opinion of the Court:
11 James Eric Verde was convicted of sexual abuse of a child. At trial, the court admitted evidence of two prior uncharged sexual assaults by Verde. On appeal to the court of appeals, Verde challenged the admis*676sibility of that evidence under Utah Rule of Evidence 404(b), asserting either that it was not offered for a non-character purpose or that its probative value was substantially outweighed by a risk of unfair prejudice. The court of appeals deemed the evidence admissible for two non-character purposes and affirmed. We reverse Verde's conviction and remand for a new trial, but leave the door open for the State to offer its evidence on grounds different from those adopted by the court of appeals or the trial court.
I
[ 2 In 2005, Verde was charged with sexually abusing NH., a twelve-year-old boy. The charge was based on an incident that occurred in the summer of 20083, when Verde allegedly put his hand down N.H.'s pants and fondled his genitalia. Verde pled not guilty.
3 Prior to trial, the State filed a motion in limine, asking the trial court to allow testimony from three men who claimed that Verde had sexually assaulted them when they were eighteen years old. The State contended that the testimony was admissible under rule 404(b) for the non-character purposes of demonstrating Verde's "knowledge, intent, plan, modus operandi and/or absence of mistake or accident." Verde challenged the admissibility of the evidence on the ground that it was not relevant to any controverted issues in a manner untethered to his character. The trial court granted the State's motion as to two of the witnesses, concluding that the evidence was admissible to prove Verde's specific intent. The court noted that the evidence could also be admitted to prove "a pattern of behavior," and that Verde "prepared and planned to meet minor males with a motive of enticing them into sexual relationships."
14 At trial, NH. testified that he met Verde in the fall of 2001 when Verde moved into N.H.'s neighborhood. According to N.H., Verde took him to a carnival on the day they met, and the two spent extensive time together thereafter-with NH. playing video games or basketball at Verde's home, riding Verde's ATVs, or working in Verde's yard for pay.
T5 N.H. further testified that Verde sexually abused him in the summer of 2008 when he was at Verde's home. According to N.H.'s testimony, Verde sat by N.H. on the couch and put "his hand down [N.H.'s] pants" and "touched [his] penis and testicles." N.H. said that he told Verde to stop, and that Verde said something like "don't be cool" and then moved to a chair. In December 2004, N.H. reported these events to his mother.
T6 The State also presented evidence at trial that Verde had engaged in sexual misconduct with two eighteen-year-old males in 2002 and 2004.1 The first witness, J.T.S., testified that Verde first approached him when he was fifteen years old and working as a grocery store bagger. J.T.S. claimed that Verde initiated a conversation with him, gave him a pair of sunglasses, and invited him to play basketball. J.T.S. did not see Verde again until he was eighteen years old. At that time, Verde expressed interest in a car J.T.S. was selling and insisted that J.T.S. come to his house so he could test drive the car.
T7 J.T.S. testified that he went to Verde's home that evening. When J.T.S. realized that Verde was not interested in purchasing the car, J.T.S. attempted to leave. Verde then pulled on J.T.S.s leg and refused to let him go. According to J.T.S., Verde then rubbed J.T.S.'s leg, unbuttoned his jeans, and groped his genitals. J.T.S. testified that he tried to stop Verde "many times," but that he responded with force, frightening J.T.S. He immediately reported the incident to the police and his parents, but no charges were filed.
18 M.A. testified to a similar incident. According to M.A., he met Verde at the gym in 2002 when he was eighteen years old. Verde allegedly approached M.A. and invited him home, where Verde groped M.A.'s groin "close enough to his genitals to arouse him." MH. terminated this encounter and later *677reported the incident to police, again without charges ever being brought.
T9 After the State presented its case, Verde testified on his own behalf, denying that he ever sat next to N.H. on the couch or touched N.H. in a sexual manner. Verde presented witnesses who testified about N.H.'s lack of credibility, one saying that N.H. "pathologically lief{d]." Verde also testified that he never had any sexual contact with M.A. or J.T.S.
T 10 The jury found Verde guilty, and he appealed. In the court of appeals, Verde pressed his argument that the evidence of uncharged sexual misconduct should not have been admitted because it served no purpose other than to show that Verde's conduct conformed to a propensity to commit sexual crimes. State v. Verde, 2010 UT App 30, ' 15, 227 P.3d 840.
1 11 The court of appeals affirmed, holding that the 404(b) evidence was admissible to establish Verde's specific intent, or alternatively, to rebut Verde's theory that NH. fabricated his story. Id. 918, 19 n. 6. Although Verde never actually disputed intent, the court of appeals deemed the evidence admissible to establish Verde's specific intent, a required element of sexual abuse of a child, regardless of the nature of the case or Verde's defenses. Id. 118. The court based this holding on the so-called "not guilty rule," under which intent is per se controverted once a defendant pleads not guilty to a specific-intent crime. Id. In light of this holding, the court of appeals did not address the State's alternative argument that the trial court properly admitted the bad acts evidence for the additional purpose of proving Verde's pattern of conduct, preparation, or plan of enticing and exploiting teenage males. Yet the court did recognize "at least one additional ground for admitting the prior bad acts evidence." Id. 119 n. 6. Because Verde claimed that N.H. invented the alleged misconduct "after not being paid for catching a stray cat," the court held that prior bad acts evidence was admissible to rebut Verde's defense of fabrication. Id.
