We granted leave to appeal in this case to determine the scope of an employer’s vicarious liability for quid pro quo sexual harassment affecting public services under Michigan’s Civil Rights Act (CRA).1 Specifically, we consider whether Wayne County and its sheriffs department may be held vicariously liable for a civil rights claim under MCL 37.2103(i) based on a criminal act of a deputy sheriff committed during working hours but plainly beyond the scope of his employment. We hold that defendants may not be held vicariously hable for quid pro quo *6sexual harassment affecting public services under traditional principles of respondeat superior. Accordingly, we reverse the Court of Appeals’ judgment and reinstate the circuit court’s order granting summary disposition in defendants’ favor.
I. FACTS AND PROCEDURAL HISTORY
In August 2001, Livingston County deputy sheriffs arrested plaintiff, Tara Katherine Hamed, on a warrant for unpaid child support. Because plaintiff also had outstanding warrants for probation violations in Wayne County, the Livingston County deputies later transferred plaintiff to the custody of Wayne County. Wayne County deputies transported plaintiff to the Wayne County jail. When plaintiff arrived at the jail, Deputy Reginald Johnson was the only officer on duty in the inmate registry area.2 While alone with plaintiff, Johnson subjected her to sexually charged comments and offers for better treatment in exchange for sexual favors. Plaintiff resisted these advances, but Johnson transferred plaintiff into an area of the jail not subject to surveillance cameras and sexually assaulted her. Shortly thereafter, a female deputy transported plaintiff to another part of the jail. After her release, plaintiff reported the incident to departmental authorities. The Wayne County Sheriffs Department terminated Johnson’s employment, and the state subsequently charged Johnson with criminal sexual conduct, of which he was ultimately convicted.3
*7In 2003, plaintiff filed a complaint against Johnson, Wayne County, the Wayne County Sheriffs Department, and the Wayne County Sheriff, among others, alleging various claims of gross negligence.4 In 2006, plaintiff moved to amend her complaint, adding civil rights claims of quid pro quo and hostile-environment sexual harassment pursuant to MCL 37.2103(i). Defendants then moved for summary disposition under MCR 2.116(C)(8) and (10), arguing that, under the CRA, jails are excluded from liability and, because defendants had no notice of Johnson’s sexually harassing conduct, they could not be vicariously liable for his actions.
The circuit court granted defendants summary disposition in two separate orders and dismissed all of plaintiffs civil rights claims. It concluded that plaintiffs hostile-environment claim failed because defendants had no prior notice that Johnson was a sexual predator. The circuit court also dismissed plaintiffs quid pro quo sexual harassment claim on the basis that defendants are not vicariously liable for the criminal acts of sheriffs department employees.5
Plaintiff then appealed the circuit court’s decision only with regard to her quid pro quo sexual harassment claim. The Court of Appeals reversed and applied this Court’s analysis in Champion v Nation Wide Security, Inc,6 to hold that “[e]mployers are vicariously liable for acts of quid pro quo sexual harassment committed by their employees when those employees use their super*8visory authority to perpetrate the harassment.”7 The Court of Appeals held that plaintiff had established a viable quid pro quo sexual harassment claim because “Johnson used his authority as a sheriffs deputy to exploit plaintiffs vulnerability .. . .”8 We granted leave to consider whether defendants may be held vicariously liable for quid pro quo sexual harassment affecting public services under MCL 37.2103(i).9
II. STANDARD OF REVIEW
We review de novo whether the Court of Appeals erred by reversing the circuit court’s grant of summary disposition.10 Whether defendants may be held vicariously liable for quid pro quo sexual harassment affecting a public service under the CRA is a question of law that we review de novo. 11 To the extent that defendants’ arguments require us to interpret the meaning of the CRA, our review is also de novo.12 When interpreting the meaning of a statute, we discern the Legislature’s intent by examining the language used.13 We read the statutory language in context and as a whole, considering the plain and ordinary meaning of every word.14 If the language is clear and unambiguous, then we apply the statute as written without judicial construction.15
*9III. analysis
A. QUID PRO QUO SEXUAL HARASSMENT UNDER THE CRA
The CRA recognizes that freedom from discrimination because of sex is a civil right.16 Accordingly, the act prohibits discrimination because of sex in employment, places of public accommodation, and public services.17 MCL 37.2103(i) broadly defines “discrimination because of sex” as follows:
Discrimination because of sex includes sexual harassment. Sexual harassment means unwelcome sexual advances, requests for sexual favors, and other verbal or physical conduct or communication of a sexual nature under the following conditions:
(i) Submission to the conduct or communication is made a term or condition either explicitly or implicitly to obtain employment, public accommodations or public services, education, or housing.
(ii) Submission to or rejection of the conduct or communication by an individual is used as a factor in decisions affecting the individual’s employment, public accommodations or public services, education, or housing.
(Hi) The conduct or communication has the purpose or effect of substantially interfering with an individual’s employment, public accommodations or public services, education, or housing, or creating an intimidating, hostile, or offensive employment, public accommodations, public services, educational, or housing environment. [Emphasis added.]
The first two subdivisions of MCL 37.2301(i) describe quid pro quo sexual harassment, while the third subdivision *10refers to hostile-environment sexual harassment.18
A plaintiff alleging quid pro quo sexual harassment affecting public services must show by a preponderance of the evidence (1) that he or she was subjected to any of the types of unwelcome sexual conduct or communication described in the statute and (2) that the public service provider or the public service provider’s agent made submission to the proscribed conduct a term or condition of obtaining public services or used the plaintiffs submission to or rejection of the proscribed conduct as a factor in a decision affecting his or her receipt of public services.19
When the harassment was committed by an agent and the plaintiff is pursuing a civil rights claim against the principal, as in this case, a court must always “determine the extent of the employer’s vicarious liability .. . .”20 We require this analysis because the CRA specifically incorporates common-law agency principles.21 Thus, if a defendant is not vicariously liable for the acts of its agent under traditional principles of respondeat superior, the plaintiffs claim under the CRA fails as a matter of law.
