Appellant Robert Haddad appeals from the district court's judgment on his national origin discrimination claim, 42 U.S.C. Secs. 2000e et seq. (1976 & Supp. V 1981), and from the jury's verdict on his age discrimination in employment claim, 29 U.S.C. Secs. 621 et seq. (1976 & Supp. V 1981). Both claims arise from the same allegedly improper acts by appellee Lockheed. Despite the admission of improper evidence at trial, we affirm both the court's judgment and the jury's verdict.

Appellant Robert Haddad worked for appellee Lockheed California Corporation from early 1969 until his resignation in July 1979. Appellant claims that while in Lockheed's employ he was subject to a variety of forms of disparate treatment. This treatment, appellant alleged below, was the product of discrimination on the basis of national origin and age.

After filing a timely charge with the Equal Employment Opportunity Commission and receiving statutory notice of final action from the Commission, appellant initiated the present lawsuit in the Central District of California. Pursuant to 42 U.S.C. Sec. 2000e-5(f)(4) (1976), appellant's claim of discrimination based on national origin was tried by the district court judge. Appellant's age discrimination claim was submitted to a jury. Both claims were decided in favor of appellee Lockheed. Appellant brought this timely appeal.

Appellant bases his appeal on purported errors in the district court's jury instructions and evidentiary rulings. Two of appellant's claims on appeal merit only brief discussion. Appellant's third claim, involving the admission of testimony in violation of appellant's marital privilege, merits fuller consideration as it calls into question the proper standard for determining harmless error in a civil trial.

* Haddad argues that the district court improperly instructed the jury that in order to find for Haddad it had to find that Lockheed intended to discriminate. However, Haddad argued the case on a disparate treatment rather than on a disparate impact theory, and the evidence he submitted could only have supported the former claim. The district court's instruction was proper. See Heagney v. University of Washington, 642 F.2d 1157, 1163 (9th Cir.1981).

Haddad also argues that the district court improperly admitted hearsay testimony by Lockheed management regarding third parties' complaints about working with Haddad. However, this testimony was not hearsay: it was not offered to prove the truth of the complaints. See Fed.R.Evid. 801(c). Instead, this testimony was offered to show that Lockheed management had received complaints regarding Haddad. Such testimony was relevant in demonstrating Lockheed's non-discriminatory intent in its employment practices. See Fed.R.Evid. 803(3). The court's admission of the evidence was not an abuse of its discretion. See United States v. Patterson, 678 F.2d 774, 778 (9th Cir.) cert. denied, --- U.S. ----, 103 S. Ct. 219, 74 L. Ed. 2d 174 (1982).

Haddad next attacks the admission of certain testimony of his ex-wife, claiming that such evidence was protected by the confidential marital communication privilege. The privilege properly protects the marital communication about which the witness testified. See United States v. Lustig, 555 F.2d 737, 747 (9th Cir.1977). Lockheed failed to overcome the presumption in favor of applying the privilege. See In re Grand Jury Investigation of Hipes, 603 F.2d 786, 788 (9th Cir.1979). Therefore, the district court should have excluded this testimony.

Since this evidence is cumulative of other evidence in the record and the record contains no evidence to the contrary, we presume that the improper admission of this testimony had no effect on the court's decision rejecting Haddad's national origin discrimination claim. See Plummer v. Western International Hotels Co., 656 F.2d 502, 505 (9th Cir.1981). Thus, we affirm the court's decision on this claim.1

The improper admission of Haddad's ex-wife's testimony poses a more serious problem for the jury verdict on Haddad's age discrimination claim. A jury, unlike a judge, cannot be presumed to have based its verdict only on properly admitted evidence. See generally E. Cleary, McCormick on Evidence Sec. 60, at 137 (2d ed. 1972); 7 J. Moore, J.D. Lucas, Moore's Federal Practice p 61.07[2-3] (2d ed. 1982). Our task, then, is to determine whether the evidentiary error committed below affected a "substantial right" of appellant and so requires reversal of the age discrimination verdict. 28 U.S.C. Sec. 2111 (1976); Fed.R.Evid. 103(a); Fed.R.Civ.P. 61.

