Wayne and Judy Acton, guardians ad litem for James Acton, appeal the district court’s judgment at trial that the Vernonia School District’s mandatory random drug testing policy for participants in interscholastic athletics (“the Policy”) does not violate James’s right to be free from unreasonable searches, under either Article I, Section 9 of the Oregon Constitution or the Fourth Amendment. We reverse.

BACKGROUND

The District runs two schools, Washington Grade School and Vernonia High School. Several teachers in the District testified that prior to 1985 drugs and alcohol were used by only a small group of students and that there were very few discipline problems in the schools. Between 1985 and 1989, however, teachers and administrators began to perceive a marked increase in disciplinary problems, student drug use, and the glorification of drug culture. Athletic coaches noticed an increase in the number and severity of injuries, which they attributed to greater drug use. They testified to a number of incidents where students had, or were suspected to have, used drugs. They personally saw some of the problems, but were told of others. Some of the involved students were athletes and others were not. Among other things, one teacher had often seen students smoking marijuana during the school day at a coffee shop across the street from the high school. An English teacher received several essays describing and glorying in scenes of student drug and alcohol use. At one wrestling meet a boy was seriously injured when he failed to perform a basic well-drilled safety maneuver. The student’s hotel room smelled of marijuana when the coach later went to cheek on his condition. Parents reported that they had heard of drug use on a football trip, and students confronted by the principal admitted that they had used marijuana. Students even formed rowdy groups. One group called itself the “Big Elk.” It was composed largely of athletes. Another group called itself the “Drug Cartel.”

By 1989, Mr. Aultman, the principal of the grade school, and his faculty felt that they had to do something to combat the growing drug problem. They had attempted drug education programs and even a drug-sniffing dog, but they noticed no decrease in disciplinary problems or in drug use. They decided to institute the Policy.

The Policy was unanimously adopted by the Vernonia School Board in the fall of 1989. By its terms, all students who want to participate in interscholastic athletics are required to sign a form authorizing the District to perform a drug test on a urine sample provided by the student. All interscholastic student-athletes are tested at the beginning of each athletic season in which they compete. During the season, student athletes are tested at random on a weekly basis.

Boys and girls are tested in different areas of the school. Girls go to the office of Ms. King, the director of girls’ athletics, and boys *1517 go to the boys’ locker room. Each student fills out part of a specimen control form, which assigns a number to the specimen. The faculty monitor then gives the student a testing packet, which contains a cup and a vial. Boys go to the urinals in the boys’ locker room bathroom and produce a specimen in the cup. A male faculty member, either Mr. Aultman or Mr. Svenson, a coach, is seated on a bench some 12 to 15 feet away while the boy urinates. The student remains fully clothed with his back to the monitor. At no time does the monitor have a view of the student’s genitals. Aultman and Svenson testified that they do not always watch the student while he produces the specimen, but they generally listen for the normal sounds of urination. After the student produces the specimen, he returns the cup to the monitor who checks it for temperature and obvious signs of tampering. The monitor pours the specimen into the vial. The procedure for girls differs only slightly.

Samples are sent to Metrolab, a private company that specializes in drug testing by urinalysis. Security procedures protect the chain of possession and the identity of the student. The tests screen for creatinine, amphetamines, cocaine, marijuana and LSD. They are approximately 99.94% accurate. Test results are reported by telephone to authorized District personnel. Positive tests are mailed to the District’s superintendent.

If a student’s test returns positive, a second test is administered as soon as possible to confirm the results. The District notifies the student’s parents after the second positive test. If the second test is negative, no disciplinary action is taken. A student who returns two positive tests is given two options. The student may either participate in a drug counseling program for six weeks and submit to weekly drug tests or accept a suspension from the athletic program for the remainder of the current season and the entire following athletic season.

For the second offense, the Policy states that a student is suspended from participating in athletics for the remainder of the current season and the next athletic season, with no other option. The third offense draws a suspension for the remainder of the current season and the next two athletic seasons with no opportunity to reduce the penalty. Any student who refuses to submit to a drug test at any time is suspended from the team for the remainder of the athletic season.

James was a seventh grader at Washington Grade School during the 1991-92 academic year. In the fall, he tried out for the football team. At the first practice team members were given drug testing consent forms to sign and to have their parents sign. James brought the form home, but he and his parents decided that they did not want to sign the form. Because James refused to consent to drug testing, he was suspended from interscholastie athletics for the season. No evidence suggested that James has ever used drugs or that the District has any reason to suspect that he has.

James’s parents brought this action claiming that the Policy violated James’s right to be free from unreasonable government' searches under both the Fourth Amendment of the United States Constitution and Article I, Section 9 of the Oregon Constitution. After a trial, the district court rejected both claims. Acton v. Vernonia Sch. Dist., 796 F.Supp. 1354 (D.Or.1992) (Acton I). This appeal ensued.

JURISDICTION AND STANDARD OF REVIEW

The district court had jurisdiction over the Fourth Amendment claim under 28 U.S.C. § 1331 . It had supplemental jurisdiction over the Oregon , constitutional cause of action. 28 U.S.C. § 1367 (a). We have jurisdiction under 28 U.S.C. § 1291 .

