The Illinois State Police terminated William Kidd III from its employ shortly before he completed his probationary year of training and service as a state trooper. Kidd, who is African-American, filed suit under Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e-2(a). Although Kidd concedes that he was not performing at an acceptable level at the time of his discharge, he contends that his discharge was nonetheless discriminatory because the ISP did not afford him remedial assistance comparable to the training and support it provided to a white trooper who had difficulties similar to Kidd's. The district court found in favor of the ISP after a bench trial. Kidd v. Illinois, No. 89 C 8504, (N.D. Ill. June 20, 1997). We remand for further consideration.

As its name suggests, the Illinois State Police serves as the state's constabulary. Although its activities touch upon all aspects of law enforcement, the ISP is perhaps best known to us as the guardian of the state's highways, which its more than 1,500 troopers patrol.

The ISP hired Kidd as a probationary trainee in June 1988. At the same time, it hired as another prospective trooper Robert Tucker, who is white. Ultimately, the ISP would terminate both Tucker and Kidd, and in significant part for the same reason: both had remarkably poor communications skills as the result, apparently, of learning disabilities. The significant differences in the ways the ISP attempted to help the two candidates with this problem are what gave rise to Kidd's Title VII claim.

The ISP trains its troopers in three phases. In Phase I, "cadets" undergo twenty weeks of classroom training at the ISP academy in Springfield in a variety of subjects, including the Illinois Vehicular and Criminal Codes, firearms, first aid, field sobriety testing, and report writing. Trainees who complete this first phase satisfactorily are assigned to one of the ISP's district headquarters for Phase II field training. In this second phase, each trainee (now a "probationary trooper") participates in active patrol under the accompaniment and supervision of primary and alternate Field Training Officers (FTOs), typically for a period of ten to fourteen weeks.1 FTOs complete Daily Observation Reports (DORs) documenting each probationary trooper's performance during this period.2 A review board, which typically includes a candidate's primary and alternate FTOs, the District FTO Supervisor, and the State FTO Program Supervisor, determines if and when a probationary trooper will proceed to Phase III, where he will begin solo patrol duty. An FTO Supervisor will monitor the trooper during this final phase and complete Bi-Weekly Report Forms evaluating her progress. As the end of Phase III (and the trooper's one-year probationary period) approaches, a Phase III review board consisting of the District Commander, the Operations Lieutenant, the District FTO Supervisor, and the State FTO Program Supervisor will determine whether he should be advanced to permanent status or instead terminated. Senior ISP administrators review the board's assessment and make a final determination, but the decision of the review board is rarely disturbed.

Kidd, Tucker, and some eighty other cadets commenced Phase I training at the academy on June 13, 1988. That training included more than twenty hours of instruction in report writing, including form completion and narrative writing. Sergeant Frank DeBerry, the State FTO Program Supervisor, taught the class on report writing that year. In addition to the work that DeBerry assigned, cadets were required to write memoranda on an almost daily basis to their cadet supervisors. For example, in their first week at the academy, cadets were asked to write an essay explaining why they wanted to become ISP troopers. Cadets received written feedback on such assignments and, when errors were found, were often asked to turn in corrected drafts (multiple times, if necessary).

As the cadets progressed through the twenty weeks of Phase I, their supervisors at the academy completed bi-weekly observation reports rating their progress in nineteen categories, including two aspects of their report-writing skills: (1) "organization/details," and (2) "grammar/spelling/neatness/level of usage." Supervisors assigned a rating of 1 to 7 in each area, with 1 signifying that "extensive and detailed training" was "compulsory," 4 signifying that the cadet was "meeting minimal academy requirements" but that supervision was still required, and 7 meaning that the cadet was "exceed[ing] academy requirements" and required only general supervision. At no time during Phase I training did Kidd ever rate more than a 3 in either of the two designated report-writing categories. Tucker did little better, but he did manage during the final week of Phase I to rate a 4 in both categories.

The shortcomings in the writing skills of both Kidd and Tucker were apparent enough by the end of sixteenth week of Phase I that they were assigned to remedial training in both form completion and narrative writing along with ten other cadets.3 By the end of the nineteenth week, all but two of the eighty-two cadets were sufficiently skilled to pass report writing. The two who failed were Robert Tucker and William Kidd.

DeBerry acknowledged in his testimony that it is cause for alarm when any cadet fails report writing at the academy. Tr. 206. Nonetheless, because they were performing satisfactorily in other respects, both Kidd and Tucker were permitted to graduate from the academy on October 28, 1994 and advance to Phase II.4 Tucker was assigned to District 4, headquartered in Crestwood, Illinois for his field training, while Kidd was assigned to District 15, headquartered in Oak Brook.

Upon Tucker's graduation from the academy, ISP officials designed and implemented a remedial program aimed at addressing his writing difficulties. In early September, while he was still at the academy, Tucker had been sent for evaluation to a specialist in educational psychology. She had concluded that Tucker suffered from a learning disability characterized by difficulties in retaining and sequencing auditory information, which in turn contributed to his inability to spell words and construct sentences properly. On November 2, 1988, DeBerry reviewed that assessment with Sergeant Wayne Winterberg, Tucker's District FTO Supervisor, and together they developed a special "Field Training Program" for Tucker. This program was authorized by Deputy Director Ronald Grimming, who in turn sent a memo to Deputy Director William O'Sullivan, second-in-command at the ISP, outlining the following components of the program:

1. Tucker was to receive weekly remedial instruction from an educator sophisticated in learning disabilities. Tucker was expected to provide Winterberg with a weekly written summary of the time he had devoted to remedial training, the content of that training, and his progress.

2. Samples of various traffic incident reports were to be culled from the district's files and given to Tucker so that he would be aware of the proper format for these reports.

3. Tucker's supervisor would closely monitor Tucker's enforcement law efforts in the field to ensure that Tucker was not avoiding activity that would require written narratives in his reports.

4. The FTO Platoon Supervisor would meet with Tucker's FTO weekly to review Tucker's DORs, his response to auditory instruction, and his ability to remember instructions.

5. There would be an effort to minimize the time Tucker was required to spend on desk assignments or special details so that he would be exposed to a variety of police training activities.

6. Tucker would be required to proofread his reports before giving them to his FTO. His FTO would, in turn, assess their quality and photocopy them before forwarding them to headquarters.

7. Written instructions would be provided to Tucker in the event that his auditory memory disability created a training problem.

Plaintiff's Ex. 3.5 In an effort to implement this plan, Winterberg spoke with individuals at eight different agencies so that Tucker would be able to choose from a number of qualified, low-cost tutors near his home. Winterberg also obtained articles on training and supervising individuals with learning disabilities and distributed them to Tucker's FTOs.

Tucker at first demonstrated somewhat less initiative in following through with this program than his superiors had in creating it. Six weeks after a tutor was selected, Tucker had kept only three appointments and had missed a number of others. DeBerry found some of the excuses for the missed appointments unacceptable. Tucker also balked when DeBerry asked him to sign a waiver authorizing his tutor to keep the ISP apprised of his progress. Ultimately, DeBerry heeded the advice of ISP counsel that he draft a more narrow waiver for Tucker to sign and thereby avoid creating an "adversary stigma" between Tucker and the ISP.

