The United States District Court for the District of Colorado granted Petitioner John Morris’s application for a writ of habeas corpus under 28 U.S.C. § 2254, ruling that the state court violated Petitioner’s constitutional right to present a “cogent defense” during his trial for sexual assault on a child. Respondents appeal. We have jurisdiction under 28 U.S.C. §§ 1291 and 2253, and Fed. R.App. P. 4. We reverse.
The sexual assault charge against Petitioner arose out of the accusation of a 12-year-old boy (the Victim) that Petitioner had fondled him. Petitioner’s first trial ended in a hung jury. He was convicted on retrial in December 1993. The trial court’s alleged errors relate to its refusal to allow the testimony of proposed defense expert Dr. Barbara Bebensee and its restrictions on defense counsel’s cross-examination of Detective John Betz.
The state trial court based its exclusion of Dr. Bebensee’s testimony on its finding that Dr. Bebensee would essentially be expressing an opinion on the credibility of the Victim; the state appeals court agreed. We hold that Petitioner did not rebut by clear and convincing evidence the presumed correctness of that finding. See 28 U.S.C. § 2254(e)(1). We further hold that Petitioner’s due process right to present a defense was not violated by exclusion of the testimony, particularly in light of Petitioner’s failure in state court to present scientific support for the expert’s methods. As for the restrictions during the Betz cross-examination, we hold that the state courts did not unreasonably apply federal law in holding that Petitioner’s right to confrontation was not violated. See 28 U.S.C. § 2254(d)(1).
Before resolving the legal issues, we must discuss at length the factual background.
I. Background
A. The Defense’s Opening Statement
In his opening statement, defense counsel described Dr. Bebensee as “an expert in the proper techniques to be used when *1257interviewing children who are alleged victim[s] of sexual assault or actual victims of sexual assault.” ROA, Yol. 9 at 182. He said that interviewing children “is not a simple matter” and proceeded:
It is a much more complex matter than it might appear, and [Dr. Bebensee] will testify as to that it is a very delicate process. There are right and wrong ways of investigating these cases and there are right and wrong ways of interviewing children victims or witnesses in these kinds of cases, and she will be shedding some light on that through her experience and her professional knowledge. She will shed light on those issues for you to help you understand this case. It might seem very simple on its face, but these are complex cases and she will be able to illuminate for you how complex they are and the proper way of investigation.
Id. at 182-83. Defense counsel then began discussing the credibility of child witnesses:
And she will also be telling you that children do make false accusations of sexual assault. They do make false accusations. It’s not something that doesn’t happen, it does happen. And the experts in the field, they have certain investigative techniques that can assist them and aid them in discovering whether or not an allegation is false or whether or not it is valid. And she will be discussing those things with you in her testimony.
Ultimately, she will he asked to give an opinion on how this case was investigated and whether or not this case points towards a valid accusation or a false allegation or accusation. And what I expect her to say is that—
Id. at 183 (emphasis added).
At this point the prosecutor requested a bench conference, which was granted. The prosecutor objected on the ground that credibility is not a proper subject for expert testimony.
[PROSECUTOR]: I think the case law is pretty clear that testimony regarding truthfulness of the child and validity of the child’s accusation is improper and I object to him telling the jury that that’s what she’s going to say when I don’t think that she can say something that strongly. I’d be happy to get the case law.
Id. at 184. The colloquy continued as follows:
THE COURT: No, I understand the case law. What are you about to say? [DEFENSE COUNSEL]: Just that I expect her to give the opinion that she feels that this is not a valid claim.
THE COURT: Because of an investigative technique?
[DEFENSE COUNSEL]: Well, a whole lot more than that. She looks at many themes that I haven’t gone into here on opening statement.
THE COURT: I’ll permit it.
Id. (emphasis added).
In completing his remarks to the jury about Dr. Bebensee, defense counsel summarized the expected testimony:
[DEFENSE COUNSEL]: Folks, Dr. Bebensee will give an opinion on whether or not she believes that this is a valid accusation, and it’s my — I anticipate she will be telling you that this is not a valid claim, that the story, the statements don’t track. There are too many problems with the case to be considered an accurate, valid claim of sexual assault.