B. RESPONDEAT SUPERIOR
The doctrine of respondeat superior is well established in this state: An employer is generally liable for *11the torts its employees commit within the scope of their employment.22 It follows that “an employer is not liable for the torts . . . committed by an employee when those torts are beyond the scope of the employer’s business.”23 This Court has defined “within the scope of employment” to mean “ ‘engaged in the service of his master, or while about his master’s business.’ ”24 Independent action, intended solely to further the employee’s individual interests, cannot be fairly characterized as falling within the scope of employment.25 Although an act may be contrary to an employer’s instructions, liability will nonetheless attach if the employee accomplished the act in furtherance, or the interest, of the employer’s business.26
Here, there is no question that Johnson’s sexual assault of plaintiff was beyond the scope of his employment as a deputy sheriff. The sexual assault was an independent action accomplished solely in furtherance of Johnson’s own criminal interests. It cannot be said that any of the institutional defendants benefited in any way from Johnson’s criminal assault or his exercise of unlawful authority over plaintiff. In fact, Johnson’s behavior was expressly prohibited by defendants’ rules regarding treatment of detainees and defendants’ anti-discrimination policies, to say nothing of the criminal law. In short, there is no fair basis on which one could *12conclude that the sheriff or county themselves vicariously took part in the wrongful acts.
The general rule that an employer is not liable for acts of its employee outside the scope of its business, however, does not preclude vicarious liability in every instance. This Court has consistently recognized that an employer can be held liable for its employee’s conduct if “the employer ‘knew or should have known of [the] employee’s propensities and criminal record’ ” before that employee committed an intentional tort.27 This inquiry involves an analysis of whether an employer had (1) actual or constructive knowledge of prior similar conduct and (2) actual or constructive knowledge of the employee’s propensity to act in accordance with that conduct. Under this two-pronged approach, the conduct at issue may be so close in time to prior similar conduct that knowledge under the first prong gives rise to a valid inference that the conduct was foreseeable undér the second prong. Conversely, if an employee’s actions were temporally distant and the employee’s recent record suggested a change in character, foreseeability would not be established.28
We applied this principle in Brown v Brown, in which we held that the employer was not vicariously liable for a rape committed by its employee because, under the circumstances, the act was unforeseeable.29 There, the *13defendant’s employee repeatedly made sexually offensive comments to the plaintiff, a female coworker. The plaintiff reported the incidents to the defendant, yet it took no action, and the employee subsequently raped the plaintiff. In concluding that the employer was not vicariously liable, we noted that the employee had no prior criminal record and had never threatened to rape the plaintiff. We explained:
[An employer] cannot reasonably anticipate that an employee’s lewd, tasteless comments are an inevitable prelude to rape if those comments did not clearly and unmistakably threaten particular criminal activity that would have put a reasonable employer on notice of an imminent risk of harm to a specific victim. Comments of a sexual nature do not inexorably lead t,o criminal sexual conduct any more than an exasperated, angry comment inexorably results in a violent criminal assault.[30]
In summary, we have consistently held that an employer’s liability for the criminal acts of its employees is limited to those acts it can reasonably foresee or reasonably should have foreseen. This is because we should not expect employers to assume that their employees will disobey the law. Criminal conduct is inherently arbitrary and highly unpredictable. As we noted in Brown, even law enforcement agencies, which are trained in detecting and preventing crime, cannot pre*14diet the occurrence of criminal acts.31 Contrary to plaintiffs argument, our caselaw governing the imposition of vicarious liability on an employer requires more than simply the exercise of some form of authority by an employee. Thus, it would be anomalous to adopt a rule requiring employers that provide public services to protect against the criminal actions of their employees absent some indicia of foreseeability. Rather, foreseeability is a necessary element for imposing liability, and, as we recently stated in Brown, we decline to “transform the test of foreseeability into an ‘avoidability’ test that would merely judge in hindsight whether the harm could have been avoided.”32
Michigan’s well-established rules governing respondeat superior are further justified by the societal burden that imposing liability for unforeseen criminal *15actions would create. Not only would holding employers vicariously liable for such acts be unfair, but doing so would attempt to further an impossible end by requiring employers to prevent harms they cannot anticipate, which are, in essence, unpreventable. The result would be the implementation of burdensome and impractical regulations meant to oversee employee conduct. Yet because such measures are sure to fail given that criminal conduct by its nature cannot be anticipated or foreseen, employers would essentially become insurers responsible for recompensing victims for the criminal acts of their employees. The harm of adopting such a policy would also extend to potential employees with less than impeccable personal backgrounds, who would encounter barriers to employment because employers, out of an abundance of caution, would be less willing to employ these individuals out of fear that any prior indiscretion could be used in a lawsuit to impute knowledge to the employer that it did not have.33
Applying the foreseeability analysis in this case dictates the conclusion that defendants are not legally responsible for Johnson’s criminal acts. The majority of complaints against Johnson during his employment with defendants *16involved his failure to obey work-related policies, such as failure to report a change of home address, or unsatisfactory work performance, for example, temporarily leaving his work station while on duty. Some of the grievances filed against Johnson reflected more serious behavior, such as using a police vehicle without authorization to deliver baby formula to his home, allegedly making threatening calls to his landlord after receiving an eviction notice, and engaging in a physical altercation with a male inmate after an exchange of words.34 Viewed in the light most favorable to plaintiff, this past misconduct put defendants on notice of Johnson’s irresponsible and aggressive tendencies, which, at most, demonstrates that defendants were aware that Johnson had a propensity to disobey work-related protocol and engage in aggressive behavior when provoked. Defendants had no actual or constructive knowledge of prior similar criminal sexual misconduct. Even the incident of aggression did not put defendants on reasonable notice that Johnson would sexually assault an inmate; violent actions do not inevitably lead to acts of criminal sexual conduct.35 Because Johnson’s prior misconduct was not similar to the violent sexual assault he perpetrated against plaintiff, we hold that defendants may not be held vicariously liable for quid pro quo sexual harassment based on Johnson’s unforeseeable criminal act under traditional principles of respondeat superior.36
*17C. CHAMPION V NATION WIDE SECURITY, INC, AND ITS PROGENY
Plaintiff urges us to ignore these traditional common-law principles and extend the reasoning of this Court’s decision in Champion, which referred to the Second Restatement of Agency’s “aided-by-agency” exception to the rule of respondeat superior.37 We reject this argument because, for reasons we will explain, Champion wrongly applied this respondeat superior exception to the CRA.