By March 1989, however, Tucker was seeing a tutor on a twice-weekly basis, and Winterberg reported that "[h]is sentence structure and spelling are improving to an acceptable level." Plaintiff's Ex. 4 at 1.6 Although Tucker's Phase II review board had decided in February to hold him back for additional Phase II training in view of his ongoing difficulties with spelling and grammar, his FTOs were sufficiently pleased with his progress the following month that he was cleared to advance to Phase III. See id. Tucker continued to see a tutor regularly through June, 1989, when the ISP decided to terminate him. Plaintiff's Ex. 6 at 1.7

The ISP's initial efforts to address Kidd's difficulties were more modest. Upon Kidd's graduation from the academy, DeBerry advised Master Sergeant Kenneth Carter, Kidd's District FTO Supervisor, that Kidd had failed report writing at the academy and that he should be paired with an FTO who could help him in that respect. Carter in turn informed Trooper Gary Roy, Kidd's primary FTO, that there had been "problems" with Kidd at the academy, but inexplicably Carter failed to tell Roy what these problems were or what, if anything, he expected Roy to do about them. Tr. 101-02, 104-05. As a result, the ISP did little if anything during the first fourteen weeks of Kidd's Phase II training that could be characterized as an effort to remediate Kidd's writing deficiencies. Indeed, those deficiencies were not immediately apparent to Kidd's FTOs. Roy rode with Kidd for five weeks of Kidd's Phase II training, and although he observed some transpositions in Kidd's reports, he noticed nothing out of the ordinary. Tr. 102-3.

Eight to ten weeks into Phase II, however, it was becoming apparent to Carter and Kidd's FTOs that Kidd was having persistent problems with radio communication as well as report writing and form completion, particularly when he was under stress. Kidd's FTOs noted that he would transmit information over the radio (the license plate number or color of a vehicle that he had stopped, for example) that was incomplete or out of sequence. Often he would spell names phonetically--"George" would be spelled "J-E-O-R-J-E," for instance. It frequently took Kidd an extraordinarily long time to complete his reports. Even so, his reports were rife with errors and omissions. Key sections on forms were left uncompleted. Kidd repeatedly made spelling and grammatical mistakes. He often transposed letters and numbers in addresses, telephone numbers, license plates, and the like. Comments noting these problems began to surface repeatedly in Kidd's DORs. Defendant's Ex. B. Carter also issued counseling forms to Kidd noting his errors. But neither Carter nor anyone else at the ISP suggested that Kidd should seek help from a tutor.8

In early February, following the fourteenth week of Phase II training, a Phase II review board decided that Kidd (like Tucker) should not yet advance to Phase III, but should instead receive additional Phase II training. DeBerry cited as the reason Kidd's difficulties with radio communications and the thoroughness and accuracy of the forms he completed. Again the follow-up with Kidd was different than it was for Tucker. Winterberg and Czak met with Tucker to explain the reasons he was being held back in Phase II. They specifically noted the need for him to improve his spelling and grammar, and suggested that he spend additional time with his tutor. According to Kidd, his superiors held no similar meeting with him and did not in fact communicate to him the reasons for his failure to advance on schedule to Phase III. Roy would later testify that Carter told him Kidd needed to work on his report-writing; Roy did not recall any mention of Kidd's radio usage. Tr. 111, 113. And Carter said nothing to Roy about Kidd seeing a tutor. Id.

Eight months into Kidd's probationary year, the ISP began to look into the reasons why his report writing and radio communications were subpar. A short while after the decision was made to hold Kidd back for further Phase II training, Sergeant Carter asked the ISP's staff psychologist, Dr. Catherine Flanagan, to speak with Kidd about his writing and radio usage. When Flanagan agreed, Carter telephoned Kidd at home and (according to Kidd) asked him to report to headquarters immediately for an eye examination. Tr. 44. Kidd complied with the order and spoke with Dr. Flanagan, who recommended that Kidd be given a psychological evaluation at a clinic in Evanston. Flanagan suspected that Kidd had a mild learning disability with left-to-right confusion. See Plaintiff's Ex. 8. Carter and Gregory Fritz, the Operations Lieutenant for District 15, subsequently met with Kidd and relayed to him Flanagan's impressions.

Although he initially agreed to submit to the evaluation, Kidd later declined. Evanston was a long way from his home in the south suburbs, and he was reluctant to make the trip on his own time. Kidd had also spoken with his mother (a schoolteacher), who believed he had no problem. Moreover, because the need for a psychological examination apparently was never clearly explained to him (see Tr. 289; Memorandum Opinion & Order at 28), Kidd was becoming suspicious of the ISP. Upon learning that Kidd had refused to cooperate, Deputy Director O'Sullivan wrote to Kidd ordering him to submit to a psychological evaluation for the purpose of ascertaining Kidd's "suitability as an Officer." Plaintiff's Ex. 7; Defendant's Ex. J.

Kidd subsequently complied with O'Sullivan's order and submitted to an evaluation by Dr. Eric Ostrov, a specialist in forensic and police psychology at the Isaac Ray Center in Chicago. That evaluation was conducted at the ISP's expense. Consistent with O'Sullivan's letter to Kidd, Ostrov conducted his examination with the aim of determining Kidd's overall fitness for duty, although Ostrov understood that assessment to include an investigation into whether Kidd might have a learning disability. Tr. 121. After Ostrov had seen Kidd on two occasions, he concluded that Kidd was psychologically fit for duty but required some type of remedial assistance. Ostrov reached no conclusion as to whether dyslexia, a learning disability, or some other kind of problem was the source of Kidd's deficiencies in report writing and radio usage. In a letter to Flanagan, Ostrov wrote:

[C]ognitive test results were indicative of perceptual problems and marked difficulty with spelling, reading and arithmetic. His academic and perceptual motor difficulties are far greater in magnitude than would be expected given his apparent native level of intelligence and level of academic achievement.

Defendant's Ex. R at 6.

The ISP proceeded on the assumption that Kidd in fact had a learning disability. In an April 14, 1989 memorandum documenting a conversation with DeBerry, Carter indicated that the ISP had been informed by Flanagan that Kidd had a learning disability that could be corrected with six months of training. Plaintiff's Ex. 8. That training would be offered to Kidd on a "voluntary basis." Id. At that point in time, Kidd was less than two months away from completing his probationary year. In fact, on April 4, the ISP had advanced Kidd to Phase III of his training--solo patrol. Carter wrote that the ISP, notwithstanding the "voluntary" nature of the remedial training, would ask Kidd to agree in writing, "as a condition of employment" that he would participate in that training for no less than six months. Id. DeBerry, Carter, and their superiors privately agreed that if Kidd declined the offer of assistance, he would be terminated. Tr. 214-15. Carter concluded his memorandum with the observation that "I did not say anything to [Kidd] that inferred that he would be terminated if he failed to voluntarily attend." Plaintiff's Ex. 8.