112 Judge McHugh concurred, opining that the "not guilty rule" should not be used as a substitute for a meaningful inquiry into the actual purpose and relevance of evidence offered under rule 404(b). Id. 138 (McHugh, J., concurring). In Judge McHugh's view, the mere fact that "a defendant pleads not guilty should not excuse the State from identifying the precise link between the bad acts evidence and a contested issue in the trial." Id. 144. Judge McHugh also acknowledged that under current court of appeals precedent, see State v. Bradley, 2002 UT App 348, 57 P.3d 11839, the 404(b) evidence could be admitted to rebut Verde's fabrication theory; but for that precedent, however, she would have reversed and remanded for a new trial. Verde, 2010 UT App 30, 148, 227 P.3d 840 (McHugh, J., concurring).
118 Our review of the court of appeals decision on certiorari is de novo. State v. Levin, 2006 UT 50, 115, 144 P.3d 1096. That said, "[the correctness of the court of appeals' decision turns, in part, on whether it accurately reviewed the trial court's decision under the appropriate standard of review." Id. A trial court's admission of prior bad acts evidence is reviewed for abuse of discretion, but the evidence "must be serupulously examined by trial judges in the proper exercise of that discretion." State v. Decorso, 1999 UT 57, 118, 993 P.2d 837. Applying these standards, we reverse the court of appeals' holding that the 404(b) evidence was admissible to prove Verde's intent and remand for a new trial, leaving open the possibility that the trial court could determine that the State's evidence is admissible under the "doctrine of chances" as proof that N.H. did not fabricate Verde's act of abuse.
II
T 14 Our analysis must begin with the text of the governing rules of evidence. The principal rule in play here is 404(b), which states:
Evidence of other crimes, wrongs or acts is not admissible to prove the character of a person in order to show action in conformity therewith. It may, however, be admissible for other purposes, such as proof of motive, opportunity, intent, preparation, *678plan, knowledge, identity, or absence of mistake or accident.
Urax R. Evip. 404(b) (2005).2
115 Under this rule, the admissibility of prior misconduct evidence depends on its avowed purpose. When such evidence is offered to suggest action in conformity with a person's alleged bad character, it is inadmissible under the rule. When past misconduct evidence is offered for any other purpose, on the other hand, it is admissible. The rule lists examples of proper purposes-to establish motive, opportunity, intent, ete.-but the list is illustrative and not exclusive. So long as the evidence is not aimed at suggesting action in conformity with bad character, it is admissible under rule 404(b).
16 That much is clear. The difficulty in applying this simple rule, however, springs from the fact that evidence of prior bad acts often will yield dual inferences-and thus betray both a permissible purpose and an improper one. Thus, evidence of a person's past misconduct may plausibly be aimed at establishing motive or intent, but that same evidence may realistically be expected to convey a simultaneous inference that the person behaved improperly in the past and might be likely to do so again in the future. That's what makes many rule 404(b) questions so difficult: Evidence of prior misconduct often presents a jury with both a proper and an improper inference, and it won't always be easy for the court to differentiate the two inferences or to limit the impact of the evidence to the purpose permitted under the rule.
117 Yet the language and structure of rule 404(b) require the court to make such distinctions. Fidelity to the rule requires a threshold determination of whether proffered evidence of prior misconduct is aimed at proper or improper purposes. See State v. Nelson-Waggoner, 2000 UT 59, 6 P.3d 1120. If such evidence is really aimed at establishing a "defendant's propensity to commit crime," it should be excluded despite a proffered (but unpersuasive) legitimate purpose. See State v. Decorso, 1999 UT 57, ¶¶ 21-25, 993 P.2d 837. And even if 404(b) evidence appears to have a dual purpose-to be aimed at both proper and improper inferences-it may nonetheless be excluded under rule 403 if its "probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence." Utax R. Evin. 408 (2005).
Thus, when prior misconduct evidence is presented under rule 404(b), the court should carefully consider whether it is genuinely being offered for a proper, non-character purpose, or whether it might actually be aimed at sustaining an improper inference of action in conformity with a person's bad character. And even if the evidence may sustain both proper and improper inferences under rule 404(b), the court should balance the two against each other under rule 403, excluding the bad acts evidence if its tendency to sustain a proper inference is outweighed by its propensity for an improper inference or for jury confusion about its real purpose. Such weighing is essential to preserve the integrity of rule 404(b). Without it, evidence of past misconduct could routinely be allowed to sustain an inference of action in conformity with bad character-so long as the proponent of the evidence could proffer a plausible companion inference that does not contravene the rule.
119 A district court's decision to admit evidence under rule 404(b) is entitled to some deference. But such a decision can withstand our review only if the evidence falls within the bounds marked by the legal standards set forth in the rules of evidence. And the question in this case is whether the *679State's evidence fell within the permissible range.
' 20 The State seeks to defend the admissibility of the 404(b) evidence offered in Verde's trial on three grounds: (a) that it was offered to establish Verde's specific intent, (b) that it demonstrated his plan to engage in criminal activity, and (c) that it was presented to rebut Verde's charge of fabrication. We reject the first two grounds and accordingly reverse and remand for a new trial, as these were the grounds on which the evidence was admitted at trial. As to the third ground, we acknowledge that evidence of Verde's prior misconduct could potentially be admitted to rebut a charge of fabrication, but decline to affirm on that basis in the absence of any indication in the record that the district court was asked to conduct the careful weighing required to sustain the admission of such evidence in a case like this one. Thus, on this issue, we leave it to the district court on remand to decide on the admissibility of evidence of Verde's prior misconduct under the "doctrine of chances" as explained below.
