OPINION OF THE COURT
John Doe is a medical student who has multiple sclerosis. The National Board of Medical Examiners (NBME) provided testing accommodations to Doe when he took Step 1 and Step 2 of the United States Medical Licensing Examination (USMLE), as it concedes it is required to do under Title III of the Americans with Disabilities Act (“ADA”), Pub.L. 101-336, Title III, 42 U.S.C. § 12181 et seq. The accommodations provided included extra time to complete each examination. The NBME annotates the scores of examinees who receive testing accommodations if, in its judgment, the accommodations affect the comparability of the accommodated score to non-accommodated scores. The NBME follows this practice because it believes that it owes a duty of candor to the users of USMLE scores to disclose factors that may affect the meaning of an exami-nee’s scores. Although the USMLE was designed as a licensing examination, at the request of examinees, the NBME will send Step 1 and Step 2 scores to hospitals sponsoring residency and internship programs for use in evaluating candidates for admission to their programs. Examinees typically make such requests. Doe claims that, as applied to him, the NBME’s practice of flagging accommodated scores violates Title III of the ADA.
Doe is currently in the process of applying for residencies and internships in physical medicine and rehabilitation. He brought suit in the District Court for the Eastern District of Pennsylvania seeking to enjoin the NBME from annotating his scores to reflect that he received testing accommodations. By consent of the parties, Doe’s motion for a preliminary injunction prohibiting the NBME from annotating his scores was assigned to a *149Magistrate Judge (hereafter the District Court). After a three-day hearing, the District Court granted the motion, holding that Doe had standing to sue, that he had demonstrated a reasonable likelihood of success on his claim that annotating his scores violated section 302 of the ADA, and that he had demonstrated that he would be irreparably harmed absent an injunction. This expedited appeal followed (Doe must send his scores to the residency programs soon if he is to be seriously considered in the matching process that will take place in early 2000).
The critical questions on appeal are (1) whether Doe has standing to sue; (2) what section of Title III of the ADA governs Doe’s claim; (3) whether the very act of annotating Doe’s scores violates the ADA; and (4) whether Doe has proven that the additional time did not affect the comparability of his scores to non-accommodated scores, and thus that the flag imposes an inequality on him. We conclude that, although flagging sufficiently injures Doe to surmount the NBME’s argument that Doe lacks standing to sue, flagging does not constitute an ipso facto violation of Title III of the ADA. In so doing, we conclude that section 309 of Title III, 42 U.S.C. § 12189, the section specifically governing examinations, and not section 302, 42 U.S.C. § 12182 the general provision on discrimination in public accommodations, controls this case.1
We also conclude that, in order to demonstrate a reasonable likelihood of success on his claim under section 309, Doe bore the burden of showing that his scores were comparable to non-accommodated scores in terms of predicting his future success, and that he failed to meet this burden. The District Court’s conclusion that Doe had demonstrated a reasonable likelihood of success on his claims under Title III of the ADA thus was unsupported by the evidence Doe presented and the factual conclusions the Court reached. Accordingly, we hold that the District Court abused its discretion in determining that Doe had shown a reasonable likelihood of success on the merits, and we vacate the order granting the preliminary injunction.
I. Facts & Procedural History
The NBME, together with the Federation of State Medical Boards of the United States, Inc., offers the USMLE. The USMLE is a standardized multiple-choice test administered in three parts, or “Steps”. The USMLE was designed as a licensing exam meant to assess an ex-aminee’s understanding of, and ability to apply, concepts and principles that are important in health and disease and constitute the basis of safe and effective patient care. In order to obtain a license to practice medicine in the United States, an examinee must obtain a passing score on all three Steps of the USMLE. Prior to May 1999, the USMLE was provided in a written format. Since May 1999, the USMLE has been given in a computerized format. After an examinee takes the USMLE, the NBME sends a score report to the examinee. Although the USMLE was designed for use as a licensing exam, it is common practice for residency and fellowship programs to use USMLE test scores in evaluating candidates for admission to their programs. At an examinee’s request, the NBME will send a USMLE score transcript to third parties designated by the examinee, including residency and internship programs and state licensing authorities.
