Opinion for the Court by Circuit Judge ROGERS.
Dissenting opinion by Circuit Judge KAVANAUGH.
ROGERS, Circuit Judge:The National Federation of Federal Employees (“the Union”) challenges the constitutionality of a random drug testing policy applicable to all employees working at Job Corps Civilian Conservation Centers operated by the U.S. Forest Service. The district court granted summary judgment in favor of the Secretary of Agriculture and the Chief of the U.S. Forest Service (hereinafter “the Secretary”) and denied the Union’s request for a preliminary injunction. Upon de novo review, we conclude that the Secretary has failed to demonstrate “special needs” rendering the Fourth Amendment requirement of individualized suspicion impractical in the context of Job Corps employment. See Vernonia School Dist 47J v. Acton, 515 U.S. 646, 653, 115 S.Ct. 2386, 132 L.Ed.2d 564 (1995); Nat’l Treasury Emps. Union v. Von Raab, 489 U.S. 656, 665-66, 109 S.Ct. 1384, 103 L.Ed.2d 685 (1989). Although identifying governmental interests in the *486students’ abstention from drug use and in their physical safety, the Secretary offered no foundation for concluding there is a serious drug problem among staff that threatens these interests and thus renders the requirement for individualized suspicion impractical. Rather, the Secretary’s evidence to date suggests the contrary. Because the Secretary has offered a solution in search of a problem, the designation of all Forest Service Job Corps Center employees for random drug testing does not fit within the “closely guarded category of constitutionally permissible suspicionless searches,” Chandler v. Miller, 520 U.S. 305, 309, 117 S.Ct. 1295, 137 L.Ed.2d 513 (1997). Accordingly, we reverse and remand the case for proceedings consistent with this opinion.
I.
The Department of Labor (“DOL”) administers the Job Corps program at approximately 124 residential and . nonresidential centers across the United States. See 29 U.S.C. § 2887(a), (b) (2006). These centers include twenty-eight Job Corps Civilian Conservation Centers operated by the Forest Service, a unit within the Department of Agriculture (“USDA”). See id. § 2887(c)(1); 36 C.F.R. § 200.1(a) (2012); 7 C.F.R. pt. 15, subpt. A, app. (2012). As described by Larry J. Dawson, the National Director of the Forest Service Job Corps program, these Centers offer, in addition to education, vocational training and counseling, “programs of work-based learning to conserve, develop, and manage public natural resources and public recreational areas or to develop community projects in the public interest,” Decl. Larry J. Dawson ¶ 3, Nov. 5, 2010, and are located generally in “remote, rural areas,” id. ¶ 5; see 29 U.S.C. § 2887(c)(1).
All twenty-eight Forest Service Job Corps Centers are residential. Students, ages sixteen to twenty-four, live and work at the Centers except during winter and summer breaks, although some vocational training and other activities occur off site; they are prohibited from keeping personal vehicles on site. When they first enroll, students are advised of the Job Corps Zero Tolerance Policy, 29 U.S.C. § 2892(b)(2)(C)(ii) (enacted in 1998), and if they fail an initial drug test, they are placed in a special training program and must take another drug test within forty-five days; a second positive test for drug use results in the student’s expulsion from the Job Corps. Students remain subject to suspicion-based drug testing while in the program. Any Center employee can report suspicion of student drug use, and residential staff periodically search for illegal drugs and alcohol in student residential areas and in students’ luggage upon their return from winter and summer breaks. Canine units assist in these searches at some Job Corps Centers.
Prospective and incumbent Job Corps Center employees must also undergo screening. For positions “supervising] young people,” all prospective employees are subject to a “Child Care National Agency Check with Inquiries: Non Sensitive/Low Risk.” For certain positions, including directors and certain specialists, prospective employees must also undergo a “Moderate Risk Background Investigation: Moderate Risk/Publie Trust.” Drug related offenses discovered during these background checks inform suitability determinations by hiring officials. Once employed in the Job Corps, all employees are responsible for “modeling, mentoring, and monitoring” appropriate workplace behavior under DOL policy. Suppl. Decl. Larry J. Dawson ¶ 3, Jan. 27, 2011. Employees at Forest Service Job Corps Centers are also subject to reinvestigation approxi*487mately every fifteen years. See Dawson Suppl. Decl. 115.
In 1988, the USDA developed a “Plan for a Drug Free Workplace,” which called for drug testing on the basis of reasonable suspicion and of new employees in certain designated job positions; of Job Corps Center positions, only nursing occupations were designated.1 (Employees required to hold commercial driver’s licenses, such as residential staff at Job Corps Centers, were subject to random testing pursuant to Department of Transportation regulations.) Drug testing was to be conducted in accordance with guidelines promulgated by the Department of Health and Human Services (“HHS”). See Executive Order No. 12,564 § 4(d), 51 Fed. Reg. 32,889, 32,891 (Sept. 15, 1986), reprinted in 5 U.S.C. § 7301 note.2 Following a 1995 investigation by a U.S. Senate Committee that identified a drug problem among Job Corps students, see S.Rep. No. 104-118 (1995), the DOL established the Job Corps Zero Tolerance Policy and instructed that “[a]ll staff will be held accountable for actively supporting and implementing the Job Corps Zero Tolerance policy” and “must be held to the same standards of conduct described in this policy for students.” Deck Gerald A. Nagel, Drug Free Workplace Program Manager, USDA, ¶ 18, Nov. 5, 2010 (quoting 1995 DOL Job Corps Instruction No. 94-21, “Implementation of Expanded Zero Tolerance for Violence and Drugs Policy”). The DOL did not designate Job Corps employees for random drug testing. The USDA, however, in 1996 designated all Forest Service Job Corps staff positions for random drug testing. The Union, representing Forest Service Job Corps Center employees, objected to the new designation, and the new policy was not implemented. In 2003, the USDA again designated Forest Service “Job Corps Center staff’ for random testing,3 but as before the policy was not *488implemented. During collective bargaining negotiations for an agreement entered into on May 27, 2010, however, the Forest Service informed the Union that all Job Corps Center staff would be subject to the random testing program. See Nagel Decl. ¶ 19. By letter of August 30, 2010, the National Director instructed Forest Service Job Corps Center directors to “come into compliance” with the random testing policy, noting that only nurses and employees required to hold a commercial driver’s license were in compliance.4 Mem. from Larry J. Dawson to Forest Service Job Corps Center Directors 1 (Aug. 30, 2010).