A
{21 The first ground put forward by the State for admitting evidence of Verde's past misconduct is its alleged relevance to his state of mind in committing the specific intent crime of child sex abuse. This ground was embraced by the district court and affirmed by the court of appeals, which concluded that a not-guilty plea necessarily puts the question of intent at issue, opening the door to "evidence of other offenses to establish the element of intent even if the defendant has not contested his or her mental state." State v. Verde, 2010 UT App 30, 118, 227 P.3d 840 (internal quotation marks omitted). Because Verde entered a plea of not guilty, the prosecution was required to prove "not only that he 'touch[ed] the ... genitalia of a child' but also that he did so 'with intent to cause substantial emotional or bodily pain to any person or with the intent to arouse or gratify the sexual desire of any person." " Id. (alterations in original) (quoting Ut&aHK Coph § 76-5-404.1(2)). And because the prior bad acts evidence purportedly was relevant to Verde's intent, the court of appeals upheld its admissibility under the "not guilty rule." See id.
122 We find the premises of the not-guilty rule unpersuasive and accordingly reject it as a principle of Utah law. A not-guilty plea technically puts every element of a crime at issue. But the technical relevance of evidence of a defendant's intent is not enough to justify the admissibility of evidence of prior bad acts purportedly aimed at establishing intent under rule 404(b).3 Fidelity to the integrity of the rule requires a careful evaluation of the true-and predominant-purpose of any evidence proffered under rule 404(b). Thus, if proof of intent is merely a ruse, and the real effect of prior misconduct evidence is to suggest a defendant's action in conformity with alleged bad character, the ruse is insufficient and the evidence should not be admitted. That may be because the court determines that the true purpose of the evidence is an impermissible one under rule 404(b). Or it could be on the ground that any permissible purpose is outweighed by its propensity for an improper inference or for jury confusion about its real purpose.
123 Either way, the admissibility of prior misconduct evidence cannot be sustained under rule 404(b) on the mere basis of a defendant's not-guilty plea. As Judge McHugh noted in her concurring opinion below, the "not guilty rule" is an undisciplined substitute for careful analysis under rule 404(b).4 "[MJechanical reliance" on the rule does not "reflect the sort of critical evaluation" of the prosecution's purpose for offer*680ing 404(b) evidence that is required by the language and structure of the rule.5
124 We accordingly repudiate it. Instead of relying reflexively on the broad implications of a not-guilty plea, courts in Utah should evaluate the true purpose of evidence of past misconduct, determining at the threshold whether the evidence is presented for a proper purpose, or only for the purpose of suggesting an improper inference of action in conformity with alleged bad character. And even if the court finds both legitimate and improper purposes for such evidence, the court should still weigh the proper and improper uses of 404(b) evidence and exclude it under rule 4083 where the terms of that rule so require. Applying these standards, we conclude that the evidence of Verde's prior misconduct was not properly admissible to establish his specific intent-despite the fact that his not-guilty plea technically put his intent at issue.
1 25 First, we find it difficult to characterize the true purpose of the 404(b) evidence introduced at trial as permissibly aimed at establishing Verde's intent. Aside from his not guilty plea, Verde did not contest intent at trial. See id. % 17-18. Instead, his primary defense was that he never touched N.H.'s genitalia and that N.H. fabricated his testimony of that actus reus. Id. In fact, Verde offered to stipulate to his intent in his response to the State's motion in limine, asserting that "if the jury concludes that the touching of N.H. occurred, defendant is willing to stipulate that the defendant did it with the intent to arouse or gratify the sexual desire of any person." And, as even the State admits, intent is inferable from proof that Verde groped N.H.'s genitalia. In these cireumstances, it's hard to imagine a jury that would conclude that Verde committed the actus reus but with an innocent intent.
T26 Where intent is uncontested and readily inferable from other evidence, 404(b) evidence is largely tangential and duplica-tive.6 It is accordingly difficult to characterize its purpose as properly aimed at establishing intent. In context, it seems much more likely that it was aimed at sustaining an impermissible inference that Verde acted in conformity with the bad character suggested by his prior bad acts.
127 The State resists this conclusion on the ground that Verde made no "enforceable" stipulation of intent and could have reneged on his pretrial offer. But it was the State that refused Verde's offer to formally stipulate intent, and at oral argument in this court the State could identify no legitimate reason for rejecting that offer. That failure is telling. It reinforces the conclusion that the prosecution's true purpose in offering evidence of Verde's prior misconduct was to invite the jury to make the kind of character inference that is proscribed under rule 404(b).
128 In so concluding, we do not imply that the prosecution bears an obligation to accept a defendant's offer to stipulate.7 To the contrary, the prosecution retains wide discretion to reject such an offer, which it might legitimately do, for example, to preserve the right to present evidence with broad "narrative value" beyond the establishment of particular elements of a crime. As the United States Supreme Court put it, "a naked proposition in a courtroom may be no match for the robust evidence that would be used to prove it. People who hear a story interrupted by gaps of abstraction may be puzzled at the missing chapters" in the prosecution's case. Old Chief v. United States, 519 U.S. 172, 189, 117 S.Ct. 644, 136 L.Ed.2d 574 (1997). Evidence may thus be appropri*681ately aimed at completing the "missing chapters" in the prosecution's ease, and the prose-ecution may refuse an offer to stipulate to preserve its ability to tell a complete story.
129 Sometimes, however, the evidence in question has no legitimate narrative value, as in cases where it is not plausibly linked to any charged conduct. That will often be the case for evidence of prior misconduct. Such evidence may be worse than immaterial to a legitimate narrative. It may risk creating an alternative, illegitimate narrative-that the defendant has a reprehensible character, that he probably acted in conformity with it, and that he should be punished for his immoral character in any event.
