I concur in part III, D, 1 of the lead opinion, and in its disposition as it pertains to the City of Glendale’s (hereafter sometimes City) program for drag testing candidates for promotion. I agree with the lead opinion that “it is not constitutionally ‘reasonable,’ under the Fourth Amendment [to the United States Constitution], for a governmental employer to conduct suspicionless urinalysis drag testing of all current public employees seeking promotion, regardless of the nature or duties of the position at issue” (Lead opn., ante, at p. 878.) I also believe, for reasons that will become evident below, that the privacy clause of article I, section 1 of the California Constitution provides independent grounds for invalidating drag testing of all City employees in line for promotions.
I do not agree with the lead opinion and the majority that the testing of all applicants for employment within the City is constitutional under either the federal or state Constitution. As explained below, although it may be true that government employers have a somewhat greater need to test applicants than to test current employees, that need does not, by itself, constitutionally justify across-the-board testing of applicants. Nor is it true, as the lead opinion maintains, that urine testing is minimally intrusive simply because applicants for employment have been subject to medical examinations. Drag *901testing represents a significant additional invasion of these applicants’ basic rights to privacy and dignity, and the City has not carried its considerable burden of showing that such invasion is justified in the case of all applicants offered employment. The City has not demonstrated a constitutionally paramount need to include applicants for such positions as clerk-typist and city attorney in a drug testing program that has typically been confined to those seeking employment in such safety-sensitive jobs as bus driver or police officer. From the record before us, we cannot determine if the City’s universal drug testing program arises from a substantial, legitimate governmental need, or from purely political or ideological considerations, and therefore cannot determine the extent to which the program is constitutionally justified.
I.
The Fourth Amendment to the United States Constitution affirms the right of the people to “be secure in their persons, houses, papers and effects, against unreasonable searches and seizures.” In general, a government search and seizure must be justified with a warrant “on probable cause,” or must at least be backed by “ ‘some quantum of individualized suspicion.’ ” (Skinner v. Railway Labor Executives’ Assn. (1989) 489 U.S. 602, 624 [103 L.Ed.2d 639, 664, 109 S.Ct. 1402] (Skinner).) “In limited circumstances, 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, a search may be reasonable despite the absence of such suspicion.” (Ibid.)
In Skinner, the court concluded that “the collection and testing of urine” is a search under the Fourth Amendment because such urine testing “intrudes upon expectations of privacy that society has long recognized as reasonable.” (Skinner, supra, 489 U.S. at p. 617 [103 L.Ed.2d at p. 660].) The Skinner court identified two privacy interests that bring urine testing within the ambit of the Fourth Amendment. First, the very act of urine testing itself requires the employee to perform under government supervision “an excretory function traditionally shielded by great privacy.” (Id. at p. 626 [103 L.Ed.2d at p. 666].) “ ‘There are few activities in our society more personal or private than the passing of urine. Most people describe it by euphemisms if they talk about it at all. It is a function traditionally performed without public observation; indeed, its performance in public is generally prohibited by law as well as social custom.’ ” (Id. at p. 617 [103 L.Ed.2d at pp. 659-660].) Second, the Skinner court found it undisputed that “chemical analysis of urine . . . can reveal a host of private medical facts about an [individual], including whether he or she is epileptic, pregnant, or diabetic.” (Ibid.)
*902There is no question that the City’s drug test in the present case infringes on these same privacy interests. The test is conducted, according to the physician who operated the program for the City, as follows: First, the applicant is required to sign a waiver form that includes “a place for the individual to list any and all medications, drugs, etc. that the individual is taking. If any of these medications are included in the drug testing procedure, the individual will have to satisfactorily explain their usage. [<jn The individual is then asked to remove his or her clothing. A man removes his clothing to his shorts, and a woman to her bra and panties at which point they both are given a gown to wear. . . . [^D The individual is then accompanied by [a] medical representative to a restroom to give a sample of urine in order to prevent tampering with the sample. The medical representative monitoring the giving of the urine sample stands in the cubicle adjacent to the cubicle occupied by the individual to assure that there is no tampering with the sample. However, there is no direct observation of the individual giving the mine sample. [^Q The individual returns the container to the medical representative. The medical representative notes whether the sample is cold or warm. If it is cold, this will not be acceptable for this test procedure because it indicates tampering . . . .” If the sample does not appear to have been tampered with, it is tested for blood, sugar and protein as screening for medical problems, then sealed, labeled and sent to a laboratory.
Thus, the City’s test possesses the two intrusive features identified by the Skinner court. First, the aural monitoring of the applicant’s urination, after being asked to disrobe, is a basic form of privacy invasion. Second, the urinalysis and the required disclosures prior to urinalysis will reveal a host of private information, including the full array of legal medications the applicant is taking. Thus, the City may learn from such data, and from the drug test results themselves, much about the private lives of the applicants, including whether the applicant is “HIV positive,” or is taking birth control pills or high blood pressure medication. The City will also learn who is using marijuana outside the workplace for permissible medical reasons. (See Health & Saf. Code, § 11362.5.) “Because drug tests often furnish information about employee activities occurring outside of working hours, such tests may provide ... a periscope through which [the testers] can peer into an individual’s behavior in her private life, even in her own home.” (Jones v. McKenzie (D.C. Cir. 1987) 833 F.2d 335, 339 [266 App.D.C. 85], opn. mod. 878 F.2d 1476, 1477 [279 App.D.C. 19].)
