delivered the Opinion of the Court.

Defendant Jesse Bernal (Bernal) was convicted of second degree kidnapping, aggravated robbery, conspiracy to commit aggravated robbery, and second degree assault. The court of appeals affirmed his convictions in People v. Bernal, No. 98CA448 (Colo.Ct. App. Nov. 18, 1999)(not selected for official publication). We granted certiorari to consider two issues. First, Bernal contends that his due process right to a fair trial was violated by admission of testimony regarding an out-of-court identification based on an im-permissibly suggestive photo array. Second, Bernal contends that the admission of a hearsay statement by his co-defendant violated-his rights under the Confrontation Clause.1 We reverse in part, affirm in part, and remand.

With regard to the first issue, we conclude that the photo array was impermissibly suggestive. Because we find that the trial court failed to make adequate factual findings regarding the reliability of the impermissibly suggestive out-of-court identification, we remand with directions to the trial court to make such findings consistent with this opinion.

With regard to the second issue, we conclude that the co-defendant's statement offered against Bernal pursuant to CRE 804(b)(8) was inferentially inculpatory and did not satisfy the Confrontation Clause's *188requirement of trustworthiness and reliability. On this issue, we also remand with directions.

I. Facts and Procedure2

On the afternoon of December 12, 1996, two men, wearing baseball caps and sunglasses, robbed a credit union in Brighton, Colorado. Mandie McBride (McBride), an assistant manager, noticed a light blue or gray car pull up to the bank and saw two men exit the car and approach the credit union. McBride was sitting at her desk when the first man entered the credit union. At trial, McBride testified that she did not see the face of the man as he entered the credit union. The man approached her desk with a gun, grabbed her by the hair, and demanded to be taken to the vault. McBride testified that once the robber grabbed her by the hair, she did not see him anymore.3 The robber pushed McBride towards the vault, which was located next to the manager's office. McBride testified that while walking to the vault, she did not see the robber because he was behind her. When asked about the robber's facial features, McBride testified that she could not see the robber's eyes, and that she did not notice his ears, teeth, or chin. In fact, her testimony revealed that all she saw of the robber was his black hair and "puffy" cheeks.

The manager, Kathy Wagner (Wagner), was at her desk when the robber and McBride entered her office. At trial, Wagner testified that the robber was behind McBride when they entered her office. The robber instructed Wagner to go to the vault and he followed behind. When describing the robber's facial features, Wagner testified that she saw the robber's profile, but could not give any detailed description of the robber. Furthermore, the robber told the women not to look at him and struck them with the barrel of the gun when they did. Wagner opened the vault and gave him the cash in the vault, nearly $12,000. Once the man had the money, he left. During the robbery, a second man armed with a gun stood in the lobby and demanded that the two other ered-it union employees and a customer get on the floor. The robbery lasted no longer than several minutes and both women testified that everything happened quickly and that they were seared.

Immediately after the robbery, Wagner and McBride gave a statement to the police. Although the lighting in the eredit union was good and both women had been trained to deal with robberies, they were unable to give a detailed description of the robber. Both women described his appearance as a "Hispanic" male with a medium complexion. Wagner expanded on this physical description by stating that the robber had a "rough" complexion. - The eyewitnesses also de-seribed the robbers as "sounding Hispanic." The only other description given was an approximate height and weight of the robber. Other witnesses gave similar descriptions, and also described the robbers as being roughly the same height.

Soon after the robbery, the police located the car used by the robbers. Two baseball caps and two pairs of sunglasses were found in the car. Police investigators also procured physical evidence linking Bernal's co-defendant, Raymond Rodarte (Rodarte) to the car. Specifically, a hair removed from one of the baseball caps in the car was consistent with Rodarte's. The police developed the co-defendants as suspects, at least in part, from information provided by the car's owner, Daniel Tucker (Tucker). In particular, Tucker told police that he had previously met Rodarte. He also told police that his car had been stolen the night before but that a friend, Tano Torres (Torres), had a key and might have taken it. Tucker stated that when he discovered the car was missing, he paged Torres, who returned the call. Tucker's girlfriend, Walynda Marshall (Marshall) testified that Torres had called earlier that *189day; her caller ID revealed that Torres had called from a telephone identified as Ro-darte's. Tucker further told the police that he had previously met Bernal through both Rodarte and Torres.

On January 28, 1997, approximately six weeks after the robbery, the witnesses were shown a photo array that consisted of six color photographs of young men arranged in two rows of three, with Bernal appearing in the middle of the top row. Bernal's photograph appears to be the only photograph of an "Hispanic" male, although one of the other photographs could be perceived as also depicting an "Hispanic" male. Bernal's photograph also stands out from the six as the only one with a clear white background. The other photographs have a neutrally colored venetian blind backdrop. Two of the young men in the photographs have blue eyes. Each of the photographs is only of the head and neck and reveals nothing of the height or weight of the young men. After examining the photo array for approximately two minutes, Wagner selected Bernal's photograph. McBride looked at the photo array for approximately one minute before selecting Ber-nal's picture.

Based on the above evidence, Bernal and Rodarte were arrested and charged with kidnapping, aggravated robbery, conspiracy to commit aggravated robbery, and second degree assault. Bernal and Rodarte were tried separately.

Prior to trial, defense counsel filed a motion to suppress evidence of the out-of-court identification of Bernal by Wagner and McBride. At the pretrial motions hearing, the prosecution presented Detective Grose (Grose).4 Although defense counsel had subpoenaed Wagner and McBride, the trial court refused to hear their testimony. After hearing Grose's testimony, the trial court stated that it was unnecessary for it to hear the testimony of Wagner and McBride, the eyewitnesses, to make its ruling.

In ruling on the motion to suppress, the trial court inconsistently found the photo array to be both "not suggestive," and not "so impermissibly suggestive" as to require suppression of the out-of-court identification. Because the trial court's findings are extraordinarily limited, it is unclear what standard the court used to deny the motion. The trial court did state that the credit union had good lighting and, somewhat callously, that the "fact that the women were getting whacked with a gun would certainly focus their attention." However, by refusing to entertain the testimony of the eyewitnesses, the trial court implied that it did not think it necessary to address the reliability of the identification. The out-of-court identifications were thus admitted at trial.

At trial, the prosecution successfully sought to admit a hearsay statement made by Rodarte to Grose. Grose testified regarding several interviews he conducted with Ro-darte on the day that Rodarte was arrested. The statements were given at the Brighton police station. Although Grose testified regarding several statements made by Ro-darte, the statement at issue in this appeal was as follows: "He [Rodarte] admitted to taking the car, but adamantly denied being involved in the robbery." The trial court admitted the statement pursuant to CRE 804(b)(8), the "statement against interest" exception to the hearsay rule. The trial court made no findings regarding the reliability or trustworthiness of the statement pursuant to the Confrontation Clause.

Bernal was convicted on all counts and sentenced to forty-six years in prison. The court of appeals affirmed Bernal's convictions.5 With regard to the photo array, the court of appeals held that it was "not unduly suggestive" and the out-of-court identifications were thus properly admitted at trial. Bernal, No. 98CAA448, slip op. at 4. A dissenting judge disagreed, opining that the photo array was so impermissibly suggestive as to give rise to a substantial likelihood of *190misidentification. Id. at 24 (Casebolt, J., dissenting). With regard to the admission of Rodarte's statement under 804(b)8), the court of appeals held that the trial court failed to undertake the required Confrontation Clause inquiry regarding whether the statement bore "particularized guarantees of trustworthiness," and therefore erroneously admitted the statement. However, the court of appeals found that the admission of the statement was constitutional harmless error. Id. at 7.

II. The Photo Array and Out-of-Court Identification

A. Standard of Review

As an initial matter, we address the standard of review required when reviewing the trial court's conclusion that the out-of-court identification procedure did not violate Bernal's due process rights. The ultimate question as to the constitutionality of pretrial identification procedures is a mixed question of law and fact. Sumner v. Mata, 455 U.S. 591, 597, 102 S.Ct. 1303, 71 L.Ed.2d 480 (1982). Thus, while the trial court's findings of historical fact are entitled to deference, an appellate court may give different weight to those facts and may reach a different conclusion in light of the legal standard. Id.

B. Photo Lineups Generally

Suggestive lineups are disapproved because they increase the likelihood of mis-identification and have, in the past, too often brought about the conviction of the innocent. In United States v. Wade, the United States Supreme Court observed:

A major factor contributing to the high incidence of miscarriage of justice from mistaken identification has been the degree of suggestion inherent in the manner in which the prosecution presents the suspect to witnesses for pretrial identification.... Suggestion can be created intentionally or unintentionally in many subtle ways.[ ] And the dangers for the suspect are particularly grave when the witness' opportunity for observation was insubstantial, and thus his susceptibility to suggestion the greatest.

388 U.S. 218, 228-29, 87 S.Ct. 1926, 18 L.Ed.2d 1149 (1967)(footnote omitted). Furthermore, the Supreme Court has noted that, once an eyewitness has chosen a suspect from a line-up, that witness is "not likely to go back on his word later on, so that in practice the issue of identity may (in the absence of other relevant evidence) for all practical purposes be determined there and then, before the trial." Id. at 229, 87 S.Ct. 1926. The danger of misidentification is further increased in prosecutions where the vie-tim is the witness, such as in a robbery, because there is a particular hazard that a victim's understandable outrage may excite a vengeful or spiteful motive. Id. at 280, 87 S.Ct. 1926.

Subsequent experience and empirical evidence support the Supreme Court's conclusions. For example, a study of forty cases in which the convicted persons were later exonerated through DNA testing revealed that ninety percent (90%) of the convictions were obtained, at least in part, by erroneous eyewitness identifications. Gary L. Wells et al., Eyewitness Identification Procedures: Recommendations for Lineups and Photo-spreads, 22 Law & Hum. Behav. 608, 605 (1998). The study concluded that "mistaken eyewitness identification is responsible for more of these wrongful convictions than all other causes combined," and that "eyewitness identification evidence is among the least reliable forms of evidence and yet is persuasive to juries." Id. The study further demonstrated that "accuracy of description is a rather poor predictor of identification." Id. at 608. A different study revealed that "ree-ognition accuracy was found to be poorer when the perpetrator was holding a weapon." Vaughn Tooley et al., Facial Recognition: Weapon Effect and Attentional Focus, 17 J. of Applied Social Psychology 845, 854 (1987).