T 30 Absent any legitimate explanation for the prosecution's rejection of the defendant's offer to stipulate, we view this rejection to reinforce the conclusion that the prosecution's purpose was not to tell a legitimate narrative to the jury but instead to present an improper one. So, while the state was free to reject Verde's offer to stipulate, it was not free to distance itself from the probative implications of that decision, which in our view thoroughly undermine the State's position on appeal.
131 Second, even if the past misconduct evidence in this case could plausibly be deemed to have been aimed at a legitimate purpose under rule 404(b), it would still fail under the balancing framework required under rule 403. Specifically, and for all the reasons detailed above, we conclude that any legitimate tendency the 404(b) evidence had to tell a narrative of Verde's specific intent was minimal at best. And we likewise conclude that any such legitimate purpose is far outweighed by the obvious, illegitimate one of suggesting action in conformity with bad character.
1 32 We accordingly conclude that the district court abused its discretion in admitting evidence of Verde's prior misconduct to establish his specific intent. That evidence was not plausibly aimed at a proper purpose, and in any event any such proper purpose was outweighed by an illegitimate effect.
B
183 The second ground put forward by the State for admitting evidence of Verde's prior bad acts is its alleged relevance in demonstrating his "plan" to "entice teenage males to be his friends with the motive of exploiting their trust for his sexual gratification."8 This basis was embraced by the district court, but not addressed by the court of appeals. We reject this as a ground for admitting evidence of past misconduct in this case, as the evidence presented at trial did not legitimately establish a "plan" but was instead effectively aimed at demonstrating mere propensity to act in conformity with bad character.
134 Under the classic formulation of the rule, prior misconduct evidence can demonstrate a "plan" only where the defendant's "preconceived plan ... encompasses all of the acts" in an overarching design. Davin P. LEonarp, THE New WicmorE: A TrEatiss on EvipENCE: EvineEncE or Otesr Misconpucr AND EvENts § 9.4.2 (2009). This standard requires that
all the erimes-both charged and uncharged-are the product of some prior, conscious resolve in the accused's mind. The accused formulates a single, overall grand design that encompasses both the charged and uncharged offenses. That design is overarching; all the crimes are integral components or portions of the *682same plan. Each crime is a step or stage in the execution of the plan. Each is a means to achieving the same goal.9
35 This type of plan evidence is admissible because it is based on the permissible inference that, regardless of character, a person who has formulated a plan is more likely to carry out the elements of the plan. Id. § 9.1.10 Such evidence is thus relevant without implicating a forbidden inference of action in conformity with immoral character.
36 We adopted this approach in State v. Featherson, 781 P.2d 424, 429 (Utah 1989), abrogated on other grounds by State v. Do-porto, 985 P.2d 484 (Utah 1997). There, the defendant was charged with sexual assault. In the trial on that offense, the court allowed evidence of prior uncharged sexual assaults on the ground that they demonstrated that the defendant committed each act under a common plan and thus had the requisite state of mind. Id. at 425, 429. We reversed, concluding that the evidence was inadmissible because it did not "qualify as links in a chain forming a common design." Id. at 429 (internal quotation marks omitted). And absent such a scheme or plan linking the prior acts and the charged offense, we held that the evidence proved "only a propensity, proclivity, predisposition or inclination to commit" sexual assault, rendering it inadmissible under rule 404(b). marks omitted). Id. (internal quetation
'I 37 In so holding, we cited favorably People v. Tassell, 36 Cal.3d 77, 201 Cal.Rptr. 567, 679 P.2d 1, 7-8 (1984), overruled by People v. Ewoldt, 7 Cal.4th 380, 27 Cal.Rptr.2d 646, 867 P.2d 757 (1994), in which the California Supreme Court adopted the narrow or classic rule for admissibility of plan evidence. See Featherson, 781 P.2d at 429. The Tassell court concluded that evidence of a common scheme is admissible only if it reveals a single conception or plot of which the charged and uncharged crimes are individual manifestations. Absent such a grand design, talk of common plan or scheme is really nothing but the bestowing of a respectable label on a disreputable basis for admissibility-the defendants disposition. Tassell, 201 Cal.Rptr. 567, 679 P.2d at 5 (citation and internal quotation marks omitted).
1 38 Tassell was subsequently overruled by Ewoldt, 27 Cal.Rptr.2d 646, 867 P.2d at 759. In Ewoldt, the court abandoned the requirement that plan evidence reveal a single, continuing conception or plot. Id. at 767. Instead, mere similarity between uncharged and charged acts was deemed sufficient for admissibility as evidence of a plan "such a concurrence of common features that the various acts are naturally to be explained as caused by a general plan of which they are the individual manifestations." Id. at 770. (internal quotation marks omitted). Thus, in California, episodes of misconduct unlinked by any overarching plan are admitted as evidence of a "general plan" and thus that the defendant acted in conformity with that plan.
1139 The State heralds the Ewoldt rule as the more liberal or modern view and invites us to adopt it.11 We decline to do so and instead confirm our holding in Featherson. Evidence of a general plan to commit crimes with common features is perilously close to evidence of a general disposition to commit crime. Any difference between the two concepts is extremely subtle and quite likely to be lost on a jury.12 Moreover, while re*683peated commission of a crime is only weak evidence of a plan, it gives rise to a strong- and impermissible-propensity inference. A jury presented with evidence of repetitive criminal acts under the theory that the defendant had a general plan to commit similar crimes may find the forbidden inference hard to resist.13
1 40 Under the Utah standard adopted in Featherson and further clarified and confirmed here, the evidence of Verde's prior misconduct was not relevant to establish a "plan" to commit similar crimes, and its admissibility accordingly cannot be affirmed on that basis. There is no suggestion of a prior, conscious resolve on Verde's part to formulate an overarching grand design encompassing both the charged and uncharged offenses. In fact, the "victims" of Verde's past encounters were not even minors like NH. was. They were adults when Verde is alleged to have sexually assaulted them. And of course an attempt to entice an adult into a sexual relationship is hardly equivalent to the sexual enticement of a child. The age difference is highly significant, It undermines any suggestion of a plan by Verde to engage in the criminal conduct he is accused of here.
