Charles TESTERMAN, Plaintiff-Appellant, v. EDS TECHNICAL PRODUCTS CORPORATION, Defendant-Appellee

FLAUM, Circuit Judge. Charles Testerman brought this suit under the Employee Retirement Income Security Act of 1974, 29 U.S.C. § 1001 et seq., and the Age Discrimination in Employment Act, 29 U.S.C. § 621 et seq., alleging that when EDS Technical Products Corporation (“EDS”) trimmed its workforce in 1993, the company unlawfully considered Testerman’s age in deciding to include him among those employees sacrificed to the bottom line. After a thorough review of the evidence, the district court granted summary judgment for EDS. Only the ADEA claim is presented on this appeal, and we affirm. I. Mr. Testerman became an EDS employee in January 1985 when the company was acquired by General Motors, his employer since 1956. At the time of the acquisition, Testerman worked as a tape librarian and backup microfiche operator at the Allison Division’s Data Center in Indianapolis, and his job initially remained unchanged when EDS became his employer. In late 1991, however, the Data Center was reduced to a small printing facility after its two major customers, Allison Gas Turbine and Allison Transmission, transferred their business to an out-of-state facility. Testerman, along with other employees, remained at the Data Center doing temporary work until permanent positions within the Indianapolis area could be found. In May 1992, Testerman was offered a position with the Operations Group of the Allison Gas Turbine Account (the “Account”), at first on a trial basis. He was assigned to the Hotline, where he received calls from Allison Gas Turbine employees with computer difficulties and either assisted the callers himself or referred them to a service person. The new job required knowledge of computer systems with which Testerman had no experience, but he ultimately was hired on a permanent basis, and it is undisputed for the purposes of this appeal that his performance level would have ensured him continued employment had business remained strong. Business did not remain strong, however, and with salary growing faster than revenues, EDS chose to stem the former rather than rest its hopes on boosting the latter. In April 1993, the company instituted a “resource alignment plan” — its euphemism of choice for what the caselaw refers to as a reduction in force (“RIF”). Mr. Testerman, 57 years old, was one of five individuals fired *300 in June 1993 after being placed on what began as a list of 14 of the Account’s lowest-performing employees. All five who lost their jobs were over 40 years of age. Prior to joining the Hotline, Testerman earned consistently high ratings from his supervisors, although he was reprimanded occasionally for being too talkative. 1 Of the written performance reviews he received while at EDS, only one related his tenure at the Hotline. As discussed below, the significance of that review — completed in June 1993 after EDS fired him and assigning him his poorest rating ever — is a matter of some dispute, but, for the most part, Testerman does not challenge the review’s veracity to the extent it chronicles specific behavior. Testerman similarly acknowledges, at least implicitly, the facts memorialized in a one-page evaluation in which Account Manager Sara Hooper recorded her reasons for selecting Testerman for discharge. The picture that emerges is of a popular employee who talked too much on the job, who took too many smoking breaks, who failed to file required status reports, and who exhibited poor judgment on occasion, as when he kissed the hand of a female visitor to the Hotline or complained to EDS customers about EDS’s internal problems. The June 1993 review observed that Testerman “[h]as learned much in the past year” and was “[w]ell liked by the customer,” but complained that he “[njeeds guidance to remain productive” and “[s]eems quick to blame ‘EDS’ for any perceived inconvenience .... [, which] leaves customers with a bad opinion of our company.” Likewise, Hooper’s evaluation noted, “I hesitate to put [Testerman] into situations where he is alone with [customers] because of his unprofessional manner and attitude toward EDS.” In his first status report following a one-month gap, Testerman himself apologized “about the absence of status reports,” but explained that “honking my own horn isn’t an easy task for me.” The reports Testerman did manage to file provided comic relief, but could not have endeared him to his employer. Of a project for which he had volunteered — drafting procedures for the IBM 2540 card reader/punch — Testerman explained in December 1992 that it was “on hold because my mind hasn’t returned from [vacation].” More than a month later, he confessed, “Progress isn’t breath taking,” and still later, he assured his supervisor that “[j]ust like Eveready batteries, I’m still working, on the 2540 procedure that is.” Testerman still had not completed the project when EDS fired him. Within two weeks after discharging Test-erman, Testerman’s