*32Defendant appeals a judgment of conviction for two counts of first-degree sexual abuse, four counts of first-degree rape, and two counts of first-degree sodomy involving his two minor granddaughters. We reject his second assignment of error without discussion, and write to address defendant's first assignment of error, in which he argues that the trial court erred in admitting the testimony of his adult daughter that he sexually abused her when she was a child. Defendant contends that the trial court erred as a matter of law when it failed to analyze the admissibility of that evidence under the framework prescribed in State v. Baughman , 361 Or. 386, 393 P.3d 1132 (2017), a failure that then impacted the OEC 403 balancing. Defendant also argues that the trial court erred in admitting the evidence of the uncharged acts for several nonpropensity purposes under OEC 404(3) and in finding that the probative value of the evidence substantially outweighed the danger of unfair prejudice under OEC 403.1
On appeal, the state defends the trial court's admission of the evidence on one theory, to show defendant's sexual purpose in committing the charged acts, which it characterizes as a nonpropensity theory.2 Therefore, even assuming *33the trial court erred in failing to follow the Baughman framework, the state urges us to affirm because nonpropensity evidence is generally admissible under State v. Williams , 357 Or. 1, 346 P.3d 455 (2015). We agree with defendant that the trial court erred when it failed to first assess the probative value of the evidence as nonpropensity evidence under OEC 404(3) and OEC 403 and then turn to its admissibility as propensity evidence under OEC 404(4) and OEC 403 only if it was inadmissible under *366OEC 404(3). Accordingly, we reverse and remand for the trial court to follow the framework outlined in Baughman .
BACKGROUND
Defendant was charged with crimes committed against his granddaughters MB and LB. Two of the counts involved first degree sexual abuse based on allegations that defendant touched MB's vagina. Four of the counts alleged first-degree rape for sexual penetration of LB, and two of the counts alleged first-degree sodomy for deviate sexual penetration of LB. All counts were alleged to be separate criminal acts.
Before trial, defendant filed a written motion to exclude the expected testimony by his daughter, JB (who is also LB's mother) that defendant had sexually abused her in childhood. As ultimately allowed by the trial court, JB later testified that, when she was between 8 and 14 years old, defendant touched her breasts a number of times, touched her vagina one time, and put his penis in her mouth one time. She said that this occurred in the living room and in her parents' bedroom when her siblings were likely in their bedroom and her mother was at work. Defendant argued that JB's proposed testimony was inadmissible under OEC 401, OEC 404(4), and OEC 403, and under the Fourteenth Amendment to the United States Constitution and Article I, section 11, of the Oregon Constitution. The state responded that the evidence would not be relevant unless defendant "open[ed] the door." On the first day of trial, the state filed a *34trial memorandum detailing the various ways in which the evidence would be admissible if the defendant "opened the door," and it also argued for the admission of JB's testimony as relevant on "proof of motive, lack of mistake or accident as well as [a] sexual predisposition toward pre-pubescent girls, under OEC 404(4), 404(3), 403, and 401." As to its relevance to establish a lack of mistake, the state contended that "[i]f the defendant * * * puts forth a claim that any touching was innocent, or accidental, the state must * * * disprove the defense theory." The court did not rule on the state's motion to admit JB's testimony before trial.
Both MB and LB testified at trial. MB, then 19, testified that, when she was five or six years old, defendant touched her inappropriately at a family gathering. She explained that while she was sitting on his lap when he was in a recliner in the living room, defendant touched her vagina and breast. She testified that the same thing happened again that same day when, with other children around, she was on defendant's lap in a hot tub and he touched her "skin-to-skin."3
LB, then 25, testified to numerous incidents of sexual abuse beginning when she was roughly six to eight years old. She described how defendant sexually penetrated her in the bedroom and front room of his home and in his truck. She explained that defendant would take her with him to take his dog out to play in a rural area. She described how, on more than one occasion, defendant would have her sit on his lap, pull down her pants, and sexually penetrate her. LB also described how defendant put her mouth on his penis once in his bedroom and once in the front room of his home. She explained that the abuse stopped when she was a young teenager because defendant had found out that she had begun menstruating, and he told her that "it couldn't happen anymore because [she] was now a woman."
The state also called two detectives who testified that, when they interviewed him, defendant denied the conduct described by his granddaughters.
