Brickley, J.
Defendant, Albert Lee, III, was convicted of first-degree felony murder1 and sentenced to life imprisonment. We granted leave to appeal limited to:
(1) Whether the testimony of witnesses who had been hypnotized was properly admitted; and
(2) Whether similar acts evidence was properly admitted.
We hold first that the rules of People v Gonzales2 and People v Nixon3 were violated and thus that it was error for the trial court to admit the identification testimony of the hypnotized witnesses. Second, the similarities between the charged crime and the similar acts testimony satisfies the standards of People v Golochowicz, 413 Mich 298; 319 NW2d 518 (1982).
i
Procedurally this case has a long history. Following his conviction, defendant moved for a new trial which was denied. Defendant appealed by right. The Court of Appeals held that the standards of People v Gonzales, 415 Mich 615; 329 NW2d 743 (1982), modified 417 Mich 968; 336 NW2d 751 (1983), dealing with hypnotized wit[64]*64nesses were not satisfied.4 However, the Court of Appeals held that because this case was decided before Gonzales its holding was inapplicable "but that reversal is nevertheless required in a preGonzales case if defendant was prejudiced by admission of testimony as to recollections induced by hypnosis.” Thereafter the Court found that the identification procedures were not "unnecessarily suggestive or conducive to irreparable misidentification,” upholding the trial court’s admission of the testimony of hypnotized witnesses.
Judge Maher dissented on the hypnosis issue. He concluded that on the basis of the factors pronounced by this Court in People v Hampton, 384 Mich 669; 187 NW2d 404 (1971), People v Gonzales, supra, should be applied retroactively.5 Thus, he found that defendant was denied a fair trial because the eyewitness testimony was tainted by hypnosis.
Relative to the similar acts testimony, the Court of Appeals found that the facts were sufficiently substantial to justify admission to show identity. The Court held that the trial court did not abuse its discretion and that the evidence was not erroneously admitted.
A request for review was filed with this Court and it was held in abeyance pending release of another hypnosis case, People v Nixon, 421 Mich 79; 364 NW2d 593 (1984). Following the release of Nixon, the case was remanded to the Court of Appeals for reconsideration. On remand, the case was reaffirmed.6 We granted leave to appeal. 429 Mich 885 (1987)._
[65]*65II
The undisputed facts of this case are that an eleven-year-old Grand Rapids girl was abducted from her school safety patrol post on the morning of February 12, 1979. Later that same day her body was found near the Regency Park Apartments. The cause of death was strangulation.
The extensive investigation by police produced four categories of evidence: (1) eyewitness identifications, (2) physical evidence, (3) similar acts evidence, and (4) testimony of defendant’s former cellmate which is not at issue in this appeal. The propriety of the similar acts evidence is discussed below. We therefore turn to a discussion of the eyewitness identification and the physical evidence.
A
There were seven key prosecution witnesses (Jack Hill, James Vos, Jack Vos, Greg Start, David Orr, Jr., Timothy Wilcome, and Jim Bonnema), all of whom were hypnotized. Moreover, five of the witnesses were hypnotized on two separate occasions by different hypnotists. The first hypnotic session was conducted by Mr. Robert Mazur, a local hypnotist, who operates the Grand Rapids Hypnotic Clinic. Mr. Mazur is not a psychiatrist or a psychologist and did not testify at trial. Dr. Donald Rossi of the Michigan Department of State Police conducted the second hypnotic sessions. Most of the hypnotic sessions were audiotaped but not videotaped. According to the experts who listened to the tapes, the hypnosis sessions themselves were conducted without any deliberate “attempt to alter [witnesses’] memories.”
Police were notified of the incident by Jack Hill [66]*66who witnessed the last stages of the abduction. At the intersection of the abduction, Jack Hill noticed a car parked on the wrong side of the road and stopped to oifer assistance. He spoke to a black man standing outside the vehicle and noticed a young white girl in the back seat. As the black man drove off Mr. Hill attempted to follow the car. Mr. Hill described the car as a clean, shiny black two-door. He originally told the police that he thought the car might be a Chrysler; however, at trial he testified that the car was a black Grand Prix. The driver was described as a neatly dressed black man with a neat haircut. Mr. Hill did not notice any facial hair on the man.7
James Vos was driving his son, Jack Vos, to school at the time of the abduction when he noticed a black car skidding fast around a corner towards his vehicle. Mr. Vos was stopped for traffic when Mr. Hill informed them that he needed assistance pursuing a vehicle because a young girl had been abducted. James Vos described the car as a clean black two-door hardtop Pontiac Grand Prix with fancy wheel hubs and white walls. Mr. Vos noticed a black man driving the car who was approximately twenty-five to thirty years old, clean in appearance, wearing a dark leather jacket, and having an afro haircut moderate in length. However, he was unable to clearly identify the man driving the car. He did not notice anyone in the car and, in fact, stated that his main concentration was on the car and not on the person driving the car. At trial he was confident [67]*67about his testimony concerning the make and model of the car.