Id. at 185 (emphasis added).
B. The Prosecution’s Case
1. The Victim’s Story
The Victim testified that early on the morning of January 31, 1992, he awoke *1258from his sleep on the couch in his mother’s apartment when he felt Petitioner, who was living with the Victim and his mother at the time, “rubbing [the Victim’s] butt.” Id. at 198. After rubbing his buttocks for a time, Petitioner began to rub the Victim’s penis. The Victim stated that he then pretended to wake up, at which point Petitioner ceased fondling him and proceeded to choke one of the Victim’s cats to death by thrusting his fingers down the cat’s throat. The Victim then told his mother, who did nothing.
Later that day he told several other people about the incident, the police were called, and the Victim spoke with a police officer. Eighteen days later the Victim spoke with Detective Betz about the incident.
Evidence at trial revealed that the Victim had described the incident differently to different people. For example, the Victim was inconsistent with regard to (1) whether he awoke before or after Petitioner pulled his pants' down; (2) whether Petitioner rubbed his penis, his buttocks, or both; (3) whether Petitioner rubbed his buttocks with his face or hands or kissed it; (4) what prompted the incident to end; (5) who pulled up the Victim’s pants when the assault was over; (6) whether Petitioner had a knife; (7) what happened to the kitten; and (8) his accounts of what he had told various adults.
In addition to establishing inconsistencies in the Victim’s statements, the defense attempted to discredit him in other ways. First, the defense suggested that the Victim had a motive to fabricate — that his mother had had a string of live-in boyfriends, of whom Petitioner was only the latest, and the Victim resented having to share her affections. Second, the defense presented evidence that the Victim could be retaliating against Petitioner because the Victim was inordinately fond of his cats and believed that Petitioner had killed one of them. (Petitioner admitted at one point that he had killed the cat, but he said that he accidentally stepped on it, as opposed to choking it.) Third, the defense presented evidence that six years earlier the Victim had falsely accused another man of molesting him. Fourth, although the Victim stated in his first interview with the police that Petitioner had also molested a 13-year-old girl, the girl testified that Petitioner had never touched her and that she had never told the Victim that he had. Finally, the girl’s mother testified that the Victim was subject to mood swings and sometimes lied to get attention.
2. Detective Betz
Detective Betz was the lead detective in the case. The chief function of his trial testimony was to report what the Victim had said at a police station interview on February 18, 1992. But the prosecutor also attempted to use Betz to convey two propositions that would buttress the Victim’s testimony — that the Victim’s accounts of the incident were basically consistent and that Betz, an experienced officer, thought the charge was valid. Petitioner points to these latter two components of Betz’s direct testimony in contending that Dr. Bebensee should have been permitted to testify about the Victim’s inconsistencies and the inadequacy of the investigation.
As for Betz’s testimony that the Victim’s statements were basically consistent, the trial court permitted the testimony only until Petitioner objected. Betz’s testimony was as follows:
Q (By [PROSECUTOR]) Now, I think you just testified you heard [the Victim] testify on March 4,1992?
A Yes, ma’am.
Q You also heard him testify on June 12,1992?
A Yes.
*1259Q And you heard him testify yesterday?
A Yes, I did.
Q When you were interviewing [the Victim], as part of your training and experience, did you evaluate the statements that he was making to you?
A Yes, I did.
Q And did you further evaluate it when you heard him testify later at those other hearings?
A Yes.
Q And when he testified on March 4, 1992, was it consistent with what he told you?
A Yes, it was.
Q And on June 12, 1992, was that consistent with what he told you?
A Yes, ma’am.
Q And what he testified to yesterday, was that consistent with what he told you?
A I would have to say that everything was consistent as far as major—
[DEFENSE COUNSEL]: Your Honor, I object. That is a question of fact for this jury, not for this witness to determine. I object to this. I don’t think it’s a proper line of questioning at all. This is not an expert witness we have here and I don’t think it’s proper to be commenting on the evidence.