The lead opinion argues that applicants for employment in this case have diminished expectations of privacy when compared to current employees, because the preemployment drug testing is part of a medical examination *903that predated the drug testing program and is supposedly typical in the preemployment setting. “[A] medical examination, in itself, ordinarily requires the individual not only to submit a urine sample but in addition to provide a medical history and to undergo a physical examination that entails an intrusive touching of one’s body by a physician. . . . Although in this case the city’s addition of the drug screening test to its preexisting medical examination procedure was accompanied by the institution of several new security measures (such as the aural monitoring of the urinalysis process) that entailed some additional intrusion on the applicant’s privacy with regard to provision of the urine sample, the incremental intrusion on privacy attributable to these new measures appears rather minor when viewed in the context of a complete medical examination.” (Lead opn., ante, at p. 884, italics in original.)
I disagree that the “incremental” intrusion was “rather minor.” It is evident from the record that whatever physical examination had been done prior to the institution of drug testing did not compel disclosure of all the medications the applicant was taking, nor administer tests that would reveal the presence of such medication; rather, this disclosure was a particular feature of the drug screening. Although the record is somewhat unclear on this point, it appears that urinalysis before drug screening is only for the presence of blood, sugar, and protein in the urine. Moreover, even if the medical examination had required applicants to respond to a written inquiry concerning the medications they were presently taking, such an inquiry is obviously less invasive than the use of urinalysis to obtain such information.
Further, urinalysis prior to the institution of the drug testing program did not involve the monitoring of urination, aurally or otherwise. Thus, as the lead opinion concedes, “[t]he disclosure of additional private information through testing a bodily substance obtained from an individual and the increased monitoring of the process under which an individual provides a urine sample constitute intrusions upon the applicant’s constitutional privacy interests that are not insignificant or de minimis—intrusions that would not be permissible in the absence of reasonable justification.” (Lead opn., ante, at p. 897.) Or, to state the matter in stronger terms, the drug test is more than a routine physical examination of the employee; it represents a coercive governmental appropriation of information about private, and in many cases lawful, behavior, a kind of coercion which in turn necessitates the additional intrusion of monitored urination—“a type of search particularly destructive of privacy and offensive to personal dignity.” (Treasury Employees v. Von Raab (1989) 489 U.S. 656, 680 [103 L.Ed.2d 685, 711, 109 S.Ct. 1384] (dis. opn. of Scalia, J.) (Von Raab).)
Moreover, although the lead opinion cites cases approving an employer’s ability to administer medical examinations to prospective employees, it is by *904no means clear in the case law that all preemployment medical examinations, of whatever scope and for whatever reason, are constitutional.1 As one commentator has explained, medical testing has been expanding in recent years both in frequency and in breadth; its function has been gradually transformed from a means of screening out those who are presently unfit to perform a job, or who are particularly at risk of being exposed to certain occupational dangers, into a tool for assessing an employee or applicant’s general health and private habits so as to promote cost containment by the employer. (See Rothstein, Medical Screening and the Employee Health Cost Crisis (1989) pp. 1-6.) Given the inherently invasive nature of medical examinations, any expansion of their scope and intrusiveness should be viewed with vigilance rather than quiescence by courts charged with safeguarding the Fourth Amendment. The view, implicit in the lead opinion, that “all prior intrusions . . . which society has accepted, form a baseline for comparison . . . [and] [e]ach new intrusion . . . represents] only an incremental diminution of privacy overall” (Cornish & Louria, Employment Drug Testing, Preventive Searches, and the Future of Privacy (1991) 33 Wm. & Mary L.Rev. (1991) 95, 114) is one certain to allow the steady erosion of our privacy rights.
The lead opinion is of course correct that, as the United States Supreme Court noted in Skinner, the invasiveness of the drug testing process is somewhat reduced when “[t]he sample is . . . collected in a medical environment, by personnel unrelated to the . . . employer, and is thus not unlike similar procedures encountered often in the context of a regular physical examination.” (Skinner, supra, 489 U.S. at pp. 626-627 [103 L.Ed.2d at p. 666].) But the Skinner court mentioned this as only one of a number of factors weighing in favor of the constitutional legitimacy of any particular urine test: “More importantly, the expectations of privacy of covered employees are diminished by reason of their participation in an industry that is regulated pervasively to ensure safety, a goal dependent, in substantial part, on the health and fitness of covered employees.” (Id. at p. 627 [103 L.Ed.2d at p. 666], italics added.) As Skinner thereby demonstrates, the particular setting in which a drug test takes place is less important for purposes of Fourth Amendment analysis than the strength of the government interests justifying the test and the concomitant reduction of an employee’s reasonable privacy expectations in light of those interests. Thus, it is important to recognize that the City’s drug testing program is not immunized from *905constitutional challenge merely because it can be placed under the familiar aegis of a “medical examination.”
