*352 **203 This is a certified criminal appeal from an Appellate Court decision reversing a judgment of conviction arising out of allegations by the complainant, B, that her stepfather, the defendant Fernando V., sexually assaulted her repeatedly over a period of years while she was in middle school and high school. The Appellate Court reversed the judgment of conviction on the ground that the trial court improperly precluded the defendant from calling the complainant's longtime boyfriend, P, as a witness regarding his observations of certain aspects of B's behavior that the state's expert witness had testified were common symptoms of child sexual assault. See State v. Fernando V. , 170 Conn. App. 44 , 68-69, 153 A.3d 701 (2016). The Appellate Court concluded that the improper exclusion of P's testimony was not harmless because the evidence may have helped "to show that B failed to exhibit behaviors often attributed to sexual assault victims," which could have "dissuaded the jury from believing B's story generally ...." Id., at 68 , 153 A.3d 701 . We affirm the judgment of the Appellate Court.

**204 I

The following facts are relevant to this appeal. B moved to Stamford from Mexico when she was nine years old to live with her mother, brother, and the defendant, her stepfather. The defendant adopted B in 2004, when she was ten years old, and he later petitioned for her to obtain permanent residency in the United States. When B initially came to Stamford, the family lived with B's grandmother and uncle, but eventually her grandmother moved back to Mexico. B testified that she was often alone with the defendant after her grandmother's departure, and he began to act inappropriately by touching her breasts. B told her mother about the defendant's inappropriate behavior. B's mother confronted the defendant, but he denied any wrongdoing and said B was confused.

In 2006, when B was nearing her thirteenth birthday, the family moved to Norwalk. B testified that the defendant continued to touch her inappropriately after the move. According to B, she told her mother about the continuing sexual misconduct, but the defendant again denied the allegations when confronted. B testified that the abuse escalated when the defendant forced her to have sexual intercourse with him in the hallway bathroom one afternoon. She testified that the defendant thereafter continued to touch her inappropriately or to force her to have sexual intercourse on a regular basis, sometimes as often as once per week. B said that the abuse continued until approximately 2011, when she was sixteen or seventeen years old.

B explained at trial that she did not disclose immediately to her mother that the defendant was forcing her to have sex with him because she was scared of what her mother would think. She eventually disclosed the abuse to her mother in 2011, however, when her mother directly asked B whether the defendant had forced her **205 to have sex. B and her mother then *353 called the police, which resulted in the present criminal case.

The defendant was charged with one count of sexual assault in the second degree in violation of General Statutes § 53a-71 (a) (1), one count of sexual assault in the second degree in violation of § 53a-71 (a) (4), and two counts of risk of injury to a child in violation of General Statutes § 53-21 (a) (2). 1 The evidence against the defendant consisted primarily of the testimony of B and her mother, who testified as a constancy of accusation witness and also offered evidence of B's behavior during the relevant time period. Both B and her mother testified that B achieved good grades, participated in extracurricular activities, maintained employment without excessive absences, and continued to enjoy reading books and pursuing musical interests. B's mother also testified that she did not notice any personality changes in B when she was twelve or thirteen years old, but she did observe that B's disposition changed in the year before the defendant's arrest. "[S]he was more withdrawn, and I saw that she would stay in her room," "locked up," explained B's mother.

Toward the end of its case-in-chief, after B and her mother had testified, the state called an expert witness, Larry M. Rosenberg, a licensed psychologist and the clinical director of the Child Guidance Center of Southern Connecticut. Rosenberg testified about "delayed disclosure," which describes a commonly observed phenomenon in sexual abuse cases that occurs when a victim does not inform anyone of the sexual abuse for a period of time, sometimes lengthy, despite the **206 suffering and trauma experienced as a result of being abused.

The origin of the present appeal can be traced to the point in Rosenberg's testimony when he was asked by the state to opine about behavioral issues other than delayed disclosure. More specifically, Rosenberg was asked by the state about symptoms exhibited by victims of child sexual assault who have made a disclosure. Rosenberg answered that there were a variety of symptoms commonly observed in such victims, including changes in behavior, disassociation, withdrawal, depression, heightened anxiety, bad dreams, flashbacks, sleep interruption, and changes in cognitive functioning. Rosenberg elaborated the point on cross-examination, explaining that depression can manifest itself in changes in mood, irritability, and angry outbursts. He stated, "[t]he list goes on, you know, bad dreams, all sorts of things." 2 Rosenberg's expert testimony apparently was offered by the state to help the jury understand the significance of the prior testimony of B and her mother, in a manner consistent with the state's objective at trial, which was to establish the defendant's guilt. The expert testimony about delayed disclosure would help to explain why B did not immediately report *354 the most severe abuse to her mother; the testimony about common symptoms of trauma would assist the jury in understanding why B had become more withdrawn prior to the defendant's arrest.

After the conclusion of the state's case-in-chief, the defense attempted to discredit the state's version of **207 events by presenting the testimony of P, B's longtime boyfriend. Upon hearing that B and P were in a relationship, the trial court excused the jury to hear the state's objection that P's testimony was not relevant to the issue at hand. With the jury out of the courtroom, the defense made the following offer of proof relating to the admissibility of P's testimony about B's behavior:

"[Defense Counsel]: When you say you're in a relationship, are you-do you consider yourself boyfriend and girlfriend?

"[P]: Yes.

"[Defense Counsel]: And have you continuously gone out with her, or been in a relationship with her, as boyfriend and girlfriend, for four years?

"[P]: Yes, I have.

"[Defense Counsel]: Have there been any breaks in the relationship?

"[P]: No, there have not.

"[Defense Counsel]: Now, in the time period that you've been going out, as boyfriend and girlfriend, with [B], have you noticed any significant behavioral issues with her?

"[P]: No, not really.