Like his father, Jack Vos observed the car as being a Grand Prix because of the insignia and the grill on the vehicle. He was confident about the description of the car. However, unlike his father, he did observe a small girl in the back seat. He testified that she was a white girl, with blond hair, wearing a blue hat and a green coat.
On the day of the incident, February 12, 1979, Jack Hill was taken to the police department where two composite drawings were made. The first composite drawing was made prior to Mr. Hill being hypnotized. A second composite was compiled during the hypnotic session with Mr. Mazur. Jack Hill, James Vos, and Jack Vos were also all hypnotized on February 12 by Mr. Mazur.
Greg Start, a school crossing guard, gave testimony that he had seen a black car with red pinstriping and a red interior drive by his patrol corner twice. He noticed the car a second time because the driver had run a stop sign. Greg’s testimony was more specific in that he identified the car as a Grand Prix because he saw the letters on the front fender of the car. Further, he described the driver as a young black man with a mustache and an afro hair style.
The victim’s knapsack was recovered from a dumpster at the Georgetown Condominiums by the rubbish collector sometime on the afternoon of February 12, 1979. The Georgetown Condominiums are located within one mile of where the victim’s body was found. On the morning of February 12, a black car driven by a black man was lodged in a snowbank at the condominium. Two grounds maintenance men, David Orr, Jr., and Timothy Wilcome testified that they helped the man free his vehicle. Specifically, Mr. Wilcome [68]*68testified that a black man with neat hair (not an afro style), wearing a leather jacket walked up to the shed and asked for assistance to push his car out of a snowbank. Wilcome stated further that the black man smoked Kool cigarettes8 and that the car looked like a black Monte Carlo.9 Mr. Orr stated specifically that the car was a black Monte Carlo because of the emblem on the back of the car.10 Also that the car had bicentennial license plates11 and spoke-type wheels with curb feelers. Thereafter, Mr. Orr saw him get out of the car and throw something away in a nearby dumpster. Two days after the incident, February 14, 1979, Timothy Wilcome and David Orr, Jr., were hypnotized.
As the police continued to receive tips on the abductor, three mugbooks were compiled. Defendant’s picture appeared in mugbooks one and two, however he was not pictured in the third mug-book. The first photo showing (mugbook one) occurred on February 15, 1979. Jack Hill, David Orr, Jr., James Vos, Jack Vos, and Timothy Wilcome participated, but no positive identification was made. In fact, each witness identified persons other than defendant as possessing features similar to the abductor.
Mr. Jim Bonnema notified the police about see[69]*69ing the particular vehicle on the morning of February 12 after reading about the incident in the newspaper and seeing a story on television. He was hypnotized on February 16, 1979,12 and while under hypnosis he gave á license plate number. However, he was unable to positively identify the driver of the vehicle.
On February 23, 1979, Jack Hill, Jack Vos, David Orr, Jr., and Timothy Wilcome were hypnotized for the second time by Dr. Donald Rossi. During the hypnotic session with Mr. Hill, a police artist was present who attempted to sketch a third composite drawing while Mr. Hill was under hypnosis. Following the hypnotic sessions each witness participated in a second photo showing (mugbook two). Mr. Orr identified defendant among others from this mugbook. Otherwise there was no positive identification by any other witness.
On February 26, 1979, defendant went to the police department for a polygraph examination,13 at which time he agreed to participate in a corporal lineup. The participating witnesses were Jack Hill, Jack Vos, James Vos, David Orr, Jr., and Timothy Wilcome. Of these witnesses, only Timothy Wilcome picked defendant from the lineup; however, he could not make a positive identification.14
On March 15, 1979, mugbook three was shown [70]*70to Hill, Orr, Wilcome, James Vos, and Jack Vos. The witnesses did not positively identify any person (defendant was not pictured). Mr. Wilcome is the only witness who did not pick anyone from these photographs.
Defendant participated in a second lineup on March 29, 1979. At this lineup the defendant was identified by Mr. Hill, Jack Vos, and Mr. Wilcome. Mr. Orr picked someone other than defendant from this lineup as being the alleged perpetrator. James Vos also participated in the lineup, but could not identify the perpetrator.
Two months later, on June 19, 1979, Greg Start was hypnotized by Dr. Rossi.15 The next day he participated in a corporal lineup and picked defendant.16
Prior to the second lineup, March 29, 1979, one day prior to defendant’s arrest, most witnesses at one time or another had made false identifications or at least identifications of persons other than defendant. Mr. Wilcome was the only witness who picked defendant from both lineups. However, he did not make a positive identification at either.
All of the in-court identifications of defendant, except by Jack Hill, were tentative. Mr. Hill made two in-court identifications of the defendant, the [71]*71first being at the preliminary examination and the second during trial. He was extremely confident about each in-court identification.
