The opinion of the court was delivered by
S.L. REISNER, J.A.D.Defendant AO. appeals from his conviction for first-degree aggravated sexual assault on a child, C.I., and the sentence of eighteen years imprisonment with a nine-year parole bar.
We reverse the conviction on two grounds. First, we hold that defendant’s stipulation as to the admissibility at trial of polygraph test results, which he executed without benefit of counsel, was constitutionally invalid. We disagree with State v. Reyes, 237 N.J.Super. 250, 567 A.2d 287 (App.Div.1989), and in any event we conclude it is no longer good law, following the Supreme Court’s decision in State v. Domicz, 188 N.J. 285, 907 A.2d 395 (2006). Second, we conclude the trial court should have held a Rule 104 hearing pursuant to State v. Guenther, 181 N.J. 129, 854 A.2d 308 (2004), before precluding defendant from presenting evidence about an incident in which C.I., who had accused defendant of molesting her and then recanted, later accused another man of molesting her and then recanted that accusation.
*11I
We begin by briefly summarizing the chronology of this case. Defendant was living with his girlfriend, J.I., and her daughter, C.I. Defendant and his family, and J.I.’s family, are all Nigerian immigrants who speak English and Yoruba. On April 27, 2001, C.I. first reported that defendant had molested her several times during the past year. She initially told some friends at school that day and then told a school counselor later in the day. As a result, defendant was called to the Child Advocacy Center for questioning.
When defendant denied all of C.I.’s allegations and asked the police how he could clear his name, he was told that he could take a polygraph test. At this point, defendant had not been arrested and had no attorney. After waiving his Miranda,1 rights and being advised by an assistant prosecutor concerning the polygraph stipulation form, defendant signed the form and took a polygraph test. According to the State’s expert polygraph examiner, defendant failed the polygraph. He was arrested and charged with aggravated sexual assault.
C.I. subsequently recanted her allegations. Because the Division of Youth and Family Services (DYFS) concluded that the child’s family was not being supportive of her, DYFS removed the child from her home and placed her first in a shelter and then with foster parents. While in the shelter, the child accused another man of molesting her, but then recanted her allegation. In December 2002, after the child had been placed with a relative, she reaffirmed her allegations against defendant.
These were the most significant trial events. In the prosecutor’s opening statement, he highlighted expected testimony from the State’s expert on Child Sexual Assault Accommodation Syndrome (CSAAS). He told the jury that the State’s expert would explain why the lack of family support affected recantation. “You *12will hear from an expert ... Dr. Susan [Cohen Esquilin]. She will tell you there is this syndrome after children disclose, if they’re not supported by family, it is not unusual they recant or withdraw their testimony. Why is that? It’s very difficult to make this disclosure and without full family support, the child feels very uncomfortable, the child feels it’s their fault that the person has gone away---- We’re going to study that process, that lack of support.” The prosecutor also emphasized the fact that defendant failed the polygraph test.
According to C.I.’s grandmother2, at some point after C.I. reported the incidents, C.I.’s mother was reluctant to have the child examined by a State doctor at the Child Advocacy Center. Therefore, the grandmother insisted that C.I. be examined by a doctor whose offices were located above the grandmother’s store. However, this doctor refused to examine C.I. or to speak to her. The grandmother denied that there were any conversations “about soap” at the doctor’s office.
In an effort to show that C.I. was not supported by her family, the State presented testimony from a DYFS worker, who observed C.I.’s grandmother praying over her during a supervised visit on July 27, 2001. According to the DYFS worker, the grandmother placed holy oil on C.I.’s forehead “to take the evil out of her, take the spiritual demons away.” The grandmother also said that C.I. was ugly. The worker stopped the visit at that point because she believed it was inappropriate to call the child ugly and “to tell her that she had spiritual demons inside of her.” This incident occurred more than two months after the May date on which the child had recanted her accusation.
Over defense objections, and following a Rule 104 hearing, the State also presented fresh complaint evidence in the form of *13testimony from one of C.I.’s friends. According to the friend, on April 27, 2001, C.I. told a group of other children on the playground “that her father was abusing her, doing sex to her.” The friend told her teacher what C.I. had said. Immediately after this testimony, the trial judge gave the jury aTimiting instruction that the fresh complaint evidence “is not evidence that the sexual offense actually occurred or that [C.I.] is credible.”
The State also presented evidence from a school social worker who testified that on April 27, 2001, C.I. told her that her stepfather was “harassing her.” C.I. then explained that her stepfather “does it.” C.I. told the social worker that the last time an incident occurred was during spring break, which was between April 13 and April 20, 2001. A limiting instruction was also given concerning this testimony.