C. The Two-Part Test for Determining Admissibility of Out-of-Court Photographic Identification

Recognizing the inherent dangers of pretrial identification, the United States Supreme Court set out the standard for viewing photographic displays in Simmons v. United States. The Supreme Court stated:

*191[EJlach case must be considered on its own facts and ... convictions based on eye witness identification at trial following a pretrial identification by photograph will be set aside on that ground only if the photographic identification procedure was so impermissibly suggestive as to give rise to a very substantial likelihood of irreparable misidentification. _

390 U.S. 377, 384, 88 S.Ct. 967, 19 L.Ed.2d 1247 (1968). This standard has developed into a two-part analysis. First, a court must determine whether the photo array was im-permissibly suggestive, which the defendant has the burden of proving. Neil v. Biggers, 409 U.S. 188, 198, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972); United States v. Sanchez, 24 F.3d 1259, 1262 (10th Cir. 1994). If this burden is not met, no further inquiry is necessary. Sanches, 24 F.3d at 1262. Second, if the defendant's burden is met, the burden shifts to the People to show that despite the improper suggestiveness, the identification was nevertheless reliable under the "totality of the circumstances." Manson v. Brathwaite, 432 U.S. 98, 114, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977); Biggers, 409 U.S. at 199, 93 S.Ct. 375; Sanches, 24 F.3d at 1261-62. It is important to note that these two steps must be completed separately; it is only necessary to reach the second step if the court first determines that the array was impermissibly suggestive. Sanchez, 24 F.3d at 1262.

In evaluating whether a pretrial photo identification procedure is impermissibly suggestive, a number of factors may be relevant. These include the size of the array, the manner of its presentation by the officers, and the details of the photographs themselves. United States v. Wiseman, 172 F.3d 1196, 1208 (10th Cir. 1999); Sanchez, 24 F.3d at 1262. Although courts have held that a photo array with as few as six pictures is not per se a due process violation, see, e.g., United States v. Bennett, 409 F.2d 888, 898 (2d Cir. 1969), courts have recognized that the size of a photo array, specifically the number of pictures in it, is a factor affecting the weight a court gives to the irregularities in the array. Sanchez, 24 F.3d at 1262. The more pictures used in an array, the less likely it is that a minor difference, such as background color or texture, will have a prejudicial effect on selection. Id.

In contrast, when relatively few photographs are used in an array, minor differences such as background color make a picture stand out and can repeatedly draw a witness's eyes to that picture. Id. In Sanchez, the court noted:

Common sense dictates that slight irregularities are more likely to "jump out" at a witness when reviewing a single sheet of paper with only six photographs on it than at a witness reviewing a large mug book 'containing hundreds of photographs. Upon continued inspection, the witness may begin to believe that the "oddball" picture was taken under different cireum-stances than the others. This fact can suggest a number of things to the witness, the most dangerous of which is that the similar pictures were taken together to form a pool or control group, and that the one picture that stands out is the suspect.

Id. Thus, the fewer photographs used by the officers in a photo array, the closer the array must be scrutinized for suggestive irregularities. Id.

When the number of photographs shown has not been so small as to make the presentation itself unfairly suggestive, and there is nothing in the officials manner of presentation that renders the procedure surrounding the array suggestive, the principal question is whether the picture of the accused, which matches descriptions given by the witness, so stood out from all of the other photographs as to " 'suggest to an identifying witness that [that person] was more likely to be the culprit'" Jarrett v. Headley, 802 F.2d 34, 41 (2d Cir. 1986)(quoting United States v. Archibald, 734 F.2d 938, 940 (2d Cir. 1984)). In other words, the array must not be so limited that the defendant is the only one to match the witness's description of the perpetrator. United States v. Maldonado-Rivera, 922 F.2d 934, 974 (2d Cir. 1990).

The police do not have to provide a photo array containing only "exact replicas" of the defendant's picture; all that is required is that the "photos are matched by *192race, approximate age, facial hair, and a number of other characteristics." People v. Webster, 987 P.2d 836, 839 (Colo.Ct.App. 1998); People v. Borrego, 668 P.2d 21, 23 (Colo.Ct.App. 1983); United States v. Fernandez, 456 F.2d 638, 641 (2d Cir. 1972). Thus, a photo array in which the individual characteristics of the accused, such as race, stand in stark contrast to the other photographs is impermissibly suggestive. State v. Thamer, 777 P.2d 432, 435 (Utah 1989). But see Williams v. Weldon, 826 F.2d 1018, 1021 (11th Cir. 1987)("[Slimply being of a different race or ethnic group from others placed in a lineup does not necessarily make the lineup impermissibly suggestive, especially where . the other individuals had roughly the same characteristics and features of the accused.").

In United States v. Fernandes, the court found that a photographic array was imper-missibly suggestive when witnesses had de-seribed the bank robber as a light-skinned Black man with an Afro haircut, and of the six photographs presented, only that of the defendant showed a Black man having a light skin tone. 456 F.2d at 641-48. Similarly, in United States v. Gidley, 527 F.2d 1345, 1350 (5th Cir. 1976), the court found a photographic display impermissibly suggestive because none of the other pictures displayed the defendant's Asian appearance nor did any of the men pictured, except the two defendants, have long black hair. The court noted that "Ialnyone who had gotten a glimpse of two robbers with long black hair would have found only [the defendants'] pictures in this group to resemble those robbers." Id.

Although it is not required that all of the photographs in the array be uniform with respect to one given characteristic, Jarrelt, 802 F.2d at 41, a photographic array that includes a photo that is unique in a manner directly related to an important identification factor may be held impermissibly suggestive. See Grubbs v. Hannigan, 982 F.2d 1483, 1490 (10th Cir. 1983)("Although a photo-lineup is not necessarily suggestive merely because the individuals in the lineup differ in facial characteristics, ... here the differences were either strikingly apparent, such as a swollen eye, or they related to an important component of [the victim's] description of her assailant, his hair style.").

If the court finds a photo array im-permissibly suggestive, it must then proceed to the second step of the analysis and determine whether, under the totality of the cireumstances, the suggestive procedure created a very substantial likelihood of mis-identification. Manson, 432 U.S. at 116, 97 S.Ct. 2248. The factors to be considered in determining whether, despite a suggestive array, the identification was nonetheless reliable are: (1) the opportunity of the witness to view the criminal at the time of the crime; (2) the witness's degree of attention; (3) the accuracy of the witness's prior de-seription of the criminal; (4) the level of certainty demonstrated by the witness at the confrontation; and (5) the length of time between the crime and the confrontation. Biggers, 409 U.S. at 199, 93 S.Ct. 375; People v. Monroe, 925 P.2d 767, (Colo. 1996).

In conducting this analysis, a court must balance the suggestiveness of the procedures employed against indicia of reliability surrounding the identification to determine whether the identification should be suppressed. Manson, 432 U.S. at 114, 97 S.Ct. 2243; Grubbs, 982 F.2d at 1490. Reliability is the linchpin in determining the admissibility of identification testimony. Manson, 432 at 114, 97 S.Ct. 2243. As long as the totality of the cireumstances does not indicate a very substantial likelihood of irreparable misidentification, no constitutional impediment to the admission of the identification testimony exists. Id. at 116, 97 S.Ct. 2243.

D. Application of the Law to the Out of Court Identification in this Case

In this case, Bernal argues that his due process rights were violated by the admission of testimony concerning an impermissibly suggestive photo array. The determination of whether there was a violation of Bernal's due process rights requires the application of the two-part analysis described above.

Under the first part of the analysis, Bernal challenges the photographs them*193selves, arguing that they are impermissibly suggestive because, of the six men shown, he was the only obviously "Hispanic" man. Our examination of the photo array reveals it to be impermissibly suggestive. Our review of the record indicates that ethnicity was a controlling and overriding characteristic of the witnesses' description of the robber. In their initial descriptions to the police immediately after the robbery, none of the witnesses provided any sort of detailed description of the robber. Rather, their description was based on race and ethnicity: Both eyewitnesses stated that the robber was "Hispanic."6 The ethnicities of most of the other men in the photo array are conspicuously different, specifically not "Hispanic," so as to make the defendant's photo stand out as clearly different from the others.

Although it has been suggested that one of the men shown with Bernal "could possibly have been" "Hispanic", in the context of the descriptions given by the eyewitnesses in this case, this photograph is not similar to Bernal's. In this case, the witnesses chose to use the term "Hispanic." Because the term "Hispanic" can refer to people with different skin colors, eye colors, and hair colors, and even to people of different races, it might be considered to be a poor descriptive term. See generally Kevin R. Johnson, Race, Ethnicity & Nationhood: "Melting Pot" or ' "Ring of Fire" '?: Assimilation and the Mexican American Experience, 85 Cal. L. Rev. 1259, 1291-92 (1997). However, when a witness uses the term "Hispanic" to describe a suspect it may reasonably be presumed that she is using that term to denote a person with discernable and distinctive physical characteristics that comprise the stereotype of an "Hispanic" individual. Here, only Bernal's photograph is stereotypical.

Although we disagree with the use of such stereotypes to homogenize what in reality is a multi-racial, diverse population, it is undisputed that the term "Hispanic" is used as a descriptive term and has, in essence, reduced this diverse group of individuals into one group. See Wendy Chung, Ethnicity and Organizational Diversity: A Study of Social Cognition and Psychological Climate Perception 59 (1997). Specifically, the ster-ecotype has been described as "dark-complexioned with more indigenous features." Johnson, supra, at 1290-91. Some courts have described the stereotypical characteristics of someone who is "Hispanic" as "black hair, dark skin, and dark eyes." Hodgers-Durgin v. De La Vina, 199 F.3d 1037, 1038 (9th Cir. 1999); see also, e.g., United States v. Ramos, 753 F.Supp. 75, 76 (W.D.N.Y. 1990) ("light brown skin"); People v. Gonzales, 292 Ill.App.3d 280, 226 Ill.Dec. 406, 685 N.E.2d 661, 663 (1997) ("light brown skin, dark brown hair, and dark eyes"); State v. Mareno, 580 So.2d 593, 595 (La.Ct.App. 1988) ("a 'kind of Hispanic appearance'" described as "an individual 57-58", with an olive complexion, medium build, [and] brown hair"). Thus, although the photo array may not be perceived as suggestive to someone who is aware of the diversity of individuals in the "Hispanic" population, it is highly suggestive to a person who chooses to describe a person as "Hispanic" in the stereotypical sense of describing what many people consider a monolithic group. See Johnson, supra, at 1290.

Here, there is nothing to suggest that the witnesses meant anything other than the commonly understood, stereotypical characteristics of an "Hispanic" individual. Thus, we infer that "Hispanic," as used by the two eyewitnesses, was meant to describe the features of people of Mexican heritage, the predominant "Hispanic" population in the southwestern United States. See Ramos, 753 F.Supp. at 78 n. 5 ("It is apparent that appearing Mexican is the functional equivalent of appearing Hispanic."). The stark fact is that Bernal leaps out as the one person who is different because of his ethnicity, a judgment that can be made even without knowledge of what the witnesses said.