On October 13, 2010, the Union sued the Secretary, seeking a declaratory judgment that the random testing policy covering all Forest Service Job Corps Center employees violates the Fourth Amendment and an order enjoining the policy’s implementation. The Union also moved for a preliminary injunction, attaching various sworn declarations, including the declaration of Larry E. King, vice president of the Union’s Forest Service Council and a Job Corps employee since 1983, stating that neither the USDA nor the Forest Service had made any showing that random drug testing of staff was necessary for the safe operation of the Centers. Decl. Larry E. King ¶¶ 1, 3, 11-12, 17, Oct. 13, 2010. The district court granted the Secretary’s motion for summary judgment, concluding that the Secretary’s interests in preventing illegal drug use at the Job Corps Centers by both students and staff justified the intrusion on the employees’ privacy interests and Fourth Amendment rights, and it denied the Union’s request for an injunction. Nat’l Fed’n of Fed. Emps.-IAM v. Vilsack, 775 F.Supp.2d 91, 113-14 (D.D.C.2011).
The Union appeals. Our review of the grant of summary judgment is de novo, see, e.g., Moore v. Hartman, 571 F.3d 62, 66 (D.C.Cir.2009), and we must draw all justifiable inferences in favor of the non-moving party, see Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). Our review of the denial of injunctive relief is for abuse of discretion, although it remains de novo for underlying conclusions of law. See Chaplaincy of Full Gospel Churches v. England, 454 F.3d 290, 297 (D.C.Cir.2006).
II.
The Fourth Amendment to the Constitution prohibits the government from violating “[t]he right of the people to be secure in their persons ... against unreasonable searches and seizures.” U.S. Const. amend. IV. Drug testing of federal employees is a search subject to the Fourth Amendment reasonableness requirement. See Von Raab, 489 U.S. at 665, 109 S.Ct. 1384. “[A]s a general matter, warrantless searches are per se unrea*489sonable under the Fourth Amendment.” City of Ontario v. Quon, — U.S. -, 130 S.Ct. 2619, 2630, 177 L.Ed.2d 216 (2010) (quoting Katz v. United States, 389 U.S. 347, 357, 88 S.Ct. 507, 19 L.Ed.2d 576 (1967)) (internal quotation marks omitted). Among the “few specifically established and well-delineated exceptions to that general rule,” id., is an exception for circumstances in which “special needs, beyond the normal need for law enforcement, make the warrant and probable-cause requirement impracticable,” Vernonia, 515 U.S. at 653, 115 S.Ct. 2386 (quoting Griffin v. Wisconsin, 483 U.S. 868, 873, 107 S.Ct. 3164, 97 L.Ed.2d 709 (1987)) (internal quotation marks omitted). Even where the government claims “special needs,” a war-rantless search is generally unreasonable unless based on “some quantum of individualized suspicion.” Skinner v. Ry. Labor Execs.’ Ass’n, 489 U.S. 602, 624, 109 S.Ct. 1402, 103 L.Ed.2d 639 (1989). “[A] search may be reasonable despite the absence of such suspicion,” however, “where the privacy interests implicated by the search are minimal, and where an important governmental interest furthered by the intrusion would be placed in jeopardy by a requirement of individualized suspicion.” Id. Accordingly, where the government invokes “special governmental needs, beyond the normal need for law enforcement,” a court must “balance the individual’s privacy expectations against the government’s interests to determine whether it is impractical to require a warrant or some level of individualized suspicion in the particular context.” Von Raab, 489 U.S. at 665-66, 109 S.Ct. 1384. In conducting this balancing test, a court “must undertake a context-specific inquiry, examining closely the competing private and public interests advanced by the parties.” Chandler, 520 U.S. at 314, 117 S.Ct. 1295; see Von Raab, 489 U.S. at 665-66, 109 S.Ct. 1384.
The Union does not dispute that the need asserted by the Secretary — to protect Forest Service Job Corps Center students from harm caused by the use of illegal drugs by Center employees — lies beyond the normal need for law enforcement. It also does not challenge the constitutionality of random testing of Forest Service Job Corps Center nursing occupations and employees required to hold a commercial driver’s license, which include residential and recreational employees. Instead, the Union contends that the district court improperly weighed and balanced the relevant interests in upholding the random testing of all Forest Service Job Corps Center employees regardless of the requirements or responsibilities of their particular positions.