C.I. testified that defendant began living with her family in 1998. In 2001, when she was nine years old, defendant sometimes babysat for C.I. and her younger sister while C.I.’s mother was working a night shift. C.I. described several incidents in which defendant would climb on top of her and rock back and forth with his “private” in her “private.” During some of these incidents, she was dressed but could feel what defendant was doing through her clothing. She testified that on her tenth birthday, the defendant raped her. She also testified on direct examination concerning an incident when defendant attempted to have oral sex with her. At this time the lights were on and she observed “dark” “gooie stuff’ coming out of defendant’s “private.” On cross-examination, she testified that during this incident she could not see defendant’s “private” because “[i]t was dark.”
C.I. admitted that a month after the incident, she recanted her allegations. She explained that she did this because defendant’s mother “kept telling me that it didn’t happen” and that C.I. “should [t]ell them that it didn’t happen.” She also testified that in May 2001, her grandmother took her to a doctor, who did not examine her but “told me to take this thing and put in my private part.” According to C.I., it was her grandmother’s idea to use the *14“soap” “[s]o that my thing would close.” She testified that her grandmother and the doctor were speaking Yoruba.
On cross-examination, C.I. was asked whether she had ever watched the Playboy channel on television with her mother or stepfather. She testified that “when they’re watching adult stuff, he would send me to my room and I wouldn’t be able to watch ... that ... nasty stuff they were doing on TV.”
There was no medical evidence to corroborate the child’s allegations. When the child was taken to be examined at the Child Protection Center on May 2, 2001, she recanted and told the doctor that she “had lied.” Therefore, according to the Center’s policies, she was not examined.
Called as the State’s witness, Dr. A.N. testified that he refused to examine C.I. because her grandmother told his receptionist that the purpose of the examination was to determine whether she had been raped. He testified that he did not examine C.I. at all or prescribe anything for her. Instead, he recommended that she be examined by a doctor at the Child Advocacy Center. He also testified that he was from Ghana, he did not speak Yoruba, which is a Nigerian language, and he communicated with C.I.’s grandmother briefly and exclusively in English. His testimony contradicted C.I.’s testimony that her grandmother took her to an African doctor who spoke to them in Yoruba and prescribed a soap or cream to be placed in C.I.’s vagina.
C.I.’s mother, J.I., confirmed that in 2000 and 2001 there were times when defendant babysat for C.I. when she went to work. She testified that on April 27, 2001, she was called to C.I.’s school and later went to the Child Advocacy Center with the grandmother. During the drive, the grandmother was yelling at J.I. that her “bastard child” was going to get defendant in trouble and that J.I. should tell the Prosecutor that she knew nothing. J.I. contended that C.I. recanted because her grandmother told her to do so. J.I. also admitted that when C.I. asked her on April 28, 2001, where defendant was, she told her that defendant was “in the cage” because of what C.I. had said. According to J.I., before the three *15of them went to see the African doctor, the grandmother gave C.I. some aloe lotion to put in her vagina “to make everything small.” J.I. admitted that she did not mention this when she gave a statement to the police.
J.I. also admitted telling the police in May 2001 that she thought C.I. was making up the allegations because she was angry at defendant for not letting her watch television. J.I. contended that she herself felt intimidated by the grandmother because the grandmother practiced voodoo, and she had seen a voodoo shrine in the grandmother’s house with her picture near it. She also admitted that she had initially supported defendant in his defense; she denied that her change of position had anything to do with the fact that defendant eventually married another woman.
During J.I.’s testimony, the court held a Rule 104 hearing before ruling that defense counsel could not elicit testimony about an alleged conversation that J.I. had with C.I.’s second grade teacher concerning an allegation that C.I. either tried to kiss another student or otherwise was “acting out sexually” at school in some unspecified manner.
Dr. Cohen Esquilin testified about child sexual abuse accommodation syndrome (CSAAS). She testified that it was not a diagnosis. However, the elements of CSAAS help to explain why children who are abused often delay reporting the abuse. She also was asked what is “likely to happen” if the child’s family is not supportive when the child does disclose the abuse. She responded that “that’s when you get into the fifth stage which is the recantation or retraction.” On cross-examination, she confirmed that CSAAS did not address the behavior of children who had not been abused but might have claimed to be abused; the studies underlying the theory of CSAAS were based only on children “who we know have been abused.” The judge instructed the jury that they were not to “consider Dr. Cohen’s testimony as offering proof [that abuse] actually occurred in this ease.”
The State then played for the jury videotapes of police interviews of C.I. taken on April 27, 2001, in which she accused *16defendant, and on May 4, 2001, in which she recanted her accusation.
During the trial, defense counsel sought to obtain information concerning an incident at the children’s shelter in the fall of 2001, in which C.I. claimed she had been raped and then recanted the allegation. The State provided the trial judge with DYFS records and police reports concerning the incident for review in camera. After reviewing the videotape and the records, the trial judge concluded that a hearing on relevance was not required. She concluded based on the DYFS records and the Prosecutor’s Office records that the allegations were substantiated. However, she did not place any findings on the record concerning why she determined that those agencies’ findings were worthy of credence, particularly since the Prosecutor decided not to prosecute the person C.I. accused. The judge also reasoned that the incident at the shelter happened “afterwards” and did not, for example, help to explain “where she learned the sexual behavior” C.I. described with defendant. The judge also concluded that there was nothing helpful to defendant in C.I.’s psychological records, noting that the child did not tell the psychologist a materially different version of the facts in either incident.