We review the district court’s findings of fact for clear error. Fed.R.CivJP. 52(a). “A finding is ‘clearly erroneous’ when although there is evidence to support it, the reviewing court on the entire evidence is left 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). Questions of federal law are reviewed de novo, Bordallo v. Reyes, 763 F.2d 1098, 1102 (9th Cir.1985), as are questions of state law, *1518 In re McLinn, 739 F.2d 1395, 1403 (9th Cir.1984) (en banc).

DISCUSSION

The Actons assert that the District failed to prove that there actually was a drug problem. They say that there was no foundation upon which to build a drug policy. They go on to argue that even if there were a drug problem, it did not justify a random testing program. We disagree with their first contention, but agree with the second one.

Before turning to a discussion of the substantive issues, we must address the somewhat arcane question of whether we should decide this case on the basis of the Oregon Constitution or on the basis of the United States Constitution. We have held that when the state and federal constitutional provisions are “coextensive,” we can decide the federal constitutional claims because that will also decide the state constitutional claims. See Los Angeles County Bar Ass’n v. Eu, 979 F.2d 697 , 705 n. 4 (9th -Cir.1992). However, if they are not coextensive and the state constitution actually gives more protection than the federal constitution, we decide validity under the state constitution in order to avoid addressing federal constitutional claims unnecessarily. See, e.g., Ellis v. City of La Mesa, 990 F.2d 1518, 1524 (9th Cir.1993); Hewitt v. Joyner, 940 F.2d 1561, 1565 (9th Cir.1991), cert. denied, — U.S. - — , 112 S.Ct. 969 , 117 L.Ed.2d 134 (1992). Oregon presents us with a variation on these themes.

The language of Article I, Section 9 of the Oregon Constitution regarding searches and seizures is almost exactly the same as the language of the Fourth Amendment. To the extent the provisions read differently, the Oregon courts have said that the language differences are of no consequence. See State v. Flores, 280 Or. 273, 279-81 , 570 P.2d 965, 968-69 (1977). Nonetheless, Oregon insists that its constitutional provision can give more protection than the federal constitution and that it sometimes does so. See State v. Caraher, 293 Or. 741, 748-50 , 653 P.2d 942, 946-47 (1982); see also State v. Florance, 270 Or. 169 ,- 182-83, 527 P.2d 1202, 1208-09 (1974), overruled on other grounds by Caraher, 293 Or. at 748-50 , 653 P.2d at 946-47 . For example, in Nelson v. Lane County, 304 Or. 97 , 743 P.2d 692 (1987) (.Nelson II) the court found, in effect, that a roadblock stop violated the Oregon Constitution due to a lack of proper authorization, despite the fact that it would not violate the Fourth Amendment. Id. at 101-11 , 743 P.2d at 694-700 . We say “in effect” because it is not entirely clear whether the authorization requirement is a matter of general law rather than a matter of Oregon constitutional analysis. In any event, it must be' taken into account when the Oregon constitutional provision is analyzed. Furthermore, Oregon courts, or judges on those courts, have gone out of their way to insist that the federal constitutional analysis does not control Oregon search and seizure analysis. See, e.g., Nelson II, 304 Or. at 101-02 , 743 P.2d at 694 ; State v. Boyanovsky, 304 Or. 131, 137 , 743 P.2d 711, 714 (1987) (Gillette, J., specially concurring); Caraher, 293 Or. at 748-52 , 653 P.2d at 946-8 ; Nelson v. Lane County, 79 Or.App. 753, 760 , 720 P.2d 1291, 1296 (1986) (Nelson I); 1 see also State v. Ainsworth, 310 Or. 613 , 619 n. 8, 801 P.2d 749 , 752 n. 8 (1990) (aerial search questions for Oregon purposes are different from what they are for federal purposes. The result, however, appears to be the same.).

We are therefore constrained, in the first instance, to decide this case on Oregon constitutional grounds. That, however, is no mean task. Oregon law regarding random searches in general is not entirely clear. More importantly, as far as we can ascertain, the Oregon courts have never decided a random drug testing case. We are, therefore, left without a compass that clearly points us in the right direction unless we use a federal compass. We can say with absolute confidence that the Oregon Constitution will not be construed to offer less protection than the Fourth Amendment. It is highly likely that it will be found to offer more protection. We also take note of the fact that the Oregon *1519 Supreme Court itself has often decided cases by declaring that “although ... at liberty to adopt a stricter test,” it will apply Fourth Amendment analysis because that is, at least, a floor. State v. Tourtillott, 289 Or. 845, 854 , 618 P.2d 423, 427 (1980), cert. denied, 451 U.S. 972 , 101 S.Ct. 2051 , 68 L.Ed.2d 352 (1981). Having done that, the court has then gone on to state, that a particular search violates neither the Fourth Amendment nor Article I, Section 9. Id. at 859, 618 P.2d at 430; see also Flores, 280 Or. at 277-82 , 570 P.2d at 967-70 .