In Champion, this Court addressed, as a matter of first impression, whether an employer could be held vicariously liable for quid pro quo sexual harassment under the CRA. In that case, the plaintiff worked as a security guard, and her immediate supervisor scheduled her work, trained her, oversaw her performance, and was responsible for disciplining her. During a *18weekend shift, the supervisor, who had been making sexually suggestive comments to the plaintiff, led her to a remote area of the building, locked her in a room, and demanded sex. When the plaintiff refused, the supervisor forcibly raped her. The plaintiff sued her employer for quid pro quo sexual harassment under the CRA. The defendant argued that the supervisor was not acting as its agent when he raped the plaintiff because it had not authorized the rape.38
The Champion Court rejected the defendant’s argument, reasoning that “under defendant’s construction, an employer could avoid liability simply by showing that it did not authorize the sexually offensive conduct.”39 The Court indicated that the defendant’s “construction of agency principles [was] far too narrow” and briefly cited in support the aided-by-agency exception articulated in § 219(2)(d) of the Second Restatement of Agency.40 The Court further stated that
[the defendant’s view] fails to recognize that when an employer gives its supervisors certain authority over other employees, it must also accept responsibility to remedy the harm caused by the supervisors’ unlawful exercise of that authority. From his scheduling decisions that allowed him *19to work alone with [the plaintiff] to his ordering of her into a remote part of the building, [the supervisor] used his supervisory power to put [the plaintiff] in the vulnerable position that led to her rape. In fact, there is little doubt that [the supervisor] would have been unable to rape [the plaintiff] but for his exercise of supervisory authority.[41]
Citing multiple federal cases, the Champion Court held that “an employer [is] strictly liable where the supervisor accomplishes the rape through the exercise of his supervisory power over the victim.”42 The Court justified its holding on the basis that “employers rarely, if ever, authorize such conduct, [and consequently] employees would no longer have a remedy for quid pro quo sexual harassment.”43
Four years later, this Court again considered a quid pro quo sexual harassment claim in Chambers v Trettco, Inc.44 There, a supervisor subjected the plaintiff to sexually offensive conduct. After enduring this conduct for four days, the plaintiff reported the incidents to another supervisor and ultimately sued her employer for hostile-environment and quid pro quo sexual harassment. A jury returned a verdict in the plaintiffs favor, and the Court of Appeals affirmed. The Court of Appeals in Chambers referred to federal caselaw that applied the federal Civil Rights Act45 to hold that employers are vicariously liable when a supervisor victimizes a subordinate by creating a hostile work environment.
*20This Court granted leave to consider whether principles derived from federal caselaw should apply to claims brought under Michigan’s CRA. We held that courts considering claims under Michigan’s CRA must adhere to Michigan precedent and the language of the CRA.46 We clarified the law regarding sexual harassment in employment under the Michigan CRA and recognized that the “statute expressly addresses an employer’s vicarious liability for sexual harassment committed by its employees by defining ‘employer’ to include both the employer and the employer’s agents.”47 Using this definition, we determined that the Michigan CRA specifically incorporates common-law principles of respondeat superior and that “whether analyzing quid pro quo harassment or hostile environment harassment, the question is always whether it can be fairly said that the employer committed the violation — either directly or through an agent.”48
After our decisions in Champion and Chambers, this Court considered the doctrine of respondeat superior generally in Zsigo v Hurley Med Ctr.49 Although Zsigo did not involve a civil rights claim, the plaintiff sought to hold the defendant-employer vicariously liable for various intentional tort claims using the reasoning in Champion and the aided-by-agency exception to the doctrine of respondeat superior. The underlying facts involved a sexual assault perpetrated by the defen*21dant’s employee against the plaintiff, who had been admitted as a patient in the defendant hospital. The plaintiff reported the incident and subsequently sued the hospital on the basis of the employee’s actions.
We rejected the plaintiffs theory of vicarious liability and any notion that Michigan common law recognized the aided-by-agency exception or that this Court had adopted it in Champion.50 With regard to Champion’s reference to the aided-by-agency exception, we explained that Champion did not adopt the aided-by-agency exception, but referred to it “only in passing and on the basis of the very distinct facts of that civil rights matter.”51 We further explained that Champion applied only in the context of quid pro quo sexual harassment under MCL 37.2103(i) and, in such instances, “the sexual assault must be ‘accomplished through the use of the supervisor’s managerial powers.’ ”52 We ultimately rejected the plaintiffs theory of liability because it would have subjected employers to strict liability for unforeseen acts occurring outside the scope of an employee’s employment.53 Accordingly, the Zsigo Court declined to adopt the aided-by-agency exception and limited its applicability to the specific facts of the civil rights claim in Champion.
D. CHAMPION WAS WRONGLY DECIDED
Because Zsigo involved intentional tort claims, it did not provide an opportunity to address the validity of Champion in the civil rights context.54 Zsigo required us *22to consider whether Michigan common law recognized the aided-by-agency exception, given the intentional tort claims at issue. The present matter now places Champion’s continued validity squarely before us, and we conclude that Champion cannot be reconciled with Chambers, Zsigo, or the CRA itself.
First, we note that Champion’s holding was contrary to the plain language of the CRA. As we explained in Chambers, the CRA specifically incorporates common-law agency principles in its definition of “employer.”55 Michigan’s common-law agency principles, however, do not include the aided-by-agency exception,56 and the Legislature did not modify the common law by including the aided-by-agency exception in the CRA.57 The Champion Court failed to recognize this clear intent. Rather, like the dissenting justices here, the Champion Court reasoned that the remedial purpose of the civil rights law justified holding the defendant employer vicariously liable for the acts of its employee, based on an apprehension that adherence to traditional agency principles would completely foreclose employer liability for quid pro quo sexual harassment claims.58
*23Aside from failing to give effect to the Legislature’s intent, this reasoning is flawed for two additional reasons. First, it wrongly elevates the CRA’s general remedial purpose above its plain language. Such reasoning is contrary to the cornerstone of statutory interpretation, which is the rule that the plain language used is the best indicator of the Legislature’s intent.59 Second, the policy concern at the heart of Champion is fundamentally flawed because it was premised on an unfounded fear. Application of traditional agency principles does not foreclose employers from vicarious liability in the context of quid pro quo sexual harassment claims. An employer may still be liable for and act of quid pro quo sexual harassment that was committed within the scope of employment or for a foreseeable act that was committed outside the scope of employment.60 Thus, liability may certainly attach if there is sufficient cause to impute the employee’s or agent’s acts to the employer because the employer knew of the employee’s propensity to commit the type of act involved.