The Secretary maintains that for the Union to prevail in such a facial challenge it must show that no set of circumstances exist under which the policy would be valid, invoking the standard enunciated in Reno v. Flores, 507 U.S. 292, 301, 113 S.Ct. 1439, 123 L.Ed.2d 1 (1993), and citing Skinner, 489 U.S. at 632 n. 10, 109 S.Ct. 1402, as applying this standard to Fourth Amendment challenges to drug testing policies. When assessing the reasonableness of the Fourth Amendment intrusion by such policies, however, the Supreme Court has differentiated between job categories designated for testing, rather than conducting the balancing test more broadly as the Secretary appears to suggest. See, e.g., Von Raab, 489 U.S. at 677-78, 109 S.Ct. 1384. So has this court. See Harmon v. Thornburgh, 878 F.2d 484, 492-93 (D.C.Cir.1989); Nat’l Fed’n of Fed. Emps. v. Cheney, 884 F.2d 603, 611-12 (D.C.Cir.1989).5 Furthermore, to the extent the *490Secretary maintains that the court should defer to the USDA with regard to “the nature of the Job Corps program, the extent of any drug problem faced by the program, and how the needs of the program would be served and furthered by the challenged random drug testing,” Ap-pellees’ Br. 21, the Secretary relies on cases presenting familiar principles of judicial deference to reasonable interpretations and findings under statutes the department administers. See Menkes v. Dep’t of Homeland Sec., 637 F.3d 319, 329 (D.C.Cir.2011); Public Citizen, Inc. v. FAA, 988 F.2d 186, 196-97 (D.C.Cir.1993). Deference is never blind, in any event, see Am. Fed’n Gov’t Emps. v. FLRA, 778 F.2d 850, 864 (D.C.Cir.1985), and the Secretary bears a burden to establish entitlement to summary judgment, see, e.g., Celotex Corp. v. Catrett, 477 U.S. 317, 323, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986). More particularly, the Supreme Court has made clear that the constitutional question is distinct from policy questions involving otherwise constitutional administrative judgments about how best to operate a program. See, e.g., Von Raab, 489 U.S. at 665, 109 S.Ct. 1384; cf. Chandler, 520 U.S. at 317-18, 117 S.Ct. 1295.
A.
The balancing test set forth in Skinner and Von Raab requires the court, in assessing employees’ privacy interests, to determine both “the scope of the legitimate expectation of privacy at issue” and “the character of the intrusion that is complained of.” Vernonia, 515 U.S. at 658, 115 S.Ct. 2386. The court must then consider the nature of the government interests to be furthered by the drug testing policy as well as the immediacy of the government’s concerns regarding those interests and the efficacy of the policy in addressing those concerns. See id. at 660, 115 S.Ct. 2386. Finally, the court must balance the employees’ privacy interests against the government’s interests “to determine whether it is impractical to require a warrant or some level of individualized suspicion in the particular context.” Von Raab, 489 U.S. at 665-66, 109 S.Ct. 1384. Thus, even where the government asserts important interests, it must still demonstrate an immediate threat to those interests that could not practically be addressed through a suspicion-based approach in order to justify a suspicionless search under the Fourth Amendment. See Vernonia, 515 U.S. at 662-63, 115 S.Ct. 2386.
A substantial body of precedent elucidates the relevant considerations. The Supreme Court has found “compelling,” in view of documented problems, the governmental interest in ensuring public safety in railroad travel, Skinner, 489 U.S. at 620-21, 628, 634, 109 S.Ct. 1402, the “national interest in self-protection” against the importation of illegal drugs, and the public interest in preventing the promotion of potentially judgment-impaired employees to “positions where they may need to employ deadly force,” Von Raab, 489 U.S. at 670-71, 109 S.Ct. 1384. Central to these determinations were the magnitude and immediacy of the threats — the concern that “even a momentary lapse of attention [could] have disastrous consequences” for human lives and property. Von Raab, 489 U.S. at 670, 109 S.Ct. 1384; see Skinner, 489 U.S. at 628, 109 S.Ct. 1402.
Similarly, this court has upheld the random drug testing of employees in “safety-sensitive” positions, such as those responsible for maintaining and operating trains, see BNSF Ry. Co. v. Dep’t of Transp., 566 F.3d 200, 206 (D.C.Cir.2009), airplanes, see *491Aeronautical Repair Station Ass’n, Inc. v. FAA 494 F.3d 161, 174 (D.C.Cir.2007); Am. Fed’n Gov’t Emps. v. Skinner, 885 F.2d 884, 892 (D.C.Cir.1989) (“AFGE”); Cheney, 884 F.2d at 610, and motor vehicles, see Nat’l Treasury Emps. Union v. Yeutter, 918 F.2d 968, 971-72 (D.C.Cir.1990); AFGE, 885 F.2d at 892, as well as those required to carry firearms in the performance of their duties, see Cheney, 884 F.2d at 612. In the context of hazardous material inspection, this court concluded that the government’s efforts to ensure that employees “whose exclusive assigned duties are [] intimately related to the prevention of public harm [ ] are eertifiably drug-free,” even by means of random drug testing, were “a reasonable precaution against the occurrence of the feared harm.” AFGE, 885 F.2d at 891 (emphasis added). Also, in Stigile v. Clinton, 110 F.3d 801 (D.C.Cir.1997), the court upheld a policy, based on the government’s interest in ensuring protection of the President and Vice President of the United States within the White House security perimeter, authorizing the random drug testing of employees who worked in the Old Executive Office Building, which is located adjacent to the White House. Although the harm that the government was seeking to prevent was unrelated to the performance of the duties of the economists for the Office of Management and Budget, the court explained that the relevant nexus “is that between the risk posed by a drug-using employee and the evil sought to be prevented by the testing.” Id. at 805.
Beyond public safety and national security interests, the Supreme Court has also concluded that the public interest in deterring drug use by public schoolchildren is “important enough,” given the risks to their health and safety as well as the disruptive effect on the educational process as a whole, as weighed against the significantly diminished expectations of privacy enjoyed by schoolchildren. Vernonia, 515 U.S. at 661-62, 664-65, 115 S.Ct. 2386; see Board of Educ. v. Earls, 536 U.S. 822, 834, 122 S.Ct. 2559, 153 L.Ed.2d 735 (2002). Upon recounting the demonstrated problem of drug and alcohol use by student athletes who were the “leaders” of an aggressive local “drug culture,” which was responsible for disciplinary actions reaching “epidemic proportions” and which the school district had been unable to control by other means, the Court in Vemonia upheld a policy, approved by the students’ parents, requiring random drug testing of the student athletes. 515 U.S. at 649-59, 664-65, 115 S.Ct. 2386. In Earls, 536 U.S. at 830, 122 S.Ct. 2559, the Court extended its holding to competitive extracurricular activities generally, again focusing on the students’ significantly limited privacy interests in a public school environment and the “specific evidence of drugs use” by students at the schools, id. at 834, 122 S.Ct. 2559, including drugs found near school facilities and in a student’s car, id. at 835, 122 S.Ct. 2559.