Therefore, in our ensuing discussion we will, to the extent possible, refer to the guideposts set out for us by the Oregon courts. But, particularly in the specific area of urine testing — the core issue in this case— we will unabashedly refer to federal law and decide the case based upon the Fourth Amendment, confident in our view that Oregon itself would do no less.

I. Evidentiary Questions

The Actons make two basic challenges to the district court’s finding that a drug problem existed. They argue that much of the evidence upon which the court relied was inadmissible hearsay. Insofar as the court relied on admissible evidence, they contend that it shows only that the faculty in the District believed that there was a drug problem, not that there actually was one.

The district court found that the Policy is limited to the one activity that is likely to have the greatest impact “on the drug and alcohol abuse problem!,]” Acton I, 796 F.Supp. at 1363 , that there was “ample evidence in the record to support the [District’s claim that alcohol and drug use were the primary causes of disruptions and disciplinary problems during school hours,” id. at 1367 , and that “coaches have observed athletes perform poorly and unsafely while under the influence of some intoxicant.” Id. at 1363 .

The underlying evidence need not be extensively reported here. We, however, have read the record. It demonstrates that both the administrators and the faculty beheld instances of drug use and glorification. They also perceived actions of athletes and others that were so far out of the norm that use of drugs was a logical inference. Besides that, they were told of incidents by others who were concerned and who had no reason to lie. Perhaps dubbing this a “problem” is a matter of perception or definition. What appears to be a problem in one place might seem to be a minor annoyance elsewhere. Suffice it to say that drug use appeared to be more extreme than it should be and even seemed to be growing. It was sufficient to give them concern for the future of the youths they had responsibility for. Based upon the evidence, we cannot say that the concern was unreasonable. Similarly, we cannot say that the district court committed clear error in finding the facts.

However, we reemphasize that what the evidence shows, and all it shows, is that there was some drug usage in the schools, that student discipline had declined, that athletes were involved, and that there was reason to believe that one athlete had suffered an injury because of drug usage and others may have. Moreover, administrators, teachers, parents, and ultimately the school board were deeply concerned and they developed the Policy in order to deal with the “problem.”

II. Constitutional Analysis

Article I, Section 9 of the Oregon Constitution provides:

As we have already said. and as can be plainly seen, this phraseology is not materially different from that of the Fourth Amendment. In Oregon, the courts ask three questions when they are considering an Article I, Section 9 search case: was there a search; did the executive official have the authority to conduct the search; and was the search reasonable? So will we.

*1520 We first turn to the question of whether the conduct of a government official constitutes a “search” in the constitutional sense. See State v. Gerrish, 311 Or. 506, 510 , 815 P.2d 1244, 1246 (1991); Ainsworth, 310 Or. at 616-17 , 801 P.2d at 750 . The Oregon Supreme Court has stated that “Article I, section 9, protects privacy and possessory interests from unreasonable governmental intrusion.” Id. at 617 , 801 P.2d at 751 ; see also. State v. Tanner, 304 Or. 312, 319 , 745 P.2d 757, 760 (1987) (“three interests [aré] protected by section 9: privacy, property, and some sort of a ... right to be protected from undignified, forceable violations of the person”) (internal quotation and citation omitted). Any time a government inquiry intrudes upon one of these rights, it is an Article I, Section 9 “search.” See State v. Campbell, 306 Or. 157, 170-72 , 759 P.2d 1040, 1047-49 (1988); State v. Owens, 302 Or. 196, 206 , 729 P.2d 524, 530 (1986).

Because the Oregon courts have not decided a urine testing case, we turn to Fourth Amendment law to decide if the collection and testing of urine is a search. Of course, it is. See Skinner v. Railway Labor Executives’ Ass’n, 489 U.S. 602, 617 , 109 S.Ct. 1402, 1413 , 103 L.Ed.2d 639 (1989). Thus, it cannot be doubted that collection and testing pursuant to the Policy was a search within the meaning of Article I, Section 9. Given the requirements of the Fourth Amendment, it could hardly be otherwise.

The next question is whether the executive officials who executed the search had authority to do so. The Oregon Supreme Court has stated that “ ‘the function of [Article I, Section 9] is to subordinate the power of executive officers ovér the people and their houses, papers, and effects to legal controls beyond the executive branch itself.’ ” AFSCME Local 2623 v. Department of Corrections, 315 Or. 74, 82 , 843 P.2d 409, 414 (1992) (quoting State v. Weist, 302 Or. 370, 376 , 730 P.2d 26, 29 (1986)). If a search is not authorized, the court’s inquiry ends. The search violates the state constitution. State v. Atkinson, 298 Or. 1, 8-11 , 688 P.2d 832, 836-38 (1984); see Nelson II, 304 Or. at 105-06, 743 P.2d at 696-97.

Constitutionally satisfactory authorization to search may come from one of four sources. First, in most cases, “[a]bsent consent, law enforcement officials must have a warrant to search_” State v. Bridewell, 306 Or. 231, 235 , 759 P.2d 1054, 1057 (1988). A warrant is always sufficient authorization to search. Next, a warrantless search is permissible if it falls within one of the “specifically established and carefully delineated exceptions to the warrant requirement,” that is, consent, search incident to arrest, plain view, exigent circumstances, inventory search, and “stop and frisk.” Id. Third, authorization by a legislative enactment that defines and limits official authority satisfies Article I, Section 9 in some cases. AFSCME, 315 Or. at 82-83 , 843 P.2d at 414 . Finally, even if a legislative grant of authority does not sufficiently limit the scope of executive discretion, the executive itself may provide limits by establishing a systematically administered program pursuant to its statutory authority. Id.