When examinees with disabilities apply to take the USMLE, they can request that the NBME provide testing accommodations. An examinee must support such a request with evidence that he is disabled and that a particular accommodation is an appropriate accommodation for his disability. Examples of accommodations that the *150NBME has provided in the past include large type, assistance filling in answer sheets, and extra time.
When an examinee is granted a testing accommodation of extra time, the NBME flags the examinee’s transcript of scores with the statement “Testing Accommodations” on the front of the trans'cript and a comment on the back of the transcript stating: “Following review and approval of a request from the examinee, testing accommodations were provided in the administration of the examination.” The NBME flags only those testing accommodations that its experts conclude may affect the validity of a score. For example, an accommodation providing a test in large print would not be flagged. The NBME flags scores obtained under extra time accommodations because its psychometri-cians have concluded that scores obtained with extra time accommodations may not be comparable to scores obtained under standardized conditions.2 In such circumstances, according to the NBME, the extra time may under or overcompensate for the test-taker’s disability.
John Doe currently is a fourth-year medical student at the Medical College of Virginia. He was diagnosed with multiple sclerosis in the summer of 1987, when he was in college. Doe’s condition causes muscular spasticity, fíne motor problems, urgency of the bowel and bladder, and occasional incontinence. Doe does not have any learning disabilities, and his multiple sclerosis does not affect his cognitive abilities. The type, frequency, and duration of symptoms that Doe experiences vary and are unpredictable. The parties agree that Doe is disabled within the meaning of the ADA.
When Doe applied to the NBME to take Step 1 of the USMLE, he completed a NBME questionnaire in order to request testing accommodations. On that questionnaire, Doe informed the NBME that he had a physical disability. After several communications between the NBME and Doe in which Doe refused lesser accommodations, the NBME provided the following accommodations for Doe’s Step 1 examination: (1) time and one half to take the examination; and (2) a special seating 'assignment close to the restroom. Doe requested these accommodations because his condition can require him to stop and stretch his muscles frequently, taking many “micro-breaks,” and to visit the restroom often. Doe concedes that it is possible for him to continue considering questions on the exam while he takes these breaks.
Doe’s score report for Step 1 of the USMLE contained an annotation that Doe received testing accommodations for the examination. After he received the scores, Doe wrote to the NBME and requested that it remove the annotation from his scores. The NBME denied Doe’s request. When Doe applied to take Step 2 of the USMLE, he again requested testing accommodations from the NBME. Although Doe only requested time and a half for Step 2, the NBME provided Doe with double time. It did so because the computerized version of the test, which is the version of Step 2 Doe was applying to take, is designed so that the only available extra time accommodation is double time. The NBME expects to report Doe’s Step 2 scores sometime in December of this year.
Doe has sent flagged Step 1 scores to some, but not all, of the physical medicine and rehabilitation residency and internship programs to which he is applying. He has been offered interviews at some of the programs to which he applied, which review applications and make decisions regarding interviews on a rolling basis.
A flagged score effectively indicates to anyone familiar with the NBME’s policies *151regarding flagging that the examinee has a disability of some sort, because only disabled people receive testing accommodations. The NBME will respond to inquiries from third parties who have received annotated scores regarding the nature of the accommodation provided, but it will not release information regarding the disability for which the accommodation was given. In Doe’s case, it would inform residency and internship programs who made inquiries about the flag that Doe received extra time on his examinations, but it would not reveal to the programs that Doe has multiple sclerosis.
The District Court found that the NBME had not shown that it must flag the scores of accommodated examinees in order to secure the psychometric soundness of the USMLE. Significantly, however, the Court declined to conclude whether it is possible to determine psychometrically if the score of a candidate who received an accommodation of extra time is better than, worse than, or the same as the same score for a candidate who took the exam under standardized conditions.