former supervisor at the Hotline hired 31-year-old Steve Price, a transferee whose position at another EDS account had been terminated. Price inherited some of the duties for which Testerman had been responsible, but Price’s familiarity with various computer applications and hardware enabled him to perform tasks that were beyond Testerman’s capabilities. In the two reviews he received prior to joining the Hotline, Price scored “solid meets,” one level below the “solid exceeds” that Testerman received in his last two reviews at the Data Center. In addition, Mark McNett, age 27, remained at the Hotline after the RIF. McNett received “solid meets” and “and solid generally meets” ratings in 1992 and a “solid meets” in his June 1993 review, one level higher than the “solid generally meets” Test-erman earned in his post-termination review of the same month. To distill these report-card rankings: Testerman compares favorably with McNett and Price when one looks at the ratings he received before joining the Hotline, and unfavorably only if one focuses on Testerman’s final, post-termination rating. It is undisputed that some of Testerman’s coworkers referred to him in terms that reflected his age. McNett, for one, called Test-erman “Baldy Locks” and “Old Man,” and Testerman’s team leader at the Hotline addressed him as “Pops.” His supervisor in Operations stated that, as a person younger than Testerman, she hesitated to reprimand *301 him for his frequent smoke breaks because to do so smacked of “role reversal.” In addition, Testerman points out that Hooper’s last evaluation referred to him as the “senior” member of the team and that the Data Center where he worked prior to joining the Hotline was often identified as the “Old” Data Center. II. We review de novo a district court’s grant of summary judgment, reading the record in the light most favorable to the non-moving party. Courtney v. Biosound, Inc., 42 F.3d 414, 418 (7th Cir.1994). Under Federal Rule of Civil Procedure 56(c), summary judgment is appropriate only if there remains no genuine issue of material fact and the moving party is entitled to a judgment as a matter of law. Celotex Corp. v. Catrett, 477 U.S. 317, 322-28, 106 S.Ct. 2548, 2552-55, 91 L.Ed.2d 265 (1986). The non-moving party is entitled to the benefit of all reasonable inferences, Courtney, 42 F.3d at 418, and we apply this standard bearing in mind that questions of intent and credibility often are determinative in employment discrimination cases, see Wohl v. Spectrum Mfg., Inc., 94 F.3d 353, 355 n. 1 (7th Cir.1996). III. A plaintiff in an age discrimination suit may reach a jury by either of two routes: by presenting evidence that age was a “determining factor” in the discharge decision or by proceeding under the indirect, burden-shifting approach established in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973) and refined in Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 101 S.Ct. 1089, 67 L.Ed.2d 207 (1981). See Anderson v. Baxter Healthcare Corp., 13 F.3d 1120, 1122 (7th Cir.1994). Testerman tries both methods, neither of which is availing. A. In looking for discriminatory motive under the ADEA, the relevant inquiry is whether age “tipped the balance,” that is, whether age was a “but for” cause of the decision to fire the plaintiff. Umpleby v. Potter & Brumfield, Inc., 69 F.3d 209, 213 (7th Cir.1995). Testerman suggests that appellations such as “Baldy Locks,” “Pops” and “Old Man,” as well as his supervisor’s discomfort with “role reversal,” reflect exactly the kind of stereotyping that the ADEA was designed to combat. See Hazen Paper Co. v. Biggins, 507 U.S. 604, 610, 113 S.Ct. 1701, 1706, 123 L.Ed.2d 338 (1993). He argues further that, when combined with references to the “Old Data Center,” to Testerman’s knowledge of “antiquated” technologies and his “senior” status, and with the fact that length of service was among the factors considered by decisionmakers responsible for the RIF, such comments reasonably could support a finding that Testerman would not have been fired but for the presence of age bias at EDS. We cannot agree. Although it matters little that terms such as “Old Man” or “Pops” were intended as “salutations between co-workers or terms of endearment”— for the ADEA prohibits employment decisions based on “stigmatizing stereotypes,” Hazen Paper Co., 507 U.S. at 610, 113 S.Ct. at 1706, however benignly intended — , what does matter for our purposes is that none of these terms was employed by the people involved in the decision to fire Testerman. See Fuka v. Thomson Consumer Elecs., 82 F.3d 1397, 1403 (7th Cir.1996); McCarthy v. Kemper Life Ins. Cos., 924 F.2d 683, 686-87 (7th Cir.1991); La Montague v. American Convenience Prods., 750 F.2d 1405, 1412 (7th Cir.1984); cf. Shager v. Upjohn Co., 913 F.2d 398 (7th Cir.1990) (“[A]ll that these cases really stand for is the commonsense proposition that a slur is not in and of itself proof of actionable discrimination_”). Four individuals were responsible for implementing the “resource alignment” at the Allison Gas Turbine Account: the Strategic Business Unit President, John Brunet, Regional Manager Carl Wouden, and Account Managers Sara Hooper and Joseph Mennel. As Manager of the Operations side of the Account, Hooper was directly responsible for selecting Testerman for discharge. None of the language of which Testerman complains can be attributed to these decisionmakers. *302 Furthermore, Testerman’s effort to locate age bias in references to the “Old” Data Center — a Data Center that no longer existed — must be rejected for what it is: a “word game,” to quote the district court. Similarly, Testerman’s knowledge of “antiquated technologies,” placed in its proper context, turns out to have been an asset in the eyes of his employer. In the one-page summary of her reasons for selecting Testerman to be fired, Hooper weighed this expertise against Testerman’s shortcomings as an employee: “Charlie may be our last source of knowledge on some antiquated technologies (card readers, etc.) but his day-to-day productivity lags behind others in the group.” This leaves only Hooper’s lament, also included in her one-page explanation, that Testerman’s garrulousness and frequent smoke breaks provided a bad example on the part of a “sen[io]r” member of the team. Testerman would have us inteipret “senior” as a reference to his age, while EDS insists that it refers to his length of service. Summary judgment is not the place to resolve such ambiguities, and it might seem incumbent upon us to declare whether, understood as a reference to Testerman’s age, this single phrase could defeat a motion for summary judgment; but, again, when the term is read in context, any supposed ambiguity dissolves, for Hooper elaborated that “this does not set a good example for newer members of the team.” The only reasonable interpretation is that Hooper was concerned that more recent hires might learn from Testerman’s bad habits. Testerman’s final attempt to demonstrate that age was a determining factor in EDS’s decision to fire him involves instances where the company employed criteria that conceivably could have served as proxies for age. Specifically, Testerman complains that Hooper, pursuant to Brunet’s request, included length of service on the top of the evaluation sheet prepared for Testerman and the other individuals discharged, and that Wouden expressed concerns with the “skill sets” of employees who had transferred to the Hotline from the Data Center. First, we note that we would hesitate before announcing a rule that would dissuade managers contemplating staff reductions from taking length of service into account. At the same time, particularly in a case involving an employee with 37 years of experience, it would be disingenuous to deny that length of service will often correlate with age. Testerman’s failure lies not in any lack of connection between age and length of service, but in his inability to connect, even indirectly, length of service with discriminatory motive. ‘When the employer’s decision is wholly motivated by factors other than age, the problem of inaccurate and stigmatizing stereotypes disappears. This is true even if the motivating factor is correlated with age_” Hazen Paper Co., 507 U.S. at 611, 113 S.Ct. at 1706; see also Anderson, 13 F.3d at 1125 (discussing Hazen Paper Co.). Here, Wouden testified that EDS “expected people with longevity ... to have the skill sets needed .... [but that] [w]ith direct reference to the data center ... we have a lot of older equipment ... much of which went away, and therefore, that was a concern.” This testimony indicates that EDS considered length of service and the “skill sets” of its employees for their own sake, not as proxies for age, but for what they told the company about an individual’s value as an employee. Left with no direct or circumstantial evidence that age was a determining factor in the decision to fire him, Testerman must fall back upon the indirect McDonnell Douglas paradigm. B. The familiar framework established in McDonnell Douglas necessitates a three-step inquiry. First, a plaintiff must establish, by a preponderance of the evidence, a prima facie ease of discrimination. If the plaintiff makes out a prima facie case, a presumption of discrimination arises, and the burden shifts to the defendant to come forward with evidence of a “legitimate, nondiscriminatory reason” for discharging the plaintiff. McDonnell Douglas, 411 U.S. at 802, 93 S.Ct. at 1824; Burdine, 450 U.S. at 253, 101 S.Ct. at 1093-94. Finally, “the plaintiff must then have an opportunity to prove by a preponderance of the evidence that the legitimate reasons offered by the defendant were not its *303 true reasons, but were a pretext for discrimination.” Burdine, 450 U.S. at 253, 101 S.Ct. at 1093; see also Anderson, 13 F.3d at 1122-24 (discussing pretext prong of McDonnell Douglas). Throughout, “[t]he ultimate burden of persuading the trier of fact that the defendant intentionally discriminated against the plaintiff