T41 While the evidence of Verde's prior misconduct only weakly suggests a plan, it would strongly suggest to the jury the likelihood that Verde may have acted in conformity with the bad character implied by his prior acts.14 Under the cireumstances, this would pose an untenable risk of confusing jurors as to the real purpose for which the evidence was offered and of swaying jurors to base a verdict on the strong inference of action in conformity with bad character. If we adopted the Ewoldt approach of routinely admitting powerful propensity evidence under the guise of anemic evidence of a plan, we would "threaten[ ] to undo the essential protection that the character evidence prohibition is designed to afford an accused." 15 This we decline to do.
I 42 Under Ewoldt, evidence that a defendant had committed three D.U.Ls on the same road (perhaps even in the same car, with the same type of alcohol, on the same day of the week) presumably could be offered to prove the defendant had a plan to drive while intoxicated. And evidence that a defendant frequently possessed controlled substances could be offered to prove a plan to use illegal drugs. The undue prejudice inherent in proof of this sort of general plan will nearly always outweigh any legitimate probative value, and we accordingly repudiate it.
143 In support of its contrary view, the State cites rule 404(c) of the Utah Rules of Evidence as an example of the "liberal or modern" rule set forth in Ewoldt. That provision, as the State notes, expressly endorses the admission of evidence of certain prior bad acts similar to the erime in question-those involving "acts of child molestation" in a "case in which a defendant is accused of child molestation." Utax R. Evin. 404(c)(1). That provision, however, only undermines the State's position. It does so by confirming that any liberalizing trend toward greater admissibility of prior bad acts evidence may be accomplished through express amendments to our rules of evidence, see FED. R.EvID. 418, 414, 415, an avenue that counsels against the distortion of the otherwise general rule against propensity inferences under rule 404(b). We accordingly adhere to the rule embraced by this court in Feather-son, 781 P.2d at 429; see supra 1134-37, a standard the State cannot satisfy here.
C
144 Lastly, the State contends that its 404(b) evidence was admissible to prove that Verde committed the actus reus in question by rebutting Verde's theory that N.H. fabricated his testimony of the sexual assault. The fabrication question was an issue at trial. *684In his opening statement, Verde's counsel asserted that N.H. was a "pathological liar" who had invented his account of Verde's sexual abuse. The State countered by pointing to evidence of Verde's prior sexual assaults, which in its view made it more likely that N.H.'s testimony was, in fact, truthful. On appeal, the court of appeals majority coneclud-ed that rebutting the defense of fabrication was an "additional ground for admitting the prior bad acts evidence," State v. Verde, 2010 UT App 30, 119 n. 6, 227 P.3d 840 (citing State v. Bradley, 2002 UT App 348, 57 P.3d 1139), a conclusion adopted by Judge McHugh in her separate concurrence, id. {483 (McHugh, J., concurring) ("[A] proper purpose for bad acts evidence is to rebut a defense of fabrication.").
1 45 In defending the admissibility of the prior misconduct evidence on this basis, the State reasons that "while it may be plausible that one victim might fabricate such charges, it is highly unlikely that three [victims] would independently fabricate" similar accounts of unwanted sexual contact. In response, Verde argues that uncharged misconduct evidence offered to rebut a claim of fabrication is inadmissible because it "qualifies as evidence of propensity."
T46 As a threshold matter, we acknowledge the theoretical possibility that evidence of prior misconduct could be admitted under rule 404(b) to establish commission of a criminal actus reus by rebutting a charge of fabrication. Because this argument was not presented by the State in Verde's trial, however, we reject it as a ground for affirmance. To provide guidance for the parties on remand and to explain our basis for reversing the court of appeals, we clarify the legal standards that govern in this area.
147 In some cireumstances, evidence of prior misconduct can be relevant under the so-called "doctrine of chances." This doctrine defines cireumstances where prior bad acts can properly be used to rebut a charge of fabrication. It is a theory of logical relevance that "rests on the objective improbability of the same rare misfortune befalling one individual over and over." 16 Under this analysis, the State suggests that evidence of past misconduct may "tend[ ] to corroborate on a probability theory" that a witness to a charged crime has not fabricated testimony, because it is "(unlikely ... that [several] independent witnesses would ... concoct similar accusations."