THE COURT: Yeah, the objection is sustained, not because it’s an expert witness, but by reason and nature of the form of the question.
ROA, Vol. 10 at 211-12. Defense counsel did not move to strike the objectionable testimony.
Despite the sustained objection, defense counsel entered the identical territory on cross-examination. He obtained Betz’s admission that while “the major facts of the case” were the same in all the Victim’s accounts, there were differences in the details. Id. at 217. But after counsel inquired a bit further into Betz’s opinions regarding the inconsistencies, the prosecutor objected. The court sustained the objection, reminding defense counsel that he had earlier objected to such opinion testimony. Although defense counsel asserted that the court, despite his earlier objection, had allowed the opinion to get in, the court denied counsel’s request to have the record read back. In any event, defense counsel was able to alert the jury to inconsistencies in the Victim’s statements by questioning Betz extensively about matters mentioned by the Victim on other occasions but not in the statement to Betz.
With respect to the prosecutor’s attempt to use Betz to vouch for the charge, the prosecutor began the direct examination by eliciting from Betz that he was a 22-year veteran of the police force and had been a detective for four years, with “training and experience in the investigation of crime” and “training in interviewing child witnesses.” Id. at 199. Then, after Betz testified about his interview of the Victim and expressed his opinion concerning the consistency of the Victim’s statements, the direct examination concluded with the following testimony:
Q [PROSECUTOR]: You’re the person who actually is in charge of this case from the Aurora Police Department?
A Yes, ma’am.
Q It’s important to you to do your job well?
A Yes ma’am.
Q You look at a case and evaluate it?
A Ido.
Q Did you do that in this case?
A I did.
Q And would you file a case if you hadn’t evaluated it properly?
*1260A No, I would not.
Id. at 212.
Rather than raising objections to this testimony, defense counsel pursued on cross-examination Betz’s expertise and the quality of his investigation. Under defense counsel’s questioning, Betz acknowledged that (1) his training in investigating allegations of child sexual abuse consisted of only his attendance at “some seminars” or “workshops” about which he could not supply any details, id. at 237; (2) he did not make a verbatim record of his interview with the Victim, but recorded it in his own words, despite his knowledge of the importance of precision; (3) he did not investigate the Victim’s motives to fabricate the assault, although he was aware that children do make false allegations; (4) before filing charges he did not interview the 13-year-old girl whom the Victim claimed had also been molested, and he later interviewed her only after being instructed to do so by the district attorney; (5) he did not interview any other witnesses, including the Victim’s mother, his neighbors, his teachers, or his babysitters; (6) he did not contact (or obtain the Victim’s family records from) the Department of Social Services or try to determine whether the Victim had been the subject of parental neglect; (7) he did not attempt to retrieve the body of the kitten to determine how it had been killed; and (8) in fact he conducted no investigation before filing charges, other than reading the police reports and interviewing the Victim.
On the other hand, the court sustained the prosecutor’s relevance objections to several questions which, rather than focusing on what Betz had done (or omitted) in this case, related to sexual assault charges in general: (1) “Can you tell us from your training and experience what are some of the dangers and pitfalls that confront an investigator in a sexual assault on a child case?”, id. at 240; (2) “In your training, are you taught about what possible motives for false reporting should be — you should be on the lookout for?”, id. at 241; (3) “And are you also aware that recanting is not typical in these types of cases where false reports are made?”, id. at 240. When defense counsel argued that he was just attempting to inquire into Betz’s investigation and his training and skills, the court responded in a way that suggested the defense could use Dr. Bebensee to criticize the investigation:
[W]hat [Betz] did is relevant. What he didn’t do or might have done is not relevant for you. You can bring an expert in and the expert can be critical of what they did do. And the expert, assuming that one comes in, can say what they did fails to meet a standard. And I don’t have any problem with that concept, but to sit here and probe the area you’re doing through this witness, I sustain the objection.
Id. at 242.