At trial, Jack Vos made an in-court identification of the defendant as the man he picked from the second lineup. However, he testified that, while defendant looked identical to the man, he could not identify defendant as the man he saw on February 12. Greg Start positively identified defendant as the man he picked from the lineup on June 20. However, this identification occurred after considerable publicity attendant to defendant’s arrest,17 and at trial he stated that he expected to see the suspect in the lineup.
Timothy Wilcome made an in-court identification of defendant as the person he selected from both lineups. At trial he testified that he was not confident in his identification of defendant but that he looked the closest to the person he saw on February 12. At the time of trial James Vos did not make an in-court identification. Rather, he testified that he could not state, one way or the other, whether defendant was the person driving the vehicle on the morning of February 12: Jim Bonnema was unable to positively identify the defendant at trial. However, he did state that there was nothing unusual about defendant to rule him out as the driver of the Grand Prix.
[72]*72B
Our analysis on the issue of admissibility begins with two basic premises offered by the prosecution from the Rules of Evidence. First, all relevant evidence is admissible, absent prejudice, confusion, or waste of time. MRE 401, 402, 403. Second, all witnesses are presumed competent to testify. MRE 601, 602. The relevancy of an eyewitness identification is not questioned. Therefore our only concern is the admissibility of a witness’ testimony after being subjected to hypnosis.
The test to be applied is derived from our holdings in People v Gonzales and People v Nixon, supra.18 In Gonzales, we held that hypnotically induced testimony is inadmissible, but left open the question whether a hypnotized witness could testify regarding facts recalled and related prior to hypnosis. We answered the question in People v Nixon, holding that a witness may testify regarding facts " 'demonstrably recalled prior to hypnosis.’ ” The burden is on the party offering the testimony of a hypnotized witness to "establish its reliability by clear and convincing evidence.” Nixon, supra at 90. The clear and convincing standard adopted in Nixon does not, as the dissent would suggest, extend to the question whether any posthypnotic testimony is reliable.19 Rather, testimony of a hypnotized witness is reliable only when [73]*73the clear and convincing evidence offered by the proponent of the testimony demonstrates that the witness is testifying only with regard to those facts which were recalled and related prior to hypnosis.
Further, in Nixon we reaffirmed our position, quoting with approval the following language from Gonzales:
"The process of hypnosis is not a reliable means of accurately restoring forgotten incidents or repressed memory, and to permit posthypnotic testimony would unfairly denigrate the defendant’s right to cross-examination. Therefore, we hold that until hypnosis gains general acceptance in the fields of medicine and psychiatry as a method by which memories are accurately improved without undue danger of distortion, delusion or fantasy, and until the barriers which hypnosis raises to effective cross-examination are somehow overcome, the testimony of witnesses which has been tainted by hypnosis must be excluded in criminal cases.” [People v Nixon, supra at 88, quoting People v Gonzales, supra at 626-627. Emphasis in original.]
The dissent advances that this Court erred in Nixon by applying the Davis/Frye test to hypnosis.20 However, hypnosis is a scientific procedure which renders a subject susceptible to external influences which could distort the memory.21 "[S]cientific understanding of the phenomenon and of the means to control the effects of hypnosis is still [74]*74in its infancy.”22 "[N]o expert can determine whether memory retrieved by hypnosis, or any part of that memory, is truth, falsehood, or confabulation .... Such results are not scientifically reliable as accurate.”23 Accordingly, we can find no justifiable reason for overturning clear precedent established in this state.24
Defendant argues that the procedures established by this Court in People v Nixon, were not followed, so corrupting the identification process by the suggestive effect of hypnosis as to deny defendant his right to a fair trial. The prosecutor admits that the hypnotic sessions were not conducted in light of the safeguard procedures adopted by the Court in People v Nixon.25 However, the prosecutor asserts that under a totality of the circumstances, the testimony of each eyewitness was not tainted by hypnosis and, in fact, was based on prehypnotic recollection. We disagree [75]*75that such an approach is allowable under the precedent of Gonzales and Nixon.
The investigation in this case took place before Gonzales was decided. In Nixon, we ruled with regard to the application of Gonzales, holding that the principles set forth therein are not to be applied retroactively but are applicable to all cases tried after its release and to those pending on appeal where the hypnosis issue is preserved. Defendant’s case was pending on appeal when Gonzales was released, and the hypnosis issue was properly preserved. Thus defendant is entitled to application of Gonzales.
Although the rules enumerated in Gonzales and Nixon are applicable, factually the two cases differ from this case because neither involved a posthypnotic lineup identification. The only case we have dealt with involving posthypnotic lineup identification was People v Centers, 422 Mich 951 (1986). We addressed the issue of admissibility of a witness’ testimony after submitting to hypnosis to resolve a conflict in the Court of Appeals. See People v Centers, 141 Mich App 364; 367 NW2d 397 (1985), and People v McIntosh, 142 Mich App 314; 370 NW2d 337 (1985).26
Centers and McIntosh were tried together, and [76]*76both were convicted of murder and armed robbery. Following an evidentiary hearing, the trial court allowed testimony concerning the identification of the suspect made at a posthypnotic lineup. The lineup occurred eight months after the hypnotic session. Prior to hypnosis, the witness gave police a description of the suspect on two occasions. During the hypnotic session, the witness gave basically an identical description of the suspect. The hypnotic session was tape recorded and transcribed, as well as reviewed by the court during the evidentiary hearing.27 The trial court reached its decision to allow the testimony by comparing the descriptions made by the witness pre- and posthypnosis. On appeal, both defendants argued that the trial court erred in admitting testimony of an eyewitness who had been hypnotized during the police investigation.