In addition, Bernal's photograph was the only one with a clear white background. The other photographs were taken against neu*194trally colored venetian blinds. This seeming-Ty minor detail of the background compounds the already-mentioned deficiencies of the photo array because, as noted in Sanches, the likely inference that Bernal's picture was taken under different cireumstances suggests that the other photos were assembled to provide a pool or control group, as indeed they were. Wiseman, 172 F.3d at 1209. Moreover, the small number of photos in the array, which in this case consisted of six, cannot dilute the suggestiveness of the array. Id.

For these reasons, we conclude that the photo lineup is impermissibly suggestive.

Having found the pretrial identification procedure to be impermissibly suggestive, we must next proceed to the second step of the analysis, namely a determination whether, under the totality of the cireumstances, the identification was nonetheless reliable. This determination involves consideration of the five factors identified by the Supreme Court in Biggers. Manson, 432 U.S. at 114, 97 S.Ct. 2243; Biggers, 409 U.S. at 199, 93 S.Ct. 375.

As to the first two Biggers factors, the record is fairly well developed because these factors were pertinent to issues raised at trial. Thus, regarding the first Biggers factor, the opportunity of the witnesses to view the criminal at the time of the erime, the evidence showed that the robbery lasted several minutes and took place in a well-lighted area. However, there was also evidence that the robber was behind McBride during most of the robbery, and that McBride only got "slight" glances of the robber. In addition, the robber concealed his appearance with a baseball hat and sunglasses.

With regard to the second Biggers factor, the witnesses' degree of attention, the evidence showed that the witnesses had received training regarding robberies. In this case, however, the witnesses were instructed not to look at the robber, and when they did, he struck them with his weapon.

As to the rest of the Biggers factors, because the trial court failed to hear the offered testimony, the record is inadequate. Thus, regarding the third Biggers factor, the accuracy of the witnesses' prior description, the witnesses' initial descriptions to the police were very general. The witnesses described the robbers as roughly the same height, medium complexion, and "Hispanic." Wagner also described the robber as having a "rough" complexion. Because the witnesses were not allowed to testify at the suppression hearing, it is unclear what they meant by the term "Hispanic." In addition, there is nearly a foot of difference between the height of the two men arrested for the robbery. Rodarte measures five feet two inches tall, while Bernal is over six feet tall. And, contrary to Wagner's description, Ber-nal does not have a "rough" complexion. In fact, he was described as having a "very nice complexion."

Concerning the fourth Biggers factor, the level of certainty demonstrated by the witnesses at the confrontation, it is unclear what the witnesses' degree of certainty was when they chose Bernal's photograph. At the suppression hearing, Grose testified that the witnesses stated that they had made a positive identification. - When defense counsel attempted to question Grose regarding the witnesses' level of certainty, the trial court intervened and refused to permit Grose to testify, stating: "They can't be any better than positive I don't think." The evidence only showed that Wagner chose Bernal's photograph after examining the array for two minutes and that MeBride spent one minute reviewing the array before choosing Bernal.

Finally, with regard to the last Biggers factor, the length of time between the crime and the confrontation, there was a six-week gap between the robbery and the photo array. Again, because the witnesses were not allowed to give testimony, it is unclear how much weight should be placed upon this fact.

Given that the trial court failed to hear testimony and failed to make adequate factual findings, we cannot complete the review necessary to determine whether the out-of-court identification is, despite its suggestiveness, reliable. Accordingly, we remand to the trial court for a reliability determination. Upon remand, the trial court, in determining the reliability of the out-of-court identifica*195tion, should not limit itself to the specific evidence we have just discussed. Instead, the court should, in addition to the above evidence, consider any new testimony that may be offered that it finds pertinent to the Biggers factors. For example, the trial court might inquire as to what the witnesses meant when they used the term "Hispanic" to describe the robber and whether they would describe any of the other five men depicted in the photo array as "Hispanic."

III. The Admissibility of Rodarte's Statement Against Interest

In addition to the issue of the photo array, this case also involves a hearsay statement by Rodarte offered against Bernal at trial. The prosecution sought to have the statement admitted under the statement against interest exception to the hearsay rule, codified in CRE 804(b)(B).

A. CRE 804(b)(8)

The statement against interest exception states, in relevant part:

A statement which ... at the time of its making ... so far tended to subject [the declarant] to ... criminal liability ... that a reasonable man in his position would not have made the statement unless he believed it to be true. A statement tending to expose the declarant to eriminal lability and offered to exculpate the accused is not admissible unless corroborating cireum-stances clearly indicate the trustworthiness of the statement.

CRE 804(b)@8). The rule is predicated on "the principle of experience that a statement asserting a fact distinctly against one's interest is unlikely to be deliberately false or heedlessly incorrect." 5 John Wigmore, Wigmore on Evidence § 1457 (Chadbourn Rev. 1974). Thus, the statement's self-ineal-patory nature is the basis of its presumed trustworthiness. Id. Colorado Rule of Evidence 804(b)(8) may be utilized only when the declarant is unavailable to testify at trial. CRE 804(b)(@8); Blecha v. People, 962 P.2d 931, 939 (Colo. 1998). It is undisputed that Rodarte was unavailable for the purposes of CRE 804(b)(B).

Our determination in this case regarding the admissibility of Rodarte's statement is guided by our decision in People v. Newton, 966 P.2d 563 (Colo. 1998). In Newton, we discussed in great detail the seope and application of CRE 804(b)(8). More specifically, we were presented with the question of what the term "statement" in CRE 804(b)(@G) should include. We held in Newton, in contrast to the United States Supreme Court's determination regarding the federal version of 804(b)(8), that under Colorado's version of 804(b)(8) a trial court may admit related, collaterally neutral statements made by the declarant in addition to the precise statement against penal interest. Compare Williamson v. United States, 512 U.S. 594, 600-01, 114 S.Ct. 2431, 129 L.Ed.2d 476 (1994), with Newton, 966 P.2d at 566.7 Newton recognized that 804(b)(8) permits the admission of statements that both exculpate and inculpate the defendant (as opposed to the declarant) and articulated a different analysis for each type of statement against interest. Id. at 572-18.

Although we have found no precedent or other authority that explicitly engages in the task of articulating standards for defining and distinguishing the terms "exculpatory" and "inculpatory," it is clear that a statement against interest may be classified as inculpa-tory as to the defendant even though on its face the statement does not directly inculpate the defendant. One scholar has defined an inculpatory statement very generally as one that "implicates both the declarant and the defendant in criminal activity and which is admitted against the defendant." Comment, Federal Rule of Evidence 804(b)(8) and In-culpatory Statements Against Penal Interest, 66 Cal. L. Rev. 1189, 1190 n. 7 (1978) (cited in State v. Hoak, 107 Idaho 742, 692 P.2d 1174, 1178 (1984)). Many courts have recognized that a statement against interest, *196even a selfinculpatory statement, may be properly classified as inculpatory as to the defendant. These cases characterize these inculpatory statements as "inferentially" in-culpatory. See, e.g., Lilly v. Virginia, 527 U.S. 116, 131, 119 S.Ct. 1887, 144 L.Ed.2d 117 (1999) (citing Lee v. Illinois, 476 U.S. 530, 544 n. 5, 106 S.Ct. 2056, 90 L.Ed.2d 514 (1986)); Williamson, 512 U.S. at 603, 114 S.Ct. 2431; Richardson v. Marsh, 481 U.S. 200, 208, 107 S.Ct. 1702, 95 L.Ed.2d 176 (1987); Foster v. United States, 548 A.2d 1370 (D.C. 1988); Commonwealth v. Charles, 428 Mass. 672, 704 N.E.2d 1187, 1142 (1999); People v. Banks, 438 Mich. 408, 475 N.W.2d 769, 773 (1991).8 Because we conclude that Rodarte's statement inferentially inculpates Bernal, see discussion, infra, we apply the Newton analytical framework applicable to inculpatory statements.

With regard to a statement against interest that inculpates the defendant, Newton explained that the following three requirements of 804(b)(8) must be satisfied:

First, the witness must be unavailable as required by CRE 804(a). Second, the statement must tend to subject the declarant to criminal liability.... Third, the People must show by a preponderance of evidence that corroborating circumstances demonstrate the trustworthiness of the statement. In conducting this third inquiry, a trial court should limit its analysis to the cireumstances surrounding the making of the statement and should not rely on other independent evidence that also implicates the defendant. Appropriate factors for a trial court to consider include: where and when the statement was made, to whom the statement was made, what prompted the statement, how the statement was made, and what the statement contained.

966 P.2d at 576 (citing State v. Wilson, 323 Or. 498, 918 P.2d 826, 837-39 (1996)(outlining the above factors)) (emphasis added). Accordingly, Newton instructs that a trial court's assessment of the "corroborating circumstances" for an inculpatory statement, the third requirement of 804(b)(8), is the equivalent of an assessment pursuant to the Confrontation Clause. In requiring the satisfaction of a defendant's Confrontation Clause rights in order to find the third requirement of 804(b)(8) to be met, Newton recognized that a defendant's constitutional right to confrontation is directly implicated when a statement against interest inculpates the defendant under 804(b)(8):

Most courts that have required corroboration for inculpatory statements have done so out of concern that such statements comply with the Confrontation Clause.... It therefore makes sense that [804(b)@B)'sl corroboration requirement for inculpatory statements, which is rooted in the Confrontation Clause, complies with the Supreme Court's explanation in [Idaho v.] Wright [, 497 U.S. 805, 110 S.Ct. 3139, 111 L.Ed.2d 638 (1990),] that the Confrontation Clause can only be satisfied by looking to the inherent trustworthiness surrounding the making of the statement.

Id. at 575.9

B. The Confrontation Clause

Because Newton directs that the Confrontation Clause analysis is incorporated into the third 804(b)(@8) requirement when the statement inculpates a defendant, we explain the Confrontation Clause and its requirements before assessing the admissibility of Rodarte's statements.

*1971. Trustworthiness and Reliability Generally

In Ohio v. Roberts, 448 U.S. 56, 100 S.Ct. 2531, 65 L.Ed.2d 597 (1980), the United States Supreme Court determined that when an out-of-court statement fits into a "firmly rooted" exception to the hearsay rule, the reliability and trustworthiness of that statement "can be inferred without more." Roberts, 448 U.S. at 66, 100 S.Ct. 2531. The Roberts Court held that when an out-of-court statement does not fall into a firmly rooted exeeption to the hearsay rule, the statement must bear "indicia of reliability," which is indicated by "a showing of particularized guarantees of trustworthiness." Id. Both this court and the Supreme Court have held that the statement against interest exception is not a firmly rooted exception to the hearsay rule. Stevens v. People, 29 P.3d 305, 313 (Colo. 2001); Lilly, 527 U.S. at 134, 119 S.Ct. 1887.