*35The defense called a number of family members who testified that MB and LB acted normal around defendant and that they did not observe anything unusual between the two girls and defendant. Several witnesses testified that defendant did not like to be around children, was not affectionate with his grandchildren, and never had them sit on his lap or spent time alone with them. Defendant's wife testified that defendant did not take care of his own children. The witnesses also testified that it would have been impossible for defendant to sexually abuse the victims during the holiday family gatherings *367because of the number of people in the home and their visibility due to its layout.
Lastly, the defense elicited testimony from three witnesses about defendant's sexual propriety around other children. His wife testified that she had never observed any concerning behavior by defendant toward any child. Defendant's daughter-in-law likewise testified that she had never observed any concerning behavior between defendant and any other child, including her own. And LB's younger sister testified that she had sat on defendant's lap in the hot tub before and that he had hugged her in the past, but that he had never made her feel uncomfortable.
After the defense rested, the state argued that defendant had "opened the door" for JB's proposed testimony by asking several witnesses "what they had seen, how *** defendant acted around children within the residence, how the children reacted toward *** defendant, essentially implying this couldn't have happened because [MB] and [LB] were not afraid of [him]." According to the state, through this testimony, there is "some inference or implication [that] *** everybody in the family thinks [defendant] is the perfect grandfather, father, has no inappropriate contact at any point." The state further argued that the evidence was admissible under Williams "to show defendant's sexual predisposition towards a certain age of child."
In an OEC 104 hearing outside the jury's presence JB testified to the details of the sexual abuse. The state then argued that her testimony was admissible under OEC 404(4) and OEC 404(3) for two purposes: (1) "to show *** defendant's sexual predisposition towards children *** between *36five and nine" and (2) "to prove there was no accidental touching; that this was sexual touching that occurred." The state asserted that the prior conduct and charged acts were admissible under State v. Johns , 301 Or. 535, 725 P.2d 312 (1986), and that the probative value of the evidence was not substantially outweighed by any prejudice under OEC 403. Defendant disagreed that he had "opened the door" to the evidence and argued that the prejudicial effect of JB's proposed testimony "far outweigh[ed]" its probative value.
In making its ruling, the court began by stating that it was evaluating the admissibility of the testimony under
" State vs. Williams [,] *** State vs. Turnidge and State vs. Brumbach , which is a very similar case to this, as was Williams . [S]o I'm evaluating this under 404(4) and *** I'm very clearly conducting a balancing analysis under 403."
The court explained that the evidence was admissible because defendant had "open[ed] the door" to it by presenting an "impossibility defense" that he "would never have had an opportunity because he was literally never alone with children, always observed by other adults." Further, the court noted that the defense had presented evidence that "no children ever demonstrated any sort of concern for his behavior and that *** he's a loving grandfather," and also that defendant did not like to be around children. The court explained,
"[T]he need for this [evidence] *** I think under this case would classify as uncharged misconduct[.] Prior bad acts evidence. And it's clearly necessary in this case, so there's certainly a strong reason to admit it.
"And in *** analyzing the probative value versus the prejudicial impact, of course there's prejudicial impact. There's always prejudicial impact when there's evidence present that would tend to support that a defendant has committed a crime. But I don't think it's unfair in this case. In fact, I think it would be really unfair to the State not to be able to counter this impossibility defense with the testimony of the defendant's own daughter, who is in the 104 hearing testifying that the defendant *** touched her vagina, that he at some point had his penis in her mouth;
*37conduct that is completely consistent with the testimony of [MB] and [LB]."
The court continued,
"[I'm] making the decision that the probative value outweighs the prejudicial impact. And under 404(4) and 404(3), and after a 403 analysis, this evidence is admissible and it's going to be allowed.
"It's probative on motive. In other words, sexual gratification. ***
*368"It's probative on the issue of opportunity[ ] *** [t]o contradict the evidence that there could be no opportunity.
"It's probative on intent. And again going to the sexual gratification.
"It's probative value on *** some sort of a plan, in the sense of being alone with children. ***
"And the absence of mistake or accident. There was some suggestion that, oh, well, if he was in the hot tub with one of the girls, maybe he touched her, but it would have been a mistake or accident. Well, this is probative that that was not a mistake or accident."
Defendant did not request a limiting instruction. After the close of evidence, defendant was convicted on all counts.