We next review the evidence concerning the polygraph. According to a police witness, on April 27, 2001, defendant gave a statement to police in which he denied sexually assaulting C.I. Toward the end of the statement, defendant was asked how he had been treated by the prosecutor’s office and the police. Defendant replied that he felt he had been treated “like a liar” because the police believed C.I. He was then asked “Did we tell you that we are only here to seek the truth?” When he responded ‘Tes,” he was then asked “Do you agree to take a stipulated polygraph examination to determine if you’re telling the truth?” He responded “Yes.” Defendant then met with an assistant prosecutor who spoke with him for about thirty minutes to review a five-page polygraph stipulation, which defendant signed.
*17In that five-page agreement, which defendant signed without advice of counsel, defendant agreed, among other things, that the results of the polygraph examination “shall be admissible as evidence” in Grand Jury proceedings and at trial; “the polygraph examiner is acknowledged by both parties to be an expert” for purposes of the expert’s later trial testimony; that defendant “expressly waives any and all objections to the admissibility of such expert testimony;” that defendant waives “the right to introduce another polygraph expert or any other person, other than the subject himself, in reference to the original polygraph expert’s testimony;” and that “[t]he results of any other polygraph examination shall not be admissible unless covered by a separate stipulation agreement.”
The State presented testimony from Detective John Kaminskas who had administered the polygraph test to defendant shortly after midnight on April 28, 2001. He testified that there were “indications of deception” when defendant gave negative answers to questions concerning sexual activity with the child. On cross-examination he admitted that polygraph results are not considered “100 percent reliable” but they are considered in the “95 to 97 percent accuracy rate.” However, he testified that the results of tests he had done were 100% accurate, in the sense that no proof had ever been presented to him that any of the subjects whom he determined were deceptive were actually telling the truth. Kaminskas denied considering whether there might be any cultural differences concerning discussion of sexual matters that would affect defendant’s response to the polygraph. Kaminskas did not disclose to defendant that the test was not 100% accurate before defendant agreed to take the test and stipulated to its use at trial.
Defendant did not testify at the trial. However, the defense called defendant’s younger sister who lived with her mother, C.I.’s grandmother. She testified that in April 2001, C.I. used to come over to the grandmother’s house every weekend and would stay overnight in the sister’s room. At that time, the sister was a senior in high school. During these visits, C.I. never complained *18that defendant was sexually abusing her. After defendant was arrested, C.I. told the sister that defendant had not sexually assaulted her. Contrary to the testimony of C.I.’s mother, the sister also testified that there were no voodoo shrines in the grandmother’s house.
In summation, the prosecutor once again invoked Dr. Cohen Esquilin’s testimony about the importance of a child getting family support when she complains. He then argued that C.I.’s family forced her to recant. He also invoked the expert’s testimony again that “there are empirical studies based on children who are actually sexually abused and these are the elements that they display,” including “accommodation and unconvincing disclosure.” However, the thrust of the prosecutor’s comments was that the grandmother had intimidated the child into recanting. The prosecutor also invoked the polygraph results, contending that the results were “100 percent accurate” and “the machine said [defendant] lied.”
The prosecutor asked the jury to consider how poorly the child was doing in school and how much better she did after she left J.I.’s and A.O.’s home. The trial judge overruled defense counsel’s objection to this comment, although there had been no testimony to tie this to any possible sexual molestation.
In charging the jury, the judge instructed them that the polygraph result did “not by itself prove any element of the crime” but “merely indicates [that at] the time [the expert] questioned the defendant in his expert opinion the defendant was not answering truthfully the relevant questions asked.” He also thoroughly instructed the jury that Dr. Cohen Esquilin’s testimony was not to be used as proof that sexual abuse took place in this ease.