As an initial inquiry, we must determine what standard to use to determine whether the error in this case was sufficient to require reversal. Some errors involve "constitutional rights so basic to a fair trial that their infraction can never be treated as harmless error." Chapman v. California, 386 U.S. 18, 23, 87 S. Ct. 824, 827, 17 L. Ed. 2d 705 (1967); see, e.g., Payne v. Arkansas, 356 U.S. 560, 78 S. Ct. 844, 2 L. Ed. 2d 975 (1958) (coerced confession); Gideon v. Wainwright, 372 U.S. 335, 83 S. Ct. 792, 9 L. Ed. 2d 799 (1963) (no counsel); Tumey v. Ohio, 273 U.S. 510, 47 S. Ct. 437, 71 L. Ed. 749 (1927) (biased judge). A second type of constitutional error does not involve the fundamental integrity of the judicial process but does implicate the constitutional rights of the criminally accused. These errors must be shown by an appellate court to be harmless beyond a reasonable doubt. See, e.g., Chapman v. California, 386 U.S. at 24-25, 87 S.Ct. at 828-829 (1967) (drawing inferences in violation of the right against self-incrimination); Schneble v. Florida, 405 U.S. 427, 92 S. Ct. 1056, 31 L. Ed. 2d 340 (1972) (admitting testimony in violation of the confrontation clause).

It would be possible to end our inquiry here merely by distinguishing the error in the case at bar from the constitutional errors discussed in Chapman. This court first stressed the importance of such a distinction in dicta, United States v. Valle-Valdez, 554 F.2d 911, 915-19 (9th Cir.1977), and suggested that the harmlessness of trial error might be gauged under two standards: harmless beyond a reasonable doubt for constitutional errors and more probably than not harmless for non-constitutional errors. Id. The Valle-Valdez opinion, however, noted that ruling Ninth Circuit authority could be interpreted to require non-constitutional errors in criminal cases to be measured by the more rigorous "harmless beyond a reasonable doubt" standard. Id. at 916 (citing United States v. Rea, 532 F.2d 147 (9th Cir.1976); United States v. Duhart, 496 F.2d 941 (9th Cir.), cert. denied, 419 U.S. 967, 95 S. Ct. 230, 42 L. Ed. 2d 182 (1974)). Subsequent cases have relied upon the Valle-Valdez distinction between constitutional and non-constitutional errors as dispositive, and measured non-constitutional errors in criminal cases using a standard requiring reversal unless the error was more probably than not harmless. See, e.g., United States v. Berry, 627 F.2d 193, 201 (9th Cir.1980), cert. denied, 449 U.S. 1113, 101 S. Ct. 925, 66 L. Ed. 2d 843 (1981); United States v. Dixon, 562 F.2d 1138, 1143 (9th Cir.1977), cert. denied, 435 U.S. 927, 98 S. Ct. 1494, 55 L. Ed. 2d 521 (1978).2

Despite its convenience, we will not rely on the reasoning in Valle-Valdez to resolve the open question of what standard of harmlessness should prevail in a civil appeal.3 Neither the distinction between constitutional and non-constitutional error nor the strictly bifurcated standard of harmlessness it creates has been uniformly accepted.4 Courts frequently avoid relying on the distinction by finding harmlessness under all standards.5 Commentators have cautioned against making the standard for harmless error turn entirely on a distinction between constitutional and non-constitutional error.6 Finally, we are not certain that application of the more lenient standard to errors involving the improper admission of evidence in a criminal trial can be reconciled with the Supreme Court's language in Chapman: "An error in admitting plainly relevant evidence which possibly influenced the jury adversely to a litigant cannot, under Fahy [v. Connecticut, 375 U.S. 85, 84 S. Ct. 229, 11 L. Ed. 2d 171 (1963) ], be conceived of as harmless." 386 U.S. at 23-24, 87 S.Ct. at 827-828. Because of our concern that courts' adoption of this bifurcated standard of harmlessness may have occurred through misinterpretation of Chapman and Valle-Valdez, we decline to perpetuate the mistake by extending the analysis to civil cases. We therefore put this distinction to one side, and address directly the question before us: How probable must the harm from an error in a civil trial be before it affects substantial rights and thus requires reversal?7

The purpose of a harmless error standard is to enable an appellate court to gauge the probability that the trier of fact was affected by the error. See R. Traynor, supra, at 29-30. Perhaps the most important factor to consider in fashioning such a standard is the nature of the particular fact-finding process to which the standard is to be applied. Accordingly, a crucial first step in determining how we should gauge the probability that an error was harmless is recognizing the distinction between civil and criminal trials. See Kotteakos v. United States, 328 U.S. 750, 763, 66 S. Ct. 1239, 1247, 90 L. Ed. 1557 (1946); Valle-Valdez, 544 F.2d at 914-15. This distinction has two facets, each of which reflects the differing burdens of proof in civil and criminal cases. First, the lower burden of proof in civil cases implies a larger margin of error. The danger of the harmless error doctrine is that an appellate court may usurp the jury's function, by merely deleting improper evidence from the record and assessing the sufficiency of the evidence to support the verdict below. See Kotteakos, 328 U.S. at 764-65, 66 S.Ct. at 1247-48; R. Traynor,supra, at 18-22. This danger has less practical importance where, as in most civil cases, the jury verdict merely rests on a more probable than not standard of proof.