On the other hand, the Supreme Court has instructed, the merely “symbolic,” and thus insufficiently important, interest in detecting and deterring drug use by candidates for public office, who “typically do not perform high-risk, safety-sensitive tasks” and do not aid drug interdiction efforts, did not warrant intrusion on their Fourth Amendment rights. Chandler, 520 U.S. at 321-22, 117 S.Ct. 1295. “Indeed,” the Court explained, “if a need of the ‘set a good example’ genre were sufficient to overwhelm a Fourth Amendment objection, then the care [that] Court took to explain why the needs in Skinner, Von Raab, and Vemonia ranked as ‘special’ wasted many words in entirely unnecessary, perhaps even misleading, elaborations.” Id. at 322, 117 S.Ct. 1295. Similarly, this court concluded that the *492governmental concern in the general “integrity of its workforce” was insufficiently important to warrant random drug testing encompassing federal prosecutors who were not specifically “responsible for the enforcement of federal narcotics laws.” Harmon, 878 F.2d at 490-91; see also Yeutter, 918 F.2d at 974. But in the context of “critical jobs”6 in the United States Army, this court in Cheney, 884 F.2d at 614, upheld random testing of drug counselors because the government had a legitimate interest in “ensuring that its employees are allegiant to their essential mission” when their “successful performance of assigned duties may reasonably be viewed as depending on their abstinence from illicit drug use.”
In short, where the government asserts “special needs” for intruding on Fourth Amendment rights, as here, the specific context matters. In demonstrating that the governmental interests are “important enough to justify the particular search at hand, in light of other factors that show the search to be relatively intrusive upon a genuine expectation of privacy,” Vernonia, 515 U.S. at 661, 115 S.Ct. 2386, the Secretary must provide a foundation for his determination that the requirement of individualized suspicion is impractical in the Forest Service Job Corps Center context, see id. at 653, 63-64, 115 S.Ct. 2386; see also Chandler, 520 U.S. at 320, 117 S.Ct. 1295.
1. Government employees “have a serious and legitimate privacy interest in not being subject to urinalysis.” Stigile, 110 F.3d at 804; see Von Raab, 489 U.S. at 671, 109 S.Ct. 1384. Certain factors, however, may diminish their reasonable expectations of privacy at work. Upon consideration of those factors, we conclude the Forest Service Job Corps Center’ employee’s privacy interests at issue remain robust.
The “ ‘operational realities of the workplace’ may render entirely reasonable certain work-related intrusions by supervisors and co-workers that might be viewed as unreasonable in other contexts.” Von Raab, 489 U.S. at 671, 109 S.Ct. 1384 (quoting O’Connor v. Ortega, 480 U.S. 709, 717, 107 S.Ct. 1492, 94 L.Ed.2d 714 (1987) (plurality opinion)). Acknowledging that “these operational realities will rarely affect an employee’s expectations of privacy in the workplace with respect to searches of his person,” the Supreme Court observed that “certain forms of public employment may diminish privacy expectations”; “[ejmployees of the United States Mint, for example, should expect to be subject to certain routine personal searches when they leave the workplace every day. Similarly, those who join our military or intelligence services may ... also expect intrusive inquiries into their physical fitness for those special positions.” Id. (internal citations omitted). This court applied that understanding in Cheney, 884 F.2d at 613. So too, the Supreme Court suggested, the successful performance of certain employees’ duties — those involved in the interdiction of illegal drugs or required to carry firearms — may “uniquely” depend on particular attributes of “judgment and dexterity” such that these employees “reasonably should expect effective inquiry into their fitness and probity.” Von Raab, 489 U.S. at 672, 109 S.Ct. 1384. Importantly, however, such employees are “[ujnlike most private citizens or government employees in general,” id., and such *493operational realities are not characteristic of government employment, see O’Connor, 480 U.S. at 717, 107 S.Ct. 1492.
The Secretary characterizes as “operational realities” the asserted facts that Forest Service Job Corps employees “work with at-risk youth in residential settings” that are “often quite remote,” and that these employees are responsible for maintaining the Zero Tolerance Policy, ensuring the students’ safety, and driving students in emergency and other situations. Appellees’ Br. 32-33. But the Secretary offers no explanation of how these general program features and loosely ascribed staff responsibilities serve to undermine the reasonable expectations of privacy held by Job Corps employees not previously subject to random drug testing. Furthermore, this characterization consists of contested facts, for the Union proffered evidence not only that there was no staff drug problem necessitating random testing, but also that different job categories at the Job Corps Centers have different levels of responsibility, or none, for maintaining the Zero Tolerance Policy, ensuring student safety, and driving students in emergency or other situations. Job descriptions for many positions in Forest Service Job Corps Centers’ administrative staff contain no mention of these responsibilities. And employee declarations indicate that purchasing agents, for example, bear no responsibility for maintaining the Zero Tolerance Policy, have never performed CPR on students, cannot recall providing first aid to students, and have rarely if ever driven students or staff. Deck of Lance A. Hamann ¶¶ 5-8, Oct. 12, 2010; Deck Jerry D. Case ¶¶ 4-7, Oct. 12, 2010. Additionally, the National Director acknowledged that not all Forest Service Job Corps Center employees are required to have a driver’s license. See Dawson Deck ¶ 14. Viewing the evidence in the light most favorable to the Union, as we must, it is not clear that the Secretary’s description of the “nature and context” of Forest Service Job Corps employment accurately portrays the nature and context of all job categories, and thus it is not clear that the attendant “operational realities” of the workplace — undeveloped in the record — serve to diminish employees’ reasonable expectations of privacy regardless of their positions.
