Allstate Insurance Company and its related companies appeal from a judgment in favor of Stewart Turner on his wrongful discharge claim. Allstate raises three issues:

1. Whether the trial court erred by instructing the jury that, under Michigan law, Allstate had the burden of proving just cause for terminating Turner;

2. Whether the trial court erred in refusing to give Allstate's requested business judgment instruction; and

3. Whether the trial court abused its discretion in refusing to allow Allstate to introduce certain statistical evidence.

We conclude that none of the assignments of error warrant reversal and, therefore, we shall affirm the judgment entered in accordance with the jury's verdict.

Stewart Turner was employed as an insurance agent by Allstate from 1968 to 1986 under a written contract drafted by Allstate. Among the terms and conditions of the employment agreement was a provision prohibiting Turner's termination except for "just cause." An "act of dishonesty, such as, ... falsification of any Company or industry plan documents...." was specifically listed as "just cause" for employment termination.

On January 6, 1986, Allstate fired Turner on the ground that he had falsified internal company documents by submitting to the company Allstate's Customer Service Request Form falsely representing that a number of Turner's customers were entitled to a discount for homeowner's insurance policy premiums. The discount is known as the "protective device discount" ("PDD"). The discount program, which Allstate introduced in 1981, offered a five percent discount on every homeowner's policy if the customer's home was equipped with a smoke alarm on each floor excluding the basement, deadbolt locks on exterior doors, and a fire extinguisher on the premises. The savings on a typical homeowner's policy amounted to approximately $7.50 per year.

Following his termination, Turner sued Allstate for breach of contract. At trial, neither party disputed the terms of the employment agreement and Turner stipulated that Allstate's condition that he had falsified company documents was indeed the reason Allstate fired him, and not a mere pretext. Thus, the only liability issue to be litigated was whether Turner had falsified a company document.

The first and second issues raised on appeal concern, respectively, the correctness of the trial court's instructions to the jury concerning the burden of proof of just cause or its absence and Allstate's request for a business judgment instruction.

Over Allstate's objection, the trial court assigned to Allstate the burden of proving that it had just cause for discharging Turner. The court also instructed the jury that it could consider only whether Turner had falsified a company document, not whether it believed that Allstate's decision to discharge Turner was reasonable. The jury was instructed that if it found a single falsification, it was to find in favor of Allstate.

The third issue concerns the admissibility of comparative statistical data Allstate compiled relating to the percentage of Allstate customers in the Midland area, and statewide, who had been given the PDD.

Neither party disputes that Michigan law governs the contract issues in this case; however, the parties disagree as to who has the burden of proving the existence or absence of just cause for termination of an employment contract in Michigan. Turner argues that, in Michigan, the burden is assigned to the employer. Allstate insists the plaintiff has the burden of proving that Allstate lacked just cause to terminate his employment and, at trial, requested that the trial court give the following instruction:

Since plaintiff had a contract with Allstate, he must then prove that he was performing according to the terms of that contract, up to the time of his termination. That is, he must prove by a preponderance of the evidence that he was not violating the terms of the contract by falsifying company documents. If the plaintiff does not prove that he was not falsifying company documents then your inquiry is over and then you must find for the defendant.

(Emphasis added.) The district court disagreed and assigned the burden to the defendant, giving the following instruction:

The plaintiff has offered evidence that he performed his contract in all respects up to the time of discharge. This evidence is undisputed. The burden then shifted to defendant to prove that it had legal cause for the plaintiff's discharge.

* * * * * *

If the defendant proves that the plaintiff falsified at least one document within the meaning of the contract, then you shall return a verdict for the defendant.

If the defendant does not prove that the plaintiff falsified at least one document, then you shall return a verdict for the plaintiff.

We conclude that the district court did not err in so instructing the jury.

