PATRICK E. HIGGINBOTHAM, Circuit Judge:
A Texas jury found that Lone Star Gas Company discriminated against Melvin Walther because of his age. Lone Star here contends that the district court should have entered judgment notwithstanding tile jury’s verdict because Walther failed to establish a prima facie case. Alternatively, Lone Star argues that errors in the court’s jury instructions require a new trial. We find sufficient evidence to support the jury’s verdict and affirm the denial of a new trial. We are not persuaded that any error in the court’s instructions to the jury warrant reversal. We vacate the district court’s award of front pay, however, and remand for reconsideration.
I.
In 1986, Melvin Walther was a fifty-year-old regional office manager for the Green-ville region of Lone Star Gas Company. During his thirty year tenure with the company, Walther had received regular promotions and pay increases and was never told that his job performance was unsatisfactory. On the contrary, his personal file contained a number of favorable reports and letters of commendation. In 1986, however, Lone Star discharged Walther as part of a six percent reduction in force. Although Walther was first told that his job had been eliminated, he later learned that Lone Star replaced him as Greenville office manager with another employee from the Dallas office who was older than Walther.
Walther sued Lone Star, contending that he had been discharged because of his age in violation of the Age Discrimination in Employment Act, 29 U.S.C. § 623(a). Lone Star introduced evidence tending to show that Walther’s termination was part of a carefully considered program to reduce costs at the company by selectively eliminating those employees the company could most easily do without. This program responded to fundamental changes in the economic conditions of the energy industry. According to Lone Star, it terminated Walther for valid business reasons based on his performance record and not because of his age. Roy Roberts, vice-president of the Dallas distribution division, testified that he listed Walther for termination because he had received oral reports that
Walther, although competent and qualified, “was very, very weak in construction and maintenance and lacked ... leadership skills.”
Walther, on the other hand, relied heavily on evidence that several other younger, less qualified employees whose positions had been eliminated were retained and transferred or promoted to positions that Walther was qualified to fill and had in fact filled in the past. Charles Ferguson, director of regional operations for Lone Star and Walther’s immediate supervisor, testified that Walther was more qualified to fill some of these positions than those who were selected. Walther also testified that he was more qualified than these younger individuals. Thus even though Walther’s position as regional office manager was taken by an older employee, Walther argued that the fact that he was denied the opportunity to continue in his job or to fill other vacancies despite relatively strong qualifications showed that he was discharged because of his age.
Walther also sought to impeach Lone Star’s assertions that it included him in the RIF based on his performance record. Roberts, the individual listing Walther for termination, admitted that he did not consult any performance evaluations in making the decision but relied on prior oral reports. Nor did he use any other documents to compare Walther’s performance to that of other employees. Nor did he consult Ferguson, Walther’s immediate supervisor, at the time he made his decision. Of twenty-seven people on Roberts’ list for termination, only two were under the age of forty.
Walther also introduced statistical evidence showing that of seventy-four employees over the age of forty in the Greenville region, twelve were terminated in the RIF, whereas of ninety-three employees under forty in the region, only one was terminated in the RIF. The probability of obtaining this result by chance was apparently .00027. Similarly, of 1,093 employees over the age of fifty company-wide, 94 were terminated and not rehired, whereas of 3,184 employees under the age of fifty, 151 were terminated and not rehired. The probability of obtaining this result by chance was less than four in a million.
Lone Star countered with its own statistical evidence indicating that before the RIF, individuals over forty years old composed 44.67 percent of the company work force, and after the RIF, individuals over forty still constituted 44.25 percent of the workforce. Lone Star argued that this .4 percent reduction in employees over forty demonstrates that the RIF did not significantly impact older employees. Lone Star’s data also showed that although individuals from age fifty-five to sixty-nine were discharged in disproportionately large numbers, so were individuals from ages twenty to twenty-nine. The middle group, constituting employees aged thirty to fifty-four, were discharged in disproportionately small numbers. Furthermore, employees in the smaller category of age fifty to fifty-four, such as Walther, were discharged in disproportionately small numbers. It is only when one combines this smaller group with those employees aged fifty-five and over that a disproportionately large impact emerges. In sum, Lone Star sought to show that the grouping of the employees determines whether an adverse impact from the RIF is demonstrated.
II.
Lone Star first contends that because Walther was replaced by a person in the protected group — a person actually older than him — he cannot establish a prima facie case of age discrimination. This case was fully tried and a jury verdict was rendered. The Supreme Court has explained that “[wjhere the defendant has done everything that would be required of him if the plaintiff had properly made out a prima facie case, whether the plaintiff really did so is no longer relevant.”