Oregon courts analyze “criminal” searches and “administrative” searches differently. 2 Criminal searches are only permissible under Article I, Section 9 if they conform with the warrant clause or one of its recognized exceptions. Nelson II, 304 Or. at 104, 743 P.2d at 696. By contrast, the warrant clause and its exceptions do not provide the exclusive authorizations for an administrative search. Oregon courts have also upheld administrative searches against Article I, Section 9 challenge, where they have been authorized by legislative enactments or administered according to executive programs that sufficiently limit official discretion. *1521 AFSCME, 315 Or. at 82-83 , 843 P.2d at 414 ; see State v. Boyanovsky, 304 Or. 131, 133-34 , 743 P.2d 711, 712 (1987). The district court found that the Policy was properly authorized by a politically accountable body when it was adopted by a vote of the Board. Ac ton I, 796 F.Supp. at 1367 . That finding is not challenged on appeal.

The third question is whether the search was “reasonable.” Here again the analysis proceeds differently for criminal and administrative searches. Criminal searches are reasonable only if authorized by a warrant or an exception to the warrant requirement. No Oregon court has upheld a criminal search without finding that the executive official had a reasonable individualized suspicion that evidence of wrongdoing would be found on the person or at the place being searched. Nelson II, 304 Or. at 101, 743 P.2d at 694 (“searches for evidence to be used in a criminal prosecution, conducted without a warrant or suspicion of wrongdoing violate Article I, section 9”); see also Boyanovsky, 304 Or. at 133-34 , 743 P.2d at 712 (1987).

Administrative searches conducted pursuant to a warrant or an exception to the warrant requirement are also reasonable. An administrative search satisfies the reasonableness inquiry if its procedures require officers to have an individualized suspicion of wrongdoing before conducting a search. See AFSCME, 315 Or. at 83 , 843 P.2d at 414 . A suspicionless administrative search may also withstand a state constitutional challenge.

In Tourtillott, the Oregon Supreme Court held that a fixed game checkpoint, at which police stop all cars travelling on certain roads to check for hunting licenses, does not violate Article I, Section 9 even though police do not have an individualized suspicion that they will discover evidence of wrongdoing in any particular car. 289 Or. at 859, 618 P.2d at 430. In reaching its decision, the court relied upon an analysis of federal law. In particular, it referred to Delaware v. Prouse, 440 U.S. 648 , 99 S.Ct. 1391 , 59 L.Ed.2d 660 (1979). In Prouse , the Court developed a four part analysis for determining the reasonableness of random, suspicionless searches and seizures. As it said, “the permissibility of a particular law enforcement practice is judged by balancing its intrusion on the individual’s Fourth Amendment interests against its promotion of legitimate governmental interests.” Id. at 654 , 99 S.Ct. at 1396 . From Prouse , four considerations, no one of which is determinative, can be distilled: (1) the importance of the governmental interests; (2) the degree of physical and psychological intrusion on the citizen’s rights; (3) the amount of discretion the procedure vests in individual officials; and (4) the efficiency of the procedure — that is how well it contributes to the reaching of its purported goals and how necessary it is. Id. at 653-63 , 99 S.Ct. at 1396-1401 . These are the factors that the Oregon Supreme Court identified and adopted. See Tourtillott, 289 Or. at 864 , 618 P.2d at 433 . That court then determined that a random search procedure which passed muster under the factors “was not unreasonable under either the Fourth Amendment of the United States Constitution or Article I, section 9 of the Oregon Constitution.” Id. at 859, 618 P.2d at 430.

It has since been suggested that the Tour-tillott analysis was really made under the Fourth Amendment and was not a complete explication of the reach of Article I, Section 9. See, e.g. Nelson II, 304 Or. at 101-02, 743 P.2d at 694. Nonetheless, the Oregon Supreme Court has not substituted a new test, so if some nuance remains to be added to the Prouse and Tourtillott analysis, we are not aware of what that is. At any rate, we need not enter the lists on that issue. If Tourtil-lott does set forth the proper test under the Oregon Constitution, as it seems to, we must apply what amounts to the Prouse analysis. If it does not, Oregon law is unclear, and we will turn to the federal constitution for. guidance. There we find the Prouse test. Again, that will at least establish. the floor, regardless of where Oregon may later discover the ceiling.

It is clear that the Policy authorizes administrative, not criminal, searches. It is equally clear that the school administrators do not obtain search warrants and do not claim that administration of the tests fits into one of the exceptions to the warrant require *1522 ment. Furthermore, the District concedes that, as is the case with most students who are tested, it had no reason to suspect that James has ever taken drugs. Testing is totally suspicionless. As a suspicionless, administrative search program, the Policy can only withstand Article I, Section 9 challenge if the drug tests that it authorizes are reasonable, under the four factor analysis outlined in Prouse and Tourtillott.