remains at all times with the plaintiff.” Burdine, 450 U.S. at 253, 101 S.Ct. at 1093. A goal of this burden-shifting approach is “to frame the factual issue with sufficient clarity so that the plaintiff will have a full and fair opportunity to demonstrate pretext.” Id. at 255-56, 101 S.Ct. at 1094-95. EDS concedes that Testerman has established a prima facie case of age discrimination. Of the four elements of the prima facie case applicable in the RIF context— that Testerman was a member of the protected class (persons age 40 to 70), that he was meeting his employer’s legitimate expectations, that he was fired, and that younger employees were treated more favorably, see O’Connor v. Consolidated Coin Caterers, - U.S. -, 116 S.Ct. 1307, 134 L.Ed.2d 433 (1996); Weisbrot v. Medical College of Wis., 79 F.3d 677, 681 & n. 3 (7th Cir.1996); Roper v. Peabody Coal Co., 47 F.3d 925; 926 (7th Cir.1995) — only the second element, relating to Testerman’s performance, might have presented any difficulty. As noted above, however, EDS does not contend that Testerman would have been fired absent the RIF. For its part, EDS has met its burden of producing admissible evidence of a legitimate, nondiscriminatory reason for firing Testerman— that his poor performance made him expendable when the RIF took place. The propriety of summary judgment therefore turns on whether Testerman has succeeded in mustering evidence that this permissible reason is really a pretext for discrimination. Although lacking evidence of discriminatory motive on the part of EDS, Testerman nonetheless may show pretext by producing evidence that EDS’s ostensible justification is “unworthy of credence.” Burdine, 450 U.S. at 256, 101 S.Ct. at 1095; Robinson v. PPG Indus., Inc., 23 F.3d 1159, 1163 (7th Cir.1994); La Montagne, 750 F.2d at 1409. Our caselaw acknowledges that a plaintiff may accomplish this showing with evidence tending to prove that the employer’s proffered reasons are factually baseless, were not the actual motivation for the discharge in question, or were insufficient to motivate the discharge. Wolf v. Buss (America), Inc., 77 F.3d 914, 919 (7th Cir.1996), cert. denied, — U.S. -, 117 S.Ct. 175, 136 L.Ed.2d 116 (1996). These formulations are simply different ways of recognizing that when the sincerity of an employer’s asserted reasons for discharging an employee is cast into doubt, a factfinder may reasonably infer that unlawful discrimination was the true motivation: “If the only reason an employer offers for firing an employee is a lie, the inference that the real reason was a forbidden one, such as age, may rationally be drawn.” Shager v. Upjohn Co., 913 F.2d 398, 401 (7th Cir.1990). In such circumstances, summary judgment is improper. How, then, does Testerman hope to cast doubt upon the sincerity of EDS’s professed motives? He surely does not mean to argue that EDS’s stated reasons are “factually baseless,” for he does not suggest that EDS did not reduce its workforce in 1993, nor does he deny the existence of behavior on his part that supposedly shaped EDS’s low opinion of him as an employee. He does not assign himself the task of demonstrating that poor performance was not the “actual motivation” for his discharge. Although this may be the inference he would like us (or a jury) ultimately to draw, he cannot, as we have seen, point to any evidence of an alternative motive, discriminatory or otherwise, or to any instance where EDS has contradicted its stated rationale. But Testerman need not set his mark so high. In this case, involving an RIF, Tester-man should not be required to produce evidence tending to prove that EDS’s explanation is a “lie” in the sense of being a complete fabrication. Cf. Russell v. Acme-Evans Co., 51 F.3d 64, 68 (7th Cir.1995). Rather, his ultimate burden is to establish that age “tipped the balance” in favor of discharge. Umpleby, 69 F.3d at 213. Thus, he need not disprove the fact that EDS had reason to complain of his performance. In an RIF, it is foreordained that some employees will lose *304 their jobs; and, for all but the most exceptional employee, the defendant inevitably will be able to point to some imperfection as a reason for the discharge. But when age enters into the calculus to such an extent that the employer fires an employee who otherwise would have survived the RIF, the ADEA has been violated. At the same time, we deal with small gradations, with an employer’s subjective comparison of one employee to another, and it is incumbent upon us to remember that what is at issue is not the wisdom of an employer’s decision, but the genuineness of the employer’s motives. See Gustovich v. AT & T Communications, Inc., 972 F.2d 845, 848 (7th Cir.1992) (“The question is not whether the ratings were right but whether the employer’s description of its reasons is honest.”). Returning to Testerman, his task is to provide some evidence which would support an inference that, given his shortcomings as an employee, he nevertheless would not have been fired — would not