The Secretary suggests that the privacy interests of Job Corps Center employees are diminished because they were on notice of the USDA’s intention to subject them to random drug testing as early as 1996 and the Union was informed of the decision to bring the Forest Service Job Corps “into compliance” during the most recent collective bargaining negotiations.7 “[A]n applicant’s knowledge of what will be required, and when, affects the strength of his or her [privacy] interest.” Willner v. Thornburgh, 928 F.2d 1185, 1190 (D.C.Cir.1991); see Harmon, 878 F.2d at 489 & n. 6. But the USDA’s failure to implement the random drug testing policy for all Job Corps employees for more than a decade weakens the import of the 1996 notice. Furthermore, unlike in Von Raab, see 489 U.S. at 672 n. 2, 109 S.Ct. 1384; see also Willner, 928 F.2d at 1190, here the random drug testing policy applies not only to applicants for certain positions or promotions, but also to incumbent employees. *494A Job Corps employee “may decline to be tested only if she is willing to relinquish a job she already holds.” Harmon, 878 F.2d at 489; see Aeronautical Repair Station Ass’n, 494 F.3d at 174; USDA Departmental Regulation No. 4430-792-2, § 6(f).
Of course employees’ expectations of privacy may be “lessened” where “they occupy positions that require stringent background checks.” Stigile, 110 F.3d at 804. For most Forest Service Job Corps positions, applicants’ background checks include inquiries into their residential, educational, employment, and military histories, as well as any illegal drug use within the past year.8 For positions involving the supervision of young people, the background check includes investigation of applicants’ federal and state criminal histories, see 42 U.S.C. § 13041(a)-(b) (2006); Fed. R. Evid. 201(b)(2), (c)(1); these investigations are typically completed within two months after the employee’s entry on duty, see Dawson Suppl. Deck ¶ 5. Forest Service Job Corps Center employees are typically reinvestigated every fifteen years. See id. In the National Director’s opinion, the background checks of prospective Forest Service Job Corps employees are “more rigorous” than those for “most Forest Service employees,” Dawson Deck ¶ 12, suggesting that a prospective employees’ reasonable expectations of privacy may be diminished somewhat. Cf. Cheney, 884 F.2d at 615 & n. 10. Even these background checks, however, are less “stringent” than those required of Old Executive Office Building professionals with passes allowing access to areas frequented by the President and Vice President of the United States, see Stigile, 110 F.3d at 807 n. 2 (Rogers, J., concurring), or Justice Department attorney applicants, Willner, 928 F.2d at 1190-91, or members of the Army holding key positions, see Cheney, 884 F.2d at 612-13, and they are significantly less stringent than the secret and top secret national security clearance investigations required of other federal agency employees, see Hartness v. Bush, 919 F.2d 170, 173 (D.C.Cir.1990); Harmon, 878 F.2d at 492.
The record thus suggests with regard to the job positions at issue that the employees’ reasonable expectations of privacy are somewhat diminished by the pre- and post-employment background checks they must undergo, and perhaps, although to a far lesser extent, by decade-old notice of their possible inclusion in the random drug testing program. But given the relatively limited scope of their background checks and the incumbent status of the employees now subjected to random drug testing, these employees’ reasonable expectations of privacy remain more robust than the expectations of federal employees in many other positions examined by the Supreme Court and this court. Although courts have viewed the HHS Guidelines, see supra note 2, as “significantly minimizing]” the intrusion upon privacy occasioned by urinalysis, Nat’l Treasury Emps. Union v. U.S. Customs Serv., 27 F.3d 623, 629 (1994); see Von Raab, 489 U.S. at 672 n. 2, 109 S.Ct. 1384, they do not render minimal the overall intrusion on Job Corps employees’ privacy interests occasioned by the random drug testing policy. Unlike in Von Raab, where drug testing occurred only upon application for a particular position, see Von Raab, 489 U.S. at 661, 109 S.Ct. 1384; see also Willner, 928 F.2d at 1189-90, the testing pursuant to the USDA policy is random and can occur an unlimited number of times. See Nagel Deck ¶ 9. *495Even assuming this factor “ ‘would tip the scales’ only ‘in a particularly close case,’ ” U.S. Customs, 27 F.3d at 629 (quoting Harmon, 878 F.2d at 489), “[r]andom drug testing represents a greater threat to an employee’s privacy interest than does mandatory testing because of the ‘unsettling show of authority that may be associated with unexpected intrusions on privacy,’ ” id. at 629 (quoting Von Raab, 489 U.S. at 672 n. 2, 109 S.Ct. 1384).
2. The Secretary states that the drug testing policy serves two important governmental interests at the Forest Service Job Corps Centers: maintaining the Zero Tolerance Policy among students and ensuring student safety. Drug use by Job Corps employees, the Secretary reasons, could threaten to undermine the Zero Tolerance Policy because such use “could become known” to students, and because drug using employees might be “less likely to report knowledge or suspicion of a student’s drug use and could even serve as a conduit for drugs in these remote settings.” Appellees’ Br. at 37-38. Further, the Secretary states, drug use by Job Corps employees “potentially threatens the physical safety of every student at these remote sites, ... because a drug using employee is necessarily impaired in his or her ability to function in emergencies.” Id. at 38.