The lead opinion, in addition to minimizing the job applicant’s expectations of privacy, also exaggerates the nature of the government employer’s interests. It accomplishes this overstatement in two steps. First, it contends that “[i]n light of the well-documented problems that are associated with the abuse of drugs and alcohol by employees—increased absenteeism, diminished productivity, greater health costs, increased safety problems and potential liability to third parties, and more frequent turnover—an employer, private or public, clearly has a legitimate (i.e., constitutionally permissible) interest in ascertaining whether persons to be employed in any position currently are abusing drugs or alcohol.” (Lead opn., ante, at pp. 882-883, fns. omitted.) Second, it reasons that although these interests could also justify the drug testing of current employees as well as job applicants, “an employer generally need not resort to suspicionless drug testing to determine whether a current employee is likely to be absent from work or less productive or effective as a result of current drug or alcohol abuse: an employer can observe the employee at work, evaluate his or her work product and safety record, and check employment records to determine whether the employee has been excessively absent or late.” (Id. at p. 883.) Not so, they claim, for the job applicant, because the employer “may lack total confidence in the reliability of information supplied by a former employer or other references.” {Ibid.) The lead opinion therefore concludes that “[i]n view of these considerations, ... an employer has a greater need for, and interest in, conducting suspicionless drug testing of job applicants than it does in conducting such testing of current employees.” (Ibid.)
Yet even if the lead opinion is correct that employers have in general a marginally greater need to test job applicants than current employees, this fact does not justify the drug testing of all applicants for government employment. As explained below, such a conclusion is incompatible with the proper, fact-specific balancing of privacy rights and governmental interests required by the Fourth Amendment.
The principles pertinent to analyzing the sufficiency of the government’s interests are found in cases discussed at length in the lead opinion, and I will only briefly summarize those cases here. In Skinner, supra, 489 U.S. 602, and Von Raab, supra, 489 U.S. 656, the United States Supreme Court set forth the basic framework for analyzing Fourth Amendment challenges to drug testing policies and programs. The court recognized, as mentioned above, that drug testing by a government employer was a “search” within the meaning of the Fourth Amendment, but that a warrant and indeed any form *906of individualized suspicion could be dispensed with “where the Government seeks to prevent the development of hazardous conditions or to detect violations that rarely generate articulable grounds for searching any particular place or person.” (Von Raab, supra, 489 U.S. at p. 668 [103 L.Ed.2d at p. 703].) These types of government interests may “in certain limited circumstances . . . [be] sufficiently compelling to justify the intrusion on privacy entailed by conducting such searches without any measure of individualized suspicion.” (Ibid., italics added.) The Von Raab court upheld much of the United States Customs Service drug testing program at issue in that case, which involved the testing of employees prior to promotion into positions directly involved in drug interdiction, or in the handling of firearms, the misuse of which could pose a danger to public safety. Both of these interests were deemed to be “compelling” by the court. (Id. at p. 672 [103 L.Ed.2d at p. 706].) The court also agreed in principle that the drug testing of employees required to handle classified information, another part of the Custom Service’s drug testing program, was justified by “a compelling interest in protecting truly sensitive information from those who, ‘under compulsion of circumstances or for other reasons, . . . might compromise [such] information.’ ” (Id. at p. 677 [103 L.Ed.2d at p. 709].) But the court found it unclear, from the record, whether all the job classifications subject to drug testing— such as “ ‘Baggage Clerk’ ” and “ ‘Electric Equipment Repairer’ ”—were “likely to gain access to sensitive information” (id. at p. 678 [103 L.Ed.2d at p. 710]) and accordingly remanded to the court of appeals for clarification and possible limitation of the Customs Service’s drug testing program.
The lessons of Skinner and Van Raab were distilled in Harmon v. Thorn-burgh (D.C. Cir. 1989) 878 F.2d 484 [278 App.D.C. 382], a case involving a program of random drug testing for a wide range of United States Department of Justice employees. The Harmon court rejected the department’s rationale that a generalized interest in the “integrity of its workforce” could justify such testing, stating that “Von Raab . . . suggests that federal employment alone is not a sufficient predicate for mandatory urinalysis.” (Id. at p. 490.) Rather, “Von Raab . . . suggests that the government may search its employees only when a clear, direct nexus exists between the nature of the employee’s duty and the nature of the feared violation.” (Ibid., italics added.) While the court indicated that a more limited drug testing program confined to “DOJ employees having substantial responsibility for the prosecution of federal drug offenders” would be lawful (id. at p. 490, fn. omitted), it invalidated the program in question as overbroad.
The Supreme Court further elaborated on its Fourth Amendment drug testing analysis in Vernonia School Dist. 47J v. Acton (1995) 515 U.S. 646 [132 L.Ed.2d 564, 115 S.Ct. 2386] (Vernonia) a case that involved the *907lawfulness of a policy of random drug testing of high school student athletes. The court clarified that the phrase “ ‘compelling state interest,’ ” used in Skinner and Von Raab, does not describe “a fixed, minimum quantum of governmental concern, . . . [but] [r]ather ... an interest which appears important enough to justify the particular search at hand, in light of other factors which show the search to be relatively intrusive upon a genuine expectation of privacy.” (Id. at pp. [132 L.Ed.2d at p. 579, 115 S.Ct. at pp. 2394-2395].) The court’s alternate appellation for the term “compelling interest” was a “relatively high degree of government concern.” (Id. at p._[132 L.Ed.2d at p. 579, 115 S.Ct. at p. 2395].) It concluded that that standard had been met by the high school’s drug testing policy: “Deterring drug use by our Nation’s schoolchildren is at least as important as enhancing efficient enforcement of the Nation’s laws against the importation of drugs, which was the governmental concern in Von Raab . . . , or deterring drug use by engineers and trainmen, which was the governmental concern in Skinner.” (Id. at p. _ [132 L.Ed.2d at pp. 579-580, 115 S.Ct. at p. 2395].) The Vernonia court cautioned, however, “against the assumption that suspicionless drug testing will readily pass constitutional muster in other contexts. The most significant element in this case is . . . that the [drug testing] Policy was undertaken in furtherance of the government’s responsibilities, under a public school system, as guardian and tutor of children entrusted to its care.” (Id. at p. _ [132 L.Ed.2d at p. 582, 115 S.Ct. at p. 2396].)