The court's jury instruction correctly states the law as announced by the Michigan Supreme Court in Johnson v. Jessop, 332 Mich. 501, 503, 51 N.W.2d 915, 917 (1952), and Saari v. George C. Dates & Associates, Inc., 311 Mich. 624, 628, 19 N.W.2d 121, 122-23 (1945). Those cases declare that proof of "just cause" for termination of an employment relationship is to be treated as an affirmative defense on which the defendant has the burden of proof.1 A more recent decision by the Michigan Court of Appeals has reiterated the Michigan rule.2 None of these cases have been overruled.3

Allstate argues that Michigan law on the subject is not settled. It claims that at least two cases, Obey v. McFadden Corp., 138 Mich.App. 767, 776-79, 360 N.W.2d 292, 296-97 (1984), lv. denied, 422 Mich. 911 (1985), and Duke v. Pfizer, 668 F. Supp. 1031, 1040 (E.D.Mich.1987), aff'd, 867 F.2d 611 (6th Cir.1989), hold that, in a breach of employment contract action under Michigan law, the employee has the burden of proving that the employer breached the contract by discharging the employee without just cause. We do not deem these decisions dispositive.

The language in Obey, on which Allstate relies and to which the court in Duke made reference, is a quotation from the trial court's jury instructions to the effect that the plaintiff bears the burden of showing that the employer "breached a contract by terminating [the employee] without good cause." Obey, 138 Mich.App. at 776, 360 N.W.2d at 296. However, the correctness of that instruction was not at issue in Obey. Toward the close of its decision, the Obey court said that, "[h]aving [been] shown a good cause contract, ... [we] must address the question ... whether sufficient reasons had been shown to justify [plaintiff's ] termination." 138 Mich.App. at 779, 360 N.W.2d at 297 (emphasis added). It then held that the trial court should have granted the employer a directed verdict because the plaintiff conceded that he had engaged in illegal transactions for which his suspension and ultimate termination was proper. Id., 360 N.W.2d at 297. The court's subsequent statements assessing the evidence as failing to show the employer breached a "contractual duty to discharge plaintiff in a timely manner," id. at 778-79, 360 N.W.2d at 297-98, are dicta which do not contradict the Michigan Supreme Court's Johnson, Saari, and Milligan decisions on the burden of proof issue.

Duke, a federal trial court decision, likewise sheds no helpful light on the issue involved here. In discussing its understanding of Michigan law concerning the burden of proof in wrongful discharge actions, the district court stated that the "rule plaintiff excises from Rasch, supra," assigning the burden of proof of just cause to the employer, "is counter intuitive and makes little sense." 668 F. Supp. at 1040. That observation, whatever it is worth, does not comport with settled Michigan law as made explicit by the Michigan Supreme Court in Johnson, Saari, and Milligan, and reiterated by the Michigan Court of Appeals in Rasch. In all events, the district court's observation is a mere dictum since its decision did not turn on that point.

We think the district court correctly applied Michigan law in assigning the burden of proof on just cause to Allstate.

Allstate claims the trial court should have instructed the jury that it could not question Allstate's business judgment that falsification of company documents was just cause for termination under the contract. While the trial court declined to give Allstate's requested jury instruction, we think the charge it gave contained all the essential elements Allstate requested and, what is more important, correctly stated the law.

A comparison of Allstate's requested instruction and the trial court's charge confirms that the court merely distilled Allstate's proposed instruction to its essential elements by eliminating argumentative language. The instruction Allstate requested stated:

You are not to substitute your judgment for the defendant's business judgment that just cause is dishonesty, defined as the falsification of company documents, or to decide this case on what you would have done if you had been the employer.

You should keep in mind that the defendant is entitled to take all reasonable steps to insure the Allstate Insurance Company operates in a smooth, efficient and businesslike manner.

Unless you find by a preponderance of the evidence and in accordance with the other instructions that I have given, that Allstate's actions or decisions constituted a breach of contract, then the acts and decisions cannot be found to be a violation of law, whether you agree with them or not. Thus, your attention should be on whether the defendant's actions or decisions constitute a breach of contract, and not on the business decision made.