. The court also sustained objections to a question asking whether at the time of the investigation Betz had any literature providing a checklist for conducting the investigation, as well as questions about his omissions in the investigation, such as, “Did you do any investigation in this case to determine whether or not [the Victim] had any prior sexual experience?” and “Did you investigate whether there was any pornography[?]” Id. at 247, 249.
During his cross-examination concerning the conduct of the investigation, defense counsel asked Betz how he determined whether an accusation was valid. Betz replied: “I ... weighed [the Victim’s] credibility, weighed what he had to say, and evaluated that through my experience in determining whether I believed his credibility and believed what he said to me and if it was plausible.” Id. at 245-46.
*1261The prosecutor followed up on redirect. Betz essentially repeated what he had said on cross-examination, stating that in deciding whether to file a case, he weighs “the credibility of the victim ... and his or her statements to me.” Id. at 272-73. Defense counsel objected, but the trial court ruled that the door had been opened on cross-examination.
C. Dr. Bebensee
Shortly before lunch on the fourth day of trial, defense counsel called as a witness Dr. Bebensee, a professor with a doctorate in “education, counseling psychology.” ROA, Vol. 16 at 122. The defense sought to qualify her as an expert in “child psychology, child physical abuse, child sexual abuse, ... the validation criteria and detection of sexual abuse.” Id. at 149. When defense counsel informed the court that Dr. Bebensee would need extra time to set up some demonstrative exhibits, the court ordered disclosure of the exhibits to the prosecution so that it could lodge any objections that it might have before the jury returned from lunch. The prosecution had previously been informed of Dr. Bebensee’s proposed testimony by a letter sent thirty days before trial.
When court reconvened, the prosecutor expressed concern about the inappropriateness of exhibits that set forth lists “reiterating paraphrases of prior testimony of various witnesses at the various hearings,” allegedly so that Dr. Bebensee could compare the statements and comment on their consistency, credibility, and validity. Id. at 115. Defense counsel argued that the exhibits were proper demonstrative aids because they would explain the basis for Dr. Bebensee’s opinion on whether the evidence “indicates a valid claim of sexual assault.” Id. at 117. Elaborating, defense counsel explained that Dr. Bebensee would not simply testify about “the criteria that has [sic] to be followed when investigating these kinds of cases” but would also be “applying that [sic] criteria” to this case in order to “be áble to give her opinion on the stand” concerning “how the investigation was conducted, whether it was conducted properly or not, whether the information that is provided through the investigation indicates a valid claim of sexual assault or not along with a lot of other things that are involved in her testimony.” Id. at 116— 17 (emphasis added).
This response led the court to question the propriety of Dr. Bebensee’s testimony altogether. The court stated:
It appears to be, in significant part, that the expert is going to come in and do something other than merely describe in abstract what a proper investigation of a sexual assault on a child might be. It’s clearly suggested in your comments that she is going to deal not in abstract but with the particulars of the statements of various people in this case and to render an opinion as to the quality or credibility of the statements, and that gets dangerously close to, if not becoming actually an impermissible statement by an expert as to what is true or what is not true in the facts of the case.
Id. at 117-18. As a result, the trial court ruled that it would be necessary to hold a hearing outside the jury’s presence for the purpose of determining the admissibility of Dr. Bebensee’s testimony.
1. The Proffer
The Petitioner’s offer of proof with respect to Dr. Bebensee’s testimony was, as the district court noted, “less than clear.” Mem. Op. and Order at 42. A large part of it related to “validation criteria,” which Dr. Bebensee described as “certain things that are looked at in regards to the validation of — or the credibility of the whole investigation” of an allegation of sexual assault on a child. ROA, Vol. 16 at 136-37. She went on to list the criteria, which *1262include (1) when, where, how and to whom the child disclosed the alleged assault; (2) how the child told the story, including comparison of the first story to subsequent tellings; (3) “whether the telling of the story begins to look like it’s been remembered versus programmed,” id. at 137; (4) the child’s exposure to multiple episodes of abuse and any other experience the child may have had with sexual material or situations; (5) whether the offender groomed the child so that the child was more approachable; (6) the offender’s threats or other inducement for the child to keep the incident a secret; (7) the detail with which the child described the incident over time; (8) recantation; and (9) the child’s family situation. The sources Dr. Bebensee provided for her testimony included “protocols” and “checklists” describing how a proper investigation of alleged sexual assault on a child should be conducted. She testified, however, that the “purpose of [the] criteria ” was “to determine better if the child is credible or not credible,” since “children can and do lie about being sexually molested.” Id. at 141-42 (emphasis added).