United States Postal Service Bd. of Governors v. Aikens,
460 U.S. 711, 715, 103 S.Ct. 1478, 1482, 75 L.Ed.2d 403 (1983). We agree with the Sixth Circuit that when a case has been tried on the merits, a reviewing appellate court need not address the sufficiency of plaintiff’s prima facie case, and may in
stead proceed directly to the ultimate question whether plaintiff has produced sufficient evidence for a jury to find that discrimination has occurred.
Brownlow v. Edgecomb Metals Co.,
867 F.2d 960, 963 (6th Cir.1989).
Lone Star has offered a valid, nondiscriminatory explanation for Walther’s discharge: he was part of a reduction in force in which employees were laid off based on their performance records. Walther was therefore required to produce sufficient evidence from which a reasonable trier of fact could conclude that Lone Star’s explanation was pretextual.
Little v. Republic Refining Co.,
924 F.2d 93, 96-98 (5th Cir.1991);
Reeves v. General Foods Corp.,
682 F.2d 515, 521-23 (5th Cir.1982). A plaintiff may demonstrate pretext either by showing that a discriminatory motive more likely motivated the employer or that the employer’s explanation is unworthy of credence.
Amburgey v. Corhart Refractories Corp.,
936 F.2d 805, 813 (5th Cir.1991). We have recognized that motivation and intent in employment discrimination cases can often be proved only through circumstantial evidence.
Thornbrough v. Columbus & Greenville R. Co.,
760 F.2d 633, 641 (5th Cir.1985);
see also Burns v. Texas City Refining, Inc.,
890 F.2d 747, 751 (5th Cir.1989) (“It is certainly not necessary, and indeed is often impossible, for the plaintiff in a discrimination case to prove the defendant’s state of mind through direct evidence.”) Therefore, although Walther’s evidence of age discrimination is entirely circumstantial, we think it is sufficient to support the jury’s verdict.
First, we note that the fact that Walther was replaced by an older employee rather than a younger one does not necessarily show that he was not discriminated against because of his age. This is a reduction in force case. Therefore the crucial comparison is not between Walther and his replacement, but between Walther and similarly situated younger employees.
See Thornbrough,
760 F.2d at 644 (“What is suspicious in reduction-in-force cases is that the employer fired a qualified, older employee but retained a younger one.”) Although the transfer of an older employee to Walther’s position may assist Lone Star in rebutting an inference of discrimination, it does not necessarily dispel that inference entirely. “Congress never intended to give an employer license to discriminate against some employees ... merely because he favorably treats other members of the employees’ group.”
Id.
at 646 n. 20 (quoting
Connecticut v. Teal,
457 U.S. 440, 455, 102 S.Ct. 2525, 2535, 73 L.Ed.2d 130 (1982)).
Walther produced evidence that he was laid off while younger, less qualified employees were retained and transferred or promoted to positions similar to his own. Of course, the issue is not whether Walther or the retained employees were better qualified. An employer is entitled to make that decision for itself. “The ADEA was not intended to be a vehicle for judicial second guessing of business decisions, nor was it intended to transform the courts into personnel managers.”
Thornbrough,
760 F.2d at 647. Nevertheless, a plaintiff can take his case to a jury with evidence that he was
clearly
better qualified than younger employees who were retained.
Id.
Here we are confronted with more than a “bald assertion” that he was more capable.
Compare, Amburgey,
936 F.2d at 814 (affirming summary judgment when plaintiff produced only conclusory allegations that he was well qualified). Walther’s
immediate supervisor
conceded that he was more qualified than some of the younger employees who were retained and promoted. Significantly, Lone Star did not introduce any testimony that Walther was less qualified than these younger individuals.
We do not imply that an employer must “bump” younger, less experienced employees from their positions in order to make room for the older and more experienced. The ADEA does not require employers to maintain such a “bumping” policy.
See, e.g., Barnes v. GenCorp Inc.,
896 F.2d 1457, 1469 (6th Cir.1990). But when there are
vacant
positions similar to the position from which an older employee is terminated, an employer cannot consider age in deciding which employee should be transferred to fill the vacancy. If an older,
terminated employee establishes that several such vacancies were filled by younger, less qualified transfers, this is at least some evidence that age discrimination has occurred.
Walther also cast doubt on Lone Star’s justification for including him in the reduction in force. Lone Star did not point to any specific evidence in Walther’s personnel file indicating any previous dissatisfaction with his performance. Instead, Roberts could only “report” that Walther’s performance had been weak in certain areas. The jury need not have done so, but it could reasonably infer that this was “an after-the fact inspiration triggered by the necessity of fending off litigation.”