In Idaho v. Wright, 497 U.S. 805, 110 S.Ct. 3139, 111 L.Ed.2d 638 (1990), the Court explained that "particularized guarantees of trustworthiness" must be shown from the totality of the cireumstances, but that "the relevant cireumstances include only those that surround the making of the statement and that render the declarant particularly worthy of belief." Wright, 497 U.S. at 819, 110 S.Ct. 3139. The Wright Court went on to state: "In other words, if the declarant's truthfulness is so clear from the surrounding cireamstances that the test of cross-examination would be of marginal utility, then the hearsay rule does not bar admission of the statement at trial." Id. at 820, 110 S.Ct. 3189. A court may not refer to other evidence at trial when assessing the particularized guarantees of trustworthiness. Id. at 822, 110 S.Ct. 8189; see also Lilly, 527 U.S at 134-38, 119 S.Ct. 1887. We have adopted the reasoning and conclusions of these Supreme Court cases with regard to our own 804(b)(8) and the Confrontation Clause. Stevens, 29 P.3d at 310 (citing Lilly, Wright, and Roberts ).

2. Factors for Determining Trustworthiness and Reliability

Newton articulated the following factors for consideration in determining the trustworthiness of statements that inculpate the defendant: where and when the statement was made, to whom the statement was made, what prompted the statement, how the statement was made, and what the statement contained. Newton, 966 P.2d at 575-76.

Stevens expanded the factors for consideration to include: the nature and character of the statement, the relationship between the parties to the statement, the declarant's probable motivations for making the statement, and the circumstances under which the statement was made. Stevens, 29 P.3d at 314 (citing People v. Fuller, 788 P.2d 741, 745 (Colo. 1990)).10 Stevens also held that although the most important determination regarding trustworthiness is whether the statement is genuinely self-ineulpatory or whether it shifts blame to the defendant, this determination is not dispositive. Id. at 315.

In addition to consideration of the foregoing factors, a determination regarding the trustworthiness of Rodarte's statement must be informed by the Supreme Court's conclusion that when a statement against interest inculpates a criminal defendant in addition to the declarant, as in a co-defendant situation, such a statement is "inherently unreliable" because "an accomplice often has considerable interest in 'confessing and betraying his cocriminals'" Lilly, 527 U.S. at 131, 119 *198S.Ct. 1887 (citation omitted). The Supreme Court has also opined that "arrest statements by a co-defendant have traditionally been viewed with special suspicion." Lee, 476 U.S at 541, 106 S.Ct. 2056. Our cases have followed the Supreme Court's assessment of statements by a codefendant, holding that such statements are inherently and presumptively unreliable. See, e.g., Stevens, 29 P.3d at 313.

C. Application of the Law to Rodarte's Statement

1. Standard of Review

Appellate review of a possible Confrontation Clause violation is de novo. Lilly, 527 U.S. at 137, 119 S.Ct. 1887. Here, the trial court held a hearing on the admissibility of Rodarte's statement. Grose testified regarding the cireumstances surrounding the making of the statement, including such details as the layout of the police station, the procurement of Rodarte's Miranda waiver, the tone of the interviews, and Rodarte's demeanor during the interviews. In addition, the trial court admitted into evidence the transcripts of the interviews. Accordingly, we conclude that the record is adequate for a de novo review.

2. The Nature of Rodarte's Statement

The specific statement at issue was made to Grose while Rodarte was in police custody: "He [Rodarte] admitted to taking the car, but adamantly denied being involved in the robbery." This statement was made in the context of a much longer narrative given by Rodarte to Grose. A review of the record demonstrates that the evidence admitted at trial, when considered as a whole, leads to the unavoidable conclusion that Ro-darte's statement inferentially inculpated Bernal because it "implicates both the declar-ant and the defendant in criminal activity." Comment, supra, at 1190 n. 7.

Before Groge testified at trial, the employees of the credit union, other eyewitnesses to the robbery, and a police officer who arrived at the crime scene immediately after the robbery all testified. Through this testimony, both Rodarte and Bernal were identified as the men who committed the robbery. After these eyewitnesses testified, the People called Tucker, the owner of the stolen car, Marshall, Tucker's girlfriend at the time of the crimes, and Audra Ramirez, Rodarte's girlfriend at the time of the crimes. These three witnesses provided testimony that established that Tucker's car had been stolen the night before the robbery and was the car used in the robbery. These witnesses also tended to establish that Bernal and Rodarte were together the night before the robbery. After these acquaintances of Bernal and Ro-darte testified, the People called agents from the Colorado Bureau of Investigation, who established that hair samples taken from the car used in the robbery matched those from Rodarte. Finally, the People called two detectives involved in the investigation of the robbery, one of whom was Grose.

Grose testified about his interview with Rodarte, during which Rodarte made the statement at issue here: "He [Rodarte] admitted to taking the car, but adamantly denied being involved in the robbery." Immediately after Grose testified as to Rodarte's statement, he testified about the statement Ramirez gave him. Grose's recounting of Ramirez's statement established several important facts inculpatory to Bernal. In particular, Grose's summary of Ramirez's statement put Bernal and Rodarte together the night before the robbery. The night before the robbery was also the night that Tucker's car, which was used in the robbery, was stolen. Grose also established that Ramirez told him that the night before the robbery, Rodarte and Bernal left the house together in one car, but returned separately in two cars.

Rodarte's statement, coupled with Ramirez's statement, both as relayed through Grose, tended to prove that Bernal was involved in the planning and theft of the car, which in turn leads to the reasonable conclusion that he also planned and participated in the robbery the next day. Thus, the order of the testimony and the evidence that such testimony established created the links nee-essary to render Rodarte's statement inferentially inculpatory as to Bernal.

*199Turning to a consideration of the specific content of Rodarte's statement, it implicates him in the criminal activity of the car theft and exculpates him as to the robbery. The statement does not expressly deny Bernal's involvement in the theft of the car or in the robbery. Further, Rodarte's admission to the theft of the car, his express denial of involvement in the robbery, and the other evidence at trial linking Rodarte to Bernal and both men to the theft of the car, the robbery, and to both crime scenes, created for the jury the reasonable inference that Bernal participated in the robbery. Importantly, Rodarte's adamant denial of participation in the robbery created the reasonable inference that Bernal may have been the principal in the robbery.

We thus view Rodarte's statement, while facially inculpating himself as to the car theft, as attempting to shift the blame for the robbery to the only other individual implicated by the other evidence, namely Bernal. See, e.g., Williamson, 512 U.S. at 603, 114 S.Ct. 2431 ("And when seen with other evidence, an accomplice's self-inculpatory statement can inculpate the defendant directly: 'I was robbing the bank on Friday morning, coupled with someone's testimony that the declarant and the defendant drove off together Friday morning, is evidence that the defendant also participated in the robbery.").

Additionally, the fact that the prosecution offered the statement at Bernal's trial further supports our conclusion that the statement was inferentially inculpatory as to Ber-nal for purposes of an 804(b)(8) analysis. See, e.g., United States v. McCleskey, 228 F.3d 640, 644 (6th Cir. 2000)(noting that when the government seeks to introduce a statement under 804(b)(8), it tends to inculpate the defendant by "spreading or shifting onto him some, much, or all of the blame"). Finally, in his closing argument, the prosecution made explicit the inculpatory connection between Rodarte's statement regarding the car and Bernal's alleged guilt: "Once again, the last piece of evidence showing the guilt is the association. And it's the association that night when getting the car, and the next day in the bank. That's the association that we care about."

8. CRE 804(b)(8)'s Three Pronged Test Pursuant to Newton

Having determined that Rodarte's statement is ineulpatory as to Bernal, Newton compels the following analysis and conclusion. First, Rodarte was unavailable, as required. Second, the statement tended to subject Rodarte to criminal liability. Finally, because the statement is inferentially incul-patory as to Bernal, the People must show by a preponderance of evidence that the "corroborating circumstances" surrounding the making of the statement, not including other independent evidence, satisfy the Confrontation Clause's requirement of trustworthiness and reliability. Although Rodarte's statement satisfies the first two prongs of 804(b)(8), we find that the third 804(b)(8) prong of "corroborating cireumstances," as interpreted by Newton, was not satisfied.

The record reveals the following facts regarding the cireumstances surrounding the making of the statement: Detective Grose of the Brighton Police Department interviewed Rodarte the day of his arrest, January 30, 1997, while he was in police custody at the Brighton police station. Over the course of several hours on that day, which was approximately six weeks after the robbery, Grose conducted several interviews with Rodarte. Between each interview, Rodarte was placed in a holding cell at the police station. Further, Grose testified that Rodarte was worried, particularly about what kind and how much evidence the police had against him, during the interviews. Under Newton, the "where," "when," "to whom," and "what prompted" factors all indicate that the statement was untrustworthy. Newton, 966 P.2d at 575-76; see also Lilly, 527 U.S at 131, 119 S.Ct. 1887; Stevens, 29 P.3d at 313.

Similarly, the factors articulated by the Tenth Cireuit and cited with approval by Stevens lead to the conclusion that Rodarte's statement was not sufficiently trustworthy to satisfy 804(b)(8)'s "corroborating cireum-stances" requirement, which incorporates the requirements of the Confrontation Clause. A review of the interview transcripts, which amount to one continuous statement, reveals that the content of Rodarte's statement was *200contradictory in many places. In speaking about the car theft, Rodarte told Grose that he stole the car in order to sell it to four other men, whom he did not know. He further said that he did not know what the men were planning to do with the car. Later, however, he stated that he knew the men were going to do something illegal, perhaps even commit a robbery. Additionally, Ro-darte's statement is peppered with statements such as "I don't remember" and "I am not sure." Thus, the "nature and character" of the statement was contradictory, elusive, and inconsistent, none of which supports a finding that the statement is trustworthy.

Further, the "declarant's probable motivations for making the statement," as well as the "cireumstances under which the statement was made," also lead us to conclude that the statement is untrustworthy. Although he did provide a detailed account of the car theft as well as the four men who allegedly asked him to steal the car, the entirety of the statement makes it clear that Rodarte was meeting with Grose in an effort to protect himself from going to jail. At one point he tells Grose: "I'm trying to help myself out dude. I swear to God. But I don't want to go down nowhere dude. I don't want to go down on nothing."

Even Grose, a detective and witness for the prosecution, recognized the untrustwor-thiness of Rodarte's statement. At a motions hearing that addressed, among other issues, the admissibility of Rodarte's statement to Grose, Grose responded to the court's question of why Rodarte was upset during the interviews with the following explanation: "Because he was caught in several lies, and he was being confused as to time, where he was at. And he was getting frustrated because I was going back over his statements and he was changing his statements, and he was confused."

The motivation to mix falsehood with truth thus permeated Rodarte's statement and as a result we cannot conclude that his confession regarding the car theft, while denying participation in the robbery, contained the constitutionally required guarantees of trustworthiness. In contrast, we conclude that Rodarte's statement both attempted to shift blame for the robbery as well as sought to place himself in a light more favorable than Bernal by denying the greater of the two offenses, the robbery, while creating the inference that Bernal participated in that greater offense.