Tests vary along a continuum in the extent to which they are “power” or “speeded” tests. A purely power test measures an examinee’s knowledge of the subject of the exam with no time constraints. A purely speeded test measures the time in which an examinee can complete ministerial tasks. The USMLE exams are primarily power tests, but they have a speeded component as well. Some 25% of examinees have reported that they felt that they could have benefitted from more time on the examination. There was conflicting expert testimony regarding the comparability of time-accommodated scores to scores achieved under standard conditions. The NBME’s experts testified to a lack of evidence of comparability. For example, Dr. Mehrens testified that “[a]l-though research has suggested that accommodated scores tend to overpredict [success], research has certainly not informed us regarding the exact probability” of error in comparing accommodated and non-accommodated test scores. Doe’s expert, Dr. Geisinger, testified that providing extra time to individuals with disabilities leads to results comparable to tests taken under standard conditions; he acknowledged, however, that it would be difficult to determine whether Doe received any advantage from the extra time accommodation. As noted above, the District Court declined to make a finding of comparability on this evidence.
Doe believes that he will be discriminated against by residency and internship programs if he submits flagged scores. He testified at the preliminary injunction hearing, however, that he did not know whether individuals at the programs to which he had applied had any concerns about admitting persons with disabilities. He also testified that he had not been told that he would be denied admission to any program because of the annotation or because of his disability.
Pressed at oral argument to identify evidence supporting Doe’s belief, Doe’s counsel offered three bases in support of the assertion that the programs to which Doe has applied will discriminate against him. First, he offered Doe’s own experience. In 1988, Doe took admission examinations for both medical school and law school. He was accepted to one of the two medical schools to which he applied as well as to law school, and he decided on the law. After completing law school and practicing law with prestigious law firms for five years, he decided to reapply to medical school. The second time around, he applied over the course of two years and was accepted to only one of the thirty-two medical schools to which he applied.
Doe argues that the comparison between his experience applying to medical school directly from college, where his scores on the medical school admissions exam were not flagged and he was accepted at 1 of the 2 schools to which he applied, and his experience applying to medical school after practicing law for several years, where his *152scores were flagged and he was accepted to 1 of the 32 schools to which he applied, is evidence that residency and internship programs will discriminate against him. He did not, however, present any evidence of the relative selectivity of the schools to which he applied the first and the second time (which could explain the result), or evidence regarding his grades and tests scores as compared to other applicants against whom he was competing the first and second time he applied, or evidence that his success rate was lower the second time he applied because of the flag rather than as a result of some other factor — such as the possibility that he was a less attractive candidate for medical school the second time because he had practiced law for five years. He also did not present any evidence that residency and internship programs would be likely to respond to his application the same way that medical schools responded.
Second, Doe’s counsel cited testimony by Dr. Geisinger, Doe’s expert, that he believed that some programs might discriminate against Doe on the basis of his disability. More specifically, Dr. Geisinger stated that he believed some small programs might discriminate against disabled candidates because of the potential cost of accommodating a disabled resident, citing a study by Warren W. Willingham on the testing of handicapped people. After offering this opinion, however, Dr. Geisinger was asked “but there’s no research that supports anything you just said, is there?” He replied “I would say there is no empirical research.”
Third, Doe’s counsel cited the Willing-ham study referred to by Dr. Geisinger. But the Willingham study, which did not involve the USMLE, is equivocal. It states both that “overall the selection process for handicapped applicants was comparable to that for the nonhandicapped in the sense that decisions followed quite closely what one would expect from HSG and SAT scores” and that “admissions were lower than predicted for a relatively small number of visually impaired and physically handicapped students applying to smaller institutions.”
The NBME opposed Doe’s motion for a preliminary injunction by arguing that Doe lacked standing and that he had not met the requirements for a preliminary injunction. The NBME also argued that the court should recognize a communicatory privilege protecting its good faith communications to users of USMLE scores.