Two factors weigh in favor of the Policy’s constitutionality. The Policy contributed to reaching the desired goal. Every teacher who testified had noticed an improvement in discipline, a reduction in disciplinary referrals, and a decrease in drug use and the glorification of drug culture since the Policy was implemented. Thus, the efficiency factor weighs slightly in favor of the Policy. We say slightly because the exact nature of the goal is not entirely clear. If the goal was to avoid athletic injuries, testing of athletes was certainly a good way to approach that problem. If the goal was to reduce drug use in the student body in general, testing athletes seems to be a considerably more roundabout way of reaching that goal. Furthermore, if, as the testimony showed, teachers and administrators could detect behavior and physical problems that rather clearly appeared to be drug related, a program designed to target those who displayed the problems might be more efficient, although it might also present difficulties of its own when it came to discretionary enforcement. Of course, the Supreme Court has recognized that a somewhat relaxed standard of suspicion may well justify searches of students. New Jersey v. T.L.O., 469 U.S. 325, 341-2 , 105 S.Ct. 733, 743 , 83 L.Ed.2d 720 (1985). That does not mean that the Court decided that no form of individualized suspicion is required in schools. Id. at 342 n. 8, 105 S.Ct. at 743 n. 8. It does overcome some of the difficulty that is often mentioned when individualized testing is suggested. Certainly the Policy seems to have helped the District realize its goals of making the school run more smoothly. Its use and a return to normalcy coincided. Whether random testing was necessary to reach that goal is somewhat more questionable. However, on balance we believe the efficiency factor favors the District’s position.

Moreover, neither party contests the district court’s finding that drug testing under the Policy is completely random. Testing is initially required of all athletes and then conducted on a lottery basis throughout the athletic season. It vests no discretion in any District officials. This favors the Policy. That leaves importance and intrusiveness. In this ease they are dispositive. They dispose of the Policy.

In Tourtillott, the Oregon Supreme Court held' that enforcing hunting and fishing laws was a sufficient state interest to justify a checkpoint at which police ask drivers for their hunting licenses, or if none was presented, for their driving licenses and automobile registrations. 289 Or. at 864-67, 618 P.2d at 433-34. The District contends that if the state’s interest in freeing its forests and streams from unlicensed hunters and fisherman is important enough to justify a random search procedure, so is its interest in freeing its schools from the pernicious effects of illegal drugs. We do agree that keeping children away from drugs is at least as important as keeping unlicensed hunters away from the forests, but we must reject the District’s argument. The two situations are not at all comparable.

Tourtillott merely approved of a checkpoint stop. In 1976, the United States Supreme Court approved fixed immigration checkpoints in Texas and Southern California. United States v. Martinez-Fuerte, 428 U.S. 543 , 96 S.Ct. 3074 , 49 L.Ed.2d 1116 (1976). Because the level of intrusion on privacy was so limited, the Court only required the government to justify the policy with a legitimate interest. Id. at 561-62 , 96 S.Ct. at 3084-85 . The desire to exclude illegal aliens was sufficient.

As we have already said, the Oregon appellate courts have never decided a random drug testing case. The federal courts have. When confronted with a random drug testing case, the Supreme Court did not point to the kind of governmental interest — legitimate— that it found to be sufficient in its earlier checkpoint ease. It found a compelling government interest. Skinner, 489 U.S. at 628 - *1523 33, 109 S.Ct. at 1419-22 . Oregon would, no doubt, do the same when deciding the case under Article I, Section 9. But left without explicit guidance from the Oregon Courts, we will turn to a Fourth Amendment analysis. That analysis will not only uncover the legal principles, but will also flesh them out by some consideration of the facts of the relevant cases. We will also consider the private and governmental interests together. While they are separate elements, they are intertwined in any analysis to a very high degree, for they are not separate little weights to be put on a chemist’s balance. Rather, different mixtures of the values they represent will yield different results, just as different mixtures of chemicals will yield different reactions.

We start with Skinner , where the Court emphasized the need to balance governmental and privacy interests in any Fourth Amendment analysis. Id. at 619 , 109 S.Ct. at 1414 . In Skinner , the Court referred to the number of train accidents which had occurred over the last decade and which were drug or alcohol related. Id. at 607 , 109 S.Ct. at 1408 . It went on to mention the injuries and property loss that had resulted; there had even been 25 fatalities. Id. The Court said:

Id. at 624 , 109 S.Ct. at 1417 .

The Court emphasized that as a general rule “the collection and testing of urine intrudes upon expectations of privacy that society has long recognized as reasonable....” Id. at 617 , 109 S.Ct. at 1413 . As it said:

Id. (quotation and citation omitted.) However, the Court found that in the pervasively regulated railroad industry, the employees’ expectations of privacy were diminished. While they might not be exactly minimal, they were at a rather low ebb because the employees in question had been the principal focus of regulatory concern for a very long time. Id. at 627-28 , 109 S.Ct. at 1418-19 .