Thus, the record compels us to conclude that the statement is inherently untrustworthy for purposes of the "corroborating circumstances" prong of 804(b)(@8), which equates to a failure to satisfy the Confrontation Clause. Put another way, the presumption of untrustworthiness was not rebutted because it is not at all "clear from the surrounding cireumstances that the test of eross-examination would be of marginal utility" Lilly, 527 U.S. at 136, 119 S.Ct. 1887. In contrast, the nature of Rodarte's statement, coupled with the cireumstances surrounding the making of the statement, indicate that eross-examination of Rodarte would have been invaluable in undermining the credibility of Rodarte and of his statement. The statement was thus erroneously admitted at Bernal's trial.

4, - Constitutional Harmless Error

Our conclusion that Rodarte's statement is untrustworthy and was accordingly erroneously admitted against Bernal does not end the inquiry, however. A violation of the Confrontation Clause is subject to a constitutional harmless error analysis.11 Blecha, 962 P.2d at 941-42. To be classified as constitutional harmless error, a court must be confident beyond a reasonable doubt that the error did not contribute to the guilty verdict. The constitutional harmless error test "is not whether, in a trial, that occurred without the error, a guilty verdict would surely have been rendered, but whether the guilty verdict actually rendered in this trial *201was surely unattributable to the error." Id. at 942.

Because we remand on the issue of the admissibility of the impermissibly suggestive photo line-up, there are two possible resolutions to the erroneous admission of Rodarte's statement. If, on remand, the trial court finds that the out-of-court identification was not admissible and accordingly vacates Ber-nal's conviction and orders a new trial, then we direct the trial court to exclude Rodarte's statement at the new trial.

However, if the trial court concludes that the out-of-court identification was admissible, then it would be required to complete a constitutional harmless error review. The court of appeals determined that the admission of Rodarte's statement was harmless error, and in the interest of judicial economy, we will also undertake such review at this time. The evidence presented, not considering Rodarte's statement, leads us to conclude beyond a reasonable doubt that the guilty verdiet in this case was surely unattributable to the error. Specifically, the identification of Bernal by employees of the credit union (assuming that the eye-witnesses' identifications are admissible), Ramirez's statement implicating both Rodarte and Bernal in the theft of the car, and the testimony of an eyewitness in the parking lot of the credit union when the robbers were fleeing the scene, all lead us to believe that the admission of Rodarte's statement was constitutional harmless error. As stated in Wright "[The presence of corroborating evidence more appropriately indicates that any error in admitting the statement might be harmless, rather than that any basis exists for presuming the declarant to be trustworthy." Wright, 497 U.S. at 823, 110 S.Ct. 3139.

In addition, the court gave a limiting instruction to the jury regarding the purposes for which they could consider Ro-darte's statement. That instruction was given contemporaneously with Grose's testimony as well as at the conclusion of the trial, and stated that the statement could not be used to determine Bernal's guilt. Such an instruction, when given with an inferentially inculpatory statement, diminishes the possibility of reversible constitutional error. Seq, e.g., Marsh, 481 U.S. at 208, 107 S.Ct. 1702; Charles, 704 N.E.2d at 1142. Thus, we agree with the court of appeals that the other evidence at trial, coupled with the limiting instruction, renders the error of admitting Rodarte's statement constitutionally harmless.

IV. Conclusion

On the issue of the photo array and out-of-court identification, we hold that the photo array was impermissibly suggestive. On this issue we reverse the judgment of the court of appeals and remand for further factual findings by the trial court on the reliability of, and thus the admissibility of, the out-of-court identification.

On the issue of the admissibility of Ro-darte's statement against interest under CRE 804(b)(8), we find that Rodarte's statement was inferentially inculpatory as to Ber-nal. Thus, under Newton, the Confrontation Clause analysis becomes a part of the inquiry under 804(b)(8)'s third prong. We hold that the statement was improperly admitted under 804(b)(8) because the trial court failed to undertake a trustworthiness and reliability inquiry. We thus affirm the judgment of the court of appeals on this issue. If the trial court finds that the out-of-court identification is inadmissible and thus orders a new trial, we direct the trial court to exelude Rodarte's statement at the new trial. If the trial court concludes that the out-of-court identification is admissible and thus does not order a new trial, the admission of Rodarte's statement in the first trial was constitutional harmless error, and the judgment of conviction is affirmed.

dissenting:

Today the majority, in my view, erroneously analyzes two important questions of federal constitutional law: one involving almost two decades of lower court development without review by the United States Supreme Court and the other extending a very *202recent and highly limited United States Supreme Court pronouncement. While I find the majority's due process analysis of pretrial identification procedures, adopted from certain lower federal courts, to be an improvement on our own prior treatment of the subject, especially because of its mechanical ease of application and its attempt to refine the relative burdens of proof, I nevertheless find it to be irreconcilable with the constitutional underpinnings of the doctrine as articulated by the Supreme Court. Similarly, I find the majority's Confrontation Clause analysis, excluding evidence as "inferentially inculpatory," not only to result from a misreading of the controlling Supreme Court discussion of declarations against interest but in fact to stand the Supreme Court's Confrontation Clause rationale on its head. Because I do not agree that a remand is required and because both of today's holdings involve important constitutional matters with consequences far beyond the outcome of this single case, I consider it important to explain my views in some detail.

I. FACTS AND PROCEDURES

The charges against the defendant and his codefendant, Johnny Raymond , Rodarte, arose from the robbery of the American Pride Co-op Federal Credit Union in Brighton of nearly $12,000, at 2:45 in the afternoon of December 12, 1996. The charges against codefendant Rodarte were severed for trial. At the defendant's trial he was identified as one of the robbers by two bank employees, and a third employee testified to identifying Rodarte as the other robber during his trial. In addition to the testimony of bank employees who were present, the prosecution offered evidence of the police investigation, which located the car used by the robbers within minutes and later developed the code-fendants as suspects from information provided by the car's registered owner. The prosecution also offered physical evidence linking codefendant Rodarte to the car,1 Ro-darte's own admission that he had stolen the car, and testimonial evidence of the relationship between the defendant and Rodarte, in support of its theory that the defendant was the other person who committed the robbery with Rodarte. The defendant did not testify, or offer evidence of an alibi or alternate suspect, instead attempting to show discrepancies in the eyewitnesses' descriptions of the robbers.

The uncontradicted testimony from the trial indicated that four employees were working at the bank on the afternoon of the robbery. At 2:45 pm., assistant manager Mandie McBride noticed a light blue or gray car pull quickly up to the bank and saw two men get out. She then took her seat in the lobby and waited for them to enter. As the first man came in, McBride saw that he was wearing sunglasses and a baseball hat. He walked directly to McBride, pointed a handgun at her face, demanded that she take him to the vault, and pulled her up from her seat by her hair. The second robber, who was also wearing sunglasses and a baseball hat and was identified as codefendant Rodarte, confronted the two tellers, Menna Johnson and Misty Simons, and ordered them to lie on the floor. A customer who subsequently entered the bank was ordered to lie on the floor with them.

Because access to the vault room required passage through the office of the manager, Kathy Wagner, McBride led the first robber into Wagner's office, where they found her sitting behind her desk. Wagner testified that she was able to see the robber's face the whole time he was approaching and as he stood to the side and behind McBride. The robber ordered both women into the vault room, all the while yelling at them not to look at him and striking them in the head with his pistol each of the several times they ignored his order. Once inside, Wagner handed over approximately $12,000. Heeding calls of "time, time," from the robber identified as Rodarte, the first robber joined him in the lobby before the two fled the bank to their waiting car.

A witness in the area at the time of the robbery saw two men drive away from the bank in a foreign, silver/gray hatchback car. Within minutes, the police found an aban*203doned but still running car matching that description, and inside the car the police found two baseball hats and two pairs of sunglasses, matching the descriptions given by witnesses. When they contacted the car's registered owner, Daniel Tucker, he reported that the car had been stolen the night before but that a friend, Tano Torres, had a key and might have taken it. Tucker indicated that when he discovered the car was missing from outside his girlfriend's home, he paged Torres, who returned the call to the girlfriend's phone. Tucker's girlfriend testified that they received a call from Torres that day from what her caller ID identified as the telephone of Raymond Rodarte. Tucker knew Rodarte as Torres' cousin and someone he had met previously.

Tucker told the police that along with Ro-darte, Torres had introduced him to the defendant. - Rodarte's common-law wife de-seribed all three as friends, but she described the defendant and Rodarte as particularly good friends-as being "joined at the hip." Rodarte's wife conceded at trial that she had told police detective Scott Grose in an earlier interview that she, Rodarte, Torres, the defendant, and another man had all been together shooting pool on the night before the robbery, when the car was reportedly stolen, and that after returning to the Rodarte home, the men left in a single car, only to return later in two cars, the latter of which was apparently driven by the defendant.

Detective Grose testified that because Torres had a distinctive sear that had not been identified by the witnesses, he compiled separate photo arrays around Rodarte and the defendant. The Rodarte array was shown only to tellers Simons and Johnson and the customer. Only Johnson was able to identify Rodarte as the robber who remained in the lobby. The defendant's array was shown to Wagner and McBride, who both identified the defendant as the robber who forced them into the vault room. It was also shown to teller Simons, but she was unable to make any identification.

Following the photo identifications, both men were arrested and charged. The defendant was convicted of second degree kidnapping, aggravated robbery, conspiracy to commit aggravated robbery, and second degree assault, and he was sentenced to forty-six years in prison. On direct appeal his convie-tions were affirmed, but his sentence was vacated and the case remanded with directions for the trial court to resentence within the presumptive range for second degree kidnapping and state its basic reasons and primary factual considerations in imposing its new sentence on all counts.

II. EYEWITNESS IDENTIFICATIONS

Prior to trial, the defendant moved to exclude both in- and out-of-court identifications on the grounds that the out-of-court photo identifications of him were so suggestive as to violate due process of law. The trial court heard and denied the motion. The sole witness at the hearing was Detective Grose, who testified about the way he created the photo arrays and the procedures he used to present them, the physical conditions inside the credit union, and the conduct of the robbery itself, as recounted by the statements of the victims. The preliminary hearing transcript of victim Wagner's earlier testimony and cross-examination was also admitted without objection.

Although the defense had subpoenaed eyewitnesses Wagner and McBride to the motions hearing, only Wagner appeared. When the time set aside for the hearing ended, the court denied defense counsel's motion for a continuance until another day to. examine both eyewitnesses and held that their testimony was unnecessary for it to determine that the identification procedures were not "so impermissibly suggestive" as to require suppression. In its ruling, the trial court noted that the array consisted of six young men with relatively identical features, approximately the same facial characteristics, and the same amount of hair, done in about the same way. It also found that the bank had good interior lighting and that the attention of the witnesses was necessarily focused on the defendant during the erucial times.