The District Court held that Doe had standing to sue the NBME, that he had demonstrated a reasonable likelihood of showing that the practice of flagging violated his rights under the ADA, and that he had demonstrated that he would be irreparably harmed absent an injunction. It granted Doe’s motion, enjoining the NBME from annotating or flagging Doe’s scores on Step 1 and Step 2 of the USMLE. The NBME appeals from the order granting the preliminary injunction.
Pursuant to 28 U.S.C. § 636(c), an aggrieved party to a matter heard by a magistrate by consent of the parties “may appeal directly to the appropriate United States court of appeals from the judgment of the magistrate in the same manner as an appeal from any other judgment of a district court.” This Court has jurisdiction pursuant to 28 U.S.C. § 1292(a)(1), which confers jurisdiction on the Courts of Appeals to hear appeals from interlocutory orders granting injunctions.
II. Standing
The “irreducible constitutional minimum of standing” has three parts: injury in fact (a concrete harm suffered by the plaintiff that is actual or imminent), causation, and redressibility. See Lujan v. Defenders of Wildlife, 504 U.S. 555, 560, 112 S.Ct. 2130, 119 L.Ed.2d 351 (1992). Because these requirements are not pleading requirements, but are necessary elements of a plaintiffs ease, mere allegations will not support standing at the preliminary injunction stage. “[E]ach element [of *153standing] must be supported in the same way as any other matter on which the plaintiff bears the burden of proof, i.e. with the manner and degree of evidence required at the successive stages of litigation.” Id. at 561, 112 S.Ct. 2130. Doe has not adduced evidence demonstrating more than a mere possibility that he will be discriminated against by residency and internship programs if his scores are flagged. Accordingly, Doe has not demonstrated standing on this basis.
The District Court concluded that Doe had standing because he “pled infringement of the right to be free from discrimination under the [ADA].” This formulation of standing ignores the requirement that, at the preliminary injunction stage, allegations are not enough to support standing, and it incorrectly equates a violation of a statute with an injury sufficient to confer standing. The proper analysis of standing focuses on whether the plaintiff suffered an actual injury, not on whether a statute was violated. Although Congress can expand standing by enacting a law enabling someone to sue on what was already a de facto injury to that person, it cannot confer standing by statute alone. See Lujan, 504 U.S. at 578, 112 S.Ct. 2130 (noting that Congress can “ele-vat[e] to the status of a legally cognizable injur[y] concrete, de facto injuries that were previously inadequate in law.”).
Doe has identified, however, an injury in fact that provides an alternative basis for standing. He has demonstrated that the flag on his test scores identifies him as a disabled person. Being so identified harms him in the sense that, because of his justifiable and reasonable concern as a disabled person with how people who can affect his future and his livelihood, and whose judgment may be informed by the information, will perceive him, he has actively sought to avoid being so identified. We are persuaded that this injury-being identified as a disabled person against his will-is enough to establish that Doe has suffered a concrete harm as a result of the NBME's policy of flagging accommodated scores.
The jurisprudence of standing is littered with cases in which courts have dismissed actions because the injury was not personal (i.e., it accrued to third parties), or the injury was not concrete (ie., it was too theoretical), or the injury was not actual or imminent (ie., it was speculative), but that is not the case here. The injury identified is personal to Doe; he is not claiming an injury from generalized discrimination against disabled persons or suing on behalf of the disabled, he is claiming that it hurts him personally to be identified as a disabled person when he has explicitly stated that he does not want to be so identified. If his fear of discrimination were unfounded, we might call this a purely theoretical injury (i.e., one that is not concrete). But because his fear is based in reality, Doe’s injury to his interest in keeping his disability private is concrete. Similarly, the injury is actual/ imminent. Some of his score reports already have been flagged, and the others are sure to be flagged absent an injunction. Thus, we conclude that Doe has met the actual injury component of the constitutional standing requirement.