It was a term of the contract that plaintiff's employment could be terminated for falsification of company documents within the meaning of paragraph six of the written contract.

* * * * * *

The fairness or reasonableness of the contract is not an issue.

* * * * * *

If the defendant proves that plaintiff falsified at least one document within the meaning of the contract, then you shall return a verdict for the defendant.

Allstate received exactly what it sought: an instruction precluding the jury from considering the fairness or reasonableness of Allstate's business decision to reserve to itself, by contract, the right to terminate an employee for any act of dishonesty.

Allstate assigns error to the district court's refusal to receive in evidence its statistical data offered to prove, circumstantially, that Turner falsified company documents representing that certain of his customers were qualified for the PDD. We reject this argument.

Allstate's conclusion that Turner had falsely represented the number of his policyholders qualified for the discount was based on an investigation which began approximately seven months before Turner was discharged. The investigation included a telephone survey of a number of policyholders in the Midland, Michigan area who had recently received the PDD, and a statistical comparison of the percentage of Turner's customers who had been given the PDD with the percentage of such customers in other Michigan areas and statewide. The data compiled by Allstate indicated that some 74.4% of Turner's Midland area policyholders had been given the PDD, while the statewide average was about 20%. There were thirteen locations in the state other than the Midland area in which the PDD had been given to as many as 30% of the policyholders, and only five areas in which as many as 40% of the policyholders were granted the PDD. On the basis of this comparative data, Allstate concluded that Turner must have knowingly given the discount to customers in the Midland area who did not qualify.

In support of its claim, Allstate introduced the testimony of approximately a dozen Turner policyholders, and offered the statistical survey data described above. The policyholders who testified stated that either they had no recollection of Turner ever advising them of the PDD or they did not have all the devices in their homes to qualify for the discount.

While the trial court allowed Allstate to show that 74.4% of Turner's book of property insurance had the PDD, it refused to allow any evidence of the statistics for other Allstate agents or offices, thus precluding any showing of comparison. Ostensibly, Allstate's purpose in offering the copious statistical data was to show the grossly disproportionate percentage of Turner policyholders who were given the PDD as against the statewide average or the average in other areas, suggesting that Turner had falsified the company documents certifying to the eligibility of a number of his customers. However, when the comparison figures were first offered in evidence, Allstate represented that they were admissible to show the reasonableness or good faith of Allstate's conclusion that Turner was falsifying his customers' eligibility for the discount. The court ruled, correctly we think, that the evidence was irrelevant under Fed.R.Evid. 4024 for that purpose because Turner conceded Allstate's good faith in firing him and only disputed the claim that he had falsified any documents. Since there was no longer any issue of good faith firing in the case, the evidence was properly excluded on that theory. See United States v. Dunn, 805 F.2d 1275, 1281-82 (6th Cir.1986).

Allstate then offered the statistical data to show that such a disproportionately high percentage of PDD eligible customers in Turner's insurance book suggested circumstantially that the customers were not in fact eligible and that the documents declaring they were were falsified. The trial court rejected the evidence for that alternate purpose not on the grounds of irrelevancy, but because the statistical data was substantially more prejudicial than probative under Fed.R.Evid. 403.5

The trial court noted that Midland differed from Muskegon and the other cities in Allstate's survey and, therefore, could not be compared as though identical. The court also noted that there was no evidence at trial that Allstate trained its sales agents throughout the state uniformly in the matter of offering the PDD and determining whether customers were qualified for it. In fact, the court noted, the evidence was that Allstate's agents from the same district apparently had different opinions as to what qualified a customer for the PDD. Further, there was nothing to suggest that the averages presented for comparisons were based upon the percentage of customers eligible for the discount. Instead, they were based entirely upon customers who actually received the discount. Observing that there was an insufficient basis for comparing Turner's book of PDD qualified customers to Allstate's averages for other agents in other geographic areas of the state, the district court concluded that even if the evidence was minimally probative, its potential for confusing and misleading the jury outweighed its probative worth.