II
On this appeal, defendant raises the following issues for our consideration:
POINT I: THE TRIAL COURT’S EVIDENTIARY RULINGS WERE ERRONEOUS AND, AS A WHOLE, DEPRIVED DEFENDANT OF HIS STATE *19AND FEDERAL CONSTITUTIONAL RIGHTS TO A FAIR TRIAL AND DUE PROCESS OF LAW BECAUSE THEY SKEWED THE JURY’S DELIBERATIVE PROCESS ON THE CENTRAL ISSUE OF THE CHILD’S CREDIBILITY. (Partially Raised Below)
A. The Trial Court Erred In Permitting Sergeant Kaminskas To Testify As An Expert Witness And Tell The Jury That Defendant Had Failed A Lie-Detector Test, (Raised Below)
B. The Trial Court Erred In Permitting Dr. [Cohen Esquilin’s] Expert Testimony About Child Sexual Abuse Accommodation Syndrome Because This Testimony Went Beyond Permissible Limits And, Combined With The Lie-detector Results, Further Infringed Upon The Jury’s Determination Of The Child’s Credibility. (Not Raised Below)
C. The Trial Court Erred In Precluding Defendant From Cross-examining State Witnesses Or Introducing Evidence Of Other Incidents That Tended To Undermine The Credibility Of The Child’s Accusations Against Defendant And Undercut The State’s Charges Against Him. (Raised Below)
D. The Jury’s Consideration Of The Lie-Detector Results And Expert Testimony On Child Sexual Abuse Accommodation Syndrome, The Withholding From Its View Of The Child’s Recantation Of Sexual Abuse Charges Against Another Man, And The Lack Of Any Corroborating Evidence That The Alleged Assaults Had Even Occurred, Renders The Verdict Below Too Unreliable To Stand.
POINT II: REMAND IS REQUIRED BECAUSE DEFENDANT’S SENTENCE IS UNCONSTITUTIONAL, UNJUSTIFIED, AND IMPROPER.
We begin by addressing defendant’s attack on the use of the polygraph results, an issue as to which we required supplemental briefing. In State v. McDavitt, 62 N.J. 36, 297 A.2d 849 (1972), our Supreme Court, in a narrowly tailored decision, held that the results of a polygraph test were admissible where the defendant, who was represented by counsel, agreed mid-trial, to take a polygraph test. The Court did not reach the issue of whether polygraph tests should generally be admissible in criminal trials, nor did the Court address the issue of whether an unrepresented defendant may enter into such a stipulation.
Within the framework of this appeal it is unnecessary to reconsider the broad issue of admissibility vel non of polygraph test evidence in a criminal case. Indeed, the record herein is inadequate for that purpose. As heretofore noted, the general rule has been one of exclusion. However, within recent months two trial courts in criminal eases, after conducting extensive hearings on the present reliability of polygraph tests, found that such testing is now generally accepted by authorities in the field and is capable of producing highly probative evidence in a court of law when properly used by competent, experienced examiners.
*20Here we have a much more narrow issue in view of the prior stipulation as to admissibility entered into by defendant arid his attorney and the State. The circumstance under which the stipulation came into existence is also a consideration since defendant, over the prosecutor’s objection, insisted on injecting into the case his prior offer to take a lie detector test “to prove my innocence.”
[Id. at 44-45, 297 A.2d 849 (emphasis added; citations omitted).]
The Court set guidelines for the admissibility of stipulated polygraph results:
We conclude that polygraph testing has been developed to such a point of reliability that in a criminal ease when the State and defendant enter into a stipulation to have defendant submit to a polygraph test, and have the results introduced in evidence, such stipulation should be given effect. Polygraph testing has sufficient probative value to warrant admission under these circumstances.
Of course, it must appear that the stipulation is clear, unequivocal and complete, freely entered into with full knowledge of the right to refuse the test and the consequences involved in taking it. It must also appear- that the examiner is qualified and the test administered in accordance with established polygraph techniques.
[Id. at 46, 297 A.2d 849.]
State v. Powell, 98 N.J. 63, 484 A.2d 659 (1984), adopted the dissenting opinion of Judge Shebell, concluding that a defendant cannot avoid a polygraph stipulation where defendant understood the stipulation but “did not believe it.” State v. Powell, 197 N.J.Super. 191, 195, 484 A.2d 698 (App.Div.1983) (Shebell, J., dissenting). The defendant in that ease happened to have been unrepresented when he entered into the stipulation, but Powell did not address the question whether an unrepresented defendant can validly agree to a polygraph stipulation. The issue was not raised.
The issue was directly addressed in State v. Reyes, 237 N.J.Super. 250, 567 A.2d 287 (App.Div.1989). There the defendant contended “that the polygraph examination results should be suppressed because defendant was unrepresented by counsel when he entered into the stipulation.” Id. at 263, 567 A.2d 287. In rejecting defendant’s contention that the process violated his Sixth Amendment right to counsel, the panel reasoned that defendant had not yet been charged with a crime at the time he entered the stipulation, and thus his “Sixth Amendment right to counsel *21had not yet attached.” Ibid. The panel also noted that the Sixth Amendment right to counsel could be waived. Id. at 264, 567 A.2d 287 (citing Faretta v. California, 422 U.S 806, 95 S.Ct. 2525, 45 L.Ed.2d 562 (1975)). The Reyes court reasoned that McDavitt established “the terms and conditions of a polygraph stipulation and it does not prohibit the waiver of the right to counsel at the signing of the stipulation.” Ibid.