A. CONSTITUTIONAL UNDERPINNINGS OF DUE PROCESS DOCTRINE

Until 1967 the manner in which out-of— court identifications were conducted was held *204to affect the weight but not the admissibility of identification testimony at trial. Simmons v. United States, 390 U.S. 377, 382, 88 S.Ct. 967, 19 L.Ed.2d 1247 (1968); People v. Monroe, 925 P.2d 767, 770 (Colo. 1996). In that year, however, the Supreme Court decided a trilogy of cases in which it articulated constitutional grounds for the exelusion of evidence of both out-of-court identifications and subsequent in-court identifications tainted by them. See United States v. Wade, 388 U.S. 218, 87 S.Ct. 1926, 18 L.Ed.2d 1149 (1967); Gilbert v. California, 388 U.S. 263, 87 S.Ct. 1951, 18 L.Ed.2d 1178 (1967); Stovall v. Denno, 388 U.S. 293, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967). In two of the three cases, exclusion of evidence was premised on a criminal defendant's right to counsel guaranteed by the Sixth Amendment to the United States Constitution.

In Wade and Gilbert the Court sought to protect against suggestive pretrial identification procedures by permitting the presence of defense counsel, who could observe and bring to the attention of the jury anything that would affect the reliability of the witness's pretrial identification or taint subsequent in-court identifications. It therefore held that a post-indictment identification involving a lineup or showup conducted by law enforcement officials was a critical stage of the proceedings, and a failure to provide counsel at such a procedure would be sane-tioned by the exclusion of the evidence of identification. Wade, 388 U.S. at 236-37, 87 S.Ct. 1926; see also Kirby v. Illinois, 406 U.S. 682, 688-89, 92 S.Ct. 1877, 32 L.Ed.2d 411 (1972)(defendant's Sixth Amendment right attaches "at or after the time that adversary judicial proceedings have been initiated against him ... whether by way of formal charge, preliminary hearing, indictment, information, or arraignment"); People v. Anderson, 842 P.2d 621, 622 (Colo. 1992)2 Because the Court considered it meaningless to merely exclude evidence of the out-of-court identification, once the witness had observed the defendant under these circumstances, it included within the exclusionary sanction for a violation of the right to counsel any subsequent in-court identification by the witness, unless the government could establish by "clear and convincing evidence that the in-court identification was based upon observations of the suspect other than the lineup identification." Wade, 388 U.S. at 240, 87 S.Ct. 1926; Monroe, 925 P.2d at 770. Much like the Fourth Amendment exelusion-ary rule's treatment of derivative evidence, the exclusionary sanction for violation of the Sixth Amendment therefore included an exception for subsequent in-court identifications for which there was an "independent source."

The third case in the trilogy, Stovall v. Denno, involved an identification procedure to which the Sixth Amendment right to counsel was held not to apply. Nevertheless, the Court recognized that independent of any right-to-counsel claim, it would be a valid "ground of attack" that an out-of-court confrontation was "so unnecessarily suggestive and conducive to irreparable mistaken identification that [a defendant would bel denied due process of law." 388 U.S. at 802, 87 S.Ct. 1967. Because the showup in that case, however, was found not to have been "unnee-essarily suggestive," the nature of a due process violation and the scope of any sanction for such a violation was not further defined.

In part because the Supreme Court's early attempts used terms like "unnecessarily suggestive," id., and "impermissibly suggestive," see Simmons, 390 U.S. at 384, 88 S.Ct. 967 ("[Clonvictions based on eyewitness identification ... will be set aside on that ground only if the photographic identification procedure was so impermissibly suggestive as to give rise to a very substantial likelihood of irreparable misidentification."), without actually finding a due process violation and implementing a sanction, the constitutional basis and standards for excluding evidence remained unclear for a number of years. By analogy to a violation of the Sixth Amendment right to counsel, many jurisdictions, including this one, considered the wit*205ness's independent ability to make an identification only as an "independent source" or "independent basis" for allowing an in-court identification, despite an "unduly," "imper-missibly," "unnecessarily," or "unconstitutionally" suggestive out-of-court procedure. See Huguley v. People, 195 Colo. 259, 261-62, 577 P.2d 746, 747 (1978)(allowing in-court identification based on independent source following suppression of "unduly suggestive" out-of-court identification); People v. Stevens, 642 P.2d 39, 40 (Colo.App. 1981)(suppressing out-of-court identification because of "unduly suggestive" and improper procedure but remanding for consideration of independent basis for possible in-court identification); see also People v. Madonna, 651 P.2d 378, 383-84 (Colo. 1982)(allowing in-court identification after excluding "unnecessarily suggestive" photo showup); People v. Horne, 619 P.2d 53, 56 (Colo. 1980)(condoning in-court identifications after "impermissibly," "unduly," or "unconstitutionally" suggestive procedures); see generally Nathan R. Sobel, Eyewitness Identification: Legal and Practical Problems §§ 4:3-4 (2001).

At least by the time of Manson v. Brathwaite, 432 U.S. 98, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977), the Supreme Court made clear that it rejected any thought of excluding identification evidence solely as a means of deterring disfavored police conduct, as it had done with the Fourth Amendment exelusion-ary rule, concluding instead that "reliability is the linchpin in determining the admissibility of identification testimony." 482 U.S. at 114, 97 S.Ct. 2248. In contrast to a per se Sixth Amendment violation, it held that a defendant's right to due process of law was not violated by the conduct of a pretrial identification procedure alone, no matter how suggestive or disfavored. Id. (despite suggestiveness of single photo array, due process not violated if identification possesses certain indicia of reliability); see also Biggers, 409 U.S. at 198, 93 S.Ct. 375 ("[Aldmission of evidence of a showup without more does not violate due process."); People v. Smith, 620 P.2d 232, 237 (Colo. 1980)("One-on-one showups are not favored and tend to be suggestive, but are not per se violative of due process."). A due process violation would occur only by admitting very questionable identification evidence, whether of an out-of-court identification procedure that was so suggestive in the totality of the cireum-stances that it created a very substantial likelihood of misidentification or an in-court identification made unreliable by it. Manson, 432 U.S. at 116, 97 S.Ct. 2243.

The Supreme Court also made abundantly clear that the likelihood of misidentification must always depend upon the witness's ability to reliably identify the defendant, despite the corrupting effect of a suggestive identification procedure, as measured by such considerations as the witness's opportunity to see the perpetrator, his degree of attention, the accuracy of his prior descriptions, the certainty of his identification, and the time and events occurring between the crime and the identification procedure. Id.; see also Biggers, 409 U.S. at 199-200, 93 S.Ct. 375. The practical effect of this holding was to apply the Simmons standard of admissibility for in-court identifications that followed unnecessarily suggestive out-of-court identifications to the initial out-of-court identification as well. See Manson, 432 U.S. at 119-24, 97 S.Ct. 2243 (Marshall, J., dissenting)(arguing that the term "unnecessarily suggestive" was intentionally used in Stovall as a basis for suppressing unjustifiable police conduct alone; that "impermissibly suggestive" was used in Simmons in addressing the admissibility of subsequent in-court identifications where the taint was "irreparable"; and that the Biggers majority in effect overruled both holdings by dropping the word "irreparable" from the Simmons formula and applying it to both in- and out-of-court identifications). The suggestiveness of the procedures themselves, or the lack of any justifiable need for such suggestiveness, could therefore never result, in and of itself, in a due process violation or the exclusion of evidence, see Manson, 432 U.S. at 118 n. 18, 97 S.Ct. 2243, and therefore could also never be "impermissible" in any constitutional sense.

Since Manson, lower courts have generally recognized that where a witness has the ability to make a reliable identification, notwithstanding any suggestive pretrial procedures, the challenged out-of-court, as well as a subsequent in-court, identification must be *206admitted. See Smith, 620 P.2d at 237-238. Failing to grasp the full import of Biggers and Manson, however, many of these courts continued to cling to certain aspects of the constitutional violation theory that was squarely rejected by the Supreme Court in those cases. Most notably, by treating some unspecified type or degree of suggestiveness in the identification procedures themselves as "impermissible," and therefore unconstitutional in the absence of proof by the government of an "independent source" or "independent basis," these jurisdictions simply ignore Manson's interpretation of the prior Supreme Court holdings. See, e.g., United States v. Sanchez, 24 F.3d 1259 (10th Cir. 1994)(purporting to apply the "Simmons test," without acknowledging that Biggers or Manson were ever decided), Monroe, 925 P.2d at 771-72, 774-75 (acknowledging Big-gers and Manson but accounting for "the witness's own observations at the time of the crime" only after determining that "a substantial likelihood of irreparable misidentifi-cation exists" for the suggestiveness of the procedure alone).

The due process analysis of pretrial identification procedures and in-court identifications in their wake initially requires two distinct inquiries and then a "weighing" or balancing of the two. Manson, 432 U.S. at 114, 97 S.Ct. 2243. The court must weigh the suggestiveness of the pretrial procedures on one side, and the witness's independent ability to identify the defendant on the other. Id. While certain identification procedures are disfavored because they are inherently suggestive, see generally Manson, 432 U.S. at 116, 97 S.Ct. 2243 (single photo showups disfavored), no degree of suggestiveness considered apart from the witness's independent ability to make a reliable identification is in any sense "unconstitutional" or "impermissible." Not only do Biggers and Manson not dictate otherwise; their explanation and ultimate holdings are an emphatic rejection of any such reading. As the analysis and conclusion of the majority, and those courts upon which it relies, make painfully clear, applying the concept of "impermissibility" to one half of the equation, as a kind of threshold burden for the defendant prior to any balancing, is not merely a matter of semantics but goes to the very heart of the dispute resolved by the Court in Biggers and Manson.

Shifting the burden of persuasion to the prosecution to establish the reliability of identification testimony can be justified only by characterizing police conduct as being, in some sense, impermissible. See, e.g., State v. Cefalo, 396 A.2d 233, 237 (Me. 1979)("Having utilized an unfair means to establish defendant's guilt, the State must show that defendant was not harmed. by its own transgression."). Unlike violations of the Fourth or Sixth Amendment, from which the "independent source" doctrine is clearly borrowed, however, the due process test applies to both the "derivative" in-court identification and the challenged pretrial identification itself; and unlike unreasonable seizures or post-indictment lineups outside the presence of defense counsel, suggestive pretrial identifications are not unconstitutional. As Manson makes clear in its very formulation of the due process standard, in- and out-of-court identifications are not rehabilitated and rendered admissible by a showing of reliability; rather they become unconstitutional and are exelud-ed or suppressed only upon a showing that there is a "very substantial likelihood of mis-identification." Manson, 432 U.S. at 116, 97 S.Ct. 2243. Conceptually, the burden of persuasion therefore never shifts to the proponent of identification evidence challenged as impermissibly suggestive, even by a preponderance of the evidence, much less by a clear and convincing standard.3

While the Supreme Court has not spoken directly about many of the procedural aspects of a due process challenge to eyewit*207ness testimony, it has held that the ultimate determination whether a pretrial identification procedure creates a very substantial likelihood of misidentification is a mixed question of fact and law. Sumner v. Mata, 455 U.S. 591, 597, 102 S.Ct. 1303, 71 L.Ed.2d 480 (1982). Like the voluntariness of a confession, see Miller v. Fenton, 474 U.S. 104, 106 S.Ct. 445, 88 L.Ed.2d 405 (1985), this determination involves findings of historical fact, such as lighting, timing, positioning, and the witness's degree of attention, as well as the way in which the identification procedure was conducted, all of which are entitled to deference by a reviewing court; but the ultimate constitutional question concerning the likelihood of misidentification is primarily a legal matter to which no deference is owed. Summer, 455 U.S. at 597 n. 10, 102 S.Ct. 1303.