The trial court revisited Allstate's arguments for the admissibility of the statistical data many times during the course of the trial. Each time Allstate raised the issue of the exclusion of its data, it began by arguing that the evidence went to the issue of falsification but invariably returned to its initial position that the evidence was relevant to show that Allstate fired Turner in good faith. The trial court was not convinced by Allstate's shifting argument. Each time it addressed the matter, the court recognized that whether Allstate believed or had reason to believe that it had good cause to discharge Turner was irrelevant. Moreover, the court noted:

[The evidence that Allstate sought to introduce] has very little probative value [because] we're not concerned in this case so much with how many people had deadbolts and how many people had these things.

* * * * * *

... You keep wanting to get it in, and every once in a while, about every third sentence, it sneaks out that what you really want to do is you want it for its subjective aspect, and not to prove what Mr. Turner did or did not do.

It is more prejudicial because it would require a showing of what other people did. If you want to go into what they did in Detroit or what they did in Grand Rapids, then we would have to know whether the agents conducted these surveys, did this certain thing or not. We would have to go into whether there are differences in character in a community, the age of all the houses, all kinds of things.

It would just turn this case into a month-long trial, and still wouldn't change the hard fact that what counts is he did or did not falsify the records in a way which constituted dishonesty.

The court balanced the limited probative value of the evidence against the likelihood that the statistics would confuse the jury. Even the defense counsel appeared to be confused about what the evidence was being offered to prove. As the district court recognized in its opinion denying Allstate's motion for judgment notwithstanding the verdict:

[T]he probative value of the evidence depended upon too many inferences and assumptions, some of which were proven to be false:

1. The company had uniform practices and policies for interpretation of its rule;

2. The agents everywhere uniformly were informed of these interpretations;

3. That agents followed the uniform system and practice of determining the eligibility of policyholders for discounts.

The undisputed testimony from the agents called by both parties was that, in the Saginaw area, there was a great deal of confusion concerning such matters as credit card deadbolts, smoke alarms in basements, fire extinguishers in garages, and the effect of a promise to acquire protective devices for a newly purchased home. The evidence showed that there was little financial incentive, if any, for an agent to give a discount except with respect to prospective customers shopping for price. There was no basis in evidence, including the offer of proof, that agents in other areas made any effort to contact the existing policyholders to offer the discount, nor whether the policyholders reflected and the proffered statistics were new or old. Thus the probative value of the statistics was marginal at best.

Our standard of review is whether the trial court abused the discretion it enjoys under Fed.R.Evid. 403 to exclude otherwise relevant evidence whose "probative value [it finds] is substantially outweighed by the danger," in this case, of "confusion of the issues." See United States v. Zipkin, 729 F.2d 384, 389 (6th Cir.1984).

There is no question that the district court correctly ruled that the statistical data was inadmissible to prove Allstate's good faith in terminating Turner. Good faith was immaterial in the case or, in the language of Fed.R.Evid. 401,6 was not a "fact [of] consequence to the determination of the action," because Turner had conceded Allstate's good faith in firing him. It was the alleged act of misconduct in falsifying the company documents that Turner disputed. In support of its just cause defense, Allstate's burden was to prove that Turner falsified at least one company document. The testimony of Turner's policyholders was some evidence in support of its claim. In addition, the comparative statistical data indicating that Turner had a disproportionately higher number of policyholders who had been issued the PDD than was the case in other areas of the state may have had a slight tendency to show, circumstantially, that Turner's customers were not, in fact, qualified, and that he falsified documents declaring that they were. However, the district court found that

[a]lthough evidence as to how many policyholders actually had the discount in other areas as compared to plaintiff's customers has some limited probative value with respect to falsification, the probative value of the evidence depended upon too many inferences and assumptions, some of which were proven to be false:....

