recognized that the elements of falsity and actual malice are "inextricably intertwined.” (Brief of Appellant, at 44 n. 27). The Court in Masson stated,
The constitutional question we must consider here is whether, in the framework of a summary judgment motion, the evidence suffices to show that respondents acted with the requisite knowledge of falsity or reckless disregard as to truth or falsity. This inquiry in turn requires us to consider the concept of falsity; for we cannot discuss the standards for knowledge or reckless disregard without some understanding of the acts required for liability. We must consider whether the requisite falsity inheres in the attribution of words to the petitioner which he did not speak.
— U.S. at -, 111 S.Ct. at 2431. That the Court has recognized the interrelationship of the concepts, however, does not mandate the conclusion that the burden of proving falsity should be by "clear and convincing" evidence.
. This conclusion is in no way undermined by the Nuclear Regulatory Commission’s November 20, 1986 Report of Investigation. (J.A. at 1694-1735). That Report, released a month after the Fortune article was published, and relied on heavily by the majority, did not conclude that Bressler had pressured Haston, but rather that Haston’s own bosses, Higginbotham and Robison, had done so. (J.A. at 1729-30).
. Radiographic testing involves the X-raying of a weld and the examination of the X-ray by a technician. (See J.A. at 529-30, 562-66). Etzler testified that radiography is the most reliable method of testing welds. (J.A. at 529).
.Fortune attempts to make an issue out of the fact that some of the welds originally covered by NCR 5609 were located in Unit 2 and had not been tested at all. This fact was not disputed by the plaintiff at trial, but the evidence also demonstrated that those Unit 2 welds had been separated out from NCR 5609 and were covered by NCR 6420 several months before Fortune even began its investigation, (see J.A. at 740-44), and Dumaine knew this fact before the article was published. (J.A. at 399).
. The Nuclear Regulatory Commission agreed that the safety of Watts Bar was not compromised by the TVA’s disposition of the NCR 5609 welds. The NRC’s 1990 Inspection Report concluded that "TVA has adequately reviewed the issues associated with this request for alternative acceptance and that proposed alternatives to the code of record will not impair the integrity of the affected containment penetrations.” (J.A. at 1821).
. 1 The court also commented on the difficulty in deferring to the finder of fact on credibility determinations, while simultaneously upholding the court’s duty of performing an independent review of the record in order to protect speech.
[I]n a case involving core First Amendment values, we cannot ignore the risk that a jury's credibility determinations may also subvert those values. We must attempt to discharge our constitutional responsibility to protect First Amendment values without unduly trenching on the fact-finding role of the jury and trial judge. We are mindful that in New York Times, Bose, and Harte-Hanks, the Supreme Court was fashioning a process for reviewing the evidence which permits judicial protection of First Amendment values while still paying due deference to the fact-finding role of juries, and particularly the jury’s opportunity to observe the demeanor of the witnesses. Striking the proper balance is a daunting task.... The Supreme Court has left us with a line drawn between highly deferential review of credibility determinations and less deferential review of the factfinder’s evaluation of other evidence relevant to the actual malice issue. The line is a thin one, but tread it we' must to ensure that future speakers need not fear to rush in.
930 F.2d at 672.
. The term "coverup" was used three separate times in the article. This term is particularly defamatory because of its connotation in post-Watergate American society. A coverup indicates not only ethical or moral wrongdoing, but illegal activity.
. Dumaine’s testimony regarding the Myers interview is located at pages 303-306 of the joint appendix, and Fromson’s is at pages 963-65.
. Dumaine’s testimony about the Kemp interview is at pages 302 and 305-06 of the joint appendix,' and Fromson’s is at pages 965-67.
. The notes of his interviews are found at pages 1202-09 and 1249-62 of the joint appendix.
. Appellant Fortune also relies on testimony and exhibits which tend to buttress Guity’s credibility, specifically the views of Kemp, Myers, and Thero.
Fortune also relies on the deposition testimony of William Mason, Michael Kidd, and John W. Nelson. (See J.A. at 1343A-1344 (Kidd); at 1459-62, 1542-43, 1553-54 (Mason); at 1379-1402 (Nelson)). The deposition testimony of both Mason and Kidd reflect almost entirely what Mansour Guity had told them he had found in his investigation. Thus, their depositions shed little additional light on what specific evidence Guity had collected on the issues. The Nelson deposition likewise offers little in regard to the malice issue.
. Although the majority contends that the "draft” the of NRC report is properly considered as part of the malice inquiry (Opinion at 1231 n. 5), there is no evidence that Fromson saw a copy of this draft or even that he knew of its explicit contents. In fact, other than Fromson’s own testimony, which the jury was entitled to reject, the only evidence regarding the congressional investigation that was known by the reporters at the time they wrote the article is contained in the notes of the interview with Myers, which, as discussed above, consists of no more than a few sentences of conclusory statements with no supporting evidence.
By discussing the draft of the report, it seems to me that the majority is attempting artificially to implant the findings of the NRC report, which was released after the publication of the Fortune article, into the reporters’ stock of knowledge at the time they prepared the article. The evidence in the record does not support such an approach.
. Bell is inapposite. In Bell, although the court noted that the defendant had contacted the plaintiff football player’s team spokesman who declined to comment, it is clear that the court’s holding of no actual malice did not turn on the spokesman’s response. See 584 F.Supp. at 129 n. 4, 131-32.
. For Fromson’s testimony about his efforts to reach Bressler, see J.A. at 968-69, 1045, 446-47 (by deposition). Bressler admitted that he had received a message from a Fortune reporter on his answering machine. (J.A. at 755-58).
. The majority notes that Fromson testified he read the relevant portions of the article to TVA’s public relations officer who responded that TVA had no comment. Even if the jury chose to believe this testimony, it in no way undermines the jury’s finding that the statements about Bressler were made with malice. Bressler was never informed of the article’s statements, and he therefore had no opportunity to deny them. Thus, I see no analogy to the cases discussed above.
. Haston testified that Fromson told him on September 27, 1986, that he had a deadline and this was the story he was going to write. (J.A. at 638). According to the reporters’ testimony, nearly all of the investigation pertaining to NCR 5609 and Bressler took place on or after September 25, 1986. The article's first draft was completed on September 30, 1986, and the final draft, as it appeared in the magazine, was completed on October 3, 1986.
The hurry factor can cut the other way in some cases, as for example, when a newspaper reporter has “hot news.” Masson, — U.S. at -, 111 S.Ct. at 2435. That is clearly not the case here. The principal thrust of the article was not about Bressler, it was about TVA chief Admiral Steven White, and the Bressler parts of it easily could have been deleted if they could not be verified sufficiently before the deadline. In fact, the deposition testimony of Fortune editor Aloysius Ehrbar demonstrated that the portion of the article that discussed Bressler was originally not going to be part of the article itself, but would have been presented separately in a "side bar" or “box” somewhere within the main article. (J.A. at 463-64; see also J.A. at 467, 1626-45).
. Fromson also did not call Dorwin Etzler, the author of NCR 5609, as part of his fact-checking. (J.A. at 1057).