48 One court explained the thinking behind this theory as follows:
[SJuppose you lose your horse; you find it in the possession of A.; he asserts that he took the horse by mistake; but you find that about the same time he took horses belonging to several others; would not the fact that he took others about the same time be proper evidence to be considered in determining whether the particular taking was or not by mistake? The chances of mistake decrease in proportion as the alleged mistakes increase."17
A parallel explanation has been offered in terms more directly applicable here:
When one person claims rape, the unusual and abnormal element of lying by the complaining witness may be present. But when two (or more) persons tell similar stories, the chances are reduced that both are lying or that one is telling the truth and the other is coincidentally telling a similar false story.18
{49 This reasoning starts with the low baseline probability that a man would take a horse by mistake or that an innocent person would be falsely accused of sexual assault- or, to cite additional examples from actual cases, that a child would die in her sleep 19 or *685that a spouse would drown in the bathtub.20 The second step in the analysis considers the effect on these already low probabilities of additional, similar occurrences: As the number of improbable occurrences increases, the probability of coincidence decreases, and the likelihood that the defendant committed one or more of the actions increases.21 An innocent person may be falsely accused or suffer an unfortunate accident, but when several independent accusations arise or multiple similar "accidents" occur, the objective probability that the accused innocently suffered such unfortunate coincidences decreases.22 At some point, " '[tlhe fortuitous coincidence becomes too abnormal, bizarre, implausible, unusual or objectively improbable to be believed." State v. Johns, 801 Or. 585, 725 P.2d 312, 822-28 (1986) (quoting 8 Epwarp J. ImnwinkEurign, Misconpuct EviDENCE § 5:05 (1984)).23
150 Propensity inferences do not pollute this type of probability reasoning. "The question for the jury is not whether the defendant is the type of person" who, for example, "sets incendiary fires or murders his relatives. The question is whether it is objectively likely that so many fires or deaths could be attributable to natural cases. It is that objective unlikelihood that tends to prove human agency, causation, and design." 24 The inferences required follow this pattern:
-_ evidence of prior similar tragedies or accusations;
- an intermediate inference that the chance of multiple similar occurrences arising by coincidence is improbable; and
- a conclusion that one or some of the occurrences were not accidents or false accusations.25
151 Under this pattern, prior misconduct evidence may tend to prove that the defendant more likely played a role in the events at issue than that the events occurred coincidentally.26 And because the evidence tends to prove a relevant fact without relying on inferences from the defendant's character, the evidence is potentially admissible. True, there is a risk of an undue inference that the defendant committed each act because of the defendant's immoral character, but a permissible inference is also possible-under the inferential chain outlined above.
152 Many courts, in Utah and elsewhere, have employed this "doctrine of chances" reasoning to analyze the relevance of uncharged misconduct evidence when a defense of fabrication has been raised. In State v. Bradley, for example, our court of appeals reasoned that evidence of a prior, independent allegation of sexual assault decreased the probability that the charged sexual assault was fabricated, as the defendant claimed. 2002 UT App 848, 128, 57 P.3d 1139. In the court's view, the defendant's fabrication "theory [was] diminished by [the uncharged conduct evidence] because it is more difficult to believe that [two] mothers were motivated to, and were successful in, convincing their children to fabricate the allegations of sexual abuse." Id.
*68653 Probability reasoning is also the best understanding of our analysis in State v. Nelson-Waggoner, 2000 UT 59, 6 P.3d 1120. There we noted the similarities among the testimony of two women who alleged that the defendant had previously raped them and the testimony of the victim of the charged rape. Id. 125. And we concluded that the uncharged misconduct evidence was probative of whether the defendant engaged in forceful and nonconsensual sex with the victim because it "laid out a pattern of behavior." Id. While we did not explicitly refer to the doctrine of chances, the relevance of the evidence in that case was based on the low probability that multiple victims would independently accuse the defendant of similar assaults. Many other courts have adopted the doctrine in these and similar contexts.27
54 The court of appeals in this case affirmed the admissibility of evidence of Verde's prior misconduct on an alternative ground resting on a vague notion of this doctrine of chances. Without denominating the doctrine as such or elaborating on its elements, the court of appeals held that the evidence was admissible to rebut Verde's charge of fabrication. Verde, 2010 UT App 30, TT 19 n. 6, 43, 227 P.3d 840. The State urges the same result here, asserting that Verde's prior acts properly rebut his charge of fabrication because they are "completely independent of the witness to the charged crime and to each other."
155 We find the grounds put forward by the State and adopted by the court of appeals insufficient on the current record to affirm the admissibility of evidence of Verde's prior misconduct. A charge of fabrication is insufficient by itself to open the door to evidence of any and all prior bad acts. As with other questions arising under rule 404(b), care and precision are necessary to distinguish permissible and impermissible uses of evidence of prior bad acts, and to limit the factfinder's use of the evidence to the uses allowed by rule.
156 We accordingly reverse the court of appeals' decision on this issue and in so doing offer some clarifying limitations on the use of evidence to rebut a charge of fabrication to guide the parties and the district court on remand. The relevant limitations are found in the prevailing ease law on the doctrine of chances, which we adopt and explain in the paragraphs that follow.
{57 Under the doctrine of chances, evidence offered to prove actus reus must not be admitted absent satisfaction of four foundational requirements,28 which should be considered within the context of a rule 403 balancing analysis. First, materiality: The issue for which the uncharged misconduct evidence is offered "must be in bona fide dispute." 29 We have examined this requirement above and need not address it further here. See supra 1% 21-82.
158 Second, similarity: "Fach uncharged incident must be roughly similar to the charged crime." 30 The required similari*687ty here need not be as great as that necessary to prove identity under a "pattern" theory. But there must be some significant similarity between the charged and uncharged incidents to suggest a decreased likelihood of coincidence-and thus an increased probability that the defendant committed all such acts:
[The more similar, detailed, and distine-tive the various accusations, the greater is the likelihood that they are not the result of independent imaginative invention. It is less likely that two accusers would independently manufacture similar stories that are detailed and unusual than that they would coincidentally tell the same commonplace lie.31
159 Any prescription of a threshold of similarity for admitting similar accusations evidence is inevitably imprecise.32 But we can say that the similarities between the charged and uncharged incidents must be "sufficient to dispel any realistic possibility of independent invention." 33 All of the incidents must at least "fall into the same general category." 34
160 Third, independence: Where the prior uncharged conduct is an accusation of sexual assault, each accusation must be independent of the others. This is because "the probative value of similar accusations evidence rests on the improbability of chance repetition of the same event." 35 And the existence of collusion among various accusers would render ineffective the comparison with chance repetition.