And,

Thus the probative value of the statistics was marginal at best. If the statistics would have been admitted, on the other hand, the jury may well have been confused by them and may have assigned an inappropriately high decree of reliability to them.

After a careful examination of the entire record, we are inclined, indeed obligated, to defer to the trial court's assessment whether the statistical data had a serious potential for confusing the jury and being misinterpreted. The district court concluded that it did, and that, as a result, the probative value of the data was substantially outweighed by its danger for causing unfair prejudice. Fed.R.Evid. 403. We find no basis to conclude that the district court's evidentiary ruling was an abuse of discretion.

For the foregoing reasons, we AFFIRM the judgment entered below.

I respectfully dissent from the affirmance of judgment in this case against the defendant-employer, Allstate Insurance Company (Allstate). I believe that there is ample authority in Michigan to support the universal rule that the plaintiff Turner has the initial, and remaining, burden of proof to establish his stated claims in his complaint: (1) that he at all times performed his job as an insurance salesman "faithfully;" (2) that he "followed Defendant's instructions;" (3) that he was told that he would "continue to be employed by defendant so long as he performed his assigned tasks satisfactorily and not be terminated except for good cause;" and (4) that defendant "breached the total employment contract by terminating" him.

The defendant denied that plaintiff was entitled to any recovery; denied that plaintiff performed faithfully his job duties; denied that plaintiff carried out the "responsibilities that he agreed to perform;" denied that it breached any employment agreement as alleged; and then stated affirmatively that it terminated plaintiff for "good cause," claiming a breach of agreement and fraud on the part of plaintiff, and submitting a counterclaim for damages accordingly.

In light of the pleadings of both parties, it is clear to me that plaintiff had the burden to prove all of the four principal contentions herein set forth in order to establish a breach of agreement on the part of defendant Allstate. The latter, on the other hand, had to establish, by a similar burden, its claim, equivalent to fraud, for damages. The district court and the majority have confused the respective burdens imposed upon the parties on the pleadings and have prejudiced defendant by this error. In effect, plaintiff is claiming a "satisfaction" contract that he would be retained so long as he faithfully performed and carried out defendant's instructions as a commission sales agent.

Under such an arrangement, whether plaintiff satisfactorily performed and carried out his part of the bargain "is a question to be determined by the defendant and not the jury" or the trier of fact. Lynas v. Maxwell Farms, 279 Mich. 684, 273 N.W. 315, 317 (1937); see also Brown v. Chris Nelsen & Son, 10 Mich.App. 95, 158 N.W.2d 818, 820 (1968); Schroeder v. Dayton-Hudson Corp., 448 F. Supp. 910, 916 (E.D.Mich.1977). Plaintiff claimed, in essence, that he carried out his duties in a satisfactory manner--he has the burden of proving this. Defendant asked for an instruction that Turner "must prove that he was performing according to the terms of that contract up to the time of his termination." I believe that is a proper instruction and should have been given. Instead, the district court refused this proffered instruction and charged the jury that since plaintiff had presented evidence of proper performance, "the burden then shifted to defendant to prove it had legal cause for the plaintiff's discharge." In this respect, I would find this instruction and shifting of the burden of proof to be prejudicial and reversible error.1 In sum, plaintiff pled a cause of action for defendant's breach of contract by discharging him; defendant denied this and put plaintiff to his proof, claiming that plaintiff had indeed breached the employment agreement. Plaintiff had the continuing ultimate burden to prove the alleged breach of contract, essentially a question of satisfactory performance.

The majority relies first on Saari v. George C. Dates & Assocs., 311 Mich. 624, 19 N.W.2d 121 (1945), in support of its position. Saari held that plaintiff, who sued for breach of an employment contract, had failed to make out his claim in a nonjury case; the Michigan Supreme Court affirmed. Saari made out a prima facie case on his proof, but the defendant produced evidence of unsatisfactory performance of duties. The court held:

"The general rule is that the burden of proof lies on the party who 'makes the averment even though it be a negative one.' "

. . . . .