Because the injury complained of is an injury fairly traceable to the NBME that would be redressed by the relief Doe seeks, we conclude that Doe has met the constitutional standing requirement.3 This *154conclusion, however, is analytically separate from the question whether flagging in these circumstances constitutes discrimination under Title III of the ADA, to which we now turn.
III. The Preliminary Injunction
A. General Standards
This Court reviews orders granting preliminary injunctions for abuse of discretion. We review underlying findings of fact for clear error and consider questions of law de novo. See Acierno v. New Castle County, 40 F.3d 645, 652 (3d Cir.1994). A court abuses its discretion when its ruling is founded on an error of law or a misapplication of law to the facts. See Marco v. Accent Pub. Co., 969 F.2d 1547, 1548 (3d Cir.1992). Accordingly, inasmuch as the result depends upon a question of law, namely, whether the practice of flagging test scores violates the Americans with Disabilities Act, we exercise plenary review. See In re Assets of Myles Martin, 1 F.3d 1351, 1357 (3d Cir.1993).
“Four factors govern a district court’s decision whether to issue a preliminary injunction: (1) whether the movant has shown a reasonable probability of success on the merits; (2) whether the mov-ant will be irreparably injured by denial of the relief; (3) whether granting preliminary relief will result in even greater harm to the nonmoving party; and (4) whether granting the preliminary relief will be in the public interest.” American Civil Liberties Union of New Jersey v. Black Horse Pike Regional Bd. of Educ., 84 F.3d 1471, 1477 n. 2 (3d Cir.1996) (en banc). If the order granting the preliminary injunction is to be upheld, Doe must demonstrate that the District Court did not abuse its discretion in finding both that he had a reasonable probability of success on his claim that flagging his score violates the ADA, and that he would be irreparably harmed if an injunction did not issue. As we hold that Doe has not demonstrated a reasonable chance of success on his claim that flagging his scores violates the ADA, we vacate the order granting a preliminary injunction without reaching the question of irreparable harm.4
B. The Specific Controls the General
The District Court analyzed the “flag” under section 302 of the ADA, which sets forth general provisions prohibiting discrimination in public accommodations, 42 U.S.C. § 12182. It failed to consider whether section 309, the more specific statute governing discrimination by providers of examinations, effectively defines the requirements of Title III of the ADA with regard to examinations.5
In reviewing this decision, we begin with the ordinary tools of statutory *155construction. “[I]t is a commonplace of statutory construction that the specific governs the general.” Morales v. Trans World Airlines, Inc., 504 U.S. 374, 384, 112 S.Ct. 2031, 119 L.Ed.2d 157 (1992) (citing Crawford Fitting Co. v. J.T. Gibbons, Inc., 482 U.S. 437, 445, 107 S.Ct. 2494, 96 L.Ed.2d 385 (1987)); see also Fourco Glass Co. v. Transmirra Products Corp., 353 U.S. 222, 228, 77 S.Ct. 787, 1 L.Ed.2d 786 (1957) (“The law is settled that however inclusive may be the general language of a statute, it will not be held to apply to a matter specifically dealt with in another part of the same enactment.”) (citations omitted). This principle has special force when Congress has targeted specific problems with specific solutions in the context of a general statute. See HCSC-Laundry v. United States, 450 U.S. 1, 6, 101 S.Ct. 836, 67 L.Ed.2d 1 (1981) (per curiam). It applies “particularly when the twofprovi-sions] are interrelated and closely positioned, both in fact being parts of’ the same statutory scheme. See HCSC-Laundry, 450 U.S. at 6, 101 S.Ct. 836. But see Varity Corp. v. Howe, 516 U.S. 489, 511, 116 S.Ct. 1065, 134 L.Ed.2d 130 (1996) (rejecting argument that specific limitation on remedies in one provision of a statute trumped a general provision for remedies in another section).6
An analogous case suggests that the District Court erred in analyzing the case under section 302 instead of section 309. In HCSC-Laundry, a cooperative laundry formed by a group of hospitals challenged a ruling of the Internal Revenue Service. The service denied the laundry not-for-profit status under the general provisions regarding not-for-profit organizations of section 501 of the Tax Code, because a more specific provision under 501 governed the not-for-profit status of hospitals, and the laundry did not fit within that provision. In holding that the laundry could not claim not-for-profit status under the general provision, it was significant to the Court that both provisions were “parts of 501 relating to exemption of organizations from tax.” 450 U.S. at 6, 101 S.Ct. 836.