Graefenhain v. Pabst Brewing Co.,
827 F.2d 13, 21 (7th Cir.1987). The jury could also find it suspicious that Roberts neither examined any personnel files nor consulted Walther’s immediate supervisor before including Walther on the list for termination.
In many reduction in force cases, the plaintiff’s qualifications for his job are less relevant since some employees will have to be let go despite competent performance.
See, e.g., Conkwright v. Westinghouse Electric Corp.,
933 F.2d 231, 235 (4th Cir.1991). Here, however, Lone Star’s explanation for including Walther in the reduction in force was his qualifications. Walther was therefore entitled to show that this explanation was pretextual by producing evidence that younger, less qualified employees received better treatment in the layoffs.
The statistical evidence produced at trial by both sides was inconclusive. Particularly in age discrimination cases where innumerable groupings of employees are possible according to ages and divisions within the corporate structure, statistics are easily manipulated and may be deceptive. Although Lone Star made a convincing showing that the reduction in force did not target older employees if the company as a whole is considered, it does not follow that discrimination did not occur in a particular region or division.
See, e.g., Healy v. New York Life Insurance Co.,
860 F.2d 1209, 1218 (3d Cir.1988),
cert. denied,
490 U.S. 1098, 109 S.Ct. 2449, 104 L.Ed.2d 1004 (1989) (“Statistics are often unreliable because company-wide data may mask discrimination on a smaller scale.”). A particular employee may discriminate even if others in the company do not.
On the other hand, Walther’s showing that employees over the age of forty in the Greenville region were laid off in disproportionately large numbers is weakened by the reality that termination decisions were not made region by region. Roy Roberts, the person who selected Walther for termination, prepared a partial termination list for the Dallas division, of which the Green-ville region was only a small part. Furthermore, Roberts did not make the decisions on all of the layoffs in the Greenville region. The statistics from the Greenville area provide little support for an inference that Roberts considered age in making the decision to let Walther go. We are mindful of the Supreme Court’s warning that statistics “come in infinite variety” and that “their usefulness depends on all the surrounding facts and circumstances.”
International Brotherhood of Teamsters v. United States,
431 U.S. 324, 340, 97 S.Ct. 1843, 1856-57, 52 L.Ed.2d 396 (1977).
The most useful numbers were those relating to Roberts, since he was the person who actually made the decision that Walther was to be terminated. Although there was no testimony at trial concerning a statistical analysis of the ages of employees Roberts selected for termination, Roberts' termination list included twenty-five employees over the age of forty and only two employees under the age of forty. Eleven of the people on the list were over age fifty, and nine more were over age sixty. If properly supported, this evidence “could cause a reasonable trier of fact to raise an eyebrow, and proceed to assess the employer’s explanation for this outcome.”
Morgan v. Arkansas Gazette,
897 F.2d 945, 951 (8th Cir.1990).
Walther’s failure to flesh out this evidence, however, undermines its probative force. We cannot tell what the age makeup of the Dallas division was for the purpose of comparing the ages of those termi
nated with those retained. Furthermore, Roberts’ explanation for this result was that he was focusing on employees in the middle management layer to see if there were layers in the organization that could be eliminated. There was insufficient testimony for us to determine whether this explanation is well grounded in fact or pretextual.
Despite the weakness of the statistical proof, the evidence as a whole was sufficient to support the jury’s verdict. We must remember that “determinations regarding motivation and intent depend on complicated inferences from the evidence and are therefore peculiarly within the province of the factfinder.”
Thornbrough,
760 F.2d at 641. We cannot say as a matter of law that no rational juror could find that Lone Star discriminated against Walther on the basis of his age.
III.
Lone Star also argues that the district court’s instruction on statistical evidence was erroneous and that a new trial is therefore required. The trial judge has broad discretion to compose the instructions to the jury, as long as they are fundamentally accurate and not misleading.
Gates v. Shell Offshore, Inc.,
881 F.2d 216, 218 (5th Cir.1989).
The court told the jury:
Both plaintiff and defendant have introduced statistics concerning the work force reduction of the defendant company. Statistics may be introduced for the purpose of showing motive, intent or for purpose to discriminate or to demonstrate that they did not exist.
Statistics alone may establish that the reason defendant discharged the plaintiff was because of
his age. It is for the jury to make the determination of whether they do or do not in this case. (Emphasis supplied).
Lone Star contends that statistics
alone
would not support a finding of discrimination and it was error so to instruct the jury. It relies on cases such as
Plemer v. Parsons-Gilbane,
713 F.2d 1127, 1135 (5th Cir.1983), in which we stated that “although statistics are of probative value in an individual discrimination case for the purpose of showing motive, intent, or purpose, that evidence is not determinative of an employer’s reason for the action taken against the individual grievant.”