Although precedent regarding employees in “safety-sensitive” positions supports the importance of the Secretary’s interest in securing the Job Corps students’ safety, see, e.g., Von Raab, 489 U.S. at 670, 109 S.Ct. 1384; AFGE, 885 F.2d at 891, support for the Secretary’s interest in main-taming the Zero Tolerance Policy by drug testing of all Job Corps Center employees is more attenuated. This interest rests upon a connection the Secretary effectively seeks to forge between the concerns recognized in Vemonia and Cheney. See Appellees’ Br. 37-39; Mem. in Supp. Defs.’ Mot. Summ. J. 14, 18-19. In Vernonia, 515 U.S. at 661-62, 115 S.Ct. 2386, the Supreme Court noted the risks of “physical, psychological, and addictive effects” associated with adolescent drug use, and risks of immediate physical harm faced by student athletes during games or practice. In Earls, 536 U.S. at 829-30, 122 S.Ct. 2559, the Court reemphasized the students’ diminished privacy interests and the schools’ custodial and tutelary responsibilities. In view of Vemonia and Earls, the Secretary’s interest in preventing drug use among at-risk youth in a government program designed to expand their opportunities may be “important enough” to justify certain searches.9 Yet here, unlike in Vernonia and Earls, the Secretary seeks to justify the random drug testing not of Job Corps students, but of an expanded group of employees at the Forest Service Job Corps Centers. The governmental interest in the detection and deterrence of drug use by such employees is thus at a remove from that previously addressed by the Supreme Court. Indeed, in Vemonia, the Court emphasized that “[c]entral” to its decision was “the fact that the subjects of the Policy [we]re (1) children, who (2) ha[d] been committed to the temporary custody of the State as schoolmaster.” Vernonia, 515 U.S. at 654, 115 S.Ct. 2386; see id. at 665, 115 S.Ct. 2386.
*496In an effort to surmount this key distinction, the Secretary has argued that the success of the Zero Tolerance Policy in deterring drug use by Job Corps students depends on its enforcement by Job Corps employees; should these employees be “unsympathetic” to this mission “because of their own drug use,” Cheney, 884 F.2d at 614, the aims of the policy would be thwarted. See Mem. in Supp. Defs.’ Mot. Summ. J. 14. Yet notwithstanding the important governmental interest identified in Vemonia and Earls, the Supreme Court did not imply that protection of this interest would justify random drug testing of the teachers and other staff at the schools — to the contrary, it “caution[ed] against the assumption that suspicionless drug testing will readily pass constitutional muster in other contexts.” Vernonia, 515 U.S. at 665, 115 S.Ct. 2386. The extent to which the Secretary’s interest in deterring drug use among Job Corps students through the Zero Tolerance Policy justifies random drug testing of all staff hinges on whether the Secretary has laid a foundation for concluding that drug use among Job Corps staff poses a threat to this interest. See id. at 662-63, 115 S.Ct. 2386. In determining the immediacy of the Secretary’s concerns, the court first looks for “a demonstrated problem of drug abuse” among the Job Corps employees in the job categories to be subjected to drug testing “to clarify — and to substantiate — the precise hazards posed by such use.” Chandler, 520 U.S. at 319, 117 S.Ct. 1295; see Vernonia, 515 U.S. at 662-63, 115 S.Ct. 2386. Absent such a foundation for invoking the “special needs” exception, the Secretary cannot show that an important governmental interest is placed in jeopardy and thus adherence to the requirement of individualized suspicion is impractical. See Von Raab, 489 U.S. at 665-66, 109 S.Ct. 1384; Skinner, 489 U.S. at 624, 109 S.Ct. 1402.
In support of summary judgment, the Secretary stated as undisputed facts: The Forest Service Job Corps Centers are generally located in remote areas; the Centers offer a residential program in which students live and work on-site; students are advised of and subject to the Zero Tolerance Policy; any Center employee can report suspicion of drug use by a student; “[a] number of employees who do not hold commercial driver’s licenses are called upon to transport students in the course of their responsibilities”; and some employees teach welding and electrical work. Defs.’ Mot. Summ. J. (Statement of Material Facts 1-2). Notably absent from this statement of facts and the record is any indication of a serious drug problem among Job Corps Center staff. The National Director stated in his declaration that “[d]rug use has been found among [Forest Service Job Corps] employees in the past, [that] several employees have been disciplined for drug use in recent years,” and that “[r]eview of existing Forest Service disciplinary records show[ed] that eight [ ] employees have been subject to adverse actions, with penalties ranging from [fourteen-]day suspensions to removal, in recent years.” Dawson Decl. ¶ 17. Taking the declaration at face value, the small number of incidents among a workforce of several thousand over an unspecified number of years does not establish a serious problem, much less an “immediate crisis,” as in Vernonia, 515 U.S. at 663, 115 S.Ct. 2386, necessitating expansion of the random drug testing policy, compare note 9, supra; nor does the declaration suggest that any problem has not been satisfactorily addressed in a manner consistent with the individualized suspicion requirement and student safety, see King Decl. ¶¶ 11-12.
Also absent from the record is any demonstration of difficulty in maintaining the *497Zero Tolerance Policy as a result of any staff drug use during the fourteen years between the USDA’s adoption of the Policy and the implementation of random drug testing for all Forest Service Job Corps Center employees. Cf. Chandler, 520 U.S. at 319, 117 S.Ct. 1295. The record indicates that the DOL, the primary administrator of the federal Job Corps program, see 29 U.S.C. §§ 2883a, 2887(a), has never required such drug testing. The absence of such a decision by the DOL combined with the USDA’s long-delayed action and the absence of a documented problem belie the conclusion that there is so serious a staff drug problem at the Forest Service Job Corps Centers as to present “special needs” requiring suspicionless intrusion on all employees’ Fourth Amendment rights. The National Director’s instruction that the “[Forest Service] Job Corps will come into compliance” with USDA’s drug testing program made no reference to the governmental concerns articulated here; rather, it offered the tautological explanation that “[t]o date the Forest Service Job Corps has not been in full compliance with this regulation.” Mem. from Larry J. Dawson to Forest Service Job Corps Center Directors (Aug. 30, 2010). Apparently neither the USDA nor the Forest Service cited any incidents leading to the determination that random drug testing of all Forest Service Job Corps Center employees was necessary, or any drug use or other statistics to support that determination. See King Decl. ¶¶ 11-12. The Secretary has thus offered a solution in search of a problem.