T61 Fourth, frequency: The defendant must have been accused of the crime or suffered an unusual loss "more frequently than the typical person endures such losses accidentally." 36 It is this infrequency that justifies the probability analysis underlying the doctrine of chances: "The probability that any given individual who might be accused of rape or child abuse will be falsely accused of those crimes is low.... Given the infrequent occurrence of false rape and child abuse allegations relative to the entire eligible population, the probability that the same innocent person will be the object of multiple false accusations is extremely low." 37
1 62 Because the trial court is in a superior position to make an initial exercise of discretion to conduct the weighing called for under rules 404(b) and 408, we remand this case for a new trial. At the retrial of this matter, if the state chooses to pursue this theory, the district court should use the standards we have articulated to decide whether evidence of Verde's uncharged sexual assaults may be presented to the jury. Thus, the district court will have to weigh carefully the materiality of and the similarities and the differences between Mr. Verde's alleged advances to and sexual abuse of a twelve-year-old child and the alleged unwanted advances to and touching of two adults. And it will have to consider the independence and the frequency of such alleged acts. Though we have articulated standards to help the parties engage in these discussions on remand, we do so without opining on the admissibility of the prose-eution's prior misconduct evidence under the "doctrine of chances." We also emphasize that our opinion is not at all aimed at influencing the district court or at expressing a view on the ultimate viability of this theory on remand.
TH
T 63 We conclude that the court of appeals erred in affirming the admissibility of evidence of Verde's uncharged misconduct offered to prove his specific criminal intent, which was not a legitimately disputed issue at trial. We likewise hold that the State's evidence was not admissible to prove that *688Verde acted in conformity with a plan to entice and abuse young men, as the evidence did not demonstrate that Verde entertained a preconceived, overarching design to commit the acts in question. We accordingly reverse and remand for a new trial, leaving open the possibility that the district court may deem the prior misconduct evidence admissible under the "doctrine of chances," as that theory is explained above.
Justice LEE authored the opinion of the Court, in which Chief Justice DURRANT, Associate Chief Justice NEHRING, Justice DURHAM, and Justice PARRISH joined.. The State also sought to elicit testimony from a fourth alleged victim, D.J. But the court concluded that the probative value of D.J.'s duplicative testimony was substantially outweighed by risk of unfair prejudice and was thus inadmissible under rule 403.
. We quote the version of our evidentiary rules in effect when Verde was tried. Though 2011 amendments altered the language of some rules-including rules 403 and 404-these changes were intended only "to make [the rules] more easily understood and to make style and terminology consistent throughout the rules." See R. Evin. 404(b) 2011 advisory committee note. So our analysis here presumably will hold under the newly amended rules, although our discussion is addressed on its face to the rules as they stood at the time of trial.
. See Tanberg v. Sholtis, 401 F.3d 1151, 1167-68 (10th Cir.2005) (other uses of excessive force not admissible to prove intent in civil rights action where officer did not dispute that he intended to use force); Thompson v. United States, 546 A.2d 414, 422-23 (D.C.1988) (whether intent is a contested issue "depends, not on the statutory definition of the offense, but on the circumstances of the case and on the nature of the defense[s]").
. State v. Verde, 2010 UT App 30, ¶¶ 27, 29, 227 P.3d 840. (McHugh, J., concurring).
. Id. % 40 (internal quotation marks omitted).
. See State v. Shickles, 760 P.2d 291, 295 (Utah 1988) (instructing courts to examine the prosecution's "need for the evidence" when engaging in rule 403 balancing (internal quotation marks omitted)), abrogated on other grounds by State v. Doporto, 935 P.2d 484 (Utah 1997).
. See State v. Florez, 777 P.2d 452, 456 (Utah 1989) (affirming the trial court's refusal to order the prosecution to accept a stipulation because [tlhe State has the right to prove every essential element of a crime in the most convincing manner within the bounds of the rules of evidence and fair play"); State v. Bishop, 753 P.2d 439, 475 (Utah 1988) ("As a general rule, a party may not preclude his adversary's offer of proof by admission or stipulation."), overruled on other grounds by State v. Menzies, 889 P.2d 393 (Utah 1994).
. The State also alludes vaguely to the notions that Verde's past misconduct might demonstrate "preparation" for or a "pattern" of the activity he is charged with in this case, but neither of those rubrics fit this case. Evidence of "preparation" would indicate steps to facilitate the commission of the crime at issue in the trial, as where a defendant is shown to have stolen a cutting torch that is used in a subsequent burglary. See Lewis v. United States, 771 F.2d 454, 456 (10th Cir.1985). "Pattern" evidence, by contrast, is implicated where the uncharged and charged conduct is "peculiarly distinctive" and thus likely to have been committed by the same individual. State v. Featherson, 781 P.2d 424, 428-29 (Utah 1989), abrogated on other grounds by State v. Doporto, 935 P.2d 484 (Utah 1997); see also, e.g., State v. Decorso, 1999 UT 57, ¶¶ 29-35, 993 P.2d 837 (uncharged misconduct evidence admissible to prove identity because of the "signature-like" similarity between the acts). This case involves neither preparation nor a pattern of this sort, but at best a "plan" by the defendant.
. Miguel A. Mendez & Edward J. Imwinkelried, People v. Ewoldt: The California Supreme Court's About-Face on the Plan Theory for Admitting Evidence of an Accused's Uncharged Misconduct, 28 Lov, L.A. L.Rev. 473, 480-81 (1995) (footnotes omitted).