Thus, in each of the cases discussed above, the court pinpointed a sufficiently compelling interest, or relatively high degree of governmental concern that justified invading privacy rights. The kinds of compelling interests these cases enumerate are either critical public safety concerns, where “ ‘even a momentary lapse of attention can have disastrous consequences’ ” (Von Raab, supra, 489 U.S. at p. 670 [103 L.Ed.2d at p. 705]) or, as in the case of customs agents or the school district, the prevention of some kind of fundamental impairment of the particular government agency’s ability to perform its public mission. (Ibid.) On the other hand, the cases suggest that other, nonspecific governmental, symbolic interests, such as the “integrity of the workforce,” or interest in a “drug free workplace,” although laudable, do not justify the invasion of privacy inherent in a drug test. (See Harmon v. Thornburgh, supra, 878 F.2d at p. 490; see also Von Raab, supra, 489 U.S. at pp. 680-681 [103 L.Ed.2d at p. 712] (dis. opn. of Scalia, J.); Georgia Ass’n of Educators v. Harris (N.D.Ga. 1990) 749 F.Supp. 1110, 1115; Taylor v. O’Grady (7th Cir. 1989) 888 F.2d 1189,1196.) “[O]utside of the law enforcement context, the government’s legitimate interest in employee drug testing extends no further than its interest in workplace conduct and the performance of work responsibilities.” (National Treasury Employees Union v. Yeutter (D.C. Cir. 1990) 918 F.2d 968, 974 [287 App.D.C. 28].)
*908The lead opinion does not rely on purely “symbolic” interests to justify the drug testing of all applicants for government employment, but rather on a general interest in government “efficiency”—the legitimate need that a government employer, like a private employer, has in promoting a productive workforce in which absenteeism, disciplinary problems, turnover, etc., are minimized. Courts have generally viewed these interests as a rationale for drug testing with considerable skepticism. (See National Treasury Employees Union v. Yeutter, supra, 918 F.2d at p. 974 [rejecting “unsupported connection between off-duty drug use and government efficiency”].) As this court has observed, “the Fourth Amendment protection against unreasonable searches and seizures has generally been interpreted to require more than an employer interest in employee job performance ... to justify drug testing without reasonable suspicion.” (Hill v. National Collegiate Athletic Assn. (1994) 7 Cal.4th 1, 54 [26 Cal.Rptr.2d 834, 865 P.2d 633].)
The lead opinion implicitly recognizes that these efficiency interests are not as compelling as public safety interests, because the former cannot justify, as the latter can, the drug testing of promotional candidates. If that were not the case, then the lead opinion’s contention that there are alternative means of assessing job performance that render prepromotional testing unnecessary would invalidate such testing even for positions critical to public safety. After all, a current employee’s actual safety record is in all likelihood a much better predictor of future safety performance than is the off-the-job drug use tested by urinalysis. But, as the lead opinion implies, an interest in preventing accidents with disastrous public safety consequences is so compelling that testing is justified even though the amount of pertinent additional information yielded by the testing may be slight compared to alternative means of employee evaluation. But the government’s interest in employee efficiency, being of less importance, cannot surmount the argument that there are less intrusive alternative means of predicting future job performance of current employees. Behind this differential treatment of the two types of interests is the self-evident recognition that the government’s interest in preventing accidents is far weightier than its interest in promoting marginal productivity improvements in its operations.
The interest in government efficiency is not only, as a general matter, less than compelling, but is also potentially all encompassing. While the concern with public safety is relatively focused—only in certain occupations, such as bus driver, can an employee’s drug impairment represent an immediate threat to public safety—efficiency concerns are applicable to all government employees. Moreover, efficiency can logically be used to justify a wide array of intrusions into the privacy of job applicants, because so much private information might be arguably relevant in some way to predicting the *909performance of future employees—current family situation, a record of psychotherapeutic treatment, genetic makeup, a score on a lie detector test. (See generally, Larson, Employment Screening (1995) §§2.07-2.10, pp. 2-15 to 2-30.)