Moreover, even assuming the Secretary’s responsibility for maintaining the Zero Tolerance Policy and ensuring Job Corps students’ safety would suffice to whittle down the relaxed evidentiary standard where “special needs” are invoked, the Secretary has failed to show, as to newly designated staff, “an immediate, non-attenuated” nexus between “the risk posed by a drug-using employee and the evil sought to be prevented by the testing.” Stigile, 110 F.3d at 805; see Von Raab, 489 U.S. at 677-78, 109 S.Ct. 1384; Harmon, 878 F.2d at 492-93; Cheney, 884 F.2d at 611-12. Those employees whose “exclusive assigned duties are [ ] intimately related to” the enforcement of the Zero Tolerance Policy, AFGE, 885 F.2d at 891— residential staff — have long been subject to random drug testing. Even assuming that drug counselors in direct and regular contact with students and are thus likely “in a position to render harm,” Stigile, 110 F.3d at 805; see Von Raab, 489 U.S. at 671, 109 S.Ct. 1384, by undermining the Policy through their own example or through their lack of commitment to their counseling and enforcement tasks, see Cheney, 884 F.2d at 614, the nexus between the danger and the duties of other positions is far more attenuated. Contrary to the district court’s findings, see Natl Fed’n of Fed. Emps.-IAM, 775 F.Supp.2d at 110-11, evidence proffered by the Union reveals that certain positions have very limited contact with students and bear no responsibility for maintaining the Policy. “[I]t is not evident that those occupying these positions are likely to” be in a position to undermine the Zero Tolerance Policy, and “this apparent discrepancy raises ... the question whether the [USDA] has defined [the] category of [Job Corps Center staff] more broadly than is necessary to meet [the Secretary’s] purposes.” Von Raab, 489 U.S. at 678, 109 S.Ct. 1384.
A similar question of scope arises with regard to Job Corps employees’ responsibility for securing the students’ safety in emergencies, for the record fails to indicate that all employees are likely to be in a position to render the harms feared by the Secretary. See id.; Harmon, 878 F.2d at 492-93; Cheney, 884 F.2d at 611-12. Al*498though some staff are responsible for transporting students and required to hold commercial driver’s licenses (and are already subject to random drug testing under Department of Transportation regulations), others are not even required to hold valid driver’s licenses. The Union proffered evidence that employees in certain positions have very limited contact with students and so are rarely if ever in a position to administer CPR or first aid; these employees are thus unlikely to be “in a position to render harm.” Cf. AFGE, 885 F.2d at 892. For such categories of employees not previously subject to random drug testing, “the chain of causation between misconduct and injury is considerably more attenuated.” Harmon, 878 F.2d at 491. The lack of evidence of a serious drug problem among Forest Service Job Corps staff coupled with the speculative nature of the risk identified by the Secretary render the expanded random drug testing policy unjustified.
To the extent the Secretary maintains that he has a “legitimate interest in deterring drug use that might affect work performance, that employees who use drugs off the job risk performance-impairing addiction, that off-duty drug users may buy [or sell] drugs at work,” Appellees’ Br. 37-38, such speculation is, as the court explained in Yeutter, 918 F.2d at 974 (emphasis in original), insufficient to justify a Fourth Amendment intrusion. In Yeutter, the court rejected suspicion-based testing of off-duty drug use by USDA employees because the Secretary had not produced sufficient “evidence that might establish a relationship between off-the-job drug use and job performance.” Id. Similarly here, unlike in Stigile, 110 F.3d at 803-04, the Secretary asserts interests in preventing harms arising only from employees’ inadequate job performance. “Absent [] a ‘clear, direct nexus’ between the duties of’ all covered Forest Service Job Corps employees “and the nature of the feared harm[s], and absent any compelling reason to expect that drug use will result in misplaced sympathies for their responsibilities, testing these employees lacks the necessary causal connection between the employees’ duties and the feared harm[s].” Cheney, 884 F.2d at 614 (citation omitted).
Additionally, the efficacy of designating all Forest Service Job Corps Center positions for random testing is dubious inasmuch as, unlike the policies upheld in Vernonia and Earls, the expanded application of the USDA policy is at best an indirect means of detecting and deterring drug use by students. See Earls, 536 U.S. at 837-38, 122 S.Ct. 2559; Vernonia, 515 U.S. at 663, 115 S.Ct. 2386. In Cheney, the court took note of the Army’s assessment of the efficacy of drug screening “in light of its experience from fifteen years of testing its military personnel,” 884 F.2d at 611, yet distinguished, much as the Supreme Court did in Von Raab, 489 U.S. at 678, 109 S.Ct. 1384, between job positions in concluding that the record supported the reasonableness of random drug testing for some but not for others, notwithstanding the Army’s “serious interests,” Cheney, 884 F.2d at 611-12. Finally, the work setting of many Forest Service Job Corps Center employees is that of “traditional office environments,” Von Raab, 489 U.S. at 674, 109 S.Ct. 1384; see, e.g., Suppl. Decl. Lance A. Hamann ¶ 4, Feb. 16, 2011, where it is “feasible to subject employees and their work product to ... day-to-day scrutiny,” Von Raab, 489 U.S. at 674, 109 S.Ct. 1384, so the requirement of individualized suspicion for employees in these positions is not impractical. See Chandler, 520 U.S. at 321, 117 S.Ct. 1295.