The Champion Court compounded its erroneous holding by relying on federal caselaw.61 Unlike the federal civil rights act, the Michigan CRA specifically *24incorporates Michigan common-law agency principles. Hence, unlike federal courts applying the federal civil rights act, Michigan courts applying the Michigan CRA are bound by this state’s common-law agency principles. Because federal courts are not so bound, their reasoning in this context is often inapposite given that the language of the CRA must guide our decisions. For this reason, the Michigan Legislature’s choice to incorporate agency principles into the CRA forecloses reliance on federal cases when determining whether a defendant will be vicariously liable under the CRA.62
Finally, we note that Champion is contrary to both prior and subsequent caselaw. Before Champion, this Court had never held that an employer could be vicariously liable for the unforeseeable criminal acts of its employees. Subsequent caselaw attempted to limit Champion's applicability, but that caselaw merely demonstrated Champion's dubious validity. Chambers recognized that the CRA incorporates common-law agency principles, and Zsigo made it clear that the aided-by-agency exception is not a part of this state’s common law. Thus, contrary to the mandates of Chambers and Zsigo, Champion requires the application of an exception to respondeat superior in the context of quid pro quo sexual harassment claims that is not a part of this state’s common law.63 Because Champion requires a result contrary to prior and subsequent caselaw and *25contrary to the language of the CRA, it is clear that Champion is not consistent with Michigan law. Rather, when considered in the context of our jurisprudence, Champion stands as an isolated aberration that relies not on the plain language of the act, but purely on policy considerations.
E. STARE DECISIS
Our inquiry does not end simply because we have concluded that Champion was wrongly decided. Rather, we must determine whether overruling Champion is the most appropriate course of action. This is a decision that we do not undertake lightly and will make “only. . . after careful consideration of the effect of stare decisis.”64 However, we are also mindful that we are under no obligation to let stand an erroneous decision in the interest of stability and continuity.65 We consider a multifactored test when determining whether to overrule precedent. The first question is whether the decision at issue was wrongly decided.66 Having already addressed this question, we must now consider whether Champion “defies ‘practical workability’ ” and “whether reliance interests would work an undue hardship . . . ,”67 These factors weigh in favor of overruling Champion.
First, despite our attempt in Zsigo to limit Champion to claims involving quid pro quo sexual harassment *26affecting employment, the present matter demonstrates that it is not possible to limit Champion in this respect. No meaningful distinction can be drawn between the facts in Champion and those in the present matter. Both Johnson and the supervisor in Champion were able to commit the rapes through their positions of authority over their victims. In both cases, the employers’ agents had discretionary control over their victims by virtue of their positions: the supervisor in Champion was able to dictate the victim’s schedule and order her to certain parts of the building, and Johnson had the authority to constrain plaintiffs freedom and to move her to certain parts of the jail. Certainly factual distinctions exist between Champion and the present case. Johnson was not plaintiffs supervisor in an employment context, and he could not have made plaintiff come to the building where he worked, unlike the supervisor in Champion. Yet these dissimilarities do not detract from the fact that Johnson would not have been able to commit the sexual assault but for his position of authority over plaintiff, much like the supervisor in Champion.
Indeed, Champion’s distortive impact, which is manifested when a plaintiff attempts to circumvent traditional rules of respondeat superior or otherwise attempts to avoid governmental immunity by framing a claim under the CRA, is apparent in lower court decisions of this state and further demonstrates Champion’s unworkability.68 This is because there is no way to *27effectively limit the rule announced in Champion, despite our prior attempt to do so. The reasoning on which Champion justified its holding is applicable not only to every quid pro quo sexual harassment case in which a plaintiff pursues a theory of vicarious liability — regardless of whether the discriminatory conduct affected employment, public services, or accommodations — but also to intentional tort claims in which a plaintiff seeks to hold an employer vicariously liable. Under Champion, it will always be “foreseeable” that employees who possess some authority by virtue of the employment relationship will abuse the power with which they have been vested when they commit, as here, a criminal act against another in the workplace.
Second, with regard to reliance interests, we cannot conclude that Champion “has become so embedded, so accepted, so fundamental, to everyone’s expectations” that overruling it would upset any real-world reliance interests.69 For there to be reliance, knowledge of a decision “must be of the sort that causes a person or entity to attempt to conform his conduct to a certain norm before the triggering event.”70 There is no indication that plaintiff or defendants relied on Champion by conforming their conduct before the underlying event— and, given the nature of the rule in Champion, it is unclear what form such reliance could have taken. It would be illogical to conclude that defendants condoned the sexual assault because of Champion, given that Champion would have imposed vicarious liability for the unforeseeable criminal acts of defendants’ agent. Nor *28would it be reasonable to suggest that plaintiff altered her conduct in rebanee on Champion. We simply fail to see any possible way defendants and plaintiff could assert rebanee on Champion.
Further, when the decision at issue involves statutory law, the best indicator of society’s knowledge of the law, and what society reasonably relies on, is the language of the statute itself.71 As we have explained, nothing in the language of the CRA eviscerates common-law rules of respondeat superior or otherwise engrafts the aided-by-agency exception into the statute. Accordingly, a decision to overrule Champion would not create any real-world dislocations.