The Court of Appeals in Centers ruled that identification of a suspect at a corporal lineup by a witness who had been hypnotized was error requiring reversal. The Court found significant the fact that identity was central to the issue at trial, so that the testimony of the witness regarding identi[77]*77fication of the defendant was not harmless error. The Court remanded the case to the circuit court holding that
the circuit court shall not admit any posthypnotic testimony or testimony concerning posthypnotic statements unless it is demonstrated that the testimony is based solely on facts recalled and related by the witness or declarant prior to hypnosis and the party offering the testimony establishes its reliability by clear and convincing evidence. [Centers, supra at 371.]
Contrary to the holding in Centers, the McIntosh panel found that because of the similarity in the witness’ descriptions of the suspect, both pre- and posthypnosis, the identification at the lineup was based on facts recalled and related prior to the hypnotic session.
In lieu of granting leave to appeal, we vacated the decision in People v Centers and affirmed the decision of People v McIntosh, 422 Mich 951 (1986). Justice Levin dissented, stating that the prosecutor had failed to establish by clear and convincing evidence that the witness was recalling facts related prior to the hypnotic session.
In McIntosh and Centers, testimony from the eyewitness was admitted on the basis of a comparative analysis of the information given during hypnosis and the prehypnotic statements. Because the information was substantially the same, the Court found that the identification was based on information recalled and related prior to hypnosis. However, in this case, we cannot make such an analysis. We do not have prehypnotic statements from the witnesses, nor an identification of the defendant prior to the hypnotic sessions.
It is obvious in this case that the police department was in a most difficult investigative posture, [78]*78knowing as they did that an eleven-year-old girl had been abducted and that she, and possibly other potential victims, were in serious jeopardy. Thus, they devoted their efforts to extract every conceivable bit of information from the witnesses as quickly, and as thoroughly, as possible. The dissent suggests that under the rule adopted in Michigan the state must "choose between use of hypnosis during the investigative stage and the potential loss of a witness’ testimony at trial . . . .” (Post, p 108, Boyle, J., dissenting.) However, the use of hypnosis as an investigatory tool is not foreclosed by today’s decision. The error in this case resulted from the fact that the investigation was conducted with little or no thought to the preservation of evidence for testimonial purposes. It is admitted that there was virtually no attempt to establish a record of the prehypnotic state of mind of each witness.
Three witnesses were hypnotized on the day of the abduction, and three other witnesses were hypnotized two days after the incident. Furthermore, five of the seven key prosecution witnesses were hypnotized twice. This further complicates and accentuates the inherent difficulties that are present in even a single hypnotism of a witness and that have already led us to exclude hypnotically induced testimony. The administration of hypnosis to the witnesses was interspersed among the photographic and corporal lineups, making it nearly impossible to track the purported state of each witness’ memory, even if such evidence were available.
The prosecution argues and the defense experts agree that the hypnotic sessions were conducted without any deliberate attempt to alter witnesses’ memories. However, even if, as proposed by the dissent, nonsuggestiveness eliminates the unrelia[79]*79bility of posthypnotic testimony, the fact that the police originally did not have a suspect toward whom the interrogator could possibly nudge an uncertain witness, there undoubtedly was in this case, and understandably so, a strong desire to have an identification made.28 While the dissent suggests we "ignored” the lack of a suspect at the time of the "initial” hypnosis, the dissent ignores the fact that Mr. Wilcome was the only witness to identify defendant from the first lineup. The identification of defendant by Mr. Wilcome during the first lineup was not positive. Mr. Wilcome initially did not identify any individual, but rather as an afterthought picked the defendant as the man who looked the closest to the perpetrator. Further, defendant was a primary suspect29 at the [80]*80time of the second hypnosis which occurred one month prior to the identification of the defendant at the second lineup. The second hypnotic session took place after police had an opportunity to verify defendant’s alibi statement which resulted in discrepancies in times and places.
The prosecutor further argues that because there was a uniformity of reactions by all the witnesses at the lineups, the identifications were based on prehypnotic memory. We disagree; first, because even though each witness gave a general description of a black man prior to hypnosis, it was not until after the hypnotic sessions that the witnesses were given the opportunity to select or reject the defendant. Furthermore, each witness at one time or another identified persons other than defendant as the perpetrator. Lastly, there is no record of the prehypnotic descriptions of the perpetrator available upon which to make a comparison, as we did in Centers and McIntosh, to support the admission of identification testimony by the hypnotized witnesses.