In a concurring opinion, Judge D’Annunzio expressed “reservations concerning the enforceability at trial of an uncounseled polygraph stipulation.” Id. at 265, 567 A.2d 287 (D’Annunzio, J., concurring). Judge D’Annunzio noted that, as in this case, the stipulation required defendant to contract away “a number of basic rights, including the right to introduce the testimony of another expert witness” to challenge the polygraph administrator’s testimony. Judge D’Annunzio disagreed with the majority that Powell or McDavitt had addressed the validity of an uncounseled stipulation. He further reasoned that:
[r]eliance on the principles that the right to counsel does not attach until the initiation of adversary criminal proceedings and that a defendant has the right to waive counsel and his Fifth Amendment privileges is an inadequate response to the issue. There is a significant difference between giving a statement to the police which may be admitted against a defendant at trial and agreeing to the admissibility of evidence as well as to the exclusion of rebutting probative evidence.
[Id. at 267-68, 567 A.2d 287.]
He also cited decisions from other states that required polygraph stipulations to be signed by defense counsel. See State v. Valdez, 91 Ariz. 274, 371 P.2d 894, 900 (1962); State v. Renfro, 96 Wash.2d 902, 906, 639 P.2d 737 (1982), cert. denied sub nom, Renfro v. Washington, 459 U.S. 842, 103 S.Ct. 94, 74 L.Ed.2d 86 (1982); State v. Pederson, 44 Wash.App. 391, 722 P.2d 127 (1986), review denied, 107 Wash.2d 1005 (1986)3. One “likely reason for the requirement is that the decision to sign a polygraph stipulation is essentially a decision not to object to otherwise inadmissible *22evidence and thus could be classified as trial strategy, ordinarily the exclusive domain of the lawyer.” Reyes, supra, 237 N.J.Super. at 266-67 n. 1, 567 A.2d 287 (quoting Pederson, supra, 722 P.2d at 129).
Although Judge D’Annunzio concluded that admission of the polygraph in that case was harmless error due to the overwhelming evidence of defendant’s guilt, id. at 268, 567 A.2d 287, we believe his reasoning on the admissibility issue was correct and we agree with it. Like Judge D’Annunzio, we conclude that a polygraph stipulation is fundamentally different than an agreement to waive Miranda rights and make a statement to police. We also agree that while the Sixth Amendment right to counsel does not attach in the investigative stage, this does not mean that an uneounseled defendant may, at this stage, be induced to make binding decisions concerning trial strategy which ordinarily would be made in consultation with trial counsel. See Pederson, supra, 722 P.2d at 129.
As the stipulation in this case graphically illustrates, defendant not only agreed that the test results would be admissible in evidence, he also waived his right to offer contradictory evidence at trial, including his right to challenge the expertise of the polygraph examiner and his right to offer in evidence the results of any other polygraph test. This is roughly analogous to agreeing not only to give a confession to police but agreeing in advance not to recant the confession or otherwise challenge its reliability. Or, to use an example from drunk driving cases, it is analogous to a defendant agreeing to take a breathalyzer test and also agreeing not to challenge the test results. Further, as the Pederson court reasoned, in signing the stipulation, the defendant agreed “not to object to otherwise inadmissible evidence,” an exercise in “trial strategy, ordinarily the exclusive domain of the lawyer.” Pederson, supra, 722 P.2d at 129. We question whether any uncounseled defendant could understand the strategic implications of such an agreement.
*23The stipulation constituted a significant waiver of defendant’s Sixth Amendment trial rights, and it was entered into based solely on advice from an assistant prosecutor. Moreover, we find nothing in the agreement that advised defendant as to the statistical likelihood of inaccurate results, and no information as to the training of the polygraph examiner so he could make an informed decision whether to stipulate to the examiner’s expert status. If this were a consumer contract, we might deem it unconscionable.
While our Supreme Court has not yet addressed the admissibility of uncounseled polygraph stipulations, we believe the Court would hold them inadmissible. McDavitt and Powell do not answer the question before us. Rather, we base our conclusion on the Court’s recent decisions concerning the questionable reliability of polygraph tests, and the Court’s decisions carefully controlling the circumstances under which a defendant may waive the right to trial counsel.
In State v. Domicz, 188 N.J. 285, 907 A.2d 395 (2006), the Court rejected arguments for the admissibility of an unstipulated polygraph test, emphasizing the narrowness of its holding in McDavitt:
It is instructive to note that the “circumstance under which the stipulation came into existence Lin McDavitt ]” was “a consideration” for the Court in deciding that ease. In a criminal jury trial, the defendant in McDavitt, over the prosecutor’s objection, testified that he offered to submit to a polygraph examination after his burglary arrest. That testimony, this Court noted, should not have been permitted. But the door was opened. On cross-examination, in response to a question by the prosecutor, the defendant stated he would be willing to take a polygraph test that day. Eventually, the defendant and the State entered into a court-approved stipulation, agreeing to the admissibility of the results of a polygraph test. Unfortunately for the defendant, the examiner found him to be untruthful. From those unusual facts was born the McDavitt exception.