Thus, the government’s general interest in efficiency and employee productivity is both less important and potentially more far reaching than the interest in public safety and other interests that have been identified as “compelling.” These two characteristics dictate that courts adopt a certain attitude of skepticism when this interest is used to justify universal drug testing. I am unwilling to categorically state that the government’s interest in efficiency can never justify urine testing in the preemployment setting. (See O’Connor v. Ortega (1987) 480 U.S. 709, 724 [94 L.Ed.2d 714, 727-728, 107 S.Ct. 1492] [interest in ensuring efficient government operation a reason to dispense with warrant and probable cause requirements in searching office of government employee suspected of misfeasance].) It may indeed be the case that the employer has a greater need, at least in some instances, for drug testing applicants than for testing current employees. But when the government seeks to drug test all applicants offered employment solely on general efficiency grounds, I would place on it the burden of showing that the efficiency interests at stake are so substantial that they require a drug testing program of such breadth. As the court stated in Georgia Ass’n of Educators v. Harris, supra, 749 F.Supp. at page 1113, in invalidating a statute that authorized the drug testing of all applicants to state employment: “The [United States Supreme] Court’s application of the balancing test [in drug testing cases] was predicated on the [government agency’s] concrete expression of substantial or compelling interests served by drug testing and a fact-specific explanation of how testing the subject employee groups furthered those interests.” We should require nothing less of the City in the present case.
The reason for such a fact-specific showing is clear. It is basic to Fourth Amendment jurisprudence that legitimate expectations of privacy are not obliged to yield the right-of-way to every claim of governmental efficiency. As Justice Stewart stated of the operation of the Fourth Amendment in the criminal context: “[T]he mere fact that law enforcement may be made more efficient can never by itself justify disregard of the Fourth Amendment. [Citation.] . . . [T]he Fourth Amendment reflects the view of those who wrote the Bill of Rights that the privacy of a person’s home and property may not be totally sacrificed in the name of maximum simplicity in enforcement of the criminal law.” (Mincey v. Arizona (1978) 437 U.S. 385, 393 [57 L.Ed.2d 290, 301, 98 S.Ct. 2408].) So too, outside the criminal law, Fourth Amendment rights cannot be nullified simply by appeals to improving the *910productivity of government employers. Indeed, a government employer would presumably not drug-test if it were not in some sense efficient to do so, i.e., if the costs incurred by the tests were not outweighed by the benefits expected to be received by the government agency. But the Fourth Amendment does not merely enjoin the government to be a rational actor; it requires a government search outside the criminal context to be reasonable not in terms of the government’s costs and benefits, but in terms of the relative weight of societal gains and privacy losses resulting from the search, and therefore requires courts to invalidate certain types of searches and seizures despite the fact that they promote the government’s efficiency. A general, unsubstantiated claim that drug testing of all applicants is necessary to improve job performance leaves courts unable to assess the actual strength of government’s need for universal, suspicionless, testing and so renders them incapable of performing their proper function under the Fourth Amendment of balancing constitutional rights and government interests.
Moreover, as suggested above, a government agency may institute drug testing not because it expects tangible productivity gains but because of political, ideological, or symbolic motivations. (See Rothstein, Medical Screening and the Employee Health Cost Crisis, supra, at pp. 100-101; see also Von Raab, supra, 489 U.S. at pp. 686-687 [103 L.Ed.2d at p. 715] (dis. opn. of Scalia, J.).) Since the general interest in government efficiency, unlike the specific, compelling interests in public safety or the like, can in the abstract be used to justify drug testing of any and all positions, it may be employed as a pretext behind which is concealed a symbolic interest in ensuring a “drug-free workplace,” although such interest, if stated overtly, would be insufficient under the Fourth Amendment to warrant a universal testing program. Under these circumstances, a court must at least look behind the proffered generalities and determine if the government agency has made a specific showing that the drug testing of all applicants would advance the tangible governmental interests offered as justification for the program, or if these interests are merely a facade behind which to promote a purely symbolic political agenda.
What kind of showing must be made? First, at the very least, the government must demonstrate a significant problem with drug abuse, i.e., a productivity problem traceable to drug abuse. The problem must be pinpointed with some degree of specificity. It is not enough to identify drug abuse somewhere in city government. As a recent study has shown, drug use is highly stratified by occupation and industry, with over 15 percent of construction workers and waitresses, for example, reporting current drug use, and only about 2-4 percent teachers and engineers so reporting. (Substance Abuse and Mental Health Services Administration, Drug Use Among U.S. Workers: *911Prevalence and Trends By Occupation and Industry Categories (Drug Use Among U.S. Workers) (1996) pp. 17-18.) Therefore, when a city has had problems with its groundkeeping staff or its temporary recreational employees engaging in drug-abuse-related unproductive behavior, it may be legitimate for that city to adopt applicant testing among some of its employees to address this problem. But there may be little if any public benefit from drug testing clerks, secretaries, engineers, or city planners, if no similar productivity problem has been identified among those groups.
It is true that the Von Raab court implicitly rejected Justice Scalia’s dissenting argument that the Customs Service’s drug testing program was unconstitutional because there had been an inadequate showing of a drug problem in the agency. (Von Raab, supra, 489 U.S. at pp. 674-675 [103 L.Ed.2d at p. 708].) But in the present case, with less compelling, more amorphous, interests at stake than in Von Raab, the Fourth Amendment balancing of interests requires the government to make a particularized showing that such comprehensive testing is genuinely necessary for the significant improvement of government operations. The government may not conduct widespread, suspicionless searches under the Fourth Amendment without evidence that it is engaged in something more than a far-flung fishing expedition or a symbolic anti-drug crusade.
Second, the government must show that drug testing is a reasonable means of addressing the problems at hand—i.e., that a positive drug test is a good predictor of future job performance. It is evident that the government cannot claim a substantial need to drug-test without some showing that such testing actually contributes to the desired aim of improving employee productivity.