"[Defense Counsel]: Have you noticed any pronounced eating disorders?

"[P]: No, I have not.

"[Defense Counsel]: Have you noticed any suicidal thoughts?

"[P]: No, I have not.

"[Defense Counsel]: Have you noticed any severe depression?

"[P]: No, I have not.

**208 "[Defense Counsel]: Have you noticed any eating disorders?

"[P]: No, I have not.

"[Defense Counsel]: Have you noticed any anger or outbursts or violence, by her?

"[P]: No, I have not.

"[Defense Counsel]: Have you noticed any trouble with her focusing on issues or tasks at hand?

"[P]: No, I have not.

"[Defense Counsel]: And, to your knowledge, do you know if her grades have slipped, in any way, in the four years you've known her?

"[P]: No, I don't think so.

"[Defense Counsel]: And, in the four years that you've known her, have you noticed any type of interruption in her playing of the flute?

"[P]: No, I have not.

"[Defense Counsel]: And, since September, 2011, have you noticed any of the things that I just mentioned, occurring with [B]?

"[P]: No, I have not."

The defense argued that P's testimony regarding B's behavior was admissible because it was relevant in two ways: first, to impeach the credibility of B's mother, who had testified that B had become more withdrawn, and, second, as direct evidence regarding the occurrence or nonoccurrence of the behavioral changes that the state's expert witness had testified are commonly exhibited by child victims of sexual assault. The latter ground in particular was twice referenced by defense counsel in colloquy with the trial court. The state, for its part, argued categorically that the testimony was not relevant and **209 pointed out that P was not qualified to offer testimony on the *355 subject because he was not an expert witness. The state also argued that the evidence did not directly impeach the testimony of B or her mother. In addition, the state noted its concern that it could be prejudicial for the jury to hear testimony about B's romantic relationship with P.

The trial court ruled that P's testimony was inadmissible in its entirety. The court stated that "[t]he relevance of this testimony ... is collateral, at best." With respect to impeachment, it found that "[i]mpeachment is not, by this evidence, extrinsic evidence. It lends itself to-it's likely to confuse the jurors. It's not probative of any issues.... I don't see any impeachment, based upon what I've heard on this record .... [An] [o]ffer of proof has been made. It's on the record, should the matter be reviewed. It's there for the Appellate Court to look at. But before the jury, it's confusing. It's not probative, and ... the objection is sustained." Therefore, P's testimony was not presented to the jury.

The jury returned a verdict of guilty on two counts of sexual assault in the second degree and two counts of risk of injury to a child. The trial court sentenced the defendant to an effective term of ten years of incarceration and ten years of special parole. The defendant appealed from the judgment of conviction on the ground that the trial court improperly excluded P's testimony from the jury's consideration. 3

**210 State v. Fernando V. , supra, 170 Conn. App. at 46 , 153 A.3d 701 . In the Appellate Court, the state argued that the trial court properly excluded P's testimony "because it was both collateral in nature and entirely consistent with the testimony given by B and her mother." Id., at 48-49 , 153 A.3d 701 . The state also contended that, even if it was error to exclude the evidence, the error was harmless because P's testimony "did not differ materially" from the testimony of B or her mother and therefore the exclusion "had little effect on the jury ...." (Internal quotation marks omitted.) Id., at 69 , 153 A.3d 701 . The Appellate Court rejected those claims, holding that P's testimony improperly was excluded because it was relevant to "the issue of whether B had exhibited behaviors associated with some sexual assault victims, which had a clear and direct bearing on the central issue before the jury, namely, whether B had been sexually assaulted by the defendant." Id., at 67 , 153 A.3d 701 . The Appellate Court further concluded that the improper exclusion of P's testimony was not harmless because the absence of any physical evidence or witnesses to the sexual assaults meant that "[t]he case turned largely on whether the jury believed B"; id., at 69 , 153 A.3d 701 ; and P's *356 testimony, which "helped to paint B as having been an ordinary high school girl," necessarily would have "decrease[d] the likelihood in the eyes of the jury that an assault had occurred." Id., at 68 , 153 A.3d 701 . The Appellate Court consequently reversed the judgment of conviction and remanded for a new trial. Id., at 69 , 153 A.3d 701 . This certified appeal followed.

II

The state first argues that the Appellate Court improperly found that the trial court had abused its discretion **211 by excluding P's testimony. The state does not rely on the grounds it raised in the trial court or the Appellate Court but instead contends, for the first time, that P's testimony regarding B's behavior properly was excluded by the trial court because it was cumulative of other evidence in the record indicating that B "was basically 'an ordinary high school girl' ... dating, getting good grades, participating in extracurricular activities and holding down a job." (Citation omitted.) This is a new argument. The state never argued in the trial court that P's testimony about B's behavior should be excluded because it was cumulative, nor did the trial court base its ruling on that ground. The argument also was not raised or briefed by the state as an alternative ground for affirmance in the Appellate Court, and the Appellate Court, like the trial court, did not address the argument as part of its admissibility analysis. On this record, we conclude that the state has failed to preserve its belated legal theory of the inadmissibility of P's behavioral testimony based on cumulativeness, made for the first time in this court, and we decline to review the claim. Because the state has abandoned all claims other than its contention that P's testimony was cumulative; see, e.g., Samelko v. Kingstone Ins. Co. , 329 Conn. 249 , 255 n.3, 184 A.3d 741 (2018) (deeming arguments not raised and briefed in this court to be abandoned); the decision of the Appellate Court that the exclusion of P's testimony was improper effectively stands unchallenged and must be upheld.