Further, the plaintiff did not argue, nor can we conclude, that the prehypnosis composite drawing by witness Jack Hill can be considered a statement recalled and related under Nixon. The composite drawing represents the work of the artist and cannot substitute for oral testimony related and recorded.
The dissent also claims that the uniformity among witnesses of the identification of the vehicle driven by the abductor supports the conclusion that the identification was reliable. The record, however, paints a fuller picture. Jack Hill, the man who witnessed the last stages of the abduction, initially and while under hypnosis identified [81]*81the vehicle as a shiny, black Chrysler.30 Subsequent to both hypnotic sessions Mr. Hill saw de[82]*82fendant’s vehicle in the police garage and thereafter testified at trial that the vehicle was not a Chrysler.31 Further, David Orr and Timothy Wilcome, the grounds maintenance men who freed a black car from a snowbank near where the victim’s body was found, both testified that the car was a black Monte Carlo.32 The dissent states that "Orr did identify the car as a black Monte Carlo” but discounts the inconsistency in its analysis by suggesting that the purpose of Orr’s testimony was merely corroborative. The procedures employed in this case were totally lacking in safeguards to ensure the reliability of the testimony being offered.33 The standards and analysis employed by [83]*83the dissent34 suggest that reliability of posthypnotic memory can be tested on the basis of the other independent evidence offered in the case. However, this Court has never proposed that the proper test for admissibility of evidence should be based on the strength of the case against a defendant. If there is enough independent evidence to prove identity, there would be no need to introduce unreliable evidence of identification. Merely because the other evidence offered in the case corroborates defendant’s guilt or innocence does not mean that it confirms the reliability of a hypnotized witness’ identification of defendant. To establish the reliability of testimony offered by a hypnotized witness would require an evaluation of evidence that would suggest that the memory produced from hypnosis was actual memory and not merely evidence which would further corroborate defendant’s guilt or innocence.
The dissent relies on testimony from Dr. Orne to support its conclusion that the testimony received from the hypnotized witness was reliable.35 At trial, Dr. Orne was allowed to comment about the nature of the physical evidence presented and whether, in his opinion, it corroborated the hypnotized witness’ testimony. We note first that such an evaluation is beyond the expertise of Dr. Orne [84]*84and should not have been permitted. Further, the physical evidence served only to circumstantially support defendant’s guilt and did not operate to afford a basis for distinguishing between what was the witness’ actual memory and what was confabulation.
The dissent has, in our view, mischaracterized the work of this Court in Gonzales. In that case, we unmistakably expressed the view that posthypnotic testimony was "inherently unreliable.” Gonzales, supra, p 626. However, we recognize that hypnosis does not always produce unreliable information, and in some cases the technique may have no effect on memory.36 But because the science of hypnosis itself cannot aid us in distinguishing which effect is being achieved in individual cases, we have and should continue to consider as unreliable all posthypnotic testimony that is not recalled and related prior to hypnosis.
The purpose of the Davis/Frye doctrine is to determine if the scientific technique as a general rule produces reliable evidence. We cannot agree with the dissent that reliability can be determined in every case. The tenet that the trial judge can make a determination in each individual case and place the stamp of reliability and credibility on a hypnotized witness’ testimony is without merit. The dissent today would scrap the Davis/Frye decision of Gonzales that
[w]e agree with the conclusion that the process of hypnosis and its outcome is inherently unreliable. Hypnosis has not received sufficient general acceptance in the scientific community to give reasonable assurance that the results produced under even the best of circumstances will be sufficiently reliable to outweigh the risks of abuse or prejudice [Gonzales, supra, p 626 (emphasis supplied),]
[85]*85and put the trial courts in the position of determining on an ad hoc basis when hypnotically induced testimony is reliable and, therefore, admissible. This Court has over the years not deviated from application of the Davis/Frye doctrine to new scientific discoveries. The best way to achieve reliability of the results of scientific principles is to only allow admissibility of the results when the technique has gained general acceptance within the recognized field.
In a further attempt to recast Nixon from a rule that posthypnotic testimony is inherently unreliable into a rule that the trial courts can determine whether or not hypnotically induced testimony is reliable, the dissent states that "[w]hile we did say in Nixon that a witness could testify on the basis of facts recalled and related prior to hypnosis, we did not say that this was the only way” that you could establish its reliability. (Post, p 105, Boyle, J., dissenting.) First, the "recalled and related” rule did not make posthypnotic testimony reliable. Rather, because posthypnotic testimony is so unreliable, the rule ensures that a hypnotized witness only testifies with regard to prehypnotic memory if the facts from that prehypnotic memory were recalled and related prior to the hypnosis. If there is any doubt about what we said on that point in Nixon, it should be resolved by the very language of the Nixon opinion.
In order to ensure that the witness’ trial testimony is based solely on facts recalled and related prior to hypnosis, we hold that the party offering the testimony must establish its reliability by clear and convincing evidence. [Nixon, supra, p 90. Emphasis supplied.]