Third, the government must demonstrate that there are no reasonable, less invasive alternatives to drug testing in the particular government department or occupational classification in question. This showing goes to the heart of the lead opinion’s reason for distinguishing between preemployment and prepromotional drug testing—that in the former case, unlike the latter, the employer has not had adequate opportunity to observe the employee’s work performance. But that assertion is by no means invariably incontrovertible. There may be some occupations in which much is revealed about an employee’s level of performance through a resume, references, and various benchmarks of performance, particularly when the employee has been subject to extensive background checks. (See Willner v. Thornburgh (D.C. Cir. 1991) 928 F.2d 1185,1197-1198 [289 App.D.C. 93] (dis. opn. of Henderson, J.).) Applicants who have had their prior work subject to “the kind of day-to-day scrutiny that is the norm in more traditional office environments” (Von Raab, supra, 489 U.S. at p. 674 [103 L.Ed.2d at p. 707]) may not be *912able to conceal significant problems in their previous employment from prospective employers. It is therefore not self-evident that an employer has in all cases a greater need to drug-test applicants for employment than it has to test current employees. As such, the City’s claim that it has insufficient alternative means for screening applicants should be subject to a requirement of evidentiary support.
After reviewing the record below, I conclude that the City has not made the required factual showing of a substantial interest in drug-testing all applicants to whom it offers employment. First, the City has failed to demonstrate that it has a widespread drug problem requiring universal testing. The City did institute a preplacement drug testing program between November 1985 and January 1986, and found that about 20 percent of the applicants/employees tested positive for drugs, or about 10 out of 48. The occupational categories testing positive were as follows: three part-time seasonal laborers, two tree trimmers, two sanitation workers, a part-time recreational leader, a pump-plant operator and an equipment service worker. Can it really be said from these figures that drug use/abuse was pandemic in the City of Glendale’s workforce? There was no evidence, for example, of drug use among the City’s clerical or professional employees, who no doubt make up a good percentage of that workforce. If applicants for the tree trimmer job are testing positive for drugs, does that tell us anything about whether there is a need for testing in the city manager’s, or city planner’s, office?
Moreover, the City’s claim that it has documented, prior to instituting drug testing, a significant increase in the percentage of disciplinary cases related to substance abuse is quite misleading. What the record does reveal is not that the number of substance-abuse related cases increased between 1983 and 1986—these remained constant at about nine to ten—but the total number of disciplinary cases decreased, thereby increasing the percentage of disciplinary cases suspected of being related to substance abuse. It is unclear from the record what the basis was for the City’s suspicions that drugs were involved in the disciplinary cases, but even if these suspicions were well founded, the total number equals about one-half of 1 percent of the City’s workforce. We have no way of knowing if these disciplinary problems were concentrated in some departments or were widespread. This is a weak reed indeed on which to lean a program of blanket preemployment drug testing.
As for the second part of the City’s showing—that drug testing will have an actual positive effect on the government agency’s productivity—the general evidence on this score is mixed. Because urinalysis detects metabolites from past drug use rather than current impairments, the connection *913between a positive test and on-the-job work performance is not self-evident. It is true that positive preemployment test results have been correlated with higher absenteeism and firing rates. (See Normand et al., Under the Influence: Drugs and the American Workforce (1994) pp. 220-223 (Drugs and the American Workforce).) But it does not follow that preemployment drug testing will yield a reduction in absenteeism or turnover, or an overall increase in productivity. While there is a correlation between positive drug test results and some indicia of poor performance, the evidence indicates that a very high percentage of those who score positive on drug tests nonetheless perform satisfactorily on the job. For example, Normand and Salyard’s study of preemployment testing in the United States Postal Service, which is probably the most comprehensive longitudinal study of its kind done to date, showed that after 3.3 years, about 24 percent of positive testers had been fired compared to 15 percent negative testers. (Drugs and the American Workforce, supra, at pp. 220-223.) This means that about 76 percent of those positive testers performed at least satisfactorily on the job. The loss of the 76 percent, and their replacement with applicants initially judged to be less qualified, will not necessarily lead to the efficiency improvement the employer seeks. Thus, as the authors of Drugs and the American Workforce caution: “The studies do not show that positive testers performed worse than those who would have replaced them had they been rejected on the basis of drug-test information.” (Drugs and the American Workforce, supra, at p. 228.)2
Moreover, the significance of drug testing declines as drug abuse decreases in the general population, which has in fact been the case since 1980. (Drugs and the American Workforce, supra, at p. 223.) A recent study of workers ages 18-49 shows a major decline in recent drug use from 16.5 percent in 1985 (around when the City program was formulated) to 8.1 percent in 1993. (Drug Use Among U.S. Workers, supra, p. 9.) To the extent that drug use declines, the capacity of drug testing to improve employee productivity likewise diminishes.