"This court is not bound to consider claims of law not made at the trial.... In order to preserve an evidentiary ruling for review, trial counsel must object properly.... In objecting to evidence, counsel must properly articulate the basis of the objection so as to apprise the trial court of the precise nature of the objection and its real purpose, in order to form an adequate basis for a reviewable ruling.... Once counsel states **212 the authority and ground of [the] objection, any appeal will be limited to the ground asserted." (Internal quotation marks omitted.) State v. Gonzalez , 272 Conn. 515 , 539, 864 A.2d 847 (2005) ; see also Perez-Dickson v. Bridgeport , 304 Conn. 483 , 499, 43 A.3d 69 (2012) (rule that claim must be "raised and decided in the trial court ... applies equally to alternate grounds for affirmance" [internal quotation marks omitted] ). We have emphasized that "[t]hese requirements are not simply formalities. They serve to alert the trial court to potential error while there is still time for the court to act.... Assigning error to a court's evidentiary rulings on the basis of objections never raised at trial unfairly subjects the court and the opposing party to trial by ambush." (Internal quotation marks omitted.) Id., at 540, 864 A.2d 847 ; see also State v. Miranda , 327 Conn. 451 , 465, 174 A.3d 770 (2018) ("[A] party cannot present a case to the trial court on one theory and then seek appellate relief on a different one .... For this court to ... consider [a] claim on the basis of a specific legal ground not raised during trial would amount to trial by ambuscade, unfair both to the [court] and to the opposing party." [Internal quotation marks omitted.] ). *357 This reasoning applies with full force in the present case, in which the state's newly minted ground for exclusion, based on the supposedly cumulative nature of the excluded evidence, calls for a discretionary determination to be made by the trial court in the first instance. See, e.g., Motzer v. Haberli , 300 Conn. 733 , 742, 15 A.3d 1084 (2011) ("We conclude that the trial court did not abuse its discretion in excluding the proffered evidence [as cumulative]. Our rules of evidence vest trial courts with discretion to exclude relevant evidence when 'its probative value is outweighed ... by considerations of undue delay, waste of time or needless presentation of cumulative evidence.' "), quoting Conn. Code Evid. § 4-3. This particular exercise of **213 discretion was not undertaken by the trial court in this case because the state never requested a ruling on the ground now being advanced. We cannot determine whether the trial court abused an exercise of discretion that it neither made nor was asked to make. Under these circumstances, we decline to review the state's unpreserved claim. 4

Our rules of reviewability in the evidentiary context are prudential in nature, not jurisdictional, but they serve essential purposes and promote vital principles, and only in the most compelling situation will we depart from them. Legal claims, arguments and objections regarding evidentiary matters ordinarily must be made at the right time and place, because that time and place is when the opposing party has the opportunity to respond to the point or to cure the defect, and it also is when the trial judge will be required to adjudicate the disputed issue within the particularized context defined by the circumstances then existing. Adhering **214 to the requirement of specificity and contemporaneity promotes fairness between the parties and helps to ensure that trial and appellate judges remain optimally positioned to perform their respective roles. There are, of course, exceptional circumstances when this court will "consider a claim, constitutional or otherwise, that has not been raised and decided in the trial court." 5 *358 Perez-Dickson v. Bridgeport , supra, 304 Conn. at 499 , 43 A.3d 69 . Nothing about the present case qualifies the state's unpreserved evidentiary claim for such exceptional treatment.

The Appellate Court determined that P's testimony improperly was excluded because it was relevant and "probative of the central issue of this case"-B's credibility. State v. Fernando V. , supra, 170 Conn. App. at 64 , 153 A.3d 701 . In this court, the state does not challenge the Appellate Court's evidentiary holding on any basis other than the unpreserved claim of cumulativeness. Accordingly, the determination of the Appellate Court that P's testimony improperly was excluded must stand.

III

We now must decide whether the improper exclusion of P's testimony was harmless. The state makes two arguments: first, that the excluded evidence was cumulative, and, second, that the case against the defendant was very strong and any inconsistencies in B's testimony were explored on cross-examination and considered **215 by the jury. We disagree with both contentions. We view the record as the Appellate Court did and concur in its conclusion that the exclusion of P's testimony cannot be considered harmless on this record.

The law governing harmless error for nonconstitutional evidentiary claims is well settled. "When an improper evidentiary ruling is not constitutional in nature, the defendant bears the burden of demonstrating that the error was harmful.... [W]hether [an improper ruling] is harmless in a particular case depends upon a number of factors, such as the importance of the witness' testimony in the [defendant's] case, whether the testimony was cumulative, the presence or absence of evidence corroborating or contradicting the testimony of the witness on material points, the extent of cross-examination otherwise permitted, and, of course, the overall strength of the prosecution's case.... Most importantly, we must examine the impact of the ... evidence on the trier of fact and the result of the trial.... [T]he proper standard for determining whether an erroneous evidentiary ruling is harmless should be whether the jury's verdict was substantially swayed by the error.... Accordingly, a nonconstitutional error is harmless when an appellate court has a fair assurance that the error did not substantially affect the verdict." (Internal quotation marks omitted.) State v. Favoccia , 306 Conn. 770 , 808-809, 51 A.3d 1002 (2012) ; accord State v. Jordan , 329 Conn. 272 , 287-88, 186 A.3d 1 (2018) ; State v. Shaw , 312 Conn. 85 , 102, 90 A.3d 936 (2014). We have observed that cases that present the jury with a "credibility contest characterized by equivocal evidence ... [are] far more prone to harmful error." (Internal quotation marks omitted.) State v. Favoccia , supra, at 816-17, 51 A.3d 1002 .