On May 8, 1989, O'Sullivan wrote to Kidd encouraging him to begin a treatment program with Dr. Anderson Freeman, a clinical psychologist. Defendant's Ex. T. O'Sullivan explained that the ISP had "exhausted every method reasonably possible for you to reach an acceptable performance level." Id. He also warned Kidd that "[y]our full cooperation is necessary or else you leave me with no other choice than to seek your termination from the Department." Id.

Kidd agreed to seek counseling. On May 17, he signed an agreement committing himself to the following course of action:

1. Kidd would meet with Dr. Freeman before May 19, 1989, to arrange for a remedial program;

2. Kidd would pursue treatment until such time as his district commander and Dr. Freeman agreed that he was able to perform his duties efficiently or, alternatively, that every reasonable effort had been exhausted in pursuit of that goal; and

3. Kidd would submit weekly reports to his supervisor documenting his progress.

Plaintiff's Ex. 11. Kidd also acknowledged that in the event he did not honor his commitment to avail himself of Dr. Freeman's services, the agreement itself could be used against him in an administrative proceeding (in other words, as a basis for his discharge). Id. Kidd commenced treatment with Dr. Freeman on May 24 and saw him thrice weekly until he was terminated two weeks later. Kidd paid for the sessions ($75 per half-hour) himself.9 So far as the record reveals, neither Dr. Freeman nor anyone else ever actually made a diagnosis as to the nature of Kidd's difficulties in report writing and radio usage.10

Meanwhile, Kidd had proceeded with Phase III solo patrols. During this phase of their training, probationary troopers are evaluated on Bi-Weekly Report Forms which rate them in twenty-one separate categories. Kidd's Bi-Weekly Reports indicated that his performance was "acceptable" in most respects and, on occasion, "superior" in some (including his attitude). His report writing, form completion, and radio usage, were consistently cited as areas which "need[ed] improvement." However, he received no "unsatisfactory" ratings. Plaintiff's Exs. 12-14; Defendant's Ex. C.

Near the end of May, 1989, the ISP began to review the performance of its probationary troopers and to decide whom to keep on and whom to let go. The Phase III review board that convened to assess Tucker's suitability decided that he should be terminated. The principal reasons cited were his difficulty using the "Locaide" navigational device and particularly his poor writing. In the latter regard, the special "Field Training Program" had by this time been in place for more than six months; but in the ISP's view, Tucker had not made sufficient progress in overcoming his learning disability. His district commander wrote:

District # 4 personnel made every effort to help Trooper Tucker improve in the area of report writing, spelling and accuracy in reporting. In light of the effort extended to improve his performance I do not feel that Trooper Tucker has progressed to a satisfactory level in the written communication area. There is no doubt that Trooper Tucker has made an effort to improve and has improved but, it is still less than desirable for the Illinois State Police. We have also afforded Trooper Tucker the maximum amount of time to show improvement.

Plaintiff's Ex. 6 at 3; see also Defendant's Ex. BB at 84-85. Tucker was therefore advised that the ISP planned to terminate him and was given the option to instead resign, as the ISP Field Training Manual directs. See Tr. 221; O'Sullivan Dep. 29; Plaintiff's Ex. 19. Tucker chose to resign.

The outcome for Kidd was at first more favorable. On May 26, 1989, a Phase III review board convened to determine whether Kidd should be advanced beyond his probationary status and made a permanent trooper. Carter and DeBerry, among several others, served on Kidd's review board. The consensus among the board members was that Kidd's performance was unacceptable giving his poor writing skills and difficulties with radio communication. Those difficulties were attributed to Kidd's apparent learning disability, however; and the ISP's Psychological Services had given assurances that the disability was mild and could be overcome with training. Kidd was otherwise performing up to expectations. The board concluded unanimously that Kidd should be advanced to permanent status effective June 12, 1989, and he was so informed. On May 30, Kidd was assigned to Zone 2, where Sergeant Raymond Voss requested that Kidd be assigned to his work group. Plaintiff's Ex. 16.

Although the determinations of a Phase III board are rarely disturbed, senior administrators questioned the decision to advance Kidd given the consensus that he was not yet performing up to expectations. Some time after May 26, DeBerry was summoned to a meeting with Colonel Dowdy, an Assistant Deputy Director reporting to O'Sullivan, and Master Sergeant Michael Koval and Sergeant Diane Carper from the academy. At that meeting, which was called to discuss the status of several probationary troopers, Kidd's name came up. DeBerry, noting that Tucker had already had the benefit of six months of remedial training, expressed the view that no further action should be taken vis a vis Kidd until he had been given the same opportunity. As O'Sullivan later wrote:

The Academy personnel feel that because, upon discovery of the problems involving Trooper Tucker and his learning disabilities, we gave him ample opportunity to self-improve through remedial training, etc; and in as much as we have just recently become aware of Trooper Kidd's need of remedial training, we should also afford him the time period of approximately six months to seek recommended remedial training to correct any deficiencies he may have. Their feeling is, if we do not afford him that opportunity and he files suit on the administrative termination using Trooper Tucker's situation, that in all probability the Department would be required to reinstate him.

Plaintiff's Ex. 17 at 1.

Although he recognized that Kidd and Tucker had been treated differently, O'Sullivan decided nonetheless to recommend Kidd's termination. In a June 6, 1989 memorandum to ISP Director Jeremy Margolis, O'Sullivan expressed concern that Kidd, having attained permanent status, might decide to withdraw from remedial training--notwithstanding Kidd's written commitment to the contrary--and that the ISP in that event would be unable to terminate him. Plaintiff's Ex. 17 at 1-2.11 O'Sullivan would later testify, however, that he would have recommended Kidd's termination irrespective of this concern. In O'Sullivan's view, retaining Kidd would "set a precedent that anybody with a severe learning disability and dyslexia could enter the Illinois State Police" and thereby render the ISP "responsible not only for training them to be a trooper, but for training them out of this learning disability problem, this dyslexia problem." O'Sullivan Dep. 42.12 O'Sullivan was not convinced that the ISP's early and proactive intervention on Tucker's behalf was reason to forego immediate termination of Kidd. "If he files suit with the Department, utilizing the Trooper Tucker incident and wins, we are in no worse a dilemma, as far as terminating Trooper Kidd's employment, then we would be if he withdrew from the training after the probationary period ends." Plaintiff's Ex. 17 at 2.

Upon receipt of O'Sullivan's memorandum, Director Margolis agreed that Kidd should not be retained, and on June 9, 1989, Kidd was terminated. Kidd was summoned to District 15 headquarters on that date and handed a termination letter from Margolis. Plaintiff's Ex. 18. Neither the letter nor the officers at District 15 gave a reason for his discharge. Kidd was not offered the opportunity to resign. Kidd was relieved of his gun, badge, and uniform at the headquarters. By his own account, he was then driven home in his socks and t-shirt, and upon arrival there he was asked to surrender the remainder of his equipment.