Finally, further justification for overruling Champion can be found in the adverse practical consequences that would result from extending the case to the present matter. As we explained in Zsigo, “it is difficult to conceive of an instance when the [aided-by-agency] exception would not apply because an employee, by virtue of his or her employment relationship with the employer[,] is always ‘aided in accomplishing’ the tort.”72 Such an all-encompassing exception would apply equally to public-service cases.73 Consequently, adoption of the aided-by-agency exception would effec*29tively abolish the doctrine of respondeat superior in quid pro quo civil rights cases affecting public services and would result in the imposition of strict liability on governmental entities. In short, the exception would swallow the rule. Contrary to the current requirements for imposing vicarious liability, if the exception were adopted, a plaintiff would merely have to allege quid pro quo harassment and show that he or she was the victim of an intentional act by an employee in a particular custodial environment. Providers of public services would be liable for the unforeseeable criminal acts of their employees as long as claimants could couch their claims under the CRA, and the dangers of such a broad basis for seemingly unlimited strict liability, discussed earlier in this opinion, would become realities. Such a standard would apply to a wide range of public-service providers whose employees interact regularly with recipients of public services, including teachers, correctional and probation officers, physicians, nurses, and firefighters, to name a few.74 Because public entities cannot increase prices or otherwise alter business practices to absorb the increased risk of liability, a governmental agency’s only option may be to cut funding or curtail beneficial public programs. In justifying our decision to overrule Champion on this basis, we do not *30downplay the heinous nature of the crime that plaintiff suffered. However, permitting liability against defendants under these circumstances would impose too great a burden on public-service providers and on society in general, which is clearly contrary to the Legislature’s intent.75
We therefore conclude that Champion was wrongly decided and that overruling it would not interfere with legitimate reliance interests. We overrule Champion because it is inconsistent with longstanding Michigan law that employers, including public-service providers, are not vicariously liable for quid pro quo sexual harassment on the basis of the unforeseeable criminal acts of their employees.76
E RESPONSE TO THE DISSENTS
We disagree with the dissenting justices regarding whether Champion was correctly decided and should be overruled. Although the dissenting justices concede that Champion was unprecedented, they adhere to Champion’s reasoning to conclude that the exception to common-law agency principles is necessary to give effect to the broad purpose of the CRA and the Legislature’s intent in enacting it. Yet the dissenting justices’ conclusion that Champion was correctly decided for this reason ignores the fundamental flaws inherent in Champion. Notably, the dissenting opinions, like Champion, do not cite any language from the CRA to support *31this view, even though a statute’s language is the best indicator of the Legislature’s intent. Instead, the dissenting justices rely on caselaw describing the CRA as “remedial,” just as Champion did, for the proposition that “the exception to common-law agency principles established in Champion [is] necessary to give effect to the broad purpose of the CRA.. . .”77 Apparently, this “necessity” is based on the dissenting justices’ concern, as was the concern in Champion, that without the exception, discriminatory conduct would not be eradicated and the purpose of the CRA would be defeated.78 This fear vastly overstates the effect of our decision because, as we have explained, employers and public-service providers will still be vicariously liable for sexual harassment under traditional and longstanding principles of respondeat superior. In short, the dissenting justices’ reliance on Champion itself for the proposition that Champion was correctly decided lacks merit for reasons we have already explained.79
*32The dissenting justices compound their erroneous reasoning by wrongly interpreting subsequent opinions of this Court as confirming that Champion was correctly decided and as explicitly confirming that Champion adopted an exception “very similar to the aided-by-agency exception.”80 Contrary to the dissenting justices’ view, Zsigo did not expressly confirm Champion in this regard, and Chambers did not expressly hold that Champion is a valid part of Michigan’s common law, both of which the dissenting justices suggest.81 Further, although the dissenting justices acknowledge that the CRA incorporates common-law agency principles, they then ignore the explicit and unambiguous holding in Zsigo, namely that this Court has never recognized the aided-by-agency exception, or a similar rule, as part of this state’s common law. Despite Zsigo’s unambiguous holding, the dissenting justices continue to declare that Champion should be applied in sexual discrimination cases because the exception can be “narrowly tailored.”82 The Zsigo majority rejected any notion that the exception had such *33boundaries, which demonstrates that Zsigo does not support Champion in this regard. Thus, it is the dissenting justices who seek to aggressively expand the law of this state, while our holding merely reaffirms and applies traditional common-law rules that have always governed in Michigan.
Not surprisingly, using the faulty premise that Champion’s reasoning is correct, the dissenting justices advocate a straightforward application of Champion. This approach ignores an irreconcilable tension in our law. Although Champion and this case are similarly framed civil rights cases involving allegations of quid pro quo sexual harassment, the conflicting dispositions in the courts below demonstrate the tension between the multiple precedents of this Court at issue in this case. The circuit court below relied on Zsigo to grant summary disposition to defendants, recognizing that Zsigo established “a very clear bright line rule” that an employer is not liable when an employee unforeseeably acts outside the scope of his or her employment, as was the case here. The Court of Appeals reversed, relying instead on Champion, which had never been applied outside the employment context, for the proposition that a public-service provider may be vicariously hable when its employee uses his or her “authority over a subordinate as a means of subjecting that subordinate to abusive and unlawful conduct.” Thus, in this case, we are presented with conflicting principles: those of the traditional common-law rule that have guided Michigan law for more than a century as articulated in Zsigo and those underlying the rule of Champion, which inexplicably departed from the requirements that have always been held as necessary to impose respondeat superior liability. The existence of these conflicting precedents and principles cries out for clarity and compels our decision to overrule Champion.
*34Further, we disagree with the main dissent’s view that principles of stare decisis do not support overruling Champion. The main dissent applies a stare decisis test set forth in Petersen v Magna Corp83 that is not the law of this state. Because a majority of this Court did not adopt that test, and a majority of justices have agreed to the rule articulated in Robinson v Detroit,84 the test in Robinson governs this analysis. Nevertheless, overruling Champion is the right result, regardless of which test is applied.
The most basic error in the main dissent’s stare decisis analysis is its misunderstanding of why Champion is unworkable. The dissent posits that the aided-by-agency exception is “narrowly tailored” because it applies only when an agency relationship aided a supervisor in committing a wrongful act.85 According to the dissent, the exception does not apply when an agency relationship merely provided a supervisor an opportunity to accomplish a wrong. This interpretation is nothing more than a semantic exercise that demonstrates the capricious nature of Champion: An employment relationship will always provide a supervisory employee an opportunity to commit a wrong, but when does that opportunity become an “aid”? Similarly, in the public-services context, a citizen’s interaction with an employee administering public services will always arise during the administration of those services while the employee is exercising his or her authority; when are public-service employees “aided” and when are they not “aided” while exercising their authority? There is no meaningful demarcation.86 Continued *35adherence to Champion would require jurors and judges to determine vicarious liability according to their subjective whims. For this same reason, the dissent’s view that Champion provides “important guidance to trial courts” is simply wrong.87
Finally, we find unpersuasive the main dissent’s reliance on decisions from other jurisdictions that have applied the aided-by-agency exception in the context of their civil rights laws. If liability is to be imposed under Michigan law on an employer for sexual harassment committed by its employee, that liability must be mandated by the Michigan CRA.88 The aided-by-agency exception in the context of civil rights cases is not so well accepted and “nearly unanimous” as the main dissent appears to claim.89 Most states have not recognized the aided-by-agency exception in civil rights cases and, at least with respect to the jurisprudence of this Court, application of the aided-by-agency exception remains an aberration.