The state seriously underestimates the potential impact of the excluded testimony. As the Appellate Court aptly pointed out, "the state's case here was not **216 an exceedingly strong one" in light of the absence of "corroborating physical evidence or any witnesses to the alleged sexual assaults." 6 *359 State v. Fernando V. , supra, 170 Conn. App. at 68-69 , 153 A.3d 701 ; see also State v. Favoccia , supra, 306 Conn. at 809 , 51 A.3d 1002 (describing child sexual assault cases that lack physical evidence and turn "entirely on the credibility of the complainant" as "not particularly strong" [internal quotation marks omitted] ); State v. Grenier , 257 Conn. 797 , 808, 778 A.2d 159 (2001) (noting that "the state's case was not particularly strong" because child victim's "version of the events provided the only evidence of the defendant's guilt"); State v. Alexander , 254 Conn. 290 , 308, 755 A.2d 868 (2000) (noting that "the state's case was not particularly strong in that it rested on the credibility of the [child] victim" [internal quotation marks omitted] ). B's testimony was the only evidence of the defendant's guilt, and, therefore, this "case turned largely on whether the jury believed B." State v. Fernando V. , supra, 170 Conn. App. at 69 , 153 A.3d 701 . Indeed, as the state explained to the jury in **217 closing argument, "[w]hat this case really comes down to is one simple question, who do you believe?" By excluding P's testimony, the trial court deprived the defense of evidence that it could have used to cast doubt on the credibility of B's allegations. See State v. Ritrovato , 280 Conn. 36 , 57-58, 905 A.2d 1079 (2006) (holding that improper exclusion of evidence pertinent to minor victim's credibility "would have cast sufficient doubt on [her] credibility [so as] to have influenced the jury's verdict on the sexual assault charges").

To understand more particularly the nature of the potential harm caused by the exclusion of P's testimony, it is important to examine how that testimony became relevant to the state's case at trial. Evidently concerned that a lay jury might draw unwarranted adverse inferences about B's credibility from the fact that B had delayed telling her mother about being sexually assaulted, the state chose to present expert testimony at trial from Rosenberg explaining that delayed reporting is common in child sexual abuse cases and describing the psychological and emotional factors that make such a delay understandable. See State v. Favoccia , supra, 306 Conn. at 817 , 51 A.3d 1002 ( Palmer , J. , dissenting) (explaining that state may use expert testimony in child sexual abuse cases to explain victim behavior that is common but may not be known to laypersons). But Rosenberg's testimony did not stop at explaining delayed disclosure. The state also questioned Rosenberg at length about postdisclosure behavioral characteristics ("symptoms of trauma") commonly observed in teenagers and young adults who have been sexually assaulted. Rosenberg initially responded in general terms, *360 stating that "being sexually abused tends to most-most typically, but not always, reduce the level of functioning of the person who has been victimized." The state asked for greater detail: "What are some symptoms of trauma from child sexual assault, that you've seen, in your practice, **218 with victims who have made a disclosure?" Rosenberg answered by providing examples, including "disassociation, the kind of psychic numbing that can go on .... But in addition to that, typically, symptoms would be bad dreams, flashbacks of the events that had occurred ... changes in functioning with regard to sleep, with regard to cognitive functioning, with regard to school functioning. [Withdrawal] is common. Depression is common. Heightened anxiety, particularly in the face of anything that is reminiscent of the event.... And those are some of the findings, typically." During cross-examination, Rosenberg highlighted depression as a particularly common symptom and explained that depression can manifest itself in a variety of ways, including, for example, "changes in mood and irritability and angry outbursts," as well as becoming "more withdrawn." As he concluded his answer about the most common behavioral symptoms, Rosenberg made it clear that his description was not exhaustive, stating that "[t]he list goes on."

Rosenberg's testimony became the lens through which the jury reasonably could have viewed the most critical issues in the case. The state presumably elicited his testimony about "behavioral symptoms" because it wanted to lend significance to B's mother's testimony that B had become more withdrawn than usual in the year prior to the defendant's arrest. In fact, the state argued this very inference in its closing argument to the jury when it suggested that B's withdrawal was a sign that she had been sexually abused. 7 Rosenberg's **219 testimony was double-edged, however, because it provided the defense with an evidentiary basis to develop a jury argument that B's allegations of abuse should not be believed. The defense sought to raise the specter of reasonable doubt by arguing that B had not exhibited any of the many behavioral symptoms of trauma that the state's own expert said were typical and common among sexual abuse victims. Rosenberg's testimony, in other words, provided the defense with an opening to argue that the absence of such symptoms equates to an absence of abuse. The potential significance of P's testimony must be seen in this light.

With this framework in place, it becomes evident why the improper exclusion of P's testimony was not harmless. First, and most significantly, P's testimony was not cumulative because it would have presented the jury with new material, not heard from any other witness, regarding certain indicia of sexual abuse identified by Rosenberg. See State v. Favoccia , supra, 306 Conn. at 808-809 , 51 A.3d 1002 (holding that cumulativeness is factor to be considered in harmless error analysis). No other witness had been asked whether B suffered from depression, anger or outbursts *361 of violence, or if she had trouble focusing on issues or tasks at hand. These particular symptoms were among those identified by Rosenberg as common behavioral manifestations of trauma caused by sexual abuse. The evidentiary ruling under review excluded P's testimony that he did not observe B showing any of these specific symptoms of abuse during the past four years-evidence provided by no other witness. This testimony, if allowed, would have supplied defense counsel with additional grounds to argue that the abuse had never happened. New evidence is not cumulative evidence. **220 Second, the jury reasonably could have found that one significant aspect of the new information contained in P's testimony actually conflicted with the testimony of B's mother and thus could not have been duplicative of that testimony. B's mother testified that B had become more withdrawn prior to the defendant's arrest, which was made highly relevant by Rosenberg's subsequent testimony that "[d]epression can manifest itself in a variety of ways," including a victim's becoming "more withdrawn." A juror reasonably could have understood P's testimony that B did not exhibit any signs of depression as being inconsistent with the testimony of B's mother regarding B's withdrawal. See United States v. Stewart , 907 F.3d 677 , 688 (2d Cir. 2018) ("[T]he fact of the inconsistency gives the jury an insight into the [witness'] state of mind; the inconsistency shows that the witness is either uncertain or untruthful. In either event, the inconsistency calls into question the [witness'] believability." [Internal quotation marks omitted.] ), quoting 1 K. Broun, McCormick on Evidence (7th Ed. 2013) § 34, p. 209. The trial court's ruling prevented the defense from using P's testimony to challenge the mother's testimony that B had become withdrawn, which, not insignificantly, was the only behavioral symptom of trauma allegedly exhibited by B. The Appellate Court summarized the unfairness: "The state cannot have it both ways: on the one hand, introducing its own evidence of B's behavior favorable to the state's case and, on the other, seeking to prevent the defendant from presenting his own contrary evidence. B's mother provided otherwise unrebutted testimony that B was more withdrawn than usual and stayed locked up in her room. The state then elicited testimony from Rosenberg that withdrawal was common among sexual assault victims, thereby giving damning context to the mother's observation. The defendant was entitled to produce his own witness in an effort to counter **221 the state's evidence and demonstrate that B had not exhibited any behavioral characteristics that could be associated with sexual assault victims. That witness was P." State v. Fernando V. , supra, 170 Conn. App. at 64-65 , 153 A.3d 701 . We believe that a reasonable juror may have been swayed by P's testimony when assessing whether to believe the allegations of abuse.