Nixon did not, as the dissent advocates, establish a "heightened reliability standard” for hypnotically [86]*86affected testimony. (Post, p 108, Boyle, J., dissenting.) To the contrary, it established a reliability and burden threshold for testimony supported by prehypnotic actions.
We adhere to the standards adopted by this Court in Gonzales and Nixon, and hold that testimony of the hypnotized witnesses is inadmissible absent proof by clear and convincing evidence that the testimony being offered was based on facts recalled and related prior to hypnosis.
c
Where the reviewing court finds that evidence was improperly admitted it must also make a determination regarding whether the error was harmless. As defined by the court rules, "[a]n error in the admission ... of evidence ... is not ground for granting a new trial . . . unless refusal to take this action appears to the court inconsistent with substantial justice.” MCR 2.613(A). The prosecution asserts that the physical evidence corroborates the eyewitness identification and therefore that any error in the admission of testimony from the hypnotized witnesses was harmless.
Justice Levin, writing for the Court, in People v Young (After Remand), 425 Mich 470, 505; 391 NW2d 270 (1986), defined the standards for determining if an error is harmless.
As an appellate court, we do not independently evaluate this evidence. "[I]t is not the appellate court’s function to determine guilt or innocence. . . . Those judgments are exclusively for the jury . . . .” Our responsibility is to determine, how the error might have affected the jury’s decision. The inquiry is "what effect the error had or reasonably may be taken to have had upon the jury’s decision.”
[87]*87If it were clear that the erroneous admission of the . . . evidence did not prejudice [the defendant], the error would be harmless. [Citing Kotteakos v United States, 328 US 750, 763-764; 66 S Ct 1239; 90 L Ed 1557 (1946).]
The error in the instant case was prejudicial to defendant. Defendant was identified as the perpetrator of the crime only by witnesses who were hypnotized. The jury was apprised of the fact that the key witnesses for the prosecution were hypnotized, and we do not know what significance it had in their determination of defendant’s guilt. Furthermore, absent the testimony of the hypnotized witnesses, the remaining evidence does not positively link defendant to this crime. Thus we cannot conclude that the evidence against defendant was overwhelming.
The dissent criticizes our conclusion that the error in this case was not harmless for two reasons. The first is because of the "wholesale application of Nixon to all the witnesses who were hypnotized . . . .” (Post, p 105, Boyle, J., dissenting.) However, by accepting the rule as set forth in Nixon as clear precedent in this state, there is no need to separate and examine individually each witness’ testimony. The dissent’s analysis and charge that we have "fail[ed] to engage in any meaningful separate analysis of the testimony of each witness” (post, p 104, Boyle, J., dissenting) is the best fortification of our decisions in Gonzales and Nixon that all posthypnotic testimony which is not based on prehypnotic memory recalled and related is unreliable.
The second is that we have "[inexplicably ignored . . . the fact that the defendant admitted to a cellmate that he 'accidentally’ killed the victim while attempting to sexually molest her.” (Post, p [88]*88105, Boyle, J., dissenting.) We note first that the cellmate stated on the record that the only reason he contacted police with any information concerning defendant’s purported confession was because he found the crime "sickening.” Yet, at the time of trial, the cellmate had pending against him two counts of criminal sexual conduct in the first degree. Further he testified that he received no promises from the state in exchange for his testimony. One month after defendant’s trial the cellmate entered into a plea bargain agreement where these charges were reduced to esc iv, a misdemeanor. Additionally, defendant testified that he observed his cellmate reading documents pertaining to his case prior to testifying.
The evidence that was particularly incriminating to the defendant was that he was one of three black men in Kent County who owned a 1976 or 1977 black Grand Prix with a red interior.37 Evidence found on the victim’s body at the time of the autopsy included a number of short black Negroid hairs and reddish-orange carpet fibers, the source being unidentifiable. Additionally, a barrette was combed out of the victim’s hair. When the police searched defendant’s car, they found several long blond hairs and feathers that allegedly came from the victim’s down mittens. Additionally, under the seat, police found a plastic barrette. The barrette found in defendant’s car and the one combed from the victim’s hair were manufactured by the same [89]*89company, although the manufacturer could not state specifically that the barrette came from the same package as the barrette found in the victim’s hair.
The prosecution brought in various expert witnesses to testify as to the similarities between the physical evidence found on the victim and in defendant’s car. However, these experts could not positively state that the evidence comparatively matched that produced from defendant’s car. As a result of the experts’ testimony, the trial judge stated:
As I understand it, Mr. Murray [defendant’s attorney] objects to Exhibits 139 through 144[38] because they apparently have no probative value other than to show what the authorities have done in an attempt to investigate this matter, and I suppose to head off any defense argument that they didn’t do enough.
But, really, I have been surprised I haven’t heard this argument a long time ago. I have been listening to this type of testimony for two days, now. It is really just a failure to link the Defendant with any of the evidence.
The dissent described the trial judge who expressed these doubts as "a scrupulously fair judge” (post, p 106, Boyle, J., dissenting). We agree.