Kidd later filed suit under Title VII contending that the ISP had discriminated against him on the basis of his race when it discharged him. The parties consented to disposition by the magistrate judge, and after a four-day trial, Judge Rosemond found in favor of the ISP.

As Kidd had no direct evidence of racial discrimination, the case was tried and analyzed under the framework for indirect evidence set out in McDonnell-Douglas Corp. v. Green, 411 U.S. 792, 802-05, 93 S. Ct. 1817, 1824-26, 36 L. Ed. 2d 668 (1973); see also Texas Dep't of Community Affairs v. Burdine, 450 U.S. 248, 252-56, 101 S. Ct. 1089, 1093-95, 67 L. Ed. 2d 207 (1981). In such cases, the plaintiff bears the initial burden of making out a prima facie case of discrimination. Kidd sought to make that case by showing that (1) he belongs to a protected class; (2) he was qualified for his position and performing satisfactorily; (3) notwithstanding his qualifications and performance, he was discharged; and (4) a similarly situated individual outside of the protected class was treated more favorably than he. See generally McDonnell-Douglas, 411 U.S. at 802, 93 S.Ct. at 1824; Burdine, 450 U.S. at 253-54 & n. 6, 101 S. Ct. at 1094 & n. 6.13 If the plaintiff establishes a prima facie case, the burden of production shifts to his former employer to articulate a legitimate, non-discriminatory reason for his discharge. McDonnell-Douglas, 411 U.S. at 802-03, 93 S.Ct. at 1824; Burdine, 450 U.S. at 254-56, 101 S. Ct. at 1094-95. Once the employer has satisfied that obligation, the burden of production returns to the plaintiff, who must show that the non-discriminatory reason the employer has given for his discharge is pretextual. McDonnell-Douglas, 411 U.S. at 804-05, 93 S.Ct. at 1825-26; Burdine, 450 U.S. at 256, 101 S.Ct. at 1095. At all times, of course, the burden of persuasion remains with the plaintiff to convince the factfinder that the employer discharged him because of his race. Id. at 253, 256, 101 S. Ct. at 1093, 1095.

The trial focused principally on the third element of the prima facie case--whether Kidd's performance at the time of his termination was satisfactory--and on the distinct but overlapping question of whether the ISP's performance-based rationale for discharging Kidd was genuine or pretextual. See Fuka v. Thomson Consumer Elecs., 82 F.3d 1397, 1404 (7th Cir.1996). Kidd pursued two alternate theories in this respect: (1) he was performing satisfactorily and the ISP's assertions to the contrary were pretextual; and (2) to the extent he was not performing satisfactorily, his shortcomings were due to the disparate remedial assistance that the ISP had afforded him versus Tucker. See Memorandum Opinion & Order at 11.

Judge Rosemond acknowledged at the outset that Tucker had been given more prompt and extensive remedial assistance than Kidd, but he concluded that the disparity did not amount to discrimination because the two were not similarly situated. Memorandum Opinion & Order at 11-12. DeBerry had been the first to see the written work of Kidd and Tucker and had noted "marked differences" between the two. Kidd had some difficulties with spelling, run-on sentences, and sentence fragments, but DeBerry believed that the training Kidd would receive in Phase II would enable him to correct those problems. Tucker, on the other hand, had trouble spelling any word of more than five letters, and in fact misspelled even simpler words--"the" and "red," for example. "Tucker had no sentence structure whatsoever and in DeBerry's words, 'there was very little if any hope of him ever writing a police report that would not be embarrassing to the State Police.' " Memorandum Opinion & Order at 12, quoting Tr. 174. The ISP accordingly had intervened earlier on Tucker's behalf because his need for remedial assistance was more easily discerned.

Judge Rosemond found further that the ISP had not withheld assistance from Kidd because of any racial animus. Kidd's FTOs had, to the contrary, commended his attitude and rapport with the public and had expressed the wish to keep him on as a trooper. In his evaluations, he had consistently scored well in areas other than report writing, forms accuracy, and radio communications. Moreover, it was DeBerry who, in Judge Rosemond's words, was the "catalyst" for the differential treatment of Kidd and Tucker, and DeBerry, who is African-American himself, had testified credibly that race played no part in his decision to intervene immediately on Tucker's behalf but not Kidd's. Not until Phase II, when Kidd's FTOs begin to notice a persistent problem in his radio communications and written work, did the ISP begin to suspect his mistakes might be due to a learning disability rather than the carelessness often seen in new recruits. When this possibility finally dawned on Kidd's superiors, the ISP intervened; and in Judge Rosemond's view, the ISP ultimately went to greater lengths on Kidd's behalf than it had for Tucker. In particular, Kidd was referred to psychologists rather than tutors without advanced degrees. Memorandum Opinion & Order at 12-14, 17.

Having concluded that the ISP did not discriminatorily withhold remedial assistance from Kidd, Judge Rosemond proceeded to consider whether Kidd's performance at the time of his discharge warranted discharge. The judge found that the ISP's professed concerns over Kidd's performance were genuine. Kidd had failed report writing at the academy and had been held back in Phase II of the training due to his difficulties with report writing and radio communication. His evaluations during Phase II noted that Kidd transposed letters and numbers, left sections of his reports blank, failed to follow correct radio procedure, and transmitted incorrect and incomplete information when using the radio. Although at week sixteen of Phase II, his evaluations indicated that his performance matched the expectations for probationary troopers in their sixteenth week of training, subsequent evaluations indicated that he was falling behind. And although Kidd eventually was permitted to move on to Phase III, his evaluations during the final month of his employment revealed that the problems with writing and radio usage noted during Phase II persisted. The testimony persuaded Judge Rosemond that accurate and complete reports and radio communications were essential to police work in general and to the public safety in particular. He therefore concluded that the deficiencies noted in Kidd's performance were of sufficient gravity to warrant termination. Memorandum Opinion & Order at 19-24.

In the district court's view, the evidence also warranted the ISP's professed concern that Kidd might not be fully committed to complete the remedial training believed necessary for him to correct his shortcomings. In a March 29, 1989 memorandum, Kidd's district supervisor had observed that Kidd tended to blame others rather than accepting responsibility for his errors. DeBerry perceived Kidd's failure to complete certain assignments as evidence of an unwillingness to correct his writing problems. The ISP similarly saw in Kidd's initial refusal to submit to a psychological evaluation a lack of initiative. Kidd himself testified at trial that he had no real problems with report writing and told his FTO that his problems were not out of the ordinary. He also admitted that he did not always take his trainers' criticisms to heart. Memorandum Opinion & Order at 26-27.

Judge Rosemond acknowledged that the ISP's own conduct was not above reproach. Rather than being candid with Kidd about the reasons it wished him to be examined, the ISP had ordered him to undergo a psychological evaluation for the stated purpose of assessing his "suitability as an Officer." It directed him to an inconvenient location for that purpose, with no assurances that the ISP would bear the cost. When Kidd questioned the need for the evaluation, he was given a "tacit ultimatum" rather than a "simple explanation." In sum, the record was "replete" with evidence that the ISP had "botched" every opportunity it was given to speak constructively with Kidd about his evident learning disability and allay his fear that he was being "set up." Memorandum Opinion & Order at 27-29.