Further compounding the harm arising from the improper exclusion of P's testimony is the fact that the state affirmatively used B's mother's testimony about B's "withdrawal" and Rosenberg's testimony about behavioral symptoms of trauma in its arguments to the jury. In its closing argument, the state attempted to focus the jury's attention on one aspect of B's behavior to support B's allegation that she had been sexually assaulted by reminding the jury that B's mother had "testified that even she noticed [B] was acting more withdrawn, spending more time alone in her room." In rebuttal closing argument, the state again pointed out that "there was testimony that showed that [B] became more withdrawn before the arrest, that she spent more time to herself. [B] herself testified that after the arrest, she felt relief, *362 that she could go home and not worry.... Rosenberg testified that symptoms from a traumatic experience, such as child sexual assault, can sometimes occur many years later." In our view, "[s]uch heavy reliance [on the withdrawal-related testimony] ... expose[s] its central role in persuading the jury to convict, as the government clearly understood that [the] statement was a powerful weapon in its arsenal." (Internal quotation marks omitted.) United States v. Stewart , supra, 907 F.3d at 689 .

After seeking to persuade the jury to infer guilt based on the mother's testimony about one of the behavioral symptoms identified by Rosenberg, the state cannot fairly argue that it was harmless to exclude P's conflicting testimony that he saw no significant behavioral **222 changes or depression in B. To the contrary, the exclusion of P's testimony deprived the defense of the ability, in its own summation to the jury, to undercut the state's argument by reminding the jury that P, who was among B's closest friends for the four years leading up to trial, had observed none of the many symptoms of sexual abuse that Rosenberg had identified. Cf. State v. Sawyer , 279 Conn. 331 , 360-61, 904 A.2d 101 (2006) (finding harm, in relevant part, because state repeatedly emphasized improperly admitted evidence in its closing argument), overruled in part on other grounds by State v. DeJesus , 288 Conn. 418 , 454-55 n.23, 953 A.2d 45 (2008) ; State v. Alexander , supra, 254 Conn. at 308 , 755 A.2d 868 (holding that prosecutor's improper remarks in closing argument were not harmless because they "directly addressed the critical issue in this case, the credibility of the victim and the defendant" [internal quotation marks omitted] ).

Lastly, the state argues, and the dissent agrees, that the defendant's ability to cross-examine B and her mother renders the error harmless. This argument ignores two important points. First, Rosenberg testified after B and her mother, and, therefore, the behavioral template provided by him was not available to the defense during the cross-examination of those key witnesses. More broadly, and perhaps more importantly, a criminal defendant is not constrained to present his defense through witnesses selected by the state. "If the accused [is] guilty, he should [nonetheless] be convicted only after a fair trial"; (internal quotation marks omitted) State v. Andrews , 313 Conn. 266 , 294, 96 A.3d 1199 (2014) ; which includes, among other things, an opportunity "to present [his] version of the facts as well as the prosecution's to the jury so it may decide where the truth lies." Washington v. Texas , 388 U.S. 14 , 19, 87 S.Ct. 1920 , 18 L.Ed.2d 1019 (1967). This does not mean that there are no limits on the defendant's right to present his defense as he wishes; see **223 State v. Wright , 320 Conn. 781 , 818-19, 135 A.3d 1 (2016) ; but, because P's testimony was admissible and could have made a substantial impact on the jury, the improper exclusion of this testimony cannot be deemed harmless. "[T]he truth is more likely to be arrived at by hearing the testimony of all persons of competent understanding who may seem to have knowledge of the facts involved in a case, leaving the credit and weight of such testimony to be determined by the jury or by the court ...." (Internal quotation marks omitted.) Washington v. Texas , supra, at 22, 87 S.Ct. 1920 . Just as "the prosecution is entitled to prove its case by evidence of its own choice," so, too, does the defendant deserve the same opportunity to defend himself. 8 Old Chief v. United States , 519 U.S. 172 , 186, 117 S.Ct. 644 , 136 L.Ed.2d 574 (1997). *363 It cannot be harmless error to "remove from the fact finder the very tools by which to make a credibility determination ...." State v. Little , 138 Conn. App. 106 , 123, 50 A.3d 360 , cert. denied, 307 Conn. 935 , 56 A.3d 713 (2012) ; see also Devincentz v. State , 460 Md. 518 , 562, 191 A.3d 373 (2018) (finding that complete exclusion of witness' testimony was not harmless error when "[t]he outcome of [the] case turned entirely on the relative credibility of the defendant and the accuser," because the exclusion "limited the jury's ability to assess [the accuser's] credibility ...."). "[W]here credibility is an issue and, thus, the jury's assessment of who is telling the truth is critical, an error affecting the jury's ability to assess a [witness'] credibility is **224 not harmless error." Devincentz v. State , supra, at 561, 191 A.3d 373 . Because P's testimony was necessary for the jury to assess B's credibility, we conclude that the exclusion of his testimony was not harmless.