Here, by analogy, both 309 and 302 are parts of Title III, which prohibits discrimination in public accommodations. We believe that the rationale of the “specific governs the general” canon counsels that we treat section 309 as Congress’s specific definition of what Title III requires in the context of examinations. Moreover, although applying section 302 would not necessarily undermine limitations created by section 309 (neither section explicitly mentions flagging), it would render 309 superfluous. If section 302 settled the question, there would have been no need to enact section 309. Accordingly, we conclude that section 309 governs in this case.
C. Section 309
Section 309 does not explicitly bar the practice of flagging the test scores of examinees who have received testing accommodations. It provides that “[a]ny person that offers examinations or courses related to applications, licensing, certification, or credentialing for secondary or post-secondary education, professional, or trade purposes shall offer such examinations or courses in a place and manner accessible to persons with disabilities or offer alternative accessible arrangements for such individuals.” 42 U.S.C. § 12189. The NBME concedes that this provision required it to accommodate Doe’s disability when he took the exam. It argues, *156however, that it is not required to keep the provision of an accommodation secret from programs that use USMLE scores to evaluate candidates.
While the Department of Justice regulations interpreting section 309 provide a useful explication of its meaning, they also make no explicit mention of the practice of flagging accommodations. They interpret the section to require that
“[t]he examination is selected and administered so as to best ensure that, when the examination is administered to an individual with a disability that impairs sensory, manual, or speaking skills, the examination results accurately reflect the individual’s aptitude or achievement level or whatever other factor the examination purports to measure, rather than reflecting the individual’s impaired sensory, manual, or speaking skills (except where those skills are the factors that the examination purports to measure).” 28 C.F.R. § 36.309(b)(l)(i).
Because he cannot point to an explicit bar on the practice of flagging, Doe argues that the annotation unfairly calls into question the validity of his scores and in effect denies him the opportunity to take the exam “in a place and manner accessible” to him. Doe reads too much into the phrase “in a [ ] manner accessible to persons with disabilities.” He would have us hold that the phrase “in a manner accessible” includes by implication the requirement that the resulting scores be declared psycho-metrically comparable to the scores of examinees who take the test under standard conditions. However, neither the language of the statute nor the regulation interpreting it sets forth or implies such a requirement.
The term “accessible” is not best understood to mean “exactly comparable.” The notion of accessibility, or best ensuring that examination results accurately reflect “aptitude or achievement level,” see 28 C.F.R. § 36.309(b)(l)(i), does not mandate that the NBME provide examinations to the disabled that yield technically equal results; it mandates changes to examina^ tions — “alternative accessible arrangements,” 42 U.S.C. § 12189 — so that disabled people who are disadvantaged by certain features of standardized examinations may take the examinations without those features that disadvantage them.
This is not a case in which the NBME refused to provide Doe with a score. The annotation does not state that Doe’s scores are invalid. Moreover, Doe has not adduced evidence that residency and internship programs would regard the annotation as a signal of invalidity. As the evidence described supra at pages 7-8 reflects, he also has not proven that his scores are comparable to non-accommodated scores, and thus that, by flagging, the NBME has imposed an inequality on him by treating the same thing differently. Indeed, the District Court explicitly refused to conclude that the Doe’s scores are comparable: “the larger issue of whether, in fact, standardized scores and scores obtained by disabled individuals for whom time-related accommodations were granted are comparable in psychometric terms ... need not be answered by me.” Doe v. National Board of Medical Examiners, 1999 WL 997141, at *12. The expert testimony was unanimous that it is not possible to know how scores of exams taken with accommodations compare to scores of exams taken under standard conditions. The annotation simply indicates that Doe’s scores are not psy-chometrically comparable to the scores of examinees who took the test without accommodations.