Against this rather minimal employee interest the Court set what it described as the compelling interest of the government in testing without a showing of individualized suspicion. It explained that determination as follows: “Employees subject to the tests discharge duties fraught with such risks of injury to others that even a momentary lapse of attention can have disastrous consequences.” Id. at 628 , 109 S.Ct. at 1419 . Given the minimal employee interest and the compelling governmental interest, the Court upheld the drug testing program.

In National Treasury Employees Union v. Von Raab, 489 U.S. 656 , 109 S.Ct. 1384 , 103 L.Ed.2d 685 (1989), the Court was presented with a somewhat different kind of safety concern. There it dealt "with certain groups of employees in the Customs Service. Those were employees who were directly involved in drug interdiction or enforcement, those who carried firearms, and those who handled classified material. The Court determined that testing the first two of these groups was valid; we will discuss the third later in this opinion.

As to the first two areas of coverage, the Court declared that employees involved in the interdiction of drugs and those who carry firearms have a diminished expectation of privacy because, considering the nature of their duties, they “reasonably should expect effective inquiry into their fitness and probity.” Id. at 672 , 109 S.Ct. at 1394. They should expect that reasonable tests will be applied to them.

To some extent the discussion of their expectations was a reiteration of the governmental interest discussion in which the Court declared that those who interdicted drugs are “our Nation’s first line of defense against *1524 one of the greatest problems affecting the health and welfare of our population.” Id. at 668, 109 S.Ct. at 1392. As such, they are subject to trickery, bribes, temptation, and extreme physical danger. Thus they must be “physically fit, and have unimpeachable integrity and judgment.” Id. at 670, 109 S.Ct. at 1393. Turning to employees who carry guns, the Court said that those “who may use deadly force plainly ‘discharge duties fraught with such risks of injury to others that even a momentary lapse of attention can have disastrous consequences.’ ” Id. (citation omitted). Given that analysis of interests, the Court, not surprisingly, determined that the suspicionless testing of those employees was proper.

Our cases have approved the expansion of drug testing to still more employees, but that has always been in the context of truly serious concerns of a safety nature. In AFGE Local 1533 v. Cheney, 944 F.2d 503 (9th Cir.1991), for example, we pointed out that people who are required to hold top secret security clearances are in a position to obtain information “the unauthorized disclosure of which reasonably could be expected to cause exceptionally grave damage to the national security!” Id. at 504 (internal quotation omitted). Those individuals, we said, have positions which subject them to regular close review of their personal lives, as a result of which their expectations of privacy are much attenuated. Id. at 507 . The compelling nature of the government’s interest, on the other hand, is found in the gravity of the risk. It takes no hierophant to see that the very safety of this country and its citizens can be seriously compromised by a drug-afflicted person with access to top secret materials. See also Department of the Navy v. Egan, 484 U.S. 518, 527-28 , 108 S.Ct. 818, 824 , 98 L.Ed.2d 918 (1988).

Our other cases are to the same effect. See IBEW, Local 124.5 v. United States Nuclear Regulatory Comm’n, 966 F.2d 521 (9th Cir.1992) (IBEW v. NRC) (diminished expectation of privacy conceded, and the danger of catastrophic harm in a nuclear accident is compelling); International Bhd. of Teamsters v. Department of Transp., 932 F.2d 1292 (9th Cir.1991) (privacy expectations of truckdrivers are markedly reduced, and given the size of commercial trucks and the possible danger from their loads, the government’s interest is compelling); IBEW, Local 1245 v. Skinner, 913 F.2d 1454 (9th Cir.1990) (the privacy interest of gas pipeline workers is diminished in this regulated industry, and the terrible accidents that can occur show a compelling government interest); Bluestein v. Skinner, 908 F.2d 451 (9th Cir.1990) (diminished expectation of privacy conceded, and the interest of the government in preventing air crash disasters is compelling), cert. denied, 498 U.S. 1083 , 111 S.Ct. 954 , 112 L.Ed.2d 1042 (1991).

The nature of the Fourth Amendment analysis is further illuminated by the Court’s comments on those who cannot necessarily be covered by random testing. In Von Raab , for example, although the Court approved testing of frontline and gun-toting workers, it did not as readily accept the notion that for security reasons a compelling government interest justifies the testing of all accountants, animal caretakers, attorneys, baggage clerks, coop students, electrical equipment repairers, mail clerks and messengers in the Customs Department. 489 U.S. at 678, 109 S.Ct. at 1397. The Court returned the case for further findings regarding them. Similarly, in IBEW v. NRC, in a concurring opinion joined by two judges of the three-judge panel, we expressed serious doubt that the danger of nuclear accidents justified drug testing of all clerical workers. 966 F.2d at 528-29 . See also Taylor v. O’Grady, 888 F.2d 1189 , 1199 (7th Cir.1989) (while prison employees in direct contact with prisoners can be tested, other prison personnel cannot be — “generalized interest in the integrity of the work force” is not enough); Harmon v. Thornburgh, 878 F.2d 484, 492-93 (D.C.Cir.1989) (while attorneys with access to top secret information can be tested, other attorneys in the DOJ who merely deal with grand juries and prosecute criminal cases cannot be, as a group), cert. denied, 493 U.S. 1056 , 110 S.Ct. 865 , 107 L.Ed.2d 949 (1990).