To say that the ISP lacked finesse or tact in the art of personnel management is to give them too much credit. ISP's whole approach cast a punitive and subversive light on their actions, created distrust and doubt, and ultimately led to resentment and retaliation.

Id. at 28.

Although it viewed the ISP's handling of Kidd with "disdain," the district court found nothing "disingenuous" in the reasons it had articulated for his termination. It further concluded that discriminatory intent did not account for the disparity in remedial training made available to Kidd and Tucker. Memorandum Opinion & Order at 29.

Because this case was tried to the bench, we have the benefit of the memorandum opinion reflecting the district court's factual findings and legal analysis as we assess the judgment in favor of the ISP. To the extent Kidd challenges any of the district court's legal conclusions, our review is, of course, plenary. E.g., Chemtool, Inc. v. Lubrication Techs., Inc., 148 F.3d 742, 744-45 (7th Cir.1998). With one exception, however, Kidd's appeal focuses on the district court's findings of fact, and our examination of those findings is constrained. We will reverse the lower court's factual determinations only when the court has committed clear error. FED.R.CIV.P. 52(a); Chemtool, 148 F.3d at 744. Consequently, our review, although by no means "abject," is "deferential." Carr v. Allison Gas Turbine Div., General Motors Corp., 32 F.3d 1007, 1008 (7th Cir.1994). Specifically, clear error requires more than our own belief that we would have weighed the evidence differently had we sat as the trier of fact. Our review of the record as a whole instead must leave us "with the definite and firm conviction that a mistake has been committed." United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 542, 92 L. Ed. 746 (1948), quoted in Anderson v. City of Bessemer City, N.C., 470 U.S. 564, 573, 105 S. Ct. 1504, 1511, 84 L. Ed. 2d 518 (1985). "Where there are two permissible views of the evidence, the factfinder's choice between them cannot be clearly erroneous." Id. at 574, 105 S.Ct. at 1511.

To the extent that the district court's findings rest on credibility determinations, they command even greater deference from us, "for only the trial judge can be aware of the variations in demeanor and tone of voice that bear so heavily on the listener's understanding of and belief in what is said." Anderson, 470 U.S. at 575, 105 S.Ct. at 1512, citing Wainwright v. Witt, 469 U.S. 412, 105 S. Ct. 844, 83 L. Ed. 2d 841 (1985); FED.R.CIV.P. 52(a). Thus, when a particular factual finding rests on the judge's decision to credit one witness over another, that finding will "virtually never" amount to clear error absent an inconsistency with the court's other findings or an evident problem with the plausibility of the credited testimony. Anderson, 470 U.S. at 575, 105 S.Ct. at 1512. However, as the Court recognized in Anderson:

Documents or objective evidence may contradict the witness' story; or the story itself may be so internally inconsistent or implausible on its face that a reasonable factfinder would not credit it. Where such factors are present, the court of appeals may well find clear error even in a finding purportedly based on a credibility determination.

Anderson, 470 U.S. at 575, 105 S.Ct. at 1512; see also, e.g., United States v. Evans, 113 F.3d 1457, 1461-62 (7th Cir.1997).

We turn to the legal question first. As we have noted previously, proving that he was qualified to retain his position is one of the elements that the plaintiff must ordinarily establish in order to make out a prima facie case of discrimination under McDonnell-Douglas. Here, Judge Rosemond found that Kidd was not qualified to become a state trooper at the time of his discharge, given his continuing problems with writing and radio communications. Memorandum Opinion & Order at 19-24. Kidd does not challenge the sufficiency of the evidence supporting this finding, but he contends that the court was wrong to require proof of his qualification. The essence of Kidd's claim is that he was denied remedial assistance on par with that given to Tucker. Kidd thus reasons that he need not show that his performance came up to ISP standards, so long as he is able to show that the ISP discriminated against him in terms of the training and other assistance it gave him to meet those standards. See, e.g., Cosgrove v. Sears, Roebuck & Co., 9 F.3d 1033, 1040-41 (2d Cir.1993); Vaughn v. Edel, 918 F.2d 517, 522-23 (5th Cir.1990); Griffin v. City of Omaha, 785 F.2d 620, 627-28 (8th Cir.1986).

Kidd mischaracterizes the district court's analysis, however. The court acknowledged and gave full consideration to Kidd's contention that the ISP discriminatorily withheld remedial training from Kidd. Memorandum Opinion & Order at 11-17. Only after the district court had concluded that Kidd and Tucker were not similarly situated in terms of their apparent need for remedial assistance, and that the ISP had not invidiously postponed granting such assistance to Kidd, did the court require that Kidd prove that his qualifications and performance were adequate at the time of his discharge. See id. at 17 ("We therefore find ISP did not discriminatorily withhold remedial training from Kidd. Accordingly, Kidd must prove he was qualified and performing satisfactorily as part of his prima facie case."). It is quite plain, in sum, that the district court did not labor under the mistaken impression that Kidd was obligated to prove that his performance measured up to ISP standards even if the ISP had discriminatorily denied to him assistance that would have enabled him to meet those standards.

Kidd's real quarrel is thus with the district court's threshold finding that the ISP did not discriminate against him in terms of its efforts to help him overcome the shortcomings in his written and radio communications. That finding, of course, is one of fact (Pullman-Standard v. Swint, 456 U.S. 273, 287-90, 102 S. Ct. 1781, 1789-91, 72 L. Ed. 2d 66 (1982)), and it rests on three subsidiary factual determinations: (1) that the ISP intervened on Kidd's behalf much later than it did on Tucker's because the need to diagnose and remediate Tucker's problems was apparent earlier (Memorandum Opinion & Order at 11-14, 17); (2) that when it finally became apparent to the ISP that Kidd's difficulties were out of the ordinary, the ISP took steps to lend him assistance that were on par with, if not superior to, the help it had provided to Tucker (id. at 14);14 and (3) that although the manner in which the ISP intervened on Kidd's behalf "cast a punitive and subversive light on their actions" (id. at 28), the negative tone of its efforts was not due to any racial bias (id.). Of these three, we regard the first as key to the district court's judgment, as it addresses the most compelling evidence of disparate treatment--namely, the six-month delay in affording Kidd remedial assistance akin to the intervention on Tucker's behalf. The delay in coming to Kidd's aid meant that by the time Kidd had been evaluated and referred for remedial assistance, his one-year period of probation was nearly at an end. In other words, Kidd was never afforded, as Tucker had been, "the maximum amount of time to show improvement" (see Plaintiff's Ex. 6 at 3) before the decision to discharge him was made. Indeed, the fact that Kidd commenced remedial training so late in his probationary year in turn supplied the ISP with a principal justification for its decision to discharge Kidd--i.e., the concern that it would be difficult to discharge Kidd once he had attained tenure even if he breached his written commitment to obtain remedial help.15