In the absence of a statutory proscription against annotating the test scores of examinees who receive accommodations, we do not view the annotation on Doe’s score — or its implications as just described — as itself constituting a denial of access. If Doe were to establish either that his scores are psychometrieally comparable to the scores of candidates who take the test under standard time condi*157tions, or that his scores will be ignored by the programs to which they are reported, he might have demonstrated a reasonable likelihood of success on this claim. He has not met these evidentiary burdens. It may be that Doe will be able to develop a fuller record at final hearing. On the current record, however, he has not shown a reasonable likelihood that he will prevail.
D. Section 302
Although we have concluded that section 309 defines the requirements of the Title III of the ADA as applied to examinations, we also note that nothing in section 302, the section under which the case was decided by the District Court, gives us reason to believe that Doe would have demonstrated a reasonable likelihood of success under section 302 if it were the appropriate section to apply. Section 302 provides that “[N]o individual shall be discriminated against on the basis of disability in the full and equal enjoyment of the goods, services, facilities, privileges, advantages, or accommodations of any place of public accommodation by any person who owns, leases (or leases to), or operates a place of public accommodation.” 42 U.S.C. § 12182. Our cases construing section 302 hold that “[t]he plain meaning of Title III is that a public accommodation is a place, leading to the conclusion that ‘[i]t is all of the services which the public accommodation offers, not all of the services which the lessor of the public accommodation offers!, ] which falls within the scope of Title III.’ ” See Ford v. Schering-Plough Corp., 145 F.3d 601, 612-13 (3d Cir.1998); see also Menkowitz v. Pottstown Memorial Medical Center, 154 F.3d 113, 122 (3d Cir.1998) (“We look for ... some nexus between the services or privileges denied and the physical place of the ... public accommodation.”).
Assuming that the service of reporting á score is bundled with the service of offering the examination and thus has the requisite direct nexus to a public accommodation, we do not believe that Doe has demonstrated a reasonable likelihood of showing that this service has been provided to him in a manner that is discriminatory or unequal under the terms of Title III.7 The District Court held that Doe had demonstrated a reasonable likelihood of successfully showing that flagging violates the general prohibition on discrimination in section 302, on the theory that the NBME provided Doe a service unequal to the service offered to other test takers. We reject this conclusion for the same reason that we rejected Doe’s argument under section 309: Doe has not demonstrated that flagging his score makes the service that the NBME provided to him substantively unequal to the service it provides to other examinees. Like other examinees, Doe took the exam and received a score. Doe has not demonstrated that his score is comparable to the scores of candidates who take the exam under standard conditions and thus that flagging his score imposes an inequality on him.8
*158Doe’s final argument under section 302 is that identifying him as a disabled person violates the general prohibition on discrimination in section 302 because it facilitates discrimination against him by third parties, namely, residency and internship programs. There are several difficulties with this argument. First, there is no provision of Title III that explicitly requires confidentiality from providers of public accommodation. By way of contrast, Title I of the ADA, regarding disabilities and the employment relationship, does require employers to protect the confidentiality of their employees with disabilities, with certain specific exceptions. See 42 U.S.C. §§ 12112(d)(3)(B), (4)(C). Second, if residency and internship programs were to discriminate against Doe as a result of his disability, such discrimination would not necessarily be attributable to the NBME. Finally, as noted above, Doe has not established that •he is likely to suffer discrimination at the hands of residency and internship programs as a result of an annotation to his scores.
IV. Conclusion
For the foregoing reasons, the order of the District Court granting a preliminary injunction will be vacated. The parties shall bear their own costs.