At this point in the hearing, defense counsel asked whether he should offer Dr. Bebensee as an expert. In response, the court expressed concern about the admissibility of Dr. Bebensee’s opinion on the validity of the evidence and suggested that defense counsel tell the court “what you intend that the jury would hear from this witness.” Id. at 143. Defense counsel stated that Dr. Bebensee would testify, “in terms of the validation criteria,” as to how such an investigation should have been conducted and how it was conducted in this case. Id. at 143. When the court specifically asked whether “she would render opinions as to the validity of the evidence,” defense counsel did not answer directly but replied:
She would be able to render an opinion as to how the investigation was conducted, whether it was conducted properly or not.
She would be able to render an opinion as to whether the reports and statements made by the police personnel are inconsistent or consistent and whether or not significant details change. She can give an opinion regarding the analysis of the emotional, behavioral, and mental characteristics of the child, family background, previous reporting, and things along those lines.
She can give an opinion concerning corroborative statements and whether there was appropriate interview and investigative procedures that were used in this particular case, and we would be asking her to do that.
Id. at 144. He further asserted that Dr. Bebensee’s testimony “would help the jury to understand the meaning of the contradictory information” it received from the witnesses. Id. at 148.
The prosecutor then questioned Dr. Be-bensee. She elicited Dr. Bebensee’s admission that she had not interviewed any of the witnesses personally but based her testimony on a forensic analysis of the police reports and transcripts of prior testimony. Furthermore, she had conducted no controlled experiments involving the validation criteria she described. Although Dr. Bebensee testified that she thought that other psychologists — Sgroi (whom she claimed had conducted a controlled experiment using the validation criteria in 2,000 cases), Delipse, Kelly-James, Jones and Underwagner — might have conducted such experiments, she was unable to substantiate her claim at the time. Nor has the defense provided such substantiation on any subsequent occasion, either on direct appeal or habeas review.
The prosecutor contended that Dr. Be-bensee was employing “statement ... validity assessment,” a process in which a *1263witness’s statements are analyzed “sentence by sentence.” Id. at 170, 160. She pointed to one of Dr. Bebensee’s exhibits, which appeared to reflect such an analysis, and provided several articles describing and discrediting the technique. Dr. Be-bensee’s testimony regarding validation criteria, the prosecutor argued, was unreliable and inappropriate in that it would have involved her opinion on the credibility of the Victim and/or the accusation.
Dr. Bebensee strongly denied that she was engaging in statement validity assessment in analyzing this case. Then, when asked by the prosecutor whether she was going to tell the jury that the police interview of the Victim was not done well and therefore resulted in an invalid report, Dr. Bebensee said: “I’m not going to say the report is not valid. I’m going to say that there are things that should have been done according to the standards in the field, even in the police field that should have been done that were not done.” Id. at 161.
2. The Trial Court’s Ruling
The trial court ruled that the admissibility of the testimony offered by Dr. Beben-see was governed by Colorado Rule of Evidence 702, which “requires the trial court to determine whether proffered evidence will assist the factfinder to either understand other evidence or to determine a fact in issue.” Id. at 187. The court noted that there had been many Colorado appellate decisions in recent years concerning expert testimony in child sexual assault cases, and that while these decisions had held admissible many varied types of expert testimony, they had held inadmissible any testimony which essentially told the jury “who[m] to believe.” Id. at 189, quoting People v. Gaffney, 769 P.2d 1081, 1087 (Colo.1989).