Unlike the voluntariness of a confession, see Jackson v. Denno, 378 U.S. 368, 84 S.Ct. 1774, 12 L.Ed.2d 908 (1964), however, the likelihood of misidentification resulting from a pretrial procedure or a subsequent in-court identification does not require a judicial determination outside the presence of the jury in every case. Watkins v. Sowders, 449 U.S. 341, 349, 101 S.Ct. 654, 66 L.Ed.2d 549 (1981). "It is the reliability of identification evidence that primarily determines its admissibility. And the proper evaluation of evidence under the instructions of the trial judge is the very task our system must assume juries can perform." Id. at 347, 101 S.Ct. 654 (citations omitted). For obvious reasons, a judicial determination outside the presence of the jury will usually be advisable, especially where matters of historical fact are in dispute or the suggestiveness of out-of-court procedures makes admissibility questionable, but it is not required, either by constitution or rule. See id.; People v. Cobbin, 692 P.2d 1069, 1073 (Colo. 1984) (identification suppression motions not among the pretrial suppression hearings required by Crim. P. 41(e) and (g)).

Furthermore, both the Supreme Court and this court, as well as courts in other federal and state jurisdictions, have relied upon testimony at trial in assessing the constitutional admissibility of identification evidence in the wake of an allegedly suggestive pretrial procedure. See, e.g., Manson, 432 U.S. at 115, 97 S.Ct. 2243; Biggers, 409 U.S. at 200, 93 S.Ct. 375; People v. Weller, 679 P.2d 1077, 1083-84 (Colo. 1984); see also United States v. de Jesus-Rios, 990 F.2d 672, 674 n. 2 (1st Cir. 1993); State v. Sims, 952 S.W.2d 286, 290 (Mo.App. 1997); State v. Keeling, 89 S.D. 436, 233 N.W.2d 586, 590 n. 2 (1975); see generalty State v. Henning, 975 S.W.2d 290 (Tenn. 1998)(providing litany of federal and state cases relying on trial testimony to uphold denials of suppression motions). While reviewing courts may not be properly situated to resolve disputed questions of historical fact. for themselves, and while they are obliged to show deference to findings of fact by trial courts, this court has often accepted as established, matters of fact not disputed in the record that are central to a lower court's rulings. People v. Rivas, 13 P.3d 315 (Colo. 2000). Particularly in this context, the distinction between law and fact is not always easily drawn, Sumner, 455 U.S. at 598, 102 S.Ct. 1303, and the ultimate balance between the suggestiveness of the identification procedure and the witness's independent ability to make a reliable identification is treated as a matter of law.

B. ADEQUACY OF TRIAL COURT FINDINGS

While it might be possible in some cases to determine from the lack of suggestiveness in the challenged identification procedure alone that suggestiveness could not have created a substantial likelihood of misidentification, see, e.g., People v. Webster, 987 P.2d 836 (Colo.App. 1998), that is not what happened in this case. Here, the trial court not only observed the photo array and heard unchallenged testimony about the neutral and formal manner in which the identification procedure was conducted; it also heard considerable testimony about the physical conditions and relevant cireumstances surrounding the robbery and the witnesses' multiple opportunities and actual observations of the robbers. Although its ruling was cursory, in addition to assessing the composition of the array itself, the trial court expressly found that the lighting inside the bank was good and that the attention of the witnesses was focused on the *208defendant during his repeated assaults to make them stop looking at him. The court's finding that the out-of-court identification was not "impermissibly suggestive" was therefore the product of a balancing of elements from both sides of the seale.

In addition to the trial court's specific findings of fact, the record of both the motions hearing and the trial contained uncontradict-ed testimony about the robbery that was supportive of the witnesses' independent ability to identify the defendant. The co-op was in a small trailer with abundant natural light and with multiple fluorescent lights, and the robbers were visible at least from the moment they noisily entered the front door, attempting to obscure their appearances only with baseball caps and sunglasses. The two women who identified the defendant observed him for several minutes, rather than merely in passing, and were actually close enough to be struck in the face and head by him several times, have clumps of hair pulled out, and smell alcohol on his breath. Each testified about her attentiveness and training to concentrate on particular facial features, see Manson, 432 U.S. at 115, 97 S.Ct. 2243 (witness's status as trained police officer is a factor in determining degree of attention), one testifying that she focused on the perpetrator's hair, skin, lips and jaw line, and the other noticing particularly his jaw and puffy cheeks. One described the six-foot, 155 pound defendant as tall with a medium build, black hair, and medium colored rather than dark skinned; while the other described him as taller than she perhaps five feet, eight inches tall-with a medium build, black hair, and skin that was not white but not dark.4 Each indicated that she was positive of the photo identification, and neither expressed any equivocation in identifying the defendant in court or during cross-examination.

Even the delay of six weeks between the robbery and the out-of-court identification procedure did not significantly weigh against the witnesses' ability to independently identify the defendant. To the contrary, in light of the traumatic nature of the encounter, its duration in time, the witnesses' numerous opportunities to view the defendant up close, their level of confidence, and the absence of any wrong choices or failures by the witnesses during the intervening six weeks to identify the defendant, a delay of six weeks is not one necessarily causing adult memories to fade. See Biggers, 409 U.S. at 203, 93 S.Ct. 375 (despite seven-month gap, identification upheld because of other strong indicia of reliability); United States ex rel. Kosik v. Napoli, 814 F.2d 1151 (7th Cir. 1987)(two month gap does not by itself raise serious question about reliability); State v. Stewart, 389 So.2d 1321, 1324 (La. 1980)(lapse of thirteen months found to be "a long time to remember a face," but was not per se unreliable); State v. Story, 646 S.W.2d 68, 71 (Mo. 1983)(seventeen month interval between illegal drug transaction and time undercover agent identified defendant's photo held not unreliable in itself).

Rather than a corporeal or photo showup of the defendant by himself, e.g., Manson, 432 U.S. at 101, 97 S.Ct. 2243 (single photo showup); Biggers, 409 U.S. at 195, 93 S.Ct. 375 (police station showup), or any situation pressuring the witness to make an identification, see Manson, 432 U.S. at 116, 97 S.Ct. 2243, an array of six photos, including the defendant, was shown to each of the three women. The witnesses were separated in different rooms of the co-op before being shown the array and read a standard set of instructions by the administering detective. Each was specifically told that the array might or might not contain a suspect and that her choice, if she made one, could not be confirmed. Each was asked not to discuss the array or her choice with anyone. Manager Wagner picked the defendant's photo after studying the pictures for about two minutes, indicating that she was positive of her - identification. _ Assistant - manager McBride picked the defendant's photo after *209about one minute, indicating that she was positive of her identification.

The array itself contained photos of six men with approximately the same youthful appearance, medium complexions, dark hair, and generally similar facial characteristics, arranged in two rows of three photos each, with the defendant appearing in the middle of the top row. By far, the most prominent feature of each individual, and the feature that Detective Grose clearly sought to match in every photo, was the unusual hairstyle (described by the trial court as "done the same goofy-looking way"), which appeared as a tuft of hair rising from the top of the head with little or no hair on the sides.5 Two photos, including that of the defendant, had differently shaded backgrounds,6 but none appeared to be a mug shot or more likely to be a photo of a police suspect. Although the two witnesses who ultimately identified the defendant indicated for various reasons and with varying degrees of confidence a belief that the robber was Hispanic, they did not describe him as having physical features that particularly appeared to be Hispanic.7 Furthermore, the varying ethnicities of several of the men in the photos were not so pronounced as to make the defendant's photo stand out from the others as one the police intended the witness to choose. Cf. People v. Mahdi, 89 Ill.App.3d 947, 45 Ill.Dec. 318, 412 N.E.2d 669 (1980)(where murderer was described as Arab, photo array not suggestive despite defendant being only Arab in array consisting of others of Puerto Rican or Mexican descent resembling defendant). In any event, however, the array was apparently not so suggestive as to influence teller Simons, who was unable to identify anyone from it. See Wiseman, 172 F.3d 1196, 1211 n. 7 (10th Cir. 1999) (failure of some witnesses to identify defendant considered indication that array was not too suggestive). Even the identifying witnesses studied the photos for some time rather than immediately leaping at the defendant's photo.

As often noted in the past, police stations are not theatrical casting offices and cannot be required to provide exact replicas of the defendant for a lineup. See United States v. Schultz, 698 F.2d 365, 367 (8th Cir. 1983); People v. Bolton, 859 P.2d 311, 319 (Colo.App. 1993); People v. Borrego, 668 P.2d 21, 23 (Colo.App. 1983). On the contrary, to the extent that the object of a photo lineup is to determine the identity of the perpetrator rather than to deceive the witness, exact replicas of the suspect would not even be desirable. If uniformity were too rigidly required in photographic lineups, they would become ineffective as an investigative tool.

Moreover, to the extent that any suggestiveness in a pretrial identification procedure appears on the face of the array, it can be brought to the attention of the jury, whether or not counsel was present during the procedure, just as well as it can be brought to the attention of a reviewing court. As has been stated by both the Supreme Court and this court:

We are content to rely on the good sense and judgment of American juries, for evidence with some element of untrustworthiness is customary grist for the jury mill. Juries are not so susceptible that they cannot measure intelligently the weight of *210identification testimony that has some questionable feature.

Manson, 432 U.S. at 116, 97 S.Ct. 2243; Monroe, 925 P.2d at 772. Not only do such concerns present less of a danger than suggestiveness that arises from the manner in which police present the array to a witness, which the jury cannot observe, but to the extent that the composition of an array presents a less difficult choice for the witness, it correspondingly loses probative foree as a test or demonstration and detracts from the persuasiveness of a subsequent in-court identification.8

Despite the care taken during the presentation of the photo array in this case, it must be considered suggestive to some degree for a number of reasons, not least because of the differently shaded background of the defendant's photo and the predisposition of the eyewitnesses to identify an individual they thought might be Hispanic. The constitutionally significant question, however, is whether that suggestiveness, considered with all other relevant cireumstances, created a "very substantial likelihood of misidentification" at that time and later when the witnesses identified the defendant at trial,. The factors, or questions of historical fact, from which this determination should be made and for which the trial court is primarily responsible, such as lighting, timing, positioning, the extent of the defendant's disguise, the number of times the witnesses looked at him, and the confidence expressed by the witnesses in their identifications, were not in dispute. The ultimate balance is a mixed question of fact and law for this court. Sumner, 455 U.S. at 597, 102 S.Ct. 1303.