The judgment of the Appellate Court is affirmed.

In this opinion PALMER and D'AURIA, Js., concurred.

DISSENT

ROBINSON, C.J., with whom KAHN, J., joins, dissenting.

I respectfully disagree with the conclusion in part III of the majority opinion, which concludes that the Appellate Court properly determined that the trial court's exclusion of testimony from P, the longtime boyfriend of B, the victim in this case, requires reversal of the judgment of conviction rendered against B's stepfather, the defendant, Fernando V. 1 State v. Fernando V. , 170 Conn. App. 44 , 153 A.3d 701 (2016). Even if the trial court improperly excluded P's testimony to the effect that B did not exhibit certain behaviors that may or may not be indicative of trauma from sexual abuse, I nevertheless have a fair assurance that this evidentiary error was harmless because it did not substantially sway the jury's verdict. I reach this conclusion particularly in light of circumstantial evidence corroborating B's allegations, the collateral nature of P's testimony, and the fact that other evidence-namely, the cross-examination testimony of B and her mother, G-provided support for the defendant's argument near-identical to that which would have been provided by P's testimony.

**225 Because I would reverse the judgment of the Appellate Court, I respectfully dissent.

I begin by noting my substantial agreement with the factual and procedural history recited in part I of the majority opinion. I also agree with part II of the majority opinion, which declines to consider the state's arguments that the trial court did not abuse its discretion when it precluded *364 P from testifying. 2 Finally, I agree **226 that, "[w]hen an improper evidentiary ruling is *365 not constitutional in nature, the defendant bears the burden of demonstrating that the error was harmful.... [W]hether [an improper ruling] is harmless in a particular case depends upon a number of factors, such as the importance of the witness' testimony in the prosecution's case, whether the testimony was cumulative, the presence or absence of evidence corroborating or contradicting the testimony of the witness on material points, the extent of cross-examination otherwise permitted, and, of course, the overall strength of the prosecution's case.... Most importantly, we must examine the impact of the ... evidence on the trier of fact and the result of the trial.... [T]he proper standard for determining whether an erroneous evidentiary ruling is harmless should be whether the jury's verdict was substantially swayed by the error.... Accordingly, a nonconstitutional error is harmless when an appellate court has a fair assurance that the error did not substantially affect the verdict." (Internal quotation marks omitted.) State v. Eleck , 314 Conn. 123 , 129, 100 A.3d 817 (2014) ; accord State v. Ritrovato , 280 Conn. 36 , 56-57, 905 A.2d 1079 (2006) (discussing factors in context of **228 exclusion of evidence). I part company from the majority with respect to its application of these well settled principles to the record in the present case, and specifically its conclusion that the exclusion of P's testimony requires a new trial. 3 *366 Given the lack of physical evidence in the present case, I acknowledge that the defendant's theory of the case focused on impeaching the credibility of B, a theory borne out in his closing argument as he described her testimony as "inconsistent," "contradictory," **229 "incomplete," and "noncorroborative evidence." See, e.g., State v. Osimanti , 299 Conn. 1 , 20-21, 6 A.3d 790 (2010) (reviewing summations to discern significant factual issues in case). Beginning with the importance of P's proposed testimony to that defense, I note that P testified in an offer of proof that he had been in a relationship with B "continuously" over the preceding four years, with no breaks, and that he considered them to be "boyfriend and girlfriend ...." P testified that, over that four year period, he had not noticed "any significant behavioral issues" with B, nor any "pronounced eating disorders," "suicidal thoughts," "severe depression," "anger or outbursts or violence," or "trouble with her focusing on issues or tasks at hand ...." 4 P also did not think that B's grades had "slipped, in any way, in the four years [he had] known her," and he had not "noticed any type of interruption in her playing of the flute," which was her main extracurricular activity Finally, P denied that the defendant had ever forbidden him from "dating," "talking to," "seeing," or "being alone" with B. The defendant offered P's testimony for several reasons: (1) to establish whether he had seen "behavior that has been testified to [that] may or may not be common with certain individuals"; (2) to indicate the nature of B's relationship with the defendant; and (3) to impeach the testimony of B and G.

Given these arguments, P's proposed testimony must be understood in the context of the earlier testimony of the state's expert witness, Larry M. Rosenberg, who is the clinical director of the Child Guidance Center of Southern Connecticut, an outpatient mental health clinic. Rosenberg had testified about the concept of delayed disclosure of sexual abuse. In connection with that topic, Rosenberg also testified about behavioral signs of the trauma resulting from sexual assault-such **230 as *367 withdrawal, depression, sleep disturbances, and declines in cognitive and educational functioning. The defendant sought to use P's testimony to establish that B had not manifested those behavioral signs that Rosenberg had testified were consistent with the trauma of sexual abuse.

Although P's testimony might have been crucially important standing alone, its relative value in this case is significantly diminished for two reasons. First, whether a person shows behavioral signs of having been sexually abused is by no means definitive evidence on that point. As Rosenberg testified during both direct and cross-examination, some sexual assault victims might show those trauma signs relatively soon, while other victims might never show any of those trauma signs. Some victims might experience no change in their ability to function in the near term, and might not manifest those signs until many years later, if at all.