It can, therefore, be readily seen that although the courts have been willing — perhaps too willing — to allow employee drug testing, they have done so in cases fraught *1525 with danger where the interests of the person to be tested were attenuated. We must consider the efficacy of the Policy with these principles in mind.

As we turn to our analysis we must be ever mindful that “students [do not] shed their constitutional rights ... at the schoolhouse gate.” Tinker v. Des Moines Indep. Community Sch. Dist., 393 U.S. 503, 506 , 89 S.Ct. 733, 736 , 21 L.Ed.2d 731 (1969). School boards have:

West Virginia State Bd. of Educ. v. Barnette, 319 U.S. 624, 637 , 63 S.Ct. 1178, 1185 , 87 L.Ed. 1628 (1943).

Children are compelled to attend school, but nothing .suggests that they lose their right to privacy in their excretory functions when they do so. While they must attend classes and follow school rules, that does not indicate they have given up their basic privacy rights. It is true, of course, that after a complete application of the balancing test, the Supreme Court did allow individualized searches of students based upon a somewhat lessened standard of suspicion than that which applies elsewhere. But the Court was careful to say that “the situation is not so dire that students in the schools may claim no legitimate expectations of privacy.” T.L.O., 469 U.S. at 338 , 105 S.Ct. at 741 . In fact, the Court suggested, without holding, that a student’s interest in privacy is not of the minimal kind that had been found to justify random searches in the past. Id. at 342 n. 8, 105 S.Ct. at 743 n. 8. There simply is no sufficient basis for saying that the privacy interests of students are much less robust than the interests of people in general.

Nor can we say that the privacy interests of athletes are substantially lower than those of students in general. The District’s argument that student-athletes have a reduced expectation of privacy because of their participation in an activity with many rules analogizes to inapposite cases. Training rules and grade point average requirements are not the sort of extensive government regulation 'that has been found to diminish the expectation of privacy of workers in high risk industries or high security areas of the government. See, e.g., IBEW v. NRC, 966 F.2d at 525 (workers at nuclear power plants); AFGE, 944 F.2d at 507 (Navy civilian employees who hold “Top Secret with Access” security clearances); Bluestein, 908 F.2d at 456 n. 7 (airline workers). High school athletes do not go through extensive background cheeks in order to join their teams. Nor, in general, do their lives or the lives of others depend upon their ability to perform their roles on the football team. Participation in interscholastic sports does not significantly diminish a high school athlete’s reasonable expectation that he will not be compelled to submit to suspicionless drug testing by urinalysis. Appellee’s contention that conditions in the school locker room reduce an athlete’s expectation of privacy is similarly unavailing. Normal locker room or restroom activities áre a far cry from having an authority figure watch, listen to, and gather the results of one’s urination.

We recognize that, in some sense, students’ participation in athletics is “voluntary.” That, too, is insufficient to reduce their privacy interests to a minimal level. Participation in athletics is obviously highly desirable and encouraged, as this record shows. • The fact is that parents wish to have their children obtain the physical and mental benefits of organized sports. That, too, is part of the educational process; part of what tax dollars pay for. While separating mind and body might be a fine idea for scientific purposes, we have learned that they are inextricably intertwined. In a word, what benefits one, benefits the other. Children, students, do not have to surrender their right to privacy in order to secure their right to participate in athletics. Athletes’ right to privacy also remains robust.

*1526 But the strength of the government’s interests must be considered. Those interests themselves are not minimal. The District wishes to prevent unnecessary athletic injuries, to reduce the attraction of drugs among other students, and to improve discipline. Certainly those are worthy goals. However, worthy as they are, they suffer by comparison to the kinds of dangers that have existed when random testing has been approved. The extreme dangers and hazards involved in the prior cases are simply not present here. The prospect that an athlete might hurt himself or a competitor is real enough, but it is not a risk of the same magnitude as an airplane or train wreck, or a gas pipeline or nuclear power plant disaster. The concern that our children will fail to acquire knowledge or respect is also real enough, but it, too, does not reach the level of the concerns that have permitted suspicionless testing. All of this is true as a general proposition, and nothing in this record shows a situation or problem so exacerbated as to take it beyond the generality.

Having uncaring, disrespectful, drug-impaired children in our schools is tragic. Having them anywhere is equally tragic. In fact, it is tragic to have adults of that type in our society. It is widely conceded that what drug usage has done to the fabric of our society is a tragedy of national proportions. We do not disagree with that. However, it is not the type of potential disaster that has caused the Court or us to find a governmental interest compelling enough to permit sus-picionless testing. It has been lurking as a background condition for our determinations, but that is all. Its presence has been referred to, but the disaster or tragedy has been some additional horrible element, some terrible threat to safety that can flow from the presence of drugs — some shooting, some explosion, some crash of train, truck, or aircraft, or some breach of top secret national security. The record shows no such possibility here* Nor does it show that the incremental difficulties involved in athletics elevated the drug danger so far above the norm that the government interest became compelling.