Although the totality of the evidence makes for a stronger circumstantial case, the physical evidence was not overwhelming, even though circumstantially significant. It has a limited probative value, absent the eyewitness identification of de[90]*90fendant, and therefore we cannot conclude that all reasonable doubts would have been eliminated in the jurors’ minds.
in
We now turn to the question whether the similar acts evidence offered by the prosecution was properly admitted.
The prosecution was allowed to admit the testimony of ten-year-old Judy Sanchez and nine-year-old Amy Combs over objection. The trial court endorsed the witnesses pursuant to MRE 403 and 404. They testified that on February 6, 1979, approximately one week prior to this incident, a black man in a black car approached them while they were walking to school. Defendant did not work on February 6, 1979. He called in sick and claimed to have been home all day.
The testimony of the two girls varies somewhat, although they generally testified that the man stopped his car and put one foot outside the door. He asked them for the time and the location of Hall Street. The girls responded to the questions, and after doing so the man motioned to the girls to come closer to the car. Both girls testified that they were scared and that the man recognized their fear. Judy had apparently run from the car, but Amy had stayed longer. As Judy gradually came back, the girls testified that the man got into the car and left.
The two girls identified defendant from a mug-book consisting of six photographs.39 Both girls [91]*91identified defendant in a corporal lineup on March 29, 1979. While participating in the corporal lineup on March 29, Judy Sanchez identified two persons, one being the defendant and a second individual whose voice she felt sounded closest to the person. At trial neither of the two girls were able to positively identify the defendant as the man who approached them on February 6, 1979. On cross-examination Judy Sanchez made the following statement concerning the in-court identification of defendant:
Q. Judy, you are not saying that this man, here, Mr. Lee, Albert Lee, is the man you saw on that particular date, are you?
A. No, it [sic] looks the closest description to the guy.
Amy Combs testified that she was "pretty sure, but not positive” that Mr. Albert Lee was the man who approached her and Judy.
B
The introduction of similar acts evidence is an exception to the general rule that evidence of other crimes is inadmissible.40 In order to be admissible, the prosecution must show a high degree [92]*92of similarity between the charged and uncharged acts. These standards are to be strictly applied and have been narrowly defined by this Court in an attempt to safeguard the defendant’s right to a fair trial. The policy considerations underlying the safeguards are an attempt to avoid convicting the defendant because of prior misconduct rather than on the proofs of the charged offense. The standards governing the admissibility of similar acts testimony were set forth by this Court in People v Golochowicz, 413 Mich 298, 309; 319 NW2d 518 (1982).
[B]efore evidence of defendant’s other misconduct may be admitted: (1) there must be substantial evidence that the defendant actually perpetrated the bad act sought to be introduced; (2) there must be some special quality or circumstance of the bad act tending to prove the defendant’s identity or the motive, intent, absence of mistake or accident, scheme, plan or system in doing the act and, in light of the slightly different language of MRE 404(b) we add, opportunity, preparation and knowledge; (3) one or more of these factors must be material to the determination of the defendant’s guilt of the charged offense; and (4) the probative value of the evidence sought to be introduced must not be substantially outweighed by the danger of unfair prejudice.
The two girls testified that it was a black man in a black car that approached them on their way to school. That alone probably does not constitute substantial evidence to show that it was defendant who committed the bad act. However, there was testimony that defendant is one of three black [93]*93men in Kent County who owns a black Pontiac Grand Prix, and, that defendant was absent from work on February 6, 1979, and could not verify his whereabouts.
The facts which weigh against the substantiality of the evidence include defendant’s own testimony that he was not the man, and, additionally, the tentativeness of the girls’ identifications. Further, one of the girls testified that the man did not have a mustache. Defendant had a mustache and noticeable sideburns a week later on February 12, 1979.
Although the identification of defendant was tentative, the identification is bolstered because two witnesses identified defendant. This alone would not constitute substantial evidence. However, the identifications, coupled with the fact that the man had a black car and that defendant drives a black car, as well as the fact that defendant did not go to work on February 6, leads us to conclude that the prosecution has shown by substantial evidence that defendant was the perpetrator of the similar act.
To test whether there is a special quality or circumstance sufficient to justify admission of the evidence on the issue of identity, this Court has made the following statements:
It is the distinguishing characteristics which constitute the acts as similar within the meaning of. . . MRE 404(b), not the fact that all constitute the same crime or are violative of the same statute. The distinguishing, peculiar or special characteristics which are common to the acts and thus personalize them are said to be the defendant’s "signature” which identifies him as the perpetrator .... [People v Major, 407 Mich 394, 398-399; 285 NW2d 660 (1979).]
Where . . . the only conceivable justification for [94]*94admission of such similar-acts evidence is to prove the identity of the perpetrator, the link is forged with sufficient strength to justify admission of evidence of the separate offense only where the circumstances and manner in which the two crimes were committed are "[s]o nearly identical in method as to earmark [the charged offense] as the handiwork of the accused. Here much more is demanded than the mere repeated commission of crimes of the same class .... The [commonality of circumstances] must be so unusual and distinctive as to be like a signature. . . .”