IV CONCLUSION
Michigan law has never imposed liability on an employer for the unforeseeable criminal actions of its *36employees, except in Champion. Nor has Michigan common law incorporated an exception based on an aided-by-agency theory of liability. Accordingly, we conclude that a provider of a public service may not be held vicariously liable for quid pro quo sexual harassment affecting public services on the basis of unforeseeable criminal acts that its employee committed outside the scope of employment. Because Champion is inconsistent with our holding and with Michigan’s common and statutory law, we overrule Champion. We reverse the Court of Appeals’ judgment and reinstate the circuit court’s order granting summary disposition in favor of defendants.
YOUNG, C.J., and MARKMAN and ZAHRA, JJ., concurred with Mary Beth Kelly, J.MCL 37.2101 et seq.
Wayne County jail regulations require that a female officer be in attendance when female inmates are present. The officers who transported plaintiff to the jail informed a supervisor that Johnson was the only deputy on duty. The supervisor advised the officers to leave plaintiff with Johnson.
See MCL 750.520c(k).
The only remaining defendants are Wayne County and the Wayne County Sheriffs Department. Plaintiff received a default judgment against Johnson; Johnson, while a defendant in plaintiffs action, is not a party to this appeal. Thus, for the purposes of this opinion, our references to “defendants” encompass only the institutional defendants.
See Zsigo v Hurley Med Ctr, 475 Mich 215; 716 NW2d 220 (2006).
Champion v Nation Wide Security, Inc, 450 Mich 702; 545 NW2d 596 (1996).
Hamed v Wayne Co, 284 Mich App 681, 693; 775 NW2d 1 (2009).
Id.
Hamed v Wayne Co, 486 Mich 996 (2010).
Maiden v Rozwood, 461 Mich 109, 118; 597 NW2d 817 (1999).
See Veenstra v Washtenaw Country Club, 466 Mich 155, 159; 645 NW2d 643 (2002).
Id.
Danse Corp v Madison Hts, 466 Mich 175, 181-182; 644 NW2d 721 (2002).
Herman v Berrien Co, 481 Mich 352, 366; 750 NW2d 570 (2008).
Danse Corp, 466 Mich at 182.
MCL 37.2102(1).
MCL 37.2202 (employment); MCL 37.2302 (public accommodations and public services). For purposes of this opinion, we assume, without deciding, that the Wayne County jail is a “public service” as defined by MCL 37.2301(b).
Chambers v Trettco, Inc, 463 Mich 297, 310; 614 NW2d 910 (2000).
See id. (stating the test for quid pro quo sexual harassment in the employment context). For purposes of our analysis, we also assume, without deciding, that plaintiff can establish the elements of quid pro quo sexual harassment affecting public services.
Id. at 311.
Id. We reached this conclusion in Chambers because MCL 37.2201(a) “expressly defines ‘employer’ to include agents,” thereby incorporating common-law agency principles into the act. Chambers, 463 Mich at 311.
See, e.g., Zsigo, 475 Mich at 221; Bradley v Stevens, 329 Mich 556, 562; 46 NW2d 382 (1951); Martin v Jones, 302 Mich 355, 358; 4 NW2d 686 (1942); Davidson v Chinese Republic Restaurant Co, 201 Mich 389, 396; 167 NW 967 (1918).
Zsigo, 475 Mich at 221.
Barnes v Mitchell, 341 Mich 7, 13; 67 NW2d 208 (1954), quoting Riley v Roach, 168 Mich 294, 307; 134 NW 14 (1912).
2 Restatement Agency, 3d, § 7.07, p 201.
See Barnes, 341 Mich at 13-16 (examining cases discussing scope of employment).
McClements v Ford Motor Co, 473 Mich 373, 381; 702 NW2d 166 (2005), quoting Hersh v Kentfield Builders, Inc, 385 Mich 410, 412; 189 NW2d 286 (1971) (quotation marks omitted) (emphasis added).
This analysis does not, as the dissenting justices assert, abandon prior caselaw to hold that “an employee’s conduct is only foreseeable to an employer if the employee had recently committed the precise conduct at issue.” Post at 56 n 20. This criticism mischaracterizes the inquiry that must he undertaken, which has its roots in well-established caselaw. See McClements, 473 Mich at 381; Hersh, 385 Mich at 412.
Brown v Brown, 478 Mich 545, 554-555; 739 NW2d 313 (2007).
Id. at 555. Brown did not, as the dissenting justices state, “conclude[] that an employee’s prior violent criminal acts are generally sufficient to put a defendant on notice of the employee’s propensity to commit similar violent acts . . . Post at 57. Rather, Brown made clear that knowledge of prior violent acts potentially provides an employer notice of an employee’s violent propensities. Brown, 478 Mich at 561. The dissenting justices attempt to broaden the holding in Brown to justify their position that defendants’ knowledge of Johnson’s dissimilar prior violent act suffices to create a question of fact regarding foreseeability.
Id. at 554, citing MacDonald v PKT, Inc, 464 Mich 322, 335; 628 NW2d 33 (2001).
Brown, 478 Mich at 556. An “avoidability test” is the type of test the dissenting justices favor. In their view, defendants’ policy prohibiting male deputies from being alone with female inmates demonstrates, in itself, that the sexual assault was preventable and foreseeable. However, reliance on the policy alone to impose liability obliterates any real foreseeability requirement; an employer’s policy is irrelevant to assessing what the employer knows with respect to a specific employee. The consequence would be imposition of vicarious liability every time an employee disobeys the employer’s policy, regardless of whether the act was unforeseeable under the actual circumstances.
Rather, as we have explained, a defendant’s specific knowledge of past misconduct and propensity to act in conformity with such conduct must be the focus of a foreseeability analysis. This analysis, which the dissenting justices term a “newly imposed foreseeability analysis,’’post at 58, merely recognizes that foreseeability has always been the touchstone for when vicarious liability will be imposed. The criticism by the main dissent is not surprising, given that Justice Cavanagh has previously expressed support for what effectively amounts to the imposition of strict liability in lieu of a foreseeability analysis. See Brown, 478 Mich at 570-580 (Cavanagh, J., dissenting); Anderson v Pine Knob Ski Resort, Inc, 469 Mich 20, 30-35; 664 NW2d 756 (2003) (Cavanagh, J., dissenting).