. See also 22 CmartBs Aran Waricut, Et ac., Frperat Practice Anp Proceoure: Evipence § 5244 (Ist ed.) (''The justification for admitting evidence of other crimes to prove a plan is that this involves no inference as to the defendant's character; instead his conduct is said to be caused by his conscious commitment to a course of conduct of which the charged crime is only a part." (footnote omitted)).
. The State also asserts that we adopted this more "modern" view in State v. Nelson-Waggoner, 2000 UT 59, 6 P.3d 1120, but the best reading of that case is that the evidence was admitted not as plan evidence, but as evidence rebutting the defendant's theory that the victim fabricated the charged conduct. We discuss this theory below in paragraphs 44-62.
. Davin P. Leowarp, Txs New Wigmore A Treatise on Evipence: Evipence or Misconpuct anp Smittar Events § 9.2.2 (2009).
. See Mendez & Imwinkelried, supra 134 n. 9, at 501-03 (discussing the "(intolerable [risks [plosed by the Ewoldt [tJest").
. See id. at 501 ("'Under Ewoldt the inference that the accused committed the charged and uncharged offenses as part of one plan is so weak as to be unacceptably speculative.... In contrast, a showing of common features is highly probative of the accused's disposition to engage in the type of criminal conduct with which he is charged.").
. Id. at 500.
. Mark Cammack, Using the Doctrine of Chances to Prove Actus reus in Child Abuse and Acquaintance Rape: People v. Ewoldt Revisited, 29 U.C. Davis L. Rev. 355, 388 (1996).
. United States v. Russell, 19 F. 591, 592 (W.D.Tex.1884); see also Leonard, supra 139 n. 12, § 7.3.2 (citing United States v. Russell).
. People v. Balcom, 7 Cal.4th 414, 27 Cal. Rptr.2d 666, 867 P.2d 777, 787 (1994) (Arabian, J., concurring).
. See United States v. Woods, 484 F.2d 127, 135 (4th Cir.1973) ("[We think that the evidence would prove that a crime had been committed because of the remoteness of the possibility that so many infants in the care and custody of defendant would suffer cyanotic episodes and respira*685tory difficulties if they were not induced by the defendant's wrongdoing. ...").
. See the English "Brides in the Bath" case, Rex v. Smith, 11 Crim.App. 229, 84 L.J.K.B. 2153 (1915).
. 1 Enwarp J. UncHaroeep Miscon-puct Evipence § 4:01 (rev. ed. 2004).
. See Leonam», supra 139 n. 12, § 7.3.2 (discussing Wigmore's classic example of a hunter "mistakenly" shooting toward a hunting partner multiple times).
. Or, as one court put it: "The man who wins the lottery once is envied; the one who wins it twice is investigated." United States v. York, 933 F.2d 1343, 1350 (7th Cir.1991), overruled on other grounds by Wilson v. Williams, 182 F.3d 562, 567 (7th Cir.1999).
. 1 Imwinketriep, supra 149 n. 21, § 4:01.
. See Edward J. Imwinkelried, An Evidentiary Paradox: Defending the Character Evidence Prohibition by Upholding a Non-Character Theory of Logical Relevance, the Doctrine of Chances, 40 U. Rick. L. Rev. 419, 436 (2006).
. See id. at 436-39 (examining the doctrine's non-character rationale and refuting arguments that character inferences and implicit improbability reasoning both ultimately require a jury to use a "defendant's subjective character as a predictor of conduct").
. See, eg., Westfield Ins. Co. v. Harris, 134 F.3d 608, 615 (4th Cir.1998) ("[The more often an accidental or infrequent incident occurs, the more likely it is that its subsequent reoccurrence is not accidental or fortuitous."); United States v. York, 933 F.2d 1343, 1350 (7th Cir.1991) (discussing the doctrine of chances and reasoning that "(ilt is not every day that one's wife is murdered; it is more uncommon still that the murder occurs after the wife says she wants a divorce; and more unusual still that the jilted husband collects on a life insurance policy with a double-indemnity provision."), overruled on other grounds by Wilson v. Williams, 182 F.3d 562 (7th Cir.1999); People v. Everett, 250 P.3d 649, 656-58 (Colo.App.2010) (examining and applying the doctrine of chances); Wynn v. State, 351 Md. 307, 718 A.2d 588, 607 (1998) (examining the doctrine of chances, collecting cases applying the doctrine, and explaining that "[i]t is the objective implausibility of the occurrence, sans nefarious activity, which rebuts the claim of an innocent occurrence"); State v. Johns, 301 Or. 535, 725 P.2d 312, 321-27 (1986) (examining and applying the doctrine of chances).
. See Edward J. Imwinkelried, The Use of Evidence of an Accused's Uncharged Misconduct to Prove Mens Rea: The Doctrines Which Threaten to Engulf the Character Evidence Prohibition, 51 Oxo St. LJ. 575, 588-92 (1990); Cammack, supra ¶ 47 n. 16, at 404; see also Everett, 250 P.3d at 658-60 (adopting and applying the foundational requirements suggested in Imwinkelried, supra 150 n. 25, at 589).
. Imwinkelried, supra ¶ 57 n. 28, at 592.
. Id. at 595.
. Cammack, supra % 47 n. 16, at 404.
. Id. at 405.
. Id. at 405-06.
. Imwinkelried, supra 1 57 n. 28, at 590.
. Cammack, supra 147 n. 16, at 402; see also id. at 397-04 (explaining the "product rule" used in calculating probabilities and the necessity of independent events for purposes of the product rule).
. See Imwinkelried, supra 1 57 n. 28, at 590.
. Cammack, supra T 47 n. 16, at 396-97.