The significance of these statistics can most clearly be illustrated with numerical examples. In the United States Postal Service study cited above, the prevalence of positive tests among applicants was slightly less than 10 percent. Given a 10 percent prevalence, and the fact that approximately 25 percent of the applicants testing positive turned out to be unsatisfactory *914employees, then the drug tests, had the results been heeded, would have correctly screened out 2.5 percent (25% x 10%) of the applicant pool, and incorrectly screened out 7.5 percent. If the prevalence rate of positive testers drops to 5 percent, then the drug tests’ rate of screening out future unsatisfactory employees drops to 1.25 percent of the total applicant pool. If, as cited above in Drug Use Among U.S. Workers, the prevalence of drug use among most of those belonging to certain professional and clerical categories is lower than in the general population, between 2 and 4 percent, then drug testing will successfully screen out less than 1 percent of this applicant pool. Whether this will improve the productivity of those segments of the work force at all (given the number of potentially successful employees who will be screened out and replaced with those who might be inferior), this marginal improvement is a scant justification for overriding the substantial privacy interests of applicants offered employment.
As for the third prong of the showing, the City has not even addressed the question whether, in all cases, there is an adequate alternative mechanism for screening prospective employees. As noted, there may be occupations in which the nature of the background investigation and the “paper trail” left by the applicant produces a picture that renders drug testing superfluous, in the same sense that it is superfluous for selecting qualified promotional candidates. There are, moreover, many methods of preemployment selection, some relatively successful, that have been studied at length by organizational or industrial psychologists. (See generally, Dale, Successful Recruitment and Selection: A Practical Guide for Managers (1995) pp. 149-186). There is no showing in the record that would indicate universal drug testing results in a more productive workforce when used instead of, or in addition to, these more traditional employee selection techniques. On remand, the parties should have the opportunity to address this issue.
In sum, under the tests I propose, a government agency could justify preemployment drug testing either by demonstrating that drug impairment will compromise a compelling state interest, such as public safety, or that the government department has or likely will have a problem with drug abuse that will substantially impair its efficiency and effectiveness, that applicant testing will help remedy this problem, and that the agency lacks sufficient alternate means of applicant screening. With these alternative tests, we would be acknowledging that employers do sometimes have a greater need, as the lead opinion argues, to drug-test applicants than to test current employees, and thus would be giving employers greater latitude to do so, while nonetheless requiring the government employer to make a particularized demonstration of a special need to drug-test. Such a demonstration has been the hallmark of Fourth Amendment jurisprudence outside the criminal *915context. The City has failed to demonstrate that universal preemployment testing is necessary for the significant improvement of employee efficiency, or for the furtherance of any other substantial government interest. I would accordingly remand to permit the City to appropriately narrow its applicant testing program.
II
In addition to violating the Fourth Amendment to the United States Constitution, the City’s preemployment drug testing scheme contravenes the privacy clause of article I, section 1 of the California Constitution, which declares as an inalienable right “pursuing and obtaining . . . privacy.” Before discussing how the privacy clause applies to this case, I briefly review this court’s interpretation of the meaning of this constitutional provision.
In our first case interpreting the privacy clause, White v. Davis (1975) 13 Cal.3d 757 [120 Cal.Rptr. 94, 533 P.2d 222], we held, after reviewing the ballot arguments advocating its passage, that the clause “does not purport to prohibit all incursion into individual privacy but rather that any such intervention must be justified by a compelling interest.” (Id. at p. 775.) In White, we questioned whether there could be a compelling interest justifying a government program of surveillance of university classrooms purportedly for the purpose of uncovering dissident or subversive activity, and held that a cause of action was stated for illegal expenditure of public money on that program. (Id. at pp. 775-776.)
In Hill v. National Collegiate Athletic Assn., supra, 7 Cal.4th 1, a case in which I dissented, the majority rejected the “compelling interest” test articulated in White. In what can only be described as a fabrication out of whole cloth, the Hill court interpreted the privacy clause to create a right of action only if the following were alleged and proven: (1) a legally protected privacy interest; (2) a reasonable expectation of privacy; and (3) a serious invasion of the privacy interest amounting to “an egregious breach of the social norms underlying the privacy right.” (Hill, supra, 7 Cal.4th at pp. 35-37.) Once this threshold has been crossed, a privacy invasion may nonetheless be shown to be justified by a “competing interest” derived from “the legally authorized and socially beneficial activities of government and private entities,” the relative importance of which “is determined by their proximity to the central functions of a particular public or private enterprise.” (Id. at p. 38.) Plaintiffs can in turn contest a defense based on countervailing interest by “demonstrating the availability and use of protective measures, safeguards, and alternatives to the defendant’s conduct that would minimize the intrusion on privacy interests.” (Ibid.)
*916The lead opinion now declares, in essence, that the formidable threshold requirements originally set forth in Hill, particularly the requirement that a privacy infringement rise to the level of an “egregious breach of. . . social norms,” (Hill v. National Collegiate Athletic Assn., supra, 7 Cal.4th at p. 37) are no longer part of the state constitutional law of privacy. In its place the lead opinion would employ a balancing test similar to the one used under Fourth Amendment analysis, at least in the case of drug testing by government employers. Because there no longer appears to be support for the Hill test by a majority of this court, and therefore no stare decisis constraint to follow a privacy test with which I have disagreed from the outset, I offer in its place the test I originally proposed in my dissenting opinion in Hill.