On appeal, defendant argues that the trial court's OEC 403 balancing was flawed because the court failed to first identify whether the evidence was admissible as propensity or nonpropensity evidence before conducting the OEC 403 balancing, which requires reversal under Baughman . Defendant also argues that the trial court erred in determining that the evidence was relevant for a non-propensity purpose to show plan, opportunity, absence of mistake or intent, or motive.
At oral argument, the state narrowed its argument in response, defending only a single theory of admissibility-to show defendant's sexual purpose, which the state characterized as a nonpropensity theory. The state argues that evidence showing that defendant acted with a sexual purpose, which is an element of the crime of first-degree sexual *38abuse, is nonpropensity evidence, and is admissible under OEC 404(4) and OEC 403. Therefore, the state argues, even though the trial court evaluated admissibility under OEC 404(4), which applies to propensity evidence, we should nonetheless affirm because, as Williams instructs, evidence that is relevant for a nonpropensity purpose "generally will be admissible" under OEC 403 balancing. 357 Or. at 19, 346 P.3d 455.
STANDARD OF REVIEW
We review a trial court's determination that other acts evidence is relevant for a nonpropensity purpose under OEC 404(3)4 and OEC 404(4)5 for legal error. Baughman , 361 Or. at 406, 393 P.3d 1132. We review a trial court's determination as to whether the probative value of the other acts evidence is substantially outweighed by the danger of unfair prejudice under OEC 4036 for abuse of discretion. Id.
ANALYSIS
We begin by concluding that the trial court erred in admitting the evidence under the nonpropensity theories of plan, opportunity, absence of mistake, or motive. The state's decision not to defend those bases for admission is sound.
The prior act in this case, defendant's sexual abuse of his daughter, was not sufficiently connected in either time *39or operation to the charged acts to show that defendant was acting in accordance with a plan. See State v. Leistiko , 352 Or. 172, 189, 282 P.3d 857 (2012) (stating that "something more than the similarity required for other crimes evidence to be admissible to prove intent is *369necessary for it to be admissible to prove plan"); State v. Kaylor , 252 Or. App. 688, 700-04, 289 P.3d 290 (2012), rev. den. , 353 Or. 428, 299 P.3d 889 (2013) (the defendant's threats to the victim and another resident at the same nursing home six months before the defendant's assault on the victim while attending to his care were not sufficiently similar to demonstrate defendant's intent or plan to treat the victim abusively). Moreover, JB's testimony that defendant abused her when she was a child does not actually speak to the issue of whether he had an opportunity to be alone and unobserved with his grandchildren or other children.
Additionally, the absence of mistake or doctrine of chances does not apply in this case because defendant denied that the charged conduct occurred. State v. Tena , 362 Or. 514, 524, 412 P.3d 175 (2018) ("[T]he doctrine of chances applies only to explain whether or not an act that a defendant performed was performed intentionally. It does not apply when there is a dispute about whether the defendant performed the act at all.").7
Lastly, treating evidence that defendant sexually abused his daughter when she was a child as relevant to his motive in sexually abusing his minor granddaughters is rooted in impermissible character-based reasoning. The logical inference of such an argument is based on a character trait of defendant-a motive to sexually abuse child victims because he has a sexual interest in children in general. See State v. Davis , 290 Or. App. 244, 246, 255, 414 P.3d 887 (2018) (defendant's two handwritten notes to women expressing a desire to engage in violent sexual acts was not admissible to show that defendant acted with a sexual purpose *40in kidnapping the victim because that "inference depends upon the existence of a persistent trait intrinsic to defendant: that defendant has an interest in sexually assaulting any female stranger") (emphasis in original).8
Having determined that the trial court erred in giving weight to the prior conduct evidence under nonpropensity theories, we conclude that the OEC 403 balancing that it engaged in was flawed as a consequence. In Baughman , the Supreme Court laid out the framework that trial courts must follow in assessing the admissibility of uncharged misconduct evidence under OEC 404(3) and OEC 404(4). 361 Or. at 404-05, 393 P.3d 1132. The court explained,
"When a party objects to the admission of other acts evidence, a trial court first should determine whether the proffered evidence is relevant for one or more nonpropensity purposes, under OEC 404(3). If it is, then the court should determine, at step two, whether the probative value of that evidence is substantially outweighed by the danger of unfair prejudice under OEC 403. If the trial court determines that the evidence is relevant for a nonpropensity purpose under OEC 404(3) and admissible under OEC 403, then it need not determine whether the evidence also is admissible *370under OEC 404(4) and OEC 403. However, if a trial court determines that proferred evidence is not relevant for a nonpropensity purpose, then it must determine whether that evidence nevertheless is otherwise relevant under OEC 404(4) and, at step two, whether the probative *41value of the evidence is substantially outweighed by the danger of unfair prejudice, under OEC 403."