Second, and more significantly, P's excluded testimony regarding the absence of these signs was consistent with that of B and G during both direct and cross-examination. B testified initially on direct examination that she had enrolled in college after graduating high school, and that she had maintained a grade point average of approximately 2.9 at both schools. She also testified that she had not experienced any lengthy absences from, or other problems at school or work because of behavioral or psychological reasons, noting that her only extended absence from high school was the result of a medical problem. B testified further that she was an active member of the college band, and that nothing had prevented her from pursuing that activity. B also contradicted her direct examination testimony that she was unable to have male friends, admitting that she had boyfriends during high school and that the defendant had not forbidden her from seeing them or having them as guests in the house.

**231 G testified similarly, stating that there had been no changes in B's personality around the ages of twelve or thirteen years old, when the abuse escalated from improper touching to intercourse, because "she was always a little shy." Although G had testified on direct examination that the defendant was strict with respect to B's grades, and preferred her to go out with female rather than male friends, she also confirmed that B had boyfriends during high school, and that the defendant had not interfered with those relationships. Moreover, while G testified that, in the year prior to the defendant's arrest in this case, B had acted "more withdrawn and ... that she would stay in her room," she then testified on cross-examination that B's activities had not changed, as she continued to enjoy reading and playing the flute from middle school into college. G also testified that B had always had a "timid" demeanor since coming to the United States as a child, and that it had not worsened during high school, although she would "stay in her room more often, locked up."

The testimony of B and G provided ample support for the defendant's theory of the case, even without P's similar testimony on point. In addition to emphasizing inconsistencies in the time, place, and nature of B's allegations, 5 the defendant's *368 closing argument relied on the testimony of B and G to argue in detail that B had not manifested behaviors consistent with sexual abuse **232 trauma. For example, defense counsel described as "contradictory" B's testimony on direct examination that "she was unable to have guy friends," and that the defendant "didn't like her talking to boys," observing that she had "admitted" during cross-examination that "she did have two boyfriends during high school, and [that the defendant] never objected to her having these boyfriends. That he never forbade her ... from seeing them, coming over to the house or in any way opposed to these relationships ...." Defense counsel also emphasized that B had not testified to any "effect on [her] grades," that she had "maintained a 2.9 through 3.0 consistently from middle school to college. She testified that her employment was never affected. There was no disruption in her extracurricular activities. She plays the flute, continues to play the flute. [G] also testified the same, that there was no changes, that [B] continued in those activities."

Defense counsel argued further that there was "no testimony by [B] that there was any behavioral changes. There's no testimony from [B] that she experienced any depression. No testimony from her that she experienced any suicidal ideations. No testimony that she experienced any eating disorders. No testimony from her that she had any violent tendencies. And more importantly, no evidence that after the alleged arrest of [the defendant], in 2011, did any of this come up. Which, as the State's own expert [witness] said, commonly is something that occurs. There's no evidence of any therapy or counseling ever received by [B]."

Turning to G's testimony, defense counsel argued that it was inconsistent with that of B, positing that G had "stated that [the defendant] did not like [B] talking to boys, but admitted [B] had boyfriends since freshman year in high school. And there was no evidence by [G] that [the defendant] ever objected to [B] having those relationships with those boys." Defense counsel further **233 emphasized that G's "testimony is noncorroborative of [B's] in that she didn't see any behavioral issues with [B]. Claimed [B] was always a bit timid, even since she came to the [United States] and there was no alleged inappropriate behavior. And that there was really no change. Didn't see any of [B's] grades slip. Didn't see [B] stop playing the flute. And never saw any inappropriate behavior, whatsoever, during the entire time that they were together, between [B] and [the defendant]."

Defense counsel then compared this testimony by B and G to Rosenberg's testimony: "[Rosenberg] stated that it is more common to have some behavioral issues in alleged victims, especially in their adolescent years, and especially after the disclosure is made. He said it's common. It happens. But there's no evidence of any of that."

Similarly, defense counsel also argued that the testimony of Vicki Smetak, a Norwalk Hospital pediatrician who had examined B after her disclosure, was not corroborative. The defense argued that Smetak had made "no physical findings of assault, whatsoever," and had stated "that there was no suicidal ideation or extreme behavioral *369 issues that she noted during the exam."

I disagree with the majority's conclusion that, because "P's testimony was necessary for the jury to assess B's credibility," it therefore "cannot be harmless error to remove from the fact finder the very tools by which to make a credibility determination ...." (Internal quotation marks omitted.) That conclusion is belied by the record in the present case, insofar as the jury had numerous tools by which it could assess the credibility of B's allegations, all of which were well highlighted by the defendant's closing argument. Specifically, the cross-examination of B and G, along with Smetak's testimony, gave the defendant ample support for his behavioral arguments, even without P's testimony.

**234 Further, the persuasive value of the behavioral arguments is diminished by Rosenberg's testimony that signs of sexual abuse may or may not be present in victims in any event, rendering P's testimony not a significant addition to the evidence in the defendant's favor.

I also disagree with the majority's reliance on the lack of physical evidence in the present case in support of its conclusion that the improper exclusion of P's testimony was harmful because the state's case was not strong. I acknowledge that, "[a]lthough the absence of conclusive physical evidence of sexual abuse does not automatically render the state's case weak where the case involves a credibility contest between the victim and the defendant ... a sexual assault case lacking physical evidence is not particularly strong, especially when the victim is a minor." (Citation omitted.) State v. Ritrovato , supra, 280 Conn. at 57 , 905 A.2d 1079 . In the present case, however, the state's case was significantly strengthened by other circumstantial evidence that corroborated B's testimony-namely, that D, B's half brother and the son of the defendant, had seen B and the defendant acting secretively on two separate occasions. Specifically, D, who was fourteen years old at the time of trial, testified that, on one occasion, he went to his parents' bedroom looking for the defendant, and that no one answered when he knocked on the door. When the door finally opened, he saw the defendant and B together in the room, with B putting her belt back on at that time. D also mentioned this incident in a statement to the police that the trial court admitted into evidence pursuant to Whelan . 6 In that document, D

**235 averred the following: "What I remember is that I went to look for my dad but the room was locked . I was just about to walk away and then I heard him call me and I just saw my sister putting on her belt." (Emphasis added.)