The district court's finding that Kidd and Tucker were not similarly situated in terms of their apparent need for remedial intervention in turn rests to a significant degree upon a credibility assessment. See Memorandum Opinion & Order at 12, 13-14. As we have noted, Sergeant DeBerry, who taught report writing to both Tucker and Kidd, testified that the problems Kidd displayed in writing at the academy were much less pronounced than those Tucker had exhibited. Tr. 173-74, 176-77. DeBerry was of the belief that Kidd's shortcomings could be addressed adequately in the normal course of Phase II training, while he held out little hope that Tucker could "ever writ[e] a police report that would not be embarrassing to the State Police." Tr. 174. For that reason, DeBerry did not believe that Kidd required the type of assistance that the ISP had marshaled for Tucker beginning in November. DeBerry's testimony on that point was crucial, because he was, in the district court's words, "the catalyst for the divergent paths taken by Kidd and Tucker." Memorandum Opinion & Order at 13. Judge Rosemond credited DeBerry's testimony, and concluded accordingly that the ISP had not withheld from or delayed remedial assistance to Kidd because of his race. See Memorandum Opinion & Order at 11-14, 17. Other evidence factored into this determination, but the testimony of DeBerry, who was the first individual at the ISP to observe the relative writing skills of Kidd and Tucker and the person who determined early on that Tucker required remedial intervention, was without doubt the primary foundation for the district court's finding that the delay in aiding Kidd was nondiscriminatory.

We owe the district court's assessment of DeBerry's credibility "special deference." E.g., United States v. Ramunno, 133 F.3d 476, 481 (7th Cir.1998). Judge Rosemond heard DeBerry's testimony and had the opportunity to observe his demeanor; we have not. Even so, other objective evidence in the record stands in significant conflict with DeBerry's testimony, and the district court did not attempt to reconcile the conflict. Although we venture no opinion as to whether this evidence might render the district court's decision to credit DeBerry clearly erroneous, left unaddressed it casts enough doubt on that assessment as to require further consideration by the factfinder.

First, the critical point of DeBerry's testimony was that Kidd and Tucker early on exhibited manifestly different degrees of competence with their writing; yet the marks they received in writing at the academy suggest that the two cadets were performing comparably. Set forth below are the ratings Kidd and Tucker received in writing from the third through the twentieth week of their training. The first number in each case represents the rating assigned to the organization and details of the cadet's report writing, while the second reflects the rating given to his grammar, spelling, neatness, and level of usage. A rating of 1 indicates "[e]xtensive and detailed training [is] compulsory"; 2 indicates that "detailed training [is] essential"; 3 indicates "[a]dditional training [is] essential"; and 4 indicates that the cadet is "[m]eeting minimal Academy requirements [but] supervision [is] required." (The scale continues up to a level of 7, but neither cadet ever earned a rating higher than 4 on his report writing skills at the academy.)

Kidd          Tucker
Week 3        1/1 1/1
Week 4        1/1 1/1
Week 5        1/1 2/2
Week 6        1/1 2/2
Week 7        1/1 2/2
Week 8        1/1 2/2
Week 9        1/1 2/2
Week 10       1/1 2/2
Week 11       2/2 3/3
Week 12       2/2 3/3
Week 13       3/3 3/3
Week 14       3/3 3/3
Week 15       3/3 2/2
Week 16       3/3 2/2
Week 17       3/3 N/A
Week 18       3/3 N/A
Week 19       3/3 3/3
Week 20       3/3 4/4

Plaintiff's Exs. 25, 26. These marks certainly demonstrate that both cadets were having trouble with their written communications, and the remarks accompanying the biweekly reports on Tucker (the record copies of the reports on Kidd include no such remarks) indicate that Tucker had ongoing difficulties with such things as sentence structure and spelling. Plaintiff's Ex. 26. To that extent, the reports are consistent with DeBerry's testimony. But DeBerry also testified that Tucker's problems were much more obvious than Kidd's, and these marks simply do not bear that out. Not only did Tucker receive marks in the final week of Phase I indicating that his report writing met minimal academy standards (Kidd never received such a rating), but his marks are otherwise wholly consistent with (if not slightly better than) Kidd's. Unfortunately, the record tells us little about who determined the ratings each cadet was given in report writing and what criteria they used. The ISP points out that DeBerry was not necessarily the source of these ratings, a possibility which leaves room for the argument that the ratings did not necessarily reflect his views. Yet, it is hard to imagine that DeBerry, as the academy instructor for report writing, had no input into these ratings; and no matter who assigned them, they ostensibly reflect the ISP's official assessment of the candidates' progress. With that in mind, we have difficulty imagining how an individual who, in DeBerry's words, had little hope of writing a report that would not embarrass the ISP, managed to receive equivalent or better marks in report writing than Kidd, whom DeBerry felt simply required a bit of extra supervision in Phase II.

Second, only two students failed report writing at the academy: Kidd and Tucker. Although the fact that both failed still leaves room for the possibility that Tucker could have been perceived as worse off and in need of more intensive help than Kidd, the fact that Kidd failed is to some degree inconsistent with DeBerry's testimony that Kidd's shortcomings were not unusual and appeared amenable to correction in the normal course of Phase II training. It bears repeating that Kidd and Tucker failed report writing only after they, like a number of other cadets, had received remedial writing instruction at the academy. That fact would seem to suggest that both cadets, not just Tucker, had problems which required proactive intervention. In that vein, DeBerry himself acknowledged that any cadet's failure to pass report-writing is "cause for alarm." Tr. 206.

Third and finally, the record contains copies of some of the memoranda that Tucker and Kidd wrote while studying at the academy, and those writing samples appear to undermine DeBerry's efforts to distinguish Tucker's writing skills from Kidd's. DeBerry cited several flaws in Tucker's writing that Kidd's own written work purportedly did not exhibit: Tucker misspelled even the simplest of words--"the" and "red," for example--and had no sentence structure whatsoever. Tr. 173-74. Yet, even a cursory review of the two cadets' writing reveals that both of them had readily apparent problems of this nature.

Consider the following two excerpts from memoranda that Kidd and Tucker wrote during their first week at the academy. (We have highlighted some of the more glaring errors for ease of comparison. Question marks indicate words that the photocopying process has rendered illegible to us.) First, Kidd's memorandum:

TO:                             FROM:
TR. WILLIAM P. HENRICH          CADET WILLIAM KIDD
# 3756                          # 4079
SUBJECT: WHY I WHAT TO          DATE: 06/16/88
BE A STATE TROOPER
  SIR,

I ALWAYS WANTED TO BE A ILLINOIS STATE TROOPER, SENSE I WAS THE AGE OF 14TEEN. I NEVER USED DRUGS BECAUSE I ALWAYS WANTED TO BE A ILLINOIS STATE TROOPER. PERSONLTY, I FEEL THAT MOST OF THE CRIME IN THIS COUNTRY Y IS THE MAINLY CAUSED BY DRUG USERS. I ATTENDED WESTERN ILLINOIS UNIVERSITY WITH A MAJOR IN LAW ENFOREMENT, SO THAT I COULD BE THE BEST STATE TROOPER. I ALWAYS READ BOOKS INVOLVING THE ILLINOIS STATE TROOPERS. I ALWAYS TALK LK TO ILLINOIS STATE TROOPERS. THEY ALWAYS SPEAK HIGHLY OF THE ORGANIZATION. I NEVER HEARDED A NEGATIVE STATEMENT FROM A TROOPER. THAT IS WHY I STUDYED LAW ENFORMENT TO BE THE BEST STATE TROOPER. R.