[Id. at 312, 907 A.2d 395 (citations omitted).]
The Court also emphasized the lack of consensus as to the reliability of polygraph tests, and in light of that controversy, declined to “revisit McDavitt’s narrow holding.” Id. at 313, 907 A.2d 395.
There is a lack of scientific consensus concerning the reliability of polygraph evidence, which in turn is reflected in the disagreement among state and federal *24courts concerning the admissibility of such evidence. United States v. Scheffer, 523 U.S. 303, 309-12, 118 S.Ct. 1261, 1265-66, 140 L.Ed.2d 413, 419-21 (1998). In criminal cases, either in a jury or non-jury setting, the vast majority of states either ban polygraph evidence altogether or do not admit such evidence absent a stipulation between the State and the defendant. To our knowledge, New Mexico is the only state in which polygraph evidence is admissible without significant restriction in criminal trials, even absent a stipulation between the parties. Today, both “state and federal courts continue to express doubt about whether [polygraph] evidence is reliable.” In the more than thirty years since McDavitt, serious questions about the reliability of polygraph evidence remain.
[Id at 312-13, 907 A.2d 395 (citations omitted).]
The Court likewise declined to revisit McDavitt in State v. Castagna, 187 N.J. 293, 901 A.2d 363 (2006). In Castagna, the Court held that a defendant was entitled to cross-examine the State’s witness concerning the results of a polygraph test the witness took after entering a stipulation with advice of counsel. After taking and failing the polygraph, the witness changed her story in order to obtain a plea agreement. The Court emphasized that “[u]nlike in McDavitt, here the reliability of the polygraph test results was not important. It was [the witness’] belief that the polygraph test results revealed she had not told the truth in her second statement that was crucial.” Hence, “the trial court erred in denying defendants the right to cross-examine [her] concerning the polygraph test results, not because those results were reliable, but because the test results caused [the witness] to change her statement.” Id. at 311-12, 901 A.2d 363.
As our Court acknowledged in Domicz, supra, the Supreme Court of the United States summarized the debate over the reliability of polygraph tests in United States v. Scheffer, supra:
[T]here is simply no consensus that polygraph evidence is reliable. To this day, the scientific community remains extremely polarized about the reliability of polygraph techniques. 1 D. Faigman, D. Kaye, M. Saks, & J. Sanders, Modem Scientific Evidence 565, n. + 14-2.0, and § 14-3.0 (1997); see also 1 P. Giannelli & E. Imwinkelried, Scientific Evidence § 8-2(C), pp. 225-227 (2d ed. 1993) (hereinafter Giannelli & Imwinkelried); 1 J. Strong, McCormick on Evidence § 206, p. 909 (4th ed. 1992) (hereinafter McCormick). Some studies have concluded that polygraph tests overall are accurate and reliable. See, e.g., S. Abrams, The Complete Polygraph Handbook 190-191 (1968) (reporting the overall accuracy rate from laboratory studies involving the common “control question technique” polygraph to be “in the range of 87 percent”). Others have found that polygraph tests assess *25truthfulness significantly less accurately — that scientific field studies suggest the accuracy rate of the “control question technique” polygraph is “little better than could be obtained by the toss of a coin, ” that is, 50 percent. See Iacono & Lykken, The Scientific Status of Research on Polygraph Techniques: The Case Against Polygraph Tests, in 1 Modern Scientific Evidence, supra, § 14-5.3, p. 629 (hereinafter Iacono & Lykken).
[Scheffer, supra, 523 U.S. at 309-10, 118 S.Ct. at 1265, 140 L.Ed.2d at 419-20 (emphasis added).]
We conclude it is fundamentally unfair to permit an uncounseled defendant to stake his fate on what may be the equivalent of a coin toss, and we do not believe McDavitt can be stretched that far. See State v. Renfro, supra, 639 P.2d at 739 (‘When there is a [counseled] stipulation, the prosecution and the defense, knowing that the degree of reliability is open to question, in effect gamble that the test will prove favorable to them.”)
The Court’s carefully-crafted limitations on a defendant’s ability to waive the right to trial counsel, as set forth in State v. Crisafi, 128 N.J. 499, 608 A.2d 317 (1992), and recently expanded in State v. Reddish, 181 N.J. 553, 859 A.2d 1173 (2004), support our conclusion. In Crisafi, the Court recognized the critical importance of trial counsel.
The Sixth Amendment “embodies a realistic recognition of the obvious truth that the average defendant does not have the professional legal skill to protect himself when brought before a tribunal with power to take his life or liberty, wherein the prosecution is presented by experienced and learned counsel.” Assistance of counsel is essential to ensuring fairness and due process in criminal prosecutions. Without counsel, the right to a fair trial would mean little, for it is through counsel that defendants secure their other rights.