D's statement to the police also averred the following regarding a second incident: "I was ... looking for my dad and my sister told me he was in the garage. I just said ok because I already checked there. So I told my friend to walk downstairs and *370 I stayed upstairs and all I saw was my dad leave my sisters room." In my view, D's testimony and statement significantly strengthened the state's case, as they provided the circumstantial smoke to the fire of B's testimony. 7 See **236 State v. Beavers , 290 Conn. 386 , 418-20, 963 A.2d 956 (2009) (improper arson expert testimony that fire was intentionally set, which was based on "assessment of the defendant's credibility," was harmless because of "enormity of the circumstantial evidence against the defendant, namely, the evidence of his motive, his opportunity, his knowledge that the fire started in the basement, his possession of fire starting supplies on the morning of the fire, his intent as shown through his prior bad acts, and the uncontroverted and properly admitted expert evidence that refuted his attempt to blame the fire on [his son's] smoking"); cf. State v. William C. , 267 Conn. 686 , 709, 841 A.2d 1144 (2004) (noting that "distinct dearth of evidence corroborating the testimony of the victim, and the fact that the [excluded Department of Children and Families] records would serve to contradict her testimony, often through her own words, demonstrate that the state's case against the defendant was not particularly strong").

The harmlessness of the exclusion of P's testimony is even more apparent when the present case is considered in juxtaposition with those cases in which the central issue was the complainant's credibility and this court has found harmful evidentiary error to exist. First, P's proffered testimony did not pertain directly to the veracity of the complainant or the allegations themselves, **237 but only to whether B had shown certain behaviors that Rosenberg had testified might-or might not be-present *371 in a person experiencing the trauma of having been sexually assaulted. In contrast, cases where this court has found harmful evidentiary error involve improper evidence that more directly bolsters or undercuts the veracity of the complainant's testimony. See State v. Favoccia , 306 Conn. 770 , 807-11, 51 A.3d 1002 (2012) (admission of improper expert testimony that indirectly vouched for teenage victim's credibility was harmful when there was evidence that "battered [victim's] veracity [and] would give any reasonable juror pause," including testimony by complainant's father "that he 'did not know whether to believe' her allegations against the defendant" because, as corroborated by testimony of his long-term girlfriend, it was factually impossible for victim's allegations to be true); State v. Ritrovato , supra, 280 Conn. at 57-58 , 905 A.2d 1079 (improper preclusion of defendant from questioning victim about her claim of virginity was harmful when it pertained to her truthfulness and "this emotionally charged subject was mentioned repeatedly ... during the state's case-in-chief" and, given lack of corroborating or physical evidence, testimony on this subject "would have cast sufficient doubt on [victim's] credibility to have influenced the jury's verdict on the sexual assault charges"); State v. Iban C. , 275 Conn. 624 , 641-45, 881 A.2d 1005 (2005) (improper expert bolstering via diagnosis of "child sexual abuse" was harmful as to one count of risk of injury to child in which "state's case rested almost entirely on the victim's credibility" with no physical or medical evidence, but was harmless with respect to second count of risk of injury to child, to which defendant had confessed); State v. William C. , supra, 267 Conn. at 707-708 , 841 A.2d 1144 (improper exclusion of Department of Children and Families records was harmful because "the information contained in [those] records evince[d], if believed by the trier of fact, a pattern of vacillations **238 with regard to the very allegations of abuse for which the defendant was standing trial," as well as victim's statements "that she would lie if she thought it necessary, and statements of the victim's physician as to the victim's capacity to distort reality and come to believe her distortions"); State v. Grenier , 257 Conn. 797 , 806-808, 778 A.2d 159 (2001) (expert testimony that improperly described child victim's accusations as "very credible" was harmful in case with no physical or medical evidence, and no corroboration beyond constancy of accusation, because it "struck at the heart of the central-indeed, the only-issue in the case, namely, the relative credibility of [the victim] and the defendant" [internal quotation marks omitted] ).

Finally, I observe there was no report of jury deadlock in this case to "indicate that the fact finder itself did not view the state's case against the defendant as particularly strong." State v. Angel T. , 292 Conn. 262 , 294, 973 A.2d 1207 (2009) ; see also State v. Favoccia , supra, 306 Conn. at 813-14 , 51 A.3d 1002 (concluding that deadlock followed by split verdict "indicates that the case was a close one in the eyes of the jury, making it more likely that the improper evidence might have tipped the balance"); State v. Angel T. , supra, at 294, 973 A.2d 1207 ("[t]he jury's deadlock in the present case renders more troubling its split verdict, following the Chip Smith charge, because the split verdict suggests that the jury had doubts concerning the victim's credibility as a general matter, as it failed to credit her testimony about the defendant's earlier attempts to molest her"). Instead, the jury in the present case returned a verdict of guilty on all counts after deliberating for several hours. In contrast to deadlock reports, this rapid verdict suggests that the trier of fact did not view this case as particularly close, an *372 assessment with which I wholeheartedly agree. 8 **239 Because the exclusion of P's testimony was, at most, harmless error, I conclude that the Appellate Court improperly reversed the trial court's judgment of conviction. I would, therefore, reverse the judgment of the Appellate Court and remand the case to that court with direction to affirm.

Accordingly, I respectfully dissent.