3. Upon balancing the employees’ privacy interests against the Secretary’s interests “to determine whether it is impractical to require a warrant or some *499level of individualized suspicion in the particular context,” Von Raab, 489 U.S. at 665-66, 109 S.Ct. 1384, we conclude that the Secretary has failed to demonstrate “special needs” justifying random drug testing of all Forest Service Job Corps Center employees. Although the newly designated employees’ privacy interests may be somewhat diminished by their required background checks and the Secretary’s adherence to the HHS Guidelines in administering the tests, they are not as diminished as other privacy interests the courts have examined and remain robust. Conversely, even assuming that subjecting all Job Corps employees to random drug testing is premised on important governmental interests, the lack of a foundation for “special needs” to intrude on their Fourth Amendment rights significantly undermines these interests. “A demonstrated problem of drug abuse, while not in all cases necessary to the validity of a testing regime, would shore up an assertion of special need for a suspicionless general search program.” Chandler, 520 U.S. at 319, 117 S.Ct. 1295 (citation omitted); see Earls, 536 U.S. at 835, 122 S.Ct. 2559. The Secretary has made no such showing with regard to the newly designated positions, and his generalized assertions of need are contradicted by evidence from the National Director of the Forest Service Job Corps program; nor has the Secretary demonstrated the requisite nexus between the stated governmental interests and all Forest Service Job Corps Center staff positions.
Accordingly, because the Secretary’s designation of all Forest Service Job Corps Center employees for random drug testing under the USDA policy “does not fit within the closely guarded category of constitutionally permissible suspicionless searches,” Chandler, 520 U.S. at 309, 117 S.Ct. 1295, we reverse the grant of summary judgment and remand the case to the district court for proceedings consistent with this opinion. We also reverse the denial of the Union’s request for a preliminary injunction, because the denial was based solely on the likelihood of the Secretary’s success on the merits and the loss of constitutional protections constitutes irreparable injury. See Mills v. Dist. of Columbia, 571 F.3d 1304, 1312 (D.C.Cir.2009).
. See Executive Order No. 12,564 § 3(a), 51 Fed. Reg. 32,889, 32,890 (Sept. 15, 1986), reprinted in 5 U.S.C. § 7301 note (requiring agency heads to “establish a program to test for the use of illegal drugs by employees in sensitive positions,” as determined by "the nature of the agency’s mission and its employees’ duties ... and the danger to the public health and safety or national security that could result from the failure of an employee adequately to discharge his or her position”).
. The HHS "Mandatory Guidelines for Federal Workplace Drug Testing Programs," 73 Fed. Reg. 71,858 (Nov. 25, 2008) ("HHS Guidelines”), provide that employees to be tested report to a collection site, where they produce a urine sample within an enclosed stall without direct visual observation. Id. at 71,863. Samples may be tested only for specified drugs, id. at 71,880, and positive tests are reported only after a second test using a different method confirms the result, id. at 71,-893-94.
. USDA Departmental Regulation No. 4430-792-2 establishes the policy and procedures for the Drug-Free Federal Workplace Program and Workplace Drug and Alcohol Testing Program. It requires random testing of specified positions, including all Job Corps Center staff as follows:
Job Corps Center Staff (Includes any occupational series in which the incumbent may perform the duties described below) (Subject to applicant testing).
Each Center staff member see [sic] students every’ day, and each staff member is responsible for the safety of every student, including administering CPR and/or first aid whenever needed. Also, each staff member is required to possess a valid driver's license to transport students in cases of emergency, to and from work sites, etc. Drug usage by Center staff members could result in the loss of students’ lives or injury to the students. Also, all Center staff personnel are responsible for administering the Zero Tolerance for Drug Policy. Improper or illegal drug use is inconsistent with assisting others in becoming and remaining drug-free.
USDA Departmental Regulation No. 4430-792-2, app. A, § 14 (Aug. 25, 2003). Job Corps nursing occupations are also subject to random testing. See id. § 16.
. Forest Service Job Corps Center employees fall roughly into five categories. (1) Administrative staff includes file clerks, automation clerks, and computer assistants; nurses and medical records technicians; supply technicians and purchasing agents; and cooks. (2) Educational staff includes classroom teachers and driver's education teachers. (3) Counseling staff consists of both guidance counselors and drug and alcohol counselors. (4) Vocational staff members teach students certain trades, such as information technology, culinary arts, urban forestry, welding, brick masonry, carpentry, and electrical work. (5) Residential and recreational employees include social service assistants and recreation assistants, who are primarily responsible for monitoring student residential areas and transporting students as necessary, and who are subject to random drug testing by reason of being required to have commercial driver's licenses. Additionally, outside independent contractors provide various services at the Job Corps Centers, including vocational training.
. Our dissenting colleague paints with a broad brush without regard to precedent from the Supreme Court, and this court, on the *490particularity of the Fourth Amendment inquiry-
. The Department of Defense ("DoD”) Directive identified "critical jobs” as those "sufficiently critical to the DoD mission or protection of public safety that screening to detect the presence of drugs is warranted as a job-related requirement.” Cheney, 884 F.2d at 605 n. 2.
. The district court noted, see Nat’l Fed’n of Fed. Emps.-IAM, 775 F.Supp.2d at 108 & n. 9, that an agency’s decision to designate positions for drug testing falls within the management rights clause of the Federal Service Labor-Management Relations Act, 5 U.S.C. § 7106(a)(1), and conflicting proposals are therefore non-negotiable. See U.S. Dep’t of. the Interior Minerals Mgmt. Serv. v. FLRA, 969 F.2d 1158, 1162 (D.C.Cir.1992); Nat’l Ass’n Gov’t Emps., Local R14—9 Union v. U.S. Army, 30 F.L.R.A. 1083, 1086-87 (1988).
. See OPM, Questionnaire for Non-Sensitive Positions, available at http://www.opm.gov/ forms/pd£_fill/sf85 .pdf.
. According to a study commissioned by the DOL, approximately 26 percent of all entering Job Corps students tested positive on their initial drug tests in 2004 and 2005. See Dawson Decl. ¶ 7. Directors of three Forest Service Job Corps Centers reported initial drug tests showing drugs in approximately 15, 18, and 23 percent of students at their respective sites. See Decl. Linda J. Guzik ¶ 5, Jan. 26, 2011; Decl. Raymond J. Ryan ¶ 5, Jan. 26, 2011; Decl. Cynthia S. Kopack 15, Jan. 26, 2011. No comparable evidence was offered regarding staff drug use.