Even with regard to more subjective matters like the emotional state of the witnesses, their impressions, fears, memories, and degree of attention, the eyewitnesses were subject to examination and cross-examination at trial, without startling revelation. I see no need for a remand because I do not believe the ultimate constitutional determination in this case is dependent upon any as yet unresolved questions of historical fact. The majority's directions on remand suggest that it too is less concerned with findings that must be made by a trial court than with providing the defendant an additional opportunity to develop matters that he failed to pursue at trial,. I strongly disagree with any suggestion that once a defendant demonstrates a degree of suggestiveness dubbed "impermissible," he acquires the absolute right to, in effect, depose the victims outside the presence of the jury. The extent and nature of the evidence necessary to determine the likelihood of misidentification will vary with the cireumstances of each case, and in my opinion should remain a matter within the discretion of the trial court. Should the record of both the hearing and trial be inadequate to resolve the question, however, I agree that remand for further findings is the appropriate remedy.

The record here demonstrates an intense, traumatic, and physically painful confrontation between trained bank employees and the robber, for several minutes, at extremely close quarters, in good lighting, with only a minimal attempt at disguise. It also shows a professionally conducted photographic lineup, designed to minimize suggestiveness in its presentation, resulting in well-considered, positive identifications by two of the three witnesses to whom it was shown. In my opinion, it cannot fairly be said that any suggestiveness inherent in the choice of subjects or the photos themselves created a "very substantial likelihood of misidentification" without which there can be no violation of the defendant's right to due process of law.

III. CONFRONTATION CLAUSE

Following the photographic identifications, warrants were prepared and both the defendant and codefendant Rodarte were arrested. During custodial interrogation, Rodarte admitted that he and Tano Torres had used Torres' key to take Tucker's car the night before the robbery.9 Rodarte specifically de*211nied, however, that the defendant was involved in taking the car and denied that any of the three had anything to do with the subsequent robbery. Instead, he indicated that he had been solicited by four other Hispanic men to steal a car and that after he had done so, he sold the car to them.

The People moved in limine for admission of Rodarte's out-of-court statement, expressly limited to that portion in which he implicated himself in the car theft. The offer was made for the limited purpose of demonstrating that Rodarte was personally involved in the robbery.10 The trial court allowed the statement pursuant to the exception to the hearsay rule in CRE 804(b)(8) for statements against a declarant's penal interest and instructed the jury following opening statements, contemporaneously with admission, and again at the close of the evidence that the statement could be considered only for the purpose of establishing Rodarte's guilt.

Like its federal counterpart, Rule 804(b)(8) of the Colorado Rules of Evidence allows an exception to the rule barring hearsay evidence for:

A statement which was at the time of its making so far contrary to the declarant's pecuniary or proprietary interest, or so far tended to subject him to civil or criminal liability ... that a reasonable man in his position would not have made the statement unless he believed it to be true.11

CRE 804(b)(8). There has been much discussion about the seope of a "statement" within the meaning of the rule, with the object of deciding whether that term connotes an extended declaration that is sufficiently inculpatory in the aggregate or merely those declarations or remarks within a confession that are "individually self-inculpa-tory." See Williamson v. United States, 512 U.S. 594, 114 S.Ct. 2431, 129 L.Ed.2d 476 (1994) (limiting the federal rule more along the lines of the latter); People v. Newton, 966 P.2d 563 (1998) (construing Colorado's rule to be broader and to include collaterally neutral statements). - Unlike statements shifting blame to someone else, with which those discussions were concerned, the statement admitted in this case was on its face an "individually self-inculpatory" admission.

Although those portions of statements against penal interest implicating other people may carry separate guarantees of trustworthiness in individual cases, their reliability, if any, does not derive from the reasons envisioned by the rule for statements against the declarant's own penal interest. In fact, both this court and the Supreme Court have on a number of occasions articulated reasons why the reliability of such statements is actually suspect, requiring corroboration or separate showings of trustworthiness. Lilly v. Virginia, 527 U.S. 116, 130-31, 119 S.Ct. 1887, 144 L.Ed.2d 117 (1999)("[Wle have over the years 'spoken with one voice in declaring presumptively unreliable accomplices' confessions that incriminate defendants.")(quoting Lee v. Illinois, 476 U.S. 530, 541, 106 S.Ct. 2056, 90 L.Ed.2d 514 (1986)); Stevens v. People, 29 P.3d 305, 313 (Colo. 2001)(holding that a codefendant's statements inculpating the defendant are "presumptively unreliable"). The dangers inherent in statements that seek to shift blame to another or cast the declar-ant's role in more favorable terms, however, are not present in all admissions of codefend-ants that might also prove disadvantageous to the defendant. The reasons that have been identified in these cases for presuming statements inculpating others to be unreliable are limited to statements that facially inculpate others or are intended by the de-clarant to do so. While a statement that minimizes the declarant's culpability by implicating a third party is not one that a reasonable person would refrain from making unless he believed it to be true, a state*212ment admitting the declarant's own felonious conduct, while expressly excluding others, is precisely the kind of statement envisioned by the rule.

Even a genuinely self-inculpatory statement by someone other than the defendant that facially inculpates only the declarant may be relevant in a trial of the defendant. It has long been recognized that proof that a defendant was an accomplice to a crime not only allows but in fact requires proof that someone else acted as the principal. See People v. Scheidt, 182 Colo. 374, 382, 513 P.2d 446, 450-51 (1973); People v. Knapp, 180 Colo. 280, 283-84, 505 P.2d 7, 9-10 (1973). In Williamson, the Supreme Court acknowledged that even under its narrow reading of the exception for statements against interest, a declarant's squarely self-inculpatory confession will likely be admissible under Rule 804(b)(8) against accomplices of his who are being tried under a coconspir-ator liability theory. 512 U.S. at 608, 114 S.Ct. 2431.12 While it is to longer necessary in this jurisdiction to distinguish principals of various degrees and accomplices, a person clearly remains legally accountable for the behavior of another whom he intentionally aids or abets in committing an offense. See § 18-1-603, 6 C.R.S. (2001)(Complicity).

Evidence that another person committed a crime is therefore logically relevant to the extent that other evidence tends to show he did not act alone and that the defendant was probably a person who aided him or conspired with him. Apart from liability for the behavior of another, as by complicity, evidence that the maker of a statement against penal interest and another person committed a crime, in conjunction with evidence that the defendant was likely to have been with the maker of the statement at the time of the crime, is evidence that logically tends to prove the defendant was the other person committing the crime. See People v. Haston, 69 Cal.2d 233, 70 Cal.Rptr. 419, 444 P.2d 91 (1968)(conjunction of accomplice with defendant in earlier robberies, together with his admitted involvement in charged robbery, supports inference that defendant and not some other person was accomplice in the charged offense). Even a statement that admits only the declarant's commission of a preparatory act is somewhat probative of his involvement, and through him the defendant's participation, in the charged offense.

Here, codefendant Rodarte admitted stealing the car used in the robbery in a statement expressly exculpating the defendant of any criminal involvement in either the car theft or the robbery. Along with other evidence, however, including physical evidence further linking Rodarte to the crime and testimonial evidence linking Rodarte and the defendant during the period in question, the admission made it more likely that the defendant was one of the two robbers, acting together, than would have been the case from evidence of the defendant's participation alone. Under these cireumstances, I would hold that the trial court did not abuse its discretion in finding Rodarte's statement admissible as an exception to the hearsay rule.

Neither do I believe the court's admission of the out-of-court statement violated the Confrontation Clause. Admittedly, the Confrontation Clause requires more of statements made by a declarant who is not present and subject to cross-examination than that they fall within an exception to the hearsay rule. It requires that they either fall within a "firmly rooted" exception or contain particularized guarantees of trustworthiness. In light of the construction that we have given CRE 804(b)(8), making it broader than the corresponding federal rule, *213it is clear that some statements admissible pursuant to this exception would not fall within a "firmly rooted" exception to the hearsay rule. See People v. Farrell, 34 P.3d 401, 405-06 (Colo. 2001); Stevens, 29 P.3d at 313. As noted by the Supreme Court, however, "the very fact that a statement is genuinely self-inculpatory ... is itself one of the 'particularized guarantees of trustworthiness' that makes a statement admissible under the Confrontation Clause." - Williamson, 512 U.S. at 605, 114 S.Ct. 2431. Without considering whether a hearsay exception can be subdivided into those portions that are firmly rooted and those that are not, see generally United States v. Flores, 985 F.2d 770, 775-80 (5th Cir. 1993), it is enough that the statement in this case is "genuinely self-inculpato-ry," without any suggestion of shifting blame to the defendant or seeking to place the declarant in a light more favorable than the defendant, and therefore it contains the constitutionally required guarantees of trustworthiness.

Despite the failure of the narrow statement that was admitted in this case to even mention the defendant, the majority finds it to be "inferentially inculpatory" of the defendant and therefore presumptively unreliable. It is not entirely clear to me whether the majority intends that no statement of a code-fendant offered by the prosecution at the defendant's trial can be considered genuinely self-inculpatory; that under the specific circumstances of this case, the codefendant's statement is not genuinely self-inculpatory; or that even genuinely self-inculpatory statements may not be offered against someone other than the declarant without a showing of more particularized guarantees of trustworthiness. Because the Supreme Court has held that the very fact that a statement is genuinely self-inculpatory is itself one of the particularized guarantees of trustworthiness that makes a statement admissible under the Confrontation Clause, I disagree that any more particularized showing is required for genuinely self-inculpatory statements. Because it seems clear to me that a statement becomes suspect by shifting blame only if that could have been the intent of the declar-ant, I disagree that a statement can lose its genuinely self-inculpatory character merely by being offered by the prosecution at trial. And with regard to the genuinely self-ineul-patory nature of the statement in this case, I cannot understand how a statement expressly exculpating the defendant of any involvement in either the theft of the getaway car or the robbery, and implicitly laying blame on four other Hispanic men to whom the stolen car was allegedly sold, can conceivably be characterized as attempting to shift the blame for the robbery to the defendant. See maj. op. at 198 - 199.

IV,. CONCLUSION

Because I do not consider it necessary to remand for further findings of historical fact and because I would find that admission of the codefendant's self-inculpatory statement was not an abuse of discretion or violative of the defendant's right to confrontation, I would affirm. I therefore respectfully dissent.

Justice KOURLIS and Justice RICE join in Part III of the dissent.