See also Gottlieb v. Tulane Univ. of Louisiana,
809 F.2d 278, 283 (5th Cir.1987);
Thornbrough,
760 F.2d at 646 n. 20 (5th Cir.1985) (statistical evidence, while probative, is not determinative or dispositive). Walther, on the other hand, relies on cases such as
Williams v. General Motors Corp.,
656 F.2d 120, 130 (5th Cir.1981), ce
rt. denied,
455 U.S. 943, 102 S.Ct. 1439, 71 L.Ed.2d 655 (1982), in which we stated that “circumstantial evidence, be it statistical evidence, evidence of employer subterfuge, or other forms, also would suffice [to prove intent to discriminate] so long as it serves the general purpose of supporting the general proposition and excluding other propositions.”
See also Haring v. CPC International, Inc.,
664 F.2d 1234, 1240 (5th Cir. Unit B 1981) (upholding instruction advising jury that it could infer discrimination from the statistical evidence in the record).
Although it may be the rare case in which statistical evidence alone will suffice to prove that an employee was discriminated against on the basis of age, we believe that the law allows such means of proof. The Supreme Court long ago made clear that “ ‘statistical analyses have served and will continue to serve an important role’ in cases in which the existence of discrimination is a disputed issue.”
International Brotherhood of Teamsters v. United States,
431 U.S. 324, 339, 97 S.Ct. 1843, 1856, 52 L.Ed.2d 396 (1977). We can imagine cases in which the statistical proof will support a finding of age discrimination by itself. When we asserted in
Plemer, Gottlieb,
and
Thombrough
that statistical evidence is not “determinative” or “disposi-tive” of the issue of discrimination, we meant simply that statistics
do not necessarily
show the existence or absence of discrimination. This does not mean that in any individual case, statistics
cannot
show the existence or absence of discrimination. The question of the weight to be accorded
admissible statistical evidence is for the jury to answer.
Haring,
664 F.2d at 1240.
That having been said, we caution that the statistical evidence introduced in Walther’s case was far short of the explanatory power necessary to support a finding of discrimination. Telling the jury that statistics alone could support a finding of age discrimination was unnecessary and potentially misleading, given the inadequacy of the statistical showing in this case. Indeed, there is an argument that such an instruction requires reversal because the jury might have relied solely on the inadequate statistics in reaching its decision. The Supreme Court has said in the criminal context that “a general verdict must be set aside if the jury was instructed that it could rely on any of two or more independent grounds, and one of those grounds is insufficient, because the verdict may have rested exclusively on the insufficient ground.”
Zant v. Stephens,
462 U.S. 862, 881, 103 S.Ct. 2733, 2745, 77 L.Ed.2d 235 (1983);
see also Neubauer v. City of McAllen,
766 F.2d 1567, 1575 (5th Cir.1985). This principle has its origins in
Stromberg v. California,
283 U.S. 359, 51 S.Ct. 532, 75 L.Ed. 1117 (1931), where the Court reversed a conviction when one of three possible bases for the jury’s verdict was unconstitutional.
The Court has recently limited the reach of
Stromberg,
however. In
Griffin v. United States,
- U.S. -, 112 S.Ct. 466, 116 L.Ed.2d 371 (1991), the Court explained that the
Stromberg
rule should be applied only when jurors have been left the option of relying on a
legally
inadequate theory, not a factually inadequate theory. Thus we will not reverse a verdict simply because the jury might have decided on a ground that was supported by insufficient evidence. Instead we must assume that the jury considered all of the evidence in reaching its decision.
Because the district court’s instruction on statistical proof was legally correct, although not factually supported, there was no reversible error here.
Lone Star also challenges the district court’s instructions on the elements of prima facie proof. The jury was told that Walther could make out a prima facie case of age discrimination by showing that he was adversely affected by Lone Star’s employment decision, that he was qualified for the position he held, and that younger, less qualified employees were not discharged.
If Walther proved each of these factors, the instructions stated that he was entitled to a rebuttable presumption that he was discharged by reason of age discrimination.
Lone Star contends that the instructions should have required Walther to show that he was replaced by a younger employee to make out a prima facie case. Furthermore, instead of requiring Walther to show merely that younger, less qualified employees were not discharged, the instructions should have required proof that these younger employees were similarly situated and that he was qualified to fill the jobs they received. Finally, in Lone Star’s view, there should have been some indication that Walther had to prove intentional discrimination to establish a prima facie ease.
We observe as a preliminary matter that it was inappropriate for the district court to submit the issue of prima facie case to the jury. The prima facie case method is “merely a sensible, orderly way to evaluate the evidence in light of common experience as it bears on the critical question of discrimination.”