BECAUSE I HOLD MY SELF IN A LEVEL HIGHER THAN THE AVERAGE PERSON. THE ILLINOIS STATE TROOPERS HAVE BEST IMAGINE. THE ILLINOIS STATE TROOPERS ARE AN ELITE LAW ENFORCEMENT AGENCY. .

SENSE I WAS THE AGE 14 I ALWAYS WANTED TO BE A ILLINOIS STATE TROOPER. THEY MADED A BIG IMPRESSION ON ME, BECAUSE THEY RIDE ALONE IN THEIR SQUAD CARS. NO MATTER WHAT PERIOD OF THE YEAR THEIR CARS ARE ALWAYS CLEAN. .

I PERSONLTY FEEL THAT THE ILLINOIS STATE TROOPERS HAVE THE BEST LOOKING UNIFORMS, I HAVE EVERY SEEN ON A LAW ENFORCEMENT OFFICER.

IF YOU LOOK GOOD, YOU FEEL GREAT. I STRONGLY BELIEVE, IN THAT. IF YOU TAKE PRIDE IN WHAT YOU DO, LIKE THE ILLINOIS STATE TROOPER. I PERSONLTY TRY TO TAKE PRIDE IN WHAT DO. THAT IS ONE MAJOR FACTOR WHY I WANT TO BE A ILLINOIS OIS STATE TROOPER.

I PERSONLTY FEEL THAT I CAN REALTY MAKE A DIFFERENTS AS A ILLINOIS STATE TROOPER. IF I CAN SAVE ONE PERSONS' LIFE. IF I COULD HELP ONE PERSON FROM BEING A VICTIM OF A CRIME AND HELP PEOPLE TO RESPECT THE CONSTITUTIONAL L RIGHTS OF OTHERS.

.    .    .    .    .
	Plaintiff's Ex. 28.  Now Tucker's memorandum:
TO:                           FROM:
TR. DOUGLAS FRANCIS           CADET ROBERT D. TUCKER
3507                          4123
SUBJECT: WHY I WANT           DATE: 06/14/88
TO BE A STATE POLICE
OFFICER

I WANT TO BE A STATE POLICE OFFICER BECAUSE I BELIEVE IN THE JUSTICE SYTEM IN THE U.S. I ALSO BELIVE IN LAW AND ORDER. THE IL STATE POLICE IS THE BEST LAW ENFORCEMENT AGENCY IN THE STATE OF IL. I'VE ALWAYS BEEN INTERESTED IN POLICE WORK. SO I DECIDED IF I'M GOING TO DO SOMETHING NO MATER WHAT IT IS, I WOULD DO THE BEST I CAN. THE REASON WHY I APPLIED FOR IL STATE TROOPER IS BECAUSE THEY ARE THE BEST IN THE STATE. I'VE ALWAYS TRYED AND HAVE HAD BEEN SUCCESSFUL IN MOST HIRES OF SITING HIGH STANDARDS AND GOALS. I ALWAYS GIVE MY ALL. ONCE I'VE ACCOMPLISHED A GOAL I ALWAYS TRY TO MACKE THEM BETTER. I FEEL WHEN I ACCOMPLISH THE GOAL OF BECOMING A STATE TROOPER, I WILL TRY TO DO MY DUTTYS TO THE BEST OF MY ABILITY AND ACCOMPLISH ALL MISSIONS AND TASKS ASSIGNED TO ME WITH 220% OF MY ENERGIZE. IL STATE POLICE ALSO IS A JOB WORKING WITH PEOPLE AND HELPING THEM. FROM EXPERIENCED I'VE HAD WORKING WITH PEOPLE IN THE U.S. ARMY ALSO TEACHING THEM IN MY KARATE SCHOOL I FOR MOST PART CAN COMMUNICATE WITH PEOPLE VERY WELL. TAKING ON THE DUTY AND TASK OF BEING AN IL STATE TROOPER IS A LOT OF RESPONABILITY BUT I KNOW I CAN PERFORM THE DUTIES OF IL STATE TROOPER. I FEEL WONCE I'VE EARNED THE RIGHT TO BE AN IL STATE TROOPER I WILL PERFORM THE DUTIES WITH HIGH STANDARDS.

? A STATE TROOPER IS A MULTI TALENTED ?. BUT HAVING THE EXPERNCE I DO IN SUCH ? AS SNIPPER AND DEMOLITIONS SCOOLS ? THE 3RD RANGER BN AT FT. BENNING GA. I'VE PROVEN MY SELVE BEFORE. ALSOW IN HIGH INTENCY ? BEFORE SUCH AS LIVE FIRE EXCRISES I'VE TACKEN CHARG OF. AS WELL AS TAKING CHARGE OF VARIOUS SIZE GROUPS OF PEOPLE....

Plaintiff's Ex. 27.

It is readily apparent from these two writing samples that both Kidd and Tucker had fundamental difficulties with writing: both misspelled relatively simple words (e.g., "want," "since," "make," and "once"), and both had trouble with sentence structure. Many of these errors would (we hope) be cause for alarm if the writer were a junior high school student, let alone an ISP cadet with a college education. It is true, as the ISP points out, that these writing samples were not necessarily representative of the entire body of writing that Kidd and Tucker completed at the academy nor of the more focused report writing that Kidd and Tucker would have done in the classroom under DeBerry's instruction. (The record is, unfortunately, devoid of writing samples from DeBerry's class at the academy.) Yet, we are given no reason to believe that the elementary types of errors that both cadets made in these essays would not also have appeared in the types of reports they prepared for DeBerry. And if, as DeBerry suggested, the prospect of embarrassment was the benchmark against which he judged the need to intervene on a cadet's behalf, then it is not at all apparent to us why the ISP believed that Tucker required specialized assistance but Kidd did not. Both memoranda struck us as quite discomfiting, having been prepared by prospective members of what Kidd described as an "elite law enforcement agency."