Id .
In this case, the trial court did not, as Baughman instructs, begin at step one and consider the potential nonpropensity theories of relevance and move to step two and apply the OEC 403 balancing to those theories. Rather, the trial court first considered whether the evidence was relevant under OEC 404(4). Following that, the court assessed the relevancy of the evidence due to defendant "opening the door." The trial court next applied its OEC 403 balancing to those theories combined. The trial court ended its analysis by identifying additional nonpropensity theories of relevance under OEC 404(3). Combining propensity and nonpropensity theories of relevance when conducting the OEC 403 balancing will significantly affect the outcome of that balancing. As Baughman explains,
"A trial court's decision, at step one, about whether other acts evidence is relevant for a nonpropensity purpose, will have a significant effect on whether the trial court admits that evidence at step two. At one end of the spectrum, other acts evidence that is relevant for a nonpropensity purpose under OEC 404(3) generally will be admissible under OEC 403 as long as the particular facts of the case do not demonstrate a risk of unfair prejudice that substantially outweighs the probative value of the evidence. * * * At the other end of the spectrum, when evidence is relevant only to prove a defendant's character, more significant due process concerns are implicated, and, generally, the danger of unfair prejudice will substantially outweigh the probative value of the evidence."
Id. at 405, 393 P.3d 1132 (citing Williams , 357 Or. at 19-20, 346 P.3d 455 ). Therefore, the trial court's error in finding the prior conduct evidence to be probative on multiple nonpropensity theories resulted in the trial court giving undue weight to the quantum of probative value of the uncharged conduct, which affected its OEC 403 balancing in this case.
Despite the court's error, the state contends that, even if the trial court erred in its OEC 403 balancing, we should nevertheless affirm because evidence of defendant's sexual purpose, an element of one of the charged crimes, is *42a nonpropensity theory, and as explained in Williams and Baughman , nonpropensity evidence generally will be admissible under OEC 403. Defendant takes a different position and characterizes evidence to establish defendant's sexual purpose towards the minor victims as propensity evidence. However, even assuming the state is correct that the evidence was admissible as nonpropensity evidence, we nonetheless conclude that the error was not harmless.9
Whether the testimony is labelled propensity or nonpropensity evidence, we cannot say, as a matter of law, that evidence to establish a defendant's sexual purpose is sufficiently probative as to outweigh any potential prejudicial impact. The decision *371whether to admit evidence under OEC 403 is a matter of discretion for the trial courts. See State v. Sewell , 257 Or. App. 462, 468-69, 307 P.3d 464, rev. den. , 354 Or. 389, 315 P.3d 420 (2013). The defendant in this case argues that the evidence should have been excluded under OEC 403 balancing for a number of reasons, including that the need for the evidence was low because the "jury had already heard from two separate victims recount[ing] their sexual abuse" by defendant and that "a decade's [old] prior tale of abuse" held little probative value on the issues in the case. Unlike State v. Zavala , 361 Or. 377, 393 P.3d 230 (2017), where the court concluded that the trial court's error in failing to conduct OEC 403 balancing was harmless because the defendant "did not contend that * * * the particular facts of his case demonstrated a risk of unfair prejudice that substantially *43outweighed the probative value of the evidence," defendant has raised such an argument in this case, and we cannot say that a trial court could not be persuaded to exclude the evidence. See State v. Holt , 292 Or. App. 826, 833-34, 426 P.3d 198 (2018) (concluding that the trial court's error in admitting uncharged nonpropensity evidence without conducting an OEC 403 balancing was not harmless where the court "permissibly could have exercised its discretion to admit or exclude some or all of the evidence of defendant's previous conduct with [the victim]; either decision would have fallen within the permissible range of outcomes"). Therefore, we cannot conclude that there was little likelihood that the trial court's error affected the verdict. See State v. Davis , 336 Or. 19, 32, 77 P.3d 1111 (2003) (for harmless error review, the question is whether there is little likelihood that the trial court's error affected the verdict). We reverse and remand for the trial court to follow the framework as outlined in Baughman and assess the admissibility of the challenged evidence.10
Reversed and remanded.