Aikens,
460 U.S. at 715, 103 S.Ct. at 1482. When the defendant has produced evidence of a nondiscriminatory reason for plaintiff’s discharge, and plaintiff has had an opportunity to challenge that reason as pretextual, the trier of fact should proceed directly to the ultimate issue of whether the defendant intentionally discriminated against the plaintiff.
Id.
at
715-16, 103 S.Ct. at 1482. The initial prima facie case is no longer relevant.
Under the logic of
Aikens,
it is clear that the issue of whether a plaintiff made out a prima facie case has no place in the jury room. Instructing the jury on the elements of a prima facie case, presumptions, and the shifting burden of proof is unnecessary and confusing. Instead, the court should instruct the jury to consider the ultimate question of whether defendant terminated plaintiff because of his age. To the extent our decisions before or after
Aikens
imply that the issue of prima facie case is a factual question for the jury to resolve,
see e.g., Haring,
664 F.2d at 1247;
Crum v. American Airlines,
946 F.2d 423, 429 (5th Cir.1991), we reject such implications as dictum.
Although Lone Star raises objections to the substance of the prima facie case instructions on appeal, it failed to raise these objections below. Consequently, any errors in these instructions are reversible only if they constitute plain error.
Rodrigue v. Dixilyn Corp.,
620 F.2d 537, 540 (5th Cir.1980),
cert. denied,
449 U.S. 1113, 101 S.Ct. 923, 66 L.Ed.2d 842 (1981). To show plain error, Lone Star “must establish that the instruction was an incorrect statement of law and that it was probably responsible for an incorrect verdict, leading to substantial injustice.”
Id.
at 541.
We are not persuaded that any of the alleged errors in the court’s instructions on prima facie proof were probably responsible for an incorrect verdict. The district court instructed the jury that in order for Walther to prevail, it was necessary to find that age was a determining factor in the decision to terminate him. Furthermore, the jury specifically found that Walther had proved that Lone Star’s stated motives for his discharge were pretextual and that, but for the motive of age discrimination, he would not have been terminated. With these findings, any errors in the standard for prima facie proof were harmless.
IY.
Finally, Lone Star contends that even if the jury’s verdict must be upheld, Walther should not have been awarded front pay because reinstatement remains an option. Reinstatement is the remedy of choice in age discrimination cases. Front pay will not be awarded unless the plaintiff shows that reinstatement is not feasible.
Hansard v. Pepsi-Cola Metro Bottling Co., Inc.,
865 F.2d 1461, 1469 (5th Cir.),
cert. denied,
493 U.S. 842, 110 S.Ct. 129, 107 L.Ed.2d 89 (1989). Since front pay is an equitable remedy, the district court rather than the jury should determine whether an award of front pay is appropriate, and if so, the amount of the award.
Reneau v. Wayne Griffin & Sons, Inc.,
945 F.2d 869, 870 (5th Cir.1991). We review the district court’s decision that reinstatement was not feasible for an abuse of discretion.
Deloach v. Delchamps, Inc.,
897 F.2d 815, 822 (5th Cir.1990).
In this case, the record does not indicate why the district court considered reinstatement infeasible. The court later stated simply that the litigation was “protracted and necessarily vexing.” Unless supported by specific instances of discord, we do not find this sufficient. The Court did not respond to Lone Star’s assertions that it considered Walther a qualified and competent employee capable of resuming work at the company and that there was little evidence of ill will between Lone Star officials and Walther. Lone Star never contended that Walther’s work was unacceptable. If the relationship between Walther and Lone Star has not in fact been irreparably damaged, reinstatement would be proper.
Furthermore, the district court does not appear to have considered the fact that liquidated damages were awarded to Walther in assessing whether front pay is proper. We agree with the Seventh and First Circuits that a substantial liquidated damage award may indicate that an additional award of front pay is inappropriate or excessive.
See Hybert v. Hearst Corp.,
900 F.2d 1050, 1056 (7th Cir.1990);
Wildman v. Lerner Stores Corp.,
771 F.2d 605, 616 (1st Cir.1985). We cannot tell from the
record whether the district court considered this factor in allowing sizeable awards of front pay and liquidated damages in this case.
Although the district court’s decision is entitled to considerable deference, we cannot effectively review the decision without greater explanation of its rationale. We find it appropriate to vacate the award of front pay and remand to the district court for a more thorough evaluation of these issues.
The judgment entered upon the jury verdict is AFFIRMED, except the award of front pay is VACATED, and the claim for recovery of front pay is REMANDED for further proceedings.