[128 N.J. at 509, 608 A.2d 317 (citations omitted).
Consequently, “[i]t is for the court to determine whether an accused has knowingly and intelligently waived that right and to establish the waiver on the record.” Ibid. Moreover, in Crisafi, the Court mandated trial judges to conduct an extensive and searching inquiry to determine whether a defendant has knowingly and intelligently made the decision to waive the right to counsel.
To ensure that a waiver of counsel is knowing and intelligent, the trial court should inform pro se defendants of the nature of the charges against them, the statutory defenses to those charges, and the possible range of punishment. The colloquy between the court and the defendant will test the defendant’s understanding of the *26implications of the waiver, and will provide appellate courts with an objective basis for review____
In general, the court should also inform defendants of the technical problems they may encounter in acting as their own counsel and of the risks they take if their defense is unsuccessful. Further, the court should inform the defendants that they must conduct their defense in accordance with the relevant rules of criminal procedure and evidence, that a lack of knowledge of law may impair their ability to defend themselves, and that their dual role as attorney and accused might hamper the effectiveness of their defense. Also, the court should explain to the defendants the difficulties in acting as their own counsel and should specifically advise the defendants that it would be unwise not to accept the assistance of counsel.
[Id. at 511-12, 608 A.2d 317 (citations omitted).]
In State v. Reddish, the Court amplified and extended the inquiry required by Crisafi to further protect a defendant from an insufficiently-informed waiver of Sixth Amendment rights:
We take this opportunity to amplify our directive in Crisafi, supra, that courts engage in a penetrating examination of the knowingness and intelligence of a defendant’s attempted waiver of the assistance of counsel. We encourage trial courts to explore subjects that are inherent in, or offshoots of, those identified in Crisafi. By way of illustration, those additional areas would include whether defendant will experience difficulty in separating his roles as defendant and counsel; whether defendant understands that he not only has the right not to testify, but also the right not to incriminate himself in any manner; whether he understands that he could make comments as counsel from which the jury might infer that he had knowledge of incriminating evidence (and the difficulty in avoiding such comments); and whether he fully understands that if he crosses the line separating counsel from witness, he may forfeit his right to remain silent and subject himself to cross-examination by the State.
... Further, as with all areas of inquiry regarding the decision to waive counsel, the trial court must question defendant to ascertain whether he actually understands the nature and consequences of his waiver.
[Reddish, supra, 181 N.J. at 593-94, 859 A.2d 1173 (citations omitted).]
See also State v. Figueroa, 186 N.J. 589, 593, 897 A.2d 1050 (2006).
As we have previously explained, in signing the polygraph stipulation, defendant was making litigation-related decisions that ordinarily would be made during trial preparation with the assistance of his trial counsel. By inducing defendant to sign the stipulation, the State was in effect causing him to waive important aspects of the right to trial counsel without any of the safeguards required by Crisafi and Reddish. If a defendant cannot waive the Sixth Amendment right to trial counsel without the scrupulous and *27searching inquiry our Court has required, such a waiver certainly cannot legitimately be accomplished in a law enforcement setting, with defendant’s only advice coming from an assistant prosecutor.
We conclude that defendant’s uncounseled stipulation and the polygraph results should not have been admitted at trial. Admission of this evidence was plain error because it impinged upon defendant’s Sixth Amendment right to counsel and, as discussed below, it had a clear capacity to produce an unjust result. R. 2:10-2; see State v. Macon, 57 N.J. 325, 335, 273 A.2d 1 (1971).
The devastating effect of a polygraph in tilting the balance of an otherwise close case cannot be underestimated. The impact in this case was magnified because the prosecution presented the test as a highly reliable “lie detector” and encouraged the jury to rely on it in reaching its verdict. Unlike Reyes, supra, in this ease the State’s evidence was far from overwhelming. Without the polygraph, the State’s case depended entirely on the uncorroborated testimony of a child witness who had recanted and then withdrawn her recantation. None of the State’s other evidence was admissible to prove that the child had been molested or even to prove that she was a credible witness. But the prosecutor’s closing remarks emphasized to the jury that the polygraph was “100%” accurate, thus suggesting that the jury make its decision based on the test results. Because the polygraph evidence may well have made the difference between conviction and acquittal in this case, the conviction must be reversed and the matter remanded for retrial. See Macon, supra, 57 N.J. at 335-36, 273 A.2d 1.4
We also conclude that prior to the retrial, the judge must conduct further proceedings to determine whether defendant is entitled to question witnesses about a subsequent incident in *28which C.I. accused another man of molesting her and then recanted that accusation.