have been ranked below other employees— “but for” his age. It is in this sense that Testerman’s evidence, read favorably, must indicate that EDS’s professed reasons for firing him were “insufficient to motivate the discharge.” Wolf, 77 F.3d at 919. We now turn to the question of whether Testerman succeeds, mindful of our admonition that the ADEA “does not protect merely the older worker who is perfect from the standpoint of his employer_ It protects, as a practical matter, the imperfect older worker from being treated worse than the imperfect younger one.” Shager, 913 F.2d at 403. First, Testerman points to the fact that all of the employees terminated in the course of the RIF were over 40 years old. We are dealing here with five employees, culled from an original list of fourteen, of which only two, Testerman and Fernando Espinal were from the Operations side of the Account. Two of the final five employees were 42 years old. Of the employees on the initial list of fourteen, two who were spared termination were in their 50s. One 37-year-old employee would have been on the final list but for the intervention of Allison Gas Turbine. It is not clear what Testerman would have us make of this evidence. He understandably takes umbrage at EDS’s characterization of these numbers as a “statistical argument,” for it is clear that he has no statistical argument to make. Rather, in conclusory terms, Testerman offers this evidence “as a basis from which a reasonable inference could be drawn that EDS’s proffered reason for Testerman’s firing was unworthy of credence.” We think not. For one thing, Test-erman does not dispute EDS’s contention that the 37-year-old was removed from the final termination list as a result of an important customer’s intervention. Contrary to Testerman’s suggestion, this last-minute removal tends to show that EDS was not exclusively targeting members of the protected class. At best, it tells us nothing about EDS’s motives. ■ Testerman makes much of the district court’s observation that two of the employees discharged were only 42, arguing that the ADEA establishes a “bright line” at two score years. While it is undeniable that the ADEA sets a “bright line,” in the sense that only individuals who have reached the age of 40 may invoke its protection, the statute does not require a court to shut its eyes to age disparities relevant to the ultimate question of discriminatory intent. See O’Connor, — U.S. at-, 116 S.Ct. at 1310 (“This language does not ban discrimination against employees because they are aged 40 or older; it bans discrimination against employees because of their age, but limits the protected class to those who are 40 or older.”). Although O’Connor specifically addressed the elements of a prima facie case of age discrimination, the Court grounded its logic in the ultimate burden that the ADEA places upon a plaintiff: “Because the ADEA prohibits discrimination on the basis of age and not class membership, the fact that a replacement is substantially younger than the plaintiff is a far more reliable indicator of age discrimination than is the fact that the plaintiff was replaced by someone outside the protected class.” Id. Here, we are engaged in a search for evidence that age tipped the balance from one imperfect, younger employee to another, less imperfect, older employee. It therefore seems highly relevant that, in *305 whittling down a list of 14 of its weakest employees, EDS removed from the list (and thus spared) employees in their 50s but kept on the list (and thus discharged) employees in their 40s, and would have discharged a 37-year-old employee if not for Allison Gas Turbine’s plea. Once again, the evidence most logically cuts against Testerman, but even in the best of ail possible lights, it tells us nothing. At bottom, we agree with the district court that the mere fact that the five fired employees fell within the protected class, without more, does nothing to east doubt upon EDS’s articulated reasons for discharging Testerman. Testerman next attempts to undercut EDS’s legitimate reasons with evidence that EDS retained one employee (McNett) significantly younger than he and hired another (Price) following his discharge, both of whom had worse performance histories. EDS counters that performance should be measured at the time of discharge, and that, at the time he was fired, Testerman’s performance review assigned him a rating below those of McNett and Price. This brings into play Testerman’s next evidence of pretext— the very fact that EDS completed a review of Testerman after it fired him. This post-termination review is itself evidence of pretext, Testerman asserts, especially in light of the fact that it assigned him his lowest rating ever; at the very least, it should not be relied upon as a basis for comparisons to McNett and Price. As support for the proposition that an unprecedentedly unfavorable post-termination appraisal may support a finding of pretext, Testerman cites Gunby v. Pennsylvania Electric Co., 840 F.2d 