In a case tried by the circuit judge without a jury we do not reverse unless the evidence clearly preponderates in the opposite direction. On the record before us the defendant has justified the discharge of plaintiff under terms of their contract by sufficient proof of breach of its provisions by the plaintiff.

19 N.W.2d at 123.

The majority also relies upon Johnson v. Jessop, 332 Mich. 501, 51 N.W.2d 915 (1952). Citing Milligan v. Sligh Furniture Co., 111 Mich. 629, 70 N.W. 133 (1897), the Johnson court held that "the burden of proof was, of course, upon him [the plaintiff] to prove his [employment] contract and its performance up to that time [of discharge]." 51 N.W.2d at 917. Because the defendant made a "frank indication of the real reason why he terminated plaintiff's employment"2 and did not show unsatisfactory performance, judgment for plaintiff was affirmed. Id. at 918.

I agree with the statement in Diggs v. Pepsi-Cola Metro. Bottling Co., 861 F.2d 914, 931 (6th Cir.1988), that an employer's promise to retain an employee for so long as his performance is satisfactory "does not give rise to a 'just cause' contract." (Krupansky, J., concurring and dissenting).3 Plaintiff here asserted that his contract was based on his faithful performance and following instructions, equivalent then to a satisfaction contract, and this is different from a "just cause" contract. Toussaint v. Blue Cross & Blue Shield of Michigan, 408 Mich. 579, 292 N.W.2d 880 (1980).

Plaintiff attached the contract to his complaint. It nowhere reflects this term, "good cause," but rather refers to "unsatisfactory work" as a basis for termination subject to an opportunity of notice and review except in case of "falsification of any Company documents." This latter exception, however, applied only to notice of unsatisfactory performance and review procedure. Allstate contended that what Turner did to gain commissions in granting unwarranted discounts was at least unsatisfactory performance. On this, I maintain, whether performance was or was not satisfactory, Turner had the burden of proof. On defendant's further affirmative claim that it had just cause to terminate because of fraud and falsification, involving specific bad intent on plaintiff's part, Allstate had the burden.

This vital distinction was not recognized by the trial court, nor is it recognized by the majority opinion. See Obey v. McFadden Corp., 138 Mich.App. 767, 360 N.W.2d 292 (1984); Duke v. Pfizer, Inc., 668 F. Supp. 1031 (E.D.Mich.1987), aff'd, 867 F.2d 611 (6th Cir.1989). I believe that Rasch v. City of East Jordan, 141 Mich.App. 336, 367 N.W.2d 856 (1985), is not to the contrary in requiring plaintiff to prove the contract and his satisfactory performance up to the time of termination.

I would reverse the verdict for plaintiff for the reason indicated on improper allocation of burden of proof. I also would grant a new trial because the district court improperly refused to permit defendant to introduce relevant and pertinent statistical evidence that was vital to a resolution of the issues in this seriously contested case. The district court precluded Allstate from demonstrating that Turner, under highly questionable circumstances (in light of testimony of twelve of his customers that they did not qualify for a discount), was giving almost four times as many PDD discounts as the statewide average. This statistical evidence bore upon whether the performance of Turner was normal, reasonable, and satisfactory in relation to all other Allstate commission agents in Michigan. It certainly was relevant as to whether Turner might be, in effect, shorting his employer, Allstate, or even defrauding it. It was, in my view, an abuse of discretion and prejudicial to place a burden of proof upon Allstate and then deny it an opportunity to present highly relevant proof in support of its position. The trial court was simply wrong in concluding that this evidence had "very little probative value" or that Allstate wanted it in for its "subjective aspect," whatever that is, or that its value was "marginal at best."

For these reasons, I would reverse and remand for a new trial.