In State v. Guenther 181 N.J. 129, 854 A.2d 308 (2004), the Court created a narrow exception to N.J.R.E. 608, based on principles of fairness rather than the Confrontation Clause, to permit a defendant to attack a victim’s credibility by presenting evidence of a prior false accusation. The Guenther Court limited its holding to eases in which the victim-witness’ credibility “was the central issue in the case.” Id. at 156, 854 A.2d 308. We see nothing in the Guenther opinion that would necessarily limit the holding to only a prior false accusation, as opposed to a false accusation made a short time after the accusation against the defendant. Recognizing the need to avoid “side-show” mini-trials on the validity of the prior accusation, id. at 155, 854 A.2d 308, the Court left the issue of admissibility to the trial court’s “sound discretion.” Id. at 158, 854 A.2d 308. However, the trial judge must exercise that discretion after making detailed findings:
In deciding whether to permit the impeachment of a victim-witness who allegedly made a prior false accusation, trial courts must first conduct an admissibility hearing pursuant to N.J.R.E. 104. At that hearing, the court must determine by a preponderance of the evidence whether the defendant has proven that a prior accusation charging criminal conduct was made by the victim and whether that accusation was false. That standard strikes the right balance, placing an initial burden on the defendant to justify the use of such evidence while not setting an exceedingly high threshold for its admission. We note that the admission of this type of specific conduct evidence is an exception to N.J.R.E. 608 and should be limited only to those circumstances in which the prior accusation has been shown to be false. Among the factors to be considered in deciding the issue of admissibility are:
1. whether the credibility of the victim-witness is the central issue in the case;
2. the similarity of the prior false criminal accusation to the crime charged;
3. the proximity of the prior false accusation to the allegation that is the basis of the crime charged;
4. the number of witnesses, the items of extrinsic evidence, and the amount of time required for presentation of the issue at trial; and
5. whether the probative value of the false accusation evidence will be outweighed by undue prejudice, confusion of the issues, and waste of time.
If the court, pursuant to its gate-keeping role, determines that evidence of the prior false accusation is admissible, the court has the discretion to limit the number of witnesses who will testify concerning the matter at trial. The court must ensure *29that testimony on the subject does not become a second trial, eclipsing the trial of the crimes charged.
[Id. at 157, 854 A.2d 308.]
In this case, the trial judge apparently reviewed the DYFS and Prosecutor’s files in camera and noted that DYFS and the Prosecutor had substantiated the child’s allegations, but she did not set forth on the record any facts concerning the thoroughness of the investigations or the evidentiary basis for those agencies’ conclusions. Moreover, the judge did not actually hold a Rule 104 hearing and did not make the findings required by Guenther. The second recantation may be a critical issue for the defense, see id. at 156, 854 A.2d 308, and it is the judge and not the agencies who must make the decision as to its admissibility. See also State v. R.E.B., 385 N.J.Super. 72, 81, 895 A.2d 1224 (App.Div.2006) (Failure to hold a Guenther hearing was prejudicial error where the child’s credibility was the central issue.). Prior to the retrial, the judge must hold a Guenther hearing.
Finally, since the case will be retried, we discuss two additional issues which we noted in the record. In emphasizing Dr. Cohen Esquilin’s opinion that lack of family support would lead to a child recanting her testimony, the prosecutor’s opening and closing remarks came close to suggesting that the expert’s testimony was evidence that molestation occurred in this case as opposed to being a more general explanation as to why recantation might occur in cases of child sex abuse. See State v. P.H., 178 N.J. 378, 840 A.2d 808 (2004); State v. J.Q., 130 N.J. 554, 617 A.2d 1196 (1993). The prosecutor’s closing arguments also referred to the child’s lack of progress in school as evidence that she was molested while she lived with A.O. There was no expert or other testimony to support that contention. We appreciate the difficulty of presenting a case that hinges only on a child’s testimony and the temptation to reach for extraneous evidence to bolster that testimony. But on retrial, the prosecutor shall refrain from comments about the child’s progress in school and shall limit *30comments about the expert testimony to the purpose for which it was properly introduced.5
Reversed and remanded.
Miranda v. Arizona, 384 U.S. 436, 86 S.Ct. 1602, 16 L. Ed.2d 694 (1966).
This witness was defendant's mother. While C.I. was not defendant’s biological child, witnesses confirmed that he treated C.I. as his daughter and that she regarded defendant’s mother as her grandmother. Hence, we will refer to this witness as the grandmother.
Ohio has also adopted the Valdez rule that defendant must be represented by counsel in entering into a polygraph stipulation. State v. Souel, 53 Ohio St.2d 123, 372 N.E.2d 1318 (1978).
Admitting the polygraph results may prejudice the defense in other ways. Like the admissibility of a defendant’s prior convictions, the admissibility of a failed polygraph test may also cause a defendant to refrain from testifying at trial.
Defendant's remaining contentions, concerning the admissibility of the expert’s testimony and other asserted trial errors relating to his conviction, are without sufficient merit to warrant discussion in a written opinion. R. 2:11— 3(e)(2).