In my dissent in Hill, I reviewed, in its full context, the right of privacy declared in article I, section 1 of the California Constitution. I set out the law relating to a plaintiff’s right of action thereunder as follows: “[T]he plaintiff must plead that he has a right of privacy and that it was interfered with by the defendant. The defendant may then plead, beyond simple denial, that any conduct on his part adversely affecting the right of privacy was justified by a compelling public need if it rose to the level of abridgment or that it was allowed as reasonable if it did not. The plaintiff must prove his right of privacy and the defendant’s interference therewith by shouldering the generally applicable burden of proof by a preponderance of the evidence [citation]. The defendant must prove under the same burden the justification or allowance of his conduct.” (Hill v. National Collegiate Athletic Assn., supra, 7 Cal.4th at pp. 85-86 (dis. opn. of Mosk, J.).)
In the present case, I conclude that plaintiff has properly pled and proved interference with the right of privacy. As discussed above, drug testing compromises informational privacy, one of the primary concerns of the proponents of the privacy amendment, by forcing those tested to reveal medications they are taking and other intimate medical information. In White v. Davis, supra, 13 Cal.3d at page 775, we identified as one of the principle “mischiefs” cited by the ballot arguments, the “overbroad collection and retention of unnecessary personal information by government and business interests.” I also note the ballot arguments specifically identified the intrusive gathering of information during the job application process as a type of abuse of privacy the privacy clause was designed to redress. “Each time we apply for a credit card or a life insurance policy, file a tax return, interview for a job, or get a drivers’ license, a dossier is opened and an informational profile is sketched. Modem technology is capable of monitoring, centralizing and computerizing this information which eliminates any possibility of individual privacy.” (Ballot Pamp., Proposed Stats, and Amends to Cal. Const, with arguments to voters, Gen. Elec. (Nov. 7, 1972) argument in *917favor of Prop. 11, p. 27, italics added.) Preemployment urine testing is a quintessential example of an invasion by modem technology of a realm of privacy previously undisturbed, in order to extract information that might be useful to the prospective employer. As such, preemployment dmg testing is precisely the kind of privacy invasion with which the privacy clause is concerned.
Moreover, as discussed above, the act of governmental monitoring of urination that occurs during a dmg test is also deeply invasive of individual privacy. This is, in itself, an independent reason for finding that dmg testing abridges protectible privacy interests under California Constitution article I, section 1. (See Hill, supra, 1 Cal.4th at p. 91 (dis. opn. of Mosk, J.).)
We turn then to whether there is a “compelling public need” for universal preemployment testing by the City. For reasons more fully articulated in the first part of this opinion, I do not find that the City has carried its burden of showing a compelling public need to test all applicants for employment or for promotion. The government does have a legitimate need to make its operations more efficient and effective, and to save the expenditure of tax revenues in its delivery of needed services. But it may not accomplish these general goals through any means, no matter how intrusive of privacy, and no matter how little the means chosen relate to the legitimate government need. As explained in the first part of this opinion, the City has not adequately justified the testing of applicants to all non-safety-sensitive positions. The City has failed to make an adequate showing that the testing of all applicants to employment, and not some narrower group, is necessary to substantially improve the workings of city government.
Nor do I find, under the arguably less stringent ad hoc balancing test employed by the lead opinion, that across-the-board applicant testing passes muster under the privacy clause. For reasons explained in the first part of this opinion, the testing of all applicants offered employment with the City cannot, on the record before us, survive a Fourth Amendment challenge. Neither can it survive a challenge under the privacy clause, which provides at least as much protection against government intrusions of privacy as the Fourth Amendment.
III.
In short, the City of Glendale has not carried its burden of showing that it has a compelling or substantial need to drug test each of the applicants for whom it has made job offers. In light of the fact that the law of preemployment testing was unclear at the time of trial of this case, I would remand to *918the City to determine which occupations merit preemployment testing. A job applicant could be subject to a preemployment test if (1) a compelling interest was at stake, as defined by the Skinner, Von Raab, and Vernonia, such as public safety, which would likely be compromised through drug impairment or (2) an interest in the significant improvement in worker productivity was at stake, provided it could be fairly demonstrated that drug abuse was a particular problem in the department to which the prospective employee was applying, and that drug testing could be shown to be a valid means of significantly abating the problem. As part of this determination, an assessment should be made of whether there were sufficient alternate means other than drug testing of evaluating applicants. Both the Fourth Amendment and the privacy clause of the state Constitution require the government to shoulder these evidentiary burdens in order to justify a suspicionless search that intrudes into some of the deepest recesses of an applicant’s private life.
I would accordingly partly affirm and partly reverse the decision of the Court of Appeal, and remand the case with directions to refashion the injunction against the City in a manner consistent with this opinion.
The chief case cited by the lead opinion in support of the categorical legal validity of such tests, Fowler v. New York City Dept, of Sanitation (S.D.N.Y. 1989) 704 F.Supp. 1264, was decided prior to Skinner. The Fowler court was guided in part by the mistaken view, soon to be repudiated in Skinner, that drug testing of government employees was not a “search” within the meaning of the Fourth Amendment. (Id. at p. 1271.)
Another researcher has reported no significant relationship between positive drug tests and various indicia of work performance in a hospital workplace, suggesting that the correlation-ship between drug use and job performance may vary according to occupation. (See Parish, Relation of the Pre-employment Drug Testing Result to Employment Status: A One-Year Follow Up (1989) 4 J. Gen. Internal Med. 44 [reporting no significant correlation between drug use and work performance within given occupational classifications].)