The District has alluded to the possibility that it could incur liability if it knowingly fielded a team of drug-affected students. No doubt it could. We agree that it need not do that, but our concession is a far cry from saying that the District can mandatorily test everyone. Given the standard made possible by T.L.O., the District’s argument might well justify individualized testing; on this record, it does not justify random testing.

Thus, when we mix all of the elements together — the slight weight in favor of efficiency, the privacy interest, the governmental interest, and the discretionary factor— and step back to look at the compound they yield, it becomes apparent that the Policy violates the Fourth Amendment. It follows that it also violates Article I, Section 9 of the Oregon Constitution.

It is important to understand the contours and outer limits of our holding. We do not, as we have already suggested, decide that there could be no appropriate individualized testing program. More than that, we do not say that there could never be a ease where a random search would be appropriate in a school setting. See, e.g., In re Isiah B, 176 Wis.2d 689 , 500 N.W.2d 637 (after a series of shootings on campus, rumors indicated that there would be a shootout on November 19, 1990; a random search of lockers on that date was upheld), cert. denied, — U.S.-, 114 S.Ct. 231 , 126 L.Ed.2d 186 (1993). But see Brooks v. East Chambers Consol. Indep. Sch. Dist. 730 F.Supp. 759 (S.D.Tex.1989) (initial and random drug testing of all students who wish to participate in extracurricular activities violates the Fourth Amendment), aff 'd, 930 F.2d 915 (5th Cir.1991); Derdeyn v. University of Colorado, 832 P.2d 1031 (Colo.App.1991) (random drug testing of athletes violates Fourth Amendment; governmental interest not compelling) aff'd, 863 P.2d 929 (Colo.1993); Odenheim v. Carlstadt-East Rutherford Regional Sch. Dist., 211 N.J.Super. 54 , 510 A.2d 709 (Ch.Div.1985) (drug testing of all students violates Fourth Amendment). Moreover, we do not decide whether a program based upon the truly voluntary consent of individual parents to have their child randomly tested — not, no *1527 testing, no playing — would obviate the difficulties we find with the Policy.

Before concluding, we must acknowledge that in Schaill v. Tippecanoe County Sch. Corp., 864 F.2d 1309 (7th Cir.1988),' the court upheld a random drug testing program very similar to the one we now review. Perhaps that court found a more compelling set of facts than we discover here, but we are unable to say so after reading the opinion. We believe, instead, that the Seventh Circuit has unduly minimized the privacy interests of students. It has also given undue weight to the governmental interest by focusing on the general problems generated by the drug plague, rather than upon the question of whether the danger to safety is so high as to be compelling. 3 We could fill more pages reiterating what we have already said, but, in a nutshell, we simply do not agree with the Seventh Circuit.

CONCLUSION

We understand the importance of having drug-free children, and the sorrow our society is reaping from the fact that the drugs which have invaded it have found their way into our children’s hands. We also understand the deep concerns of parents, teachers, administrators, and school boards. We have no doubt that the District and those associated with it have proceeded in all good faith.

However, we also understand the concern of our forebears and the importance of the protections given by the constitutional provisions which prohibit unreasonable searches and seizures. We are all too aware of 'the dangers to our liberties that'those provisions are designed to protect against and of the constant pressures upon them, despite centuries of living with and under their protections.

We have found that we must live with a certain amount of discomfort, even danger, if we are to maintain constitutional protections. If they are to continue to exist and flourish, we and our children must understand them, profit from them, and believe in them. They are, after all, just words, ideas, beliefs, and principles. But if we are vigilant there is a great deal of power in the .word “just.”

Given the Fourth Amendment, given our traditions, given our law, we are constrained to hold that the Policy is invalid under the .Fourth Amendment. That being so, Oregon would find it invalid under Article I, Section 9, of its Constitution. As we have already indicated, it would probably do so with less lucubration, and with less spilt ink.

REVERSED and REMANDED.

. This case was affirmed on other grounds in Nelson II. Thus, this iteration of the case is probably not controlling precedent. See Rogers v. Saylor, 88 Or.App. 480, 483-84 , 746 P.2d 718, 720 (1987), rev'd on other grounds, 306 Or. 267 , 760 P.2d 232 (1988).

. The classification of a search depends upon "[t]he purpose of the search and the consequences that flow from it.” Nelson II, 304 Or. at 104, 743 P.2d at 696. If the consequences of noncompliance with whatever standards the inspection is meant to uphold are noncriminal, the search is “civil” or "administrative.” By contrast, if offenders face criminal sanctions, the inspection implicates criminal law enforcement purposes. Id. A search in the criminal context can be administrative in nature. See Atkinson, 298 Or. at 10-11 , 688 P.2d at 837-38 (police inventory search).

. We note that the Schaill court balanced the students' privacy interests against the school district’s “substantial interest in enforcement of its proposed random urinalysis program.” 864 F.2d at 1321 (emphasis added). In contrast, subsequent Supreme Court cases have allowed random urinalysis programs only after finding compelling governmental interests that outweigh employees' diminished expectations of privacy. See Skinner, 489 U.S. at 626-33 , 109 S.Ct. at 1418-21 ; Von Raab, 489 U.S. at 670-77 , 109 S.Ct. at 1393-96 .