It is because of the combined value of those two factors, the unique and uncommonly distinctive style employed by the defendant in committing the "substantially proved” uncharged similar offense, and the same distinctive modus operandi employed in the charged offense, that the jury is permitted to infer, if it believes the evidence, that both crimes were the handiwork of the same person, the defendant.
... It is the uniqueness and the distinctiveness with which both crimes were committed, combined with the proof that the defendant committed the "like act,” that is the key. [People v Golochowicz, supra at 310-312. Citations omitted.]
The similarity between the two incidents upon which the prosecutor bases its admission are that (1) both incidents involved a black man driving a black car, (2) each occurred in the early morning hours when children would be walking to school, (3) both incidents involved young girls, and (4) both occurred in the same general area. Generally, it could be argued that these factors are not peculiar or do not constitute the defendant’s signature because they are common to many child molestation cases. However, the manner and circumstances of stalking and attracting victims can be and is, in this case, sufficiently distinctive and unique that it can pass the Golochowicz "signa[95]*95ture” requirement. Child abduction cases, in the overall scheme of criminal activity, are rather rare and unique and relate to aberrant behavior, i.e., there is really no "run of the mill” child abduction, as in the case of a run of the mill robbery, or car theft. The coaxing of these two girls to the actor’s car is in and of itself somewhat distinctive and, when coupled with the similarities in place and time, lend to the entire act a uniqueness and distinctiveness.
The central issue in this case is identity. The evidence is material because identity is the only proposition at issue. Identity is a material issue, not because it is a factor of proof in generally all criminal cases, but because the defense has made it a genuinely controverted matter. Here, defendant does not offer any explanation for the killing, i.e., that it was unintentional or that he had some innocent purpose in inviting the victim into his car, rather he offers only an alibi defense that he could not have been the person to commit the act. Thus, the evidence, if admissible at all under MRE 404(b), could be used only to show identity.
Defendant claims that he was prejudiced by the manner in which the prosecutor argued the similar acts incident. The trial judge did not abuse his discretion in finding that the probative value of the evidence outweighs its prejudicial effect.
In his closing argument, the prosecutor argued that Amy Combs and Judy Sanchez were in fact lucky to have gotten away from this man. He argued that what happened to the victim was intended by the defendant with regard to the two girls, however, it did not come about. Thus, in effect, the prosecution stressed the defendant’s bad character and his propensity to commit such an act. Although the prosecution agrees with the defense that the acts towards Judy Sanchez and [96]*96Amy Combs did not constitute a crime, he argued that the acts in effect constituted an attempt, that the defendant intended to kidnap and kill the two girls. In his closing argument the prosecutor stated:
[Tjn addition to those three eyewitnesses, you heard some testimony, quite frankly, if you listened to it, the implications of what may have happened probably had to scare you a little bit, and that was Amy Combs and Judy Sanchez, and they described an incident that sounded awfully familiar, except for the ending, as to what happened to [the victim]. . . . He said, "Come here.” Why did he ask her to "come here”? Why did he ask her to do that? Not only did he do it once, he did it a couple of times. Mr. Murphy will say, Well, there is nothing wrong with what that person did, even assuming it was Mr. Lee, but we don’t know what he wanted to do to her. Fortunately, Amy Combs ran away before he ever got a chance. But, again, who did both of them identify? Just another coincidence. It was Albert Lee.
The law says that I can bring these people in here to prove to you the identity of the person, the abductor and murderer of [the victim herein].
Again, compare all the similarities of what almost happened to those little girls to what did happen to [the victim], and it scares us.
It is clear that the prosecutor’s closing arguments were specifically designed to convey to the jury an improper purpose for the evidence. Indeed the trial judge gave a cautionary instruction to the jury concerning the limited scope for which the evidence could be considered. In light of the fact that we have ordered a new trial on the basis of the hypnosis issue, we hold that in and of itself (i.e., absent the prosecutor’s arguments) the simi[97]*97lar acts evidence is more probative than prejudicial. While the prosecutor’s use of the evidence was inappropriate in this trial, on this record the evidence was admissible. We are confident that on retrial the prosecution will confine its arguments within the permissible scope of MRE 404.
CONCLUSION
Defendant was prejudiced to the extent that reversal is required on the basis of the admission of identification testimony by hypnotized witnesses. Although the ruling of this Court in Gonzales was not available at the time of trial, defendant is entitled to the benefit of the Gonzales holding on the basis of the retroactivity analysis under Nixon. The prosecution has failed to show by clear and convincing evidence that the hypnotic process and the resulting testimony was reliable and based on the facts recalled and related prior to the hypnotic sessions.
On this record, the similar acts evidence offered by the prosecution in and of itself complies with the standards of Golochowicz.
The decision of the Court of Appeals is reversed, and we remand the matter for a new trial.
Levin, J., concurred with Brickley, J.