More important, perhaps, than our own impression of Tucker's shortcomings in writing is that of another key decisionmaker at the ISP--Deputy Director O'Sullivan. It was O'Sullivan, after all, who recommended to Margolis that Kidd be discharged, and O'Sullivan cited Kidd's inadequate writing skills as a prime reason. Plaintiff's Ex. 17 at 2. In his memorandum to Margolis, O'Sullivan listed the following flaws in Kidd's report writing: "[g]laringly misspells words, including some simple three-letter words"; "[v]ery poor sentence structure"; and "[g]rammatically incorrect". Id. (emphasis ours). These are, interestingly enough, quite similar to the difficulties that DeBerry had attributed to Tucker (see Tr. 173-74), and to that extent they stand in tension with DeBerry's assessment that Kidd's errors were not as grave as Tucker's. Of course, O'Sullivan's memorandum addressed Kidd's writing as of June 1989, while DeBerry's testimony addressed his writing proficiency as it had been upon Kidd's graduation from the academy six months earlier. Once again, however, we cannot imagine that the kinds of fundamental flaws that O'Sullivan described would not have been evident from the outset of Kidd's training at the academy. To the contrary, the writing exercise from which we have quoted above suggests that Kidd's difficulties, like Tucker's, were evident from the start.

These are not the kinds of conflicts we routinely see between the testimony of two witnesses, one of whom the district court credits and the other not, or between the testimony of a witness whom the district court believes and documentary evidence that is open to interpretations consistent with that testimony. The apparent conflict here is between the testimony of a defense witness and contrary documentary evidence which also emanates from the defendant--the writing scores that the ISP itself gave Tucker and Kidd, the failing grades that the ISP itself gave to both Tucker and Kidd, the appalling writing samples that the ISP itself received from both Tucker and Kidd. Although it may well be possible to reconcile this conflict, that cannot be done simply by "disbelieving" the objective documentary evidence, for there is no doubt that this evidence is authentic and that it sheds significant, if not definitive, light on the relative writing skills of Tucker and Kidd.

We have raised but will not resolve the seeming inconsistencies between this evidence and the district court's finding that DeBerry was credible when he testified that Tucker was in more apparent need of intensive assistance than Kidd. We are not factfinders, and the choices among competing inferences arising from the record are not for us to make. We also recognize that Judge Rosemond, as the finder of fact, was not required to expressly address all of the evidence and to resolve any and all inconsistencies among that evidence. See Greater Kansas City Laborers Pension Fund v. Superior Gen. Contractors, Inc., 104 F.3d 1050, 1056 (8th Cir.1997); Rollins v. Fort Bend Independent School Dist., 89 F.3d 1205, 1221 (5th Cir.1996); Griffin v. City of Omaha, supra, 785 F.2d at 627. If there were readily apparent explanations for the discrepancies between DeBerry's testimony and the evidence we have just summarized, then we would find ourselves in the familiar position of deferring to the district court's credibility assessment. Having reviewed the record, however, we cannot say that the explanations for the inconsistencies are manifest.

Although Rule 52 commands deference to the district court's factual findings, it does not, as the Eighth Circuit has explained, "free the district court of the burden of assuring the appellate court, through findings of fact or references in its memorandum opinion, that it has considered strongly conflicting evidence and come to grips with apparently irreconcilable conflicts." Griffin, 785 F.2d at 628 (internal quotation marks and citations omitted); see also United States v. Henderson, 58 F.3d 1145, 1152 n. 1 (7th Cir.1995); United States v. Randolph, 101 F.3d 607, 609 (8th Cir.1996); Rollins, 89 F.3d at 1221. Because the district court's memorandum opinion here does not address the evidence that conflicts with DeBerry's testimony, we cannot be sure that the court considered that evidence or, if so, how it reconciled the conflict. Only with the benefit of Judge Rosemond's thinking in this regard can we exercise meaningful review of his critical factual findings.

We have concluded that the most prudent action for us to take at this juncture is to remand the case to Judge Rosemond so that he can address the evidence we have highlighted. Although we recognize that this will postpone the resolution of an already aged lawsuit and impose additional burdens on the parties as well as the district court, we believe it is the course most respectful of the fact-finding prerogative of the district court.

We ask the district court to address one other portion of the record on remand. In additional support of his position that discrimination accounted for the differential treatment of Kidd and Tucker, Kidd presented evidence that African-American probationary troopers were discharged in numbers disproportionate to their representation in the cadet classes hired by the ISP. Plaintiff's Ex. 21.16 That evidence revealed that seventeen percent of the individuals in the nine cadet classes that entered the ISP academy from June 1986 through November 1990 were African-American, while African-Americans accounted for a much greater percentage--sixty-one percent--of those involuntarily terminated from these classes.17 The court found these statistics to be "probative of very little." Memorandum Opinion & Order at 24. "For statistics to be useful ...," the court explained, "they must show a racial or ethnic imbalance between the composition of an employer's workforce and the relevant labor market." Id. at 25. For purposes of a valid comparison, identifying the relevant labor market is "essential"; and the plaintiff's evidence was defective in that regard because it supplied no information about the percentage of African-Americans in the relevant labor pool. Id. at 25-26.

We believe that the court did not fully appreciate the purpose for which Kidd offered this evidence, however. Had Kidd proffered the statistics to establish a pattern of discriminatory hiring, the relevant comparison would be one between the racial composition of ISP's cadet classes and the racial makeup of the workforce qualified for employment as ISP troopers. See, e.g., Hazelwood School Dist. v. United States, 433 U.S. 299, 308, 97 S. Ct. 2736, 2742, 53 L. Ed. 2d 768 (1977), citing International Brotherhood of Teamsters v. United States, 431 U.S. 324, 337-38 & n. 17, 97 S. Ct. 1843, 1855 & n. 17, 52 L. Ed. 2d 396 (1977); Ibarra v. Martin, 143 F.3d 286, 291 (7th Cir.1998). But Kidd did not offer the statistics for that purpose. His aim instead was to establish a pattern of discriminatory discharges. For that purpose, the relevant comparison to be made is one between the racial makeup of the cadet classes and the racial makeup of the sub-group of cadets who were discharged. See id. at 291-92; Lilly v. Harris-Teeter Supermarket, 720 F.2d 326, 335-36 (4th Cir.1983), cert. denied, 466 U.S. 951, 104 S. Ct. 2154, 80 L. Ed. 2d 539 (1984). That is precisely the comparison that Kidd drew from the evidence. See R. 165 Ex. C (Plaintiff's Proposed Findings of Fact and Conclusions of Law) at 22-24, pp 70-73. The district court's rationale for disregarding this evidence was, consequently, misplaced. For that reason, we believe the court should look at this evidence afresh on remand.

In asking the district court to reconsider this evidence, we express no opinion as to what weight the court should give the statistics or what inference it may or should draw from them. The district court is far better situated than we to gauge the significance of these statistics in the context of all of the other evidence presented to the court. The court may take them for what they are worth.

So that the district court may promptly address the concerns we have raised, the mandate shall issue immediately. On remand, Judge Rosemond is free to request whatever additional submissions from the parties, or to conduct such further proceedings, as he believes are appropriate. We will retain jurisdiction over this appeal while we await the district court's supplemental findings.

The case is remanded to the district court for the limited purpose of making supplemental findings concerning the matters addressed herein. This court retains jurisdiction over the appeal.