The second facet of the distinction between errors in civil and criminal trials involves the differing degrees of certainty owed to civil and criminal litigants. Whereas a criminal defendant must be found guilty beyond a reasonable doubt, a civil litigant merely has a right to a jury verdict that more probably than not corresponds to the truth.

The civil litigant's lessened entitlement to veracity continues when the litigant becomes an appellant. We conclude that a proper harmless error standard for civil cases should reflect the burden of proof. Just as the verdict in a civil case need only be more probably than not true, so an error in a civil trial need only be more probably than not harmless. See Saltzburg, The Harm of Harmless Error, 59 Va.L.Rev. 988, 1018-21 (1973). In other words, when an appellate court ponders the probable effect of an error on a civil trial, it need only find that the jury's verdict is more probably than not untainted by the error.

In the case at bar, appellant had a "substantial right" to a jury determination as to whether the treatment accorded him by Lockheed was more probably than not the product of age discrimination. Our task on appeal is to determine whether the evidentiary error of which appellant complains has deprived appellant of the degree of certainty to which he is entitled. Although not beyond a reasonable doubt harmless,8 the error in this case more probably than not had no effect on the jury's rejection of Haddad's age discrimination claim. We therefore hold the admission of privileged testimony in this case to have been harmless and we affirm the jury's verdict below.

Our assessment of the probable harmlessness of Mrs. Haddad's testimony rests partially on the relation between that testimony and the factual issues underlying Haddad's age discrimination claim. Mrs. Haddad improperly testified to the effect that in 1977 Haddad told her that he intended to resign from Lockheed and go into a real estate and consulting business with his brother. This testimony might have harmed Haddad by tending to undercut his claim that he was constructively discharged in 1979 and by casting suspicion on his work attitudes during his last two years at Lockheed. However, constructive discharge was just a small part of appellant's disparate treatment claim. Moreover, although Haddad's "abrasive" personality was frequently discussed, Haddad's commitment to his work was not a point of contention. The eight-day trial focused primarily on the specific Lockheed employment practices described in Haddad's complaint.

Haddad's age discrimination claim recited a catalogue of grievances: no promotion in salary grade in ten-and-one-half years; improper distribution of merit raises to those within his salary grade; the imposition of travel restrictions and record-keeping requirements; the imposition of telephone use restrictions; the rejection of appellant's suggestions for company improvements; the referral of appellant to the company doctor. To each of these specific complaints of disparate treatment, Lockheed responded with non-discriminatory explanations. For example, a very large portion of the trial, perhaps 20 to 30 percent, dealt with appellant's salary grade. Appellant stressed this topic; it was the grievance that appellant's counsel chose to describe first in his closing remarks. Lockheed's response on this issue was to show that appellant's salary grade was dictated by the position he filled and the constraints of Lockheed's contracts with the United States Navy. Lockheed also produced testimony of specific instances wherein appellant was advised that a higher salary grade would be available to him if he agreed to apply to another division within the corporation.

Mrs. Haddad's testimony was not only largely tangential to an evaluation of Lockheed's employment practices, it was also partially cumulative of other competent testimony regarding Haddad's attitudes towards Lockheed. Even without Mrs. Haddad's testimony, the jury was probably aware that Haddad was considering alternative careers in 1977. Many witnesses testified to Haddad's dissatisfaction with his salary during and prior to 1977. Haddad himself testified that he obtained a real estate license in 1977. Haddad also testified that he engaged in consulting and in the practice of real estate in 1979, immediately after leaving Lockheed. Mrs. Haddad's testimony may have reinforced the suspicion that Haddad was planning to leave Lockheed in 1977, but it probably did not create that suspicion.

The minimal extent to which counsel emphasized Mrs. Haddad's testimony to the jury supports our conclusion that the error in this case was more probably than not harmless. Counsel for Haddad and counsel for Lockheed had numerous discussions before the judge as to whether Mrs. Haddad was to testify at all. Nevertheless, these confrontations, although perhaps showing the value Lockheed placed on Mrs. Haddad's testimony, were out of the jury's hearing. Lockheed's counsel examined Mrs. Haddad for a very short period. The purpose of this examination seems to have been to show that Haddad was under substantial personal pressure during his last two years at Lockheed. In his closing argument, Lockheed's counsel mentioned Mrs. Haddad's testimony only briefly, suggesting that it showed Haddad's emotional strain in 1977. The overwhelming thrust of Lockheed's closing argument, like the overwhelming bulk of the trial, dealt with a point by point analysis of Lockheed's employment practices.

AFFIRMED.