For a catalogue of some of the difficult questions that would confront an employer operating under the dissenting justices’ rule, see Brown, 478 Mich at 566-570 (Maekman, J., concurring). “The rule proposed by the dissent, and the unanswered questions arising from that rule, would create confusion and uncertainty among employers throughout this state .. . .” Id. at 566. And employers would not be the only ones to suffer; employees would suffer as well because, were the dissenting justices’ rule to become law, what rational employer would ever hire anybody with any history of problems in his or her background? “Why would any rational employer expose itself to the vagaries of litigation-by-hindsight ... where it fails to predict unpredictable behavior if this could all be avoided by simply firing [or failing to hire] every odd or rude or quirky employee?” Id. at 569-570. The rule the dissenting justices propose would result in those with imperfect criminal histories, or even merely a history of arrests, becoming increasingly unemployable.
The dissenting justices misrepresent the seriousness of Johnson’s past conduct, stating that he “had a specific history of violent and abusive behavior toward inmates.” Post at 56. In fact, Johnson had engaged in a single physical altercation with a male inmate in 1988, 13 years before the sexual assault in this case. Unlike the circumstances here, Johnson did not initiate the altercation with the male inmate.
Cf. Brown, 478 Mich at 555.
The dissenting justices dismiss our foreseeability analysis, concluding that Johnson’s past violent act and sexual assault of plaintiff more *17than a decade later is sufficient to create a question of fact concerning defendants’ vicarious liability. According to the dissenting justices, defendants had notice that Johnson would sexually assault a female inmate because Johnson, 13 years earlier, had engaged in a physical altercation initiated by a male inmate. In their view, any past violent conduct may create a jury-submissible question regarding foreseeability. Moreover, their contention that the question of foreseeability should have been submitted to the jury because this matter is substantially similar to Hersh is unavailing. In that case, the defendant’s employee had a criminal conviction for similar prior conduct 10 years earlier, which the employer knew about, thus establishing a factual question regarding whether the employee had “vicious propensities.” Hersh, 385 Mich at 413, 415. As we have indicated, evidence of dissimilar violent conduct is-not reasonably predictive of violent sexual conduct. Nor can it be said that a reasonable employer could genuinely have foreseen Johnson’s sexual assault of plaintiff on the basis of a single instance of entirely dissimilar violent conduct that arose as a result of provocation by a male inmate 13 years earlier. The dissenting justices fail to recognize that the temporal distance and the dissimilarity between past conduct and the conduct at issue make it unreasonable to conclude that an employer could have foreseen that Johnson would engage in quid pro quo sexual harassment or commit a criminal sexual assault.
Champion, 450 Mich at 712 n 6.
Id. at 705-707.
Id. at 713.
Id. at 712 n 6. 1 Restatement Agency, 2d, § 219(2), p 481, provides that
[a] master is not subject to liability for the torts of his servants acting outside the scope of their employment, unless:
(d) the servant purported to act or to speak on behalf of the principal and there was reliance upon apparent authority, or he was aided in accomplishing the tort by the existence of the agency relation.
Champion, 450 Mich at 712 (citation omitted).
Id. at 713-714. Since our decision in Champion, the drafters of the Third Restatement of Agency have excluded the aided-by-agency exception included in the Second Restatement of Agency.
Id. at 713.
Chambers, 463 Mich 297.
42 USC 2000e et seq.
Chambers, 463 Mich at 316.
Id. at 310.
Id. at 312.
Zsigo, 475 Mich 215. This Court did consider a single intervening civil rights case concerning quid pro quo sexual harassment, but the resolution of that case did not require application of the doctrine of respondeat superior because the plaintiff failed to establish that sexual harassment had occurred. See Corley v Detroit Bd of Ed, 470 Mich 274; 681 NW2d 342 (2004).
Zsigo, 475 Mich at 221-224.
Id. at 223-224.
Id. at 224 n 19, quoting Champion, 450 Mich at 704.
Zsigo, 475 Mich at 227.
Although Justice Young recognized that the exception to respondeat superior that Champion created was “hard to square. . . with any conventional notion of agency, and ... stands as an isolated, inexplicable *22exception” to this Court’s agency jurisprudence, the Court was constrained to merely limit the application of Champion given that no civil rights claim was at issue in Zsigo. Id. at 232 (Young, J., concurring).
See Chambers, 463 Mich at 310-311.
See Zsigo, 475 Mich at 223-224.
The common law remains in force until it is affirmatively modified. Const 1963, art 3, § 7. The Legislature is presumed to know the common law, and any abrogation of the common law must be explicit. Dawe v Dr Reuven Bar-Levav & Assoc, PC, 485 Mich 20, 28; 780 NW2d 272 (2010).
Champion, 450 Mich at 713. While the dissenting justices are correct that the purpose of a statute may be a relevant consideration, post at 38 n 3, what they fail to recognize is that this is correct only in so far as the purpose of the statute is derived from the actual language of the statute. See Mich Ed Ass’n v Secretary of State (On Rehearing), 489 Mich 194, 202-203; 801 NW2d 35 (2011) (“The clear purpose of [the statute], as *23reflected, in its language, is to mandate the separation of the government from politics in order to maintain governmental neutrality in elections, preserve fair democratic processes, and prevent taxpayer funds from being used to subsidize partisan political activities.”) (emphasis added).
See Danse Corp, 466 Mich at 181-182.
Application of traditional respondeat superior principles also does not foreclose other avenues of legal recourse, including pursuit of direct criminal and civil liability against the perpetrator.
See Champion, 450 Mich at 712 n 8, citing Karibian v Columbia Univ, 14 F3d 773 (CA 2, 1994), Kauffman v Allied Signal, Inc, 970 F2d 178 (CA 6, 1992), Horn v Duke Homes, 755 F2d 599 (CA 7, 1985), Craig v Y & Y Snacks, Inc, 721 F2d 77 (CA 3, 1983), Katz v Dole, 709 F2d 251 (CA 4, 1983), Henson v City of Dundee, 682 F2d 897 (CA 11, 1982), and Miller v Bank of America, 600 F2d 211 (CA 9, 1979).
Chambers, 463 Mich at 315-316.