1108 (3d Cir.1988), ce rt. denied, 492 U.S. 905, 109 S.Ct. 3213, 106 L.Ed.2d 564 (1989), and George v. Mobil Oil Corp., 739 F.Supp. 1577 (S.D.N.Y.1990). Neither case is apposite. In both Gunby and George, the plaintiff had an established history in a particular position, and the suspect evaluation related to skills that had been the subject of earlier evaluations. There thus was a basis for comparison between the disputed evaluation and the earlier evaluations, and the sudden decline in performance could reasonably support an inference that the later evaluation was a sham. In addition, in each case there were independent reasons to doubt the truthfulness of the disputed evaluation. See Gunby, 840 F.2d at 1117; George, 739 F.Supp. at 1580. In contrast, although Testerman quibbles on the level of emphasis and questions the wisdom of the company rules he transgressed, he does not, in the main, contest the factual allegations contained either in Hooper’s one-page evaluation of Testerman, completed just prior to discharging him, or in his post-termination review. In the absence of any basis for comparison to earlier reviews, or of any other reason to question the post-termination review, to entertain Testerman’s reading of the facts would be, in effect, to adopt a per se rule that negative post-termination reviews are always evidence of pretext. This we decline to do. We now can return to McNett and Price. As we have seen, there is nothing suspicious about EDS’s final review of Testerman, a review that belies Testerman’s assertion that he was more competent than the younger men. Reading the record in the light most favorable to Testerman, we might be inclined to disregard the actual rating Testerman received in the June 1993 review (“solid generally meets”), but this still leaves the facts memorialized both in the review and in Hooper’s one-page evaluation, which are largely undisputed. Examining EDS’s complaints regarding Testerman’s performance, and comparing them, where possible, to similar remarks relating to McNett, Testerman simply does not come out ahead. Testerman admittedly is guilty of the most egregious misbehavior he can assign to McNett — being “too chatty” and absentminded — and he has other indiscretions to his credit, in particular, his willingness to complain to EDS customers about the company’s policies and procedures. As for Price, it is undisputed that, although he assumed some of Testerman’s responsibilities upon transferring to the Hotline, he also performed tasks that Testerman could not. Nevertheless, Testerman would have us compare the ratings he received before joining the Hotline to the ratings garnered by McNett and Price during their time at EDS. *306 It is undeniable that Testerman earned good marks at the Data Center, indeed better than McNett’s and Price’s during the same time intervals. The problem, however, is that, by his own admission, when Testerman moved to the Hotline, he was asked to perform entirely new functions. As Testerman recalled during his deposition, the new job presented “a great deal of challenge because ... I had never seen a DEC VAXCluster before. I’ve never seen a Tandem before.” That EDS did not consider Testerman’s performance at the Data Center when it came time to decide whether the company should retain him at the Hotline — a position involving unrelated skills — does not suggest pretext. We are left with evidence that in the course of a RIF, EDS retained one younger employee with fewer performance problems than Testerman and permitted another, whose job at a different account had been terminated, to transfer to the Allison Gas Turbine Account, where he assumed some of Testerman’s duties as well as additional responsibilities. This simply is insufficient to satisfy the pretext prong of McDonnell Douglas. One final point needs to be addressed. Testerman objects that the district court erred by neglecting to examine the evidence in its totality. Even if no single piece of evidence could stave off summary judgment, he contends, the record, read in its entirety, is sufficient to support an inference of pretext. We respond with the observation that “[ajdding together a string of nothings still yields nothing.” Smith v. Cook County, 74 F.3d 829, 834 (7th Cir.1996). No synergies exist among Testerman’s evidentiary supports. If anything, a holistic approach undermines Testerman’s argument, for it highlights the selectivity with which he has plucked his putative evidence of pretext. To supply what is missing would be to cross the line separating reasoned inference from speculation. We therefore Affirm the judgment of the district court. Footnotes [1] From lowest to highest, EDS employed the following ratings: MR (marginal); SG (solid generally meets); SM (solid meets); SX (solid exceeds); SP (superior); and EX (exceptional). From 1987 through 1992, Testerman failed on only one occasion to receive a grade of either SP or SX, earning an SM in 1991. In the June 1993 Hotline review, Testerman received a grade of SG.