Significantly, the drafters of the Third Restatement of Agency chose to exclude the aided-by-agency exception, thereby implicitly recognizing that the exception is not consistent with generally accepted common-law agency principles. While the dissenting justices dismiss this authority as unpersuasive, they ignore the fact that the aided-by-agency exception is not a “widely accepted exception to the general rules of agency.” Post at 43-44. Only a few jurisdictions have adopted the exception wholesale into their common law, such that it applies to a typical tort claim. And, as we *25have explained, Michigan explicitly rejected the exception in Zsigo because it is inconsistent with fundamental principles of Michigan common law.
Haynie v Dep’t of State Police, 468 Mich 302, 314; 664 NW2d 129 (2003).
Robinson v Detroit, 462 Mich 439, 464; 613 NW2d 307 (2000).
Id.
Id.
The Court of Appeals decision in this case is one example, see Harried, 284 Mich App 681, as is the plaintiffs attempt in Zsigo to hold the employer vicariously liable for its employee’s unforeseeable criminal act. See also Diamond v Witherspoon, 265 Mich App 673, 690-691; 696 NW2d 770 (2005) (presuming strict vicarious liability and rejecting governmental immunity in the context of a quid pro quo sexual harassment case under the CRA when a police officer subjected detained *27individuals to sexual conduct), and Salinas v Genesys Health Sys, 263 Mich App 315; 688 NW2d 112 (2004) (rejecting the plaintiffs attempt to hold the employer vicariously liable in tort under Champion for an unforeseeable criminal act).
Robinson, 462 Mich at 466.
Id. at 467.
See id. (stating that “it is well to recall in discussing reliance, when dealing with an area of the law that is statutory..., that it is to the words of the statute itself that a citizen first looks for guidance in directing his actions”).
Zsigo, 475 Mich at 226.
In Zsigo, this Court noted that to adopt a rule contrary to that of the traditional common law would mean that the rule “applies to a broad range of employees whose duties grant them unique access to and authority over others, such as ... correctional officers,” which “could virtually ‘eviscerate[] the general scope of employment rule.’ ” Zsigo, 475 Mich at 230, quoting Poe v Forrest, 2004 VT 37, ¶ 59; 176 Vt 476, 505; 853 A2d 48 (2004) (Skoglund, J., dissenting) (some quotation marks omitted; emphasis added; alteration in original).
Artful pleading would also allow a plaintiff to avoid governmental immunity under the governmental tort liability act (GTLA), MCL 691.1401 et seq. A school district, for example, could not be vicariously liable in tort for a teacher’s sexual molestation of a student because the GTLA would bar the claim. However, if the plaintiff styled its claim as a CRA action, the school district could be vicariously liable under a theory of quid pro quo sexual harassment affecting public services. Plaintiffs preferred approach, under which public-service providers would be strictly liable for precisely the same conduct as that for which they would typically be immune, is inherently inconsistent with the Legislature’s intent. If the Legislature had intended such a result, it should have clearly abrogated the common-law rule for purposes of the CRA.
See Brown, 478 Mich at 557-558, and the discussion at page 15 of this opinion.
Because we have decided that defendants cannot be held vicariously liable for Johnson’s criminal act under the CRA, we need not address defendants’ alternative arguments that the Wayne County jail is not a “public service” within the meaning of the CRA or that the circuit court improperly permitted plaintiff to amend the complaint.
Post at 38 (Cavanagh, J., dissenting). Likewise, the author of the other dissent cites no specific language and provides no analysis in support of her accusation that our decision is somehow “contrary to the rule of law” or “results in the dismantling of the [CRA].” Post at 60 (Hathaway, J., dissenting). Rather, as we have explained at great length, our decision honors both our common-law tradition and the language of the CRA and is consistent with that statute’s purpose.
The dissenting justices’ concern in this regard is related to their failure to recognize plaintiffs claim for what it really is: an attempt to hold a governmental entity liable for an employee’s criminal action and unforeseeable intentional tort.
Indeed, the dissenting justices concede that the “bulk of [their] analysis” relies only on Champion, post at 42 n 7, which we have explained at length is a decision not supported in Michigan’s law generally, and thereby effectively admit that no other binding Michigan law supports their conclusion other than Champion. Simply because Champion was a unanimous decision decided 15 years ago does not mean it was correctly decided or that its reasoning is correct today. Not only do the dissenting justices ignore the plain language of the CRA, they also *32ignore subsequent changes in the law that have exposed the flaws in Champion’s reasoning. See pages 24-25 and note 63 of this opinion.
Post at 41.
In fact, contrary to the dissenting justices’ position, Zsigo did undermine the primary rationale of Champion. The Zsigo Court did not entirely “dispatch the exception created in Champion,” post at 41 n 6, because Zsigo was not a civil rights case.
Post at 42. The dissenting justices’ position that Champion was correctly decided on this basis relies primarily on the dissenting opinion in Zsigo, which is not binding precedent. The dissenting justices in this case disregard this criticism, asserting that the dissenting opinion in Zsigo is an “example” of Champion’s workability. Post at 42 n 7. Yet, the rationale of the dissent in Zsigo lacks any persuasive value because, like the main dissent here, the Zsigo dissent repeatedly advocated adopting the aided-by-agency exception and the Vermont Supreme Court’s decision in Doe. This Court has already explicitly rejected both as inconsistent with Michigan law.
Petersen v Magna Corp, 484 Mich 300, 313-320; 773 NW2d 564 (2009) (opinion by Marilyn Kelly, C.J.).
Robinson, 462 Mich 439.
Post at 42-43 (CAVANAGH, J., dissenting).
The main dissent counters that application of the Vermont Supreme Court’s three-pronged test would amount to "a narrowly tailored ap*35proach to applying the aided-by-agency exception . ...” Post at 48. Yet this test suffers from the same deficiencies we have already described because it makes no valid distinction between a mere “opportunity” and an “aid.”
Post at 48-49. The main dissent also misconstrues our citation of Diamond and Salinas. Those cases do not demonstrate Champion’s workability. Rather, they are examples of “artful pleading” in which the plaintiffs sought to circumvent traditional rules of respondeat superior by framing their claims under the CRA. See note 74 of this opinion.
Notably, the main dissent ignores the mandate of Chambers to consider the language of the Michigan CRA of paramount importance when interpreting the Michigan CRA as opposed to any guidance that federal caselaw may provide. Chambers, 463 Mich at 313-314.
Post at 49 (quotation marks and citation omitted).