The court proceeded to rule on the admissibility of Dr. Bebensee’s proffered testimony. It used as a guide her letter to the prosecutor summarizing her six proposed conclusions. The court also considered her voir dire testimony.
In the first three conclusions in her letter, Dr. Bebensee expressed the view that the “initial and subsequent statements” made by the Victim, family members and friends, and police personnel were “inconsistent and significant details change.” Id. at 190-91. The court stated that “the jury[ ] should be fully competent to recognize those observations without expert opinion directing them to these inconsistent statements or changes in detail.” Id. at 190. Hence, it ruled that “scientific knowledge or specialized knowledge would [not] be at all necessary to demonstrate that to the jury.” Id.
The fourth conclusion was an “[ajnalysis of emotional, behavioral and mental characteristics of the child, family background, previous reportings, etc., related to pre-alleged abuse and post-alleged abuse” to show that they “do not meet the necessary criteria established by the professional literature.” Aplt.App. at 381. The court found this statement “somewhat puzzling,” ROA, Vol. 16 at 191, and said it could not determine on what point it would be probative, even if true. The court said:
[T]here is nothing to indicate that the lack of posttraumatic — or postevent emotional behavior or mental characteristics would necessarily mean that an event as simple as the touching or fondling that’s been involved in this case did or did not occur. And again, I must say that I don’t think that that analysis would be useful to the'jury in deciding the issues that are presented in this case.
Id. at 192.
Conclusion five was that “[c]orroborative statements, reports, documents and miscellaneous findings do not support the *1264child’s statements.” Aplt.App. at 381. The court said that “[t]his sounds to me like an opinion by the expert witness that the child is not telling the truth and that type of opinion is not admissible under a long line of well established cases.” ROA, Vol. 16 at 192.
Dr. Bebensee’s final conclusion was that “[inappropriate interview and investigative procedures were used which brings into question the accuracy of the information gathered and conclusions that were drawn from that information.” Aplt.App. at 381. The court indicated that it thought such testimony would be permissible in the abstract. But the court understood this point to mean that Dr. Bebensee “would testify as to what procedures were used, what procedures should have been used, and why the difference between the ‘should have been used’ and the ‘was used’ led to, in her opinion, conclusions lacking in validity.” ROA, Vol. 16 at 193. Such testimony, the court continued, would “draw[ ] into question whether or not any research that is scientific research has been accomplished or demonstrated which would indicate that one line of techniques would more frequently result in a valid determination than another,” and “there has not been any demonstration that such conclusion would be scientific.” Id.
The court concluded:
[T]he sum of this is that the proffered testimony really is not a matter for expert opinion. I think in very large part all it does is go to [the Victim’s] credibility. I don’t think that expert testimony is necessary to evaluate [the Victim’s] credibility. It is certainly suspect. It’s certainly an area of fair argument in this case.
There are lots of inconsistencies. There are lots of reasons why the jury might not think that his testimony was credible, but I don’t think that it is at all appropriate or necessary for an expert witness to come in here and tell this jury what result they should reach. In other words, to tell the jury her conclusions that the information gathered is not credible.
There does not seem to be any scientific principles or there don’t seem to be any scientific principles involved in this case that cannot be understood by the jury. And while Dr. Beben[s]ee might be more qualified than the police officers who investigated this case in taking a fresh case and assimilating and interpreting the data available, I just don’t see that her testimony would be assis-tive to the jury in the context of this particular case.
It would really get down to her opinion as to the credibility of the evidence where we have the accusation made by the victim and the denial by the defendant as to being the primary essence of this case....
Id. at 194-95. The court ruled Dr. Beben-see’s testimony inadmissible.
The defense objected, citing its need for Dr. Bebensee’s testimony in light of the court’s limitations on defense counsel’s cross-examination of Detective Betz regarding his training and the investigation in the case. But the court refused to “entertain further arguments.” Id. at 196. The court did, however, allow defense counsel to clarify on the record his understanding of the court’s ruling (during which defense counsel reminded the court of their discussion about Dr. Bebensee’s forthcoming testimony when the court limited Betz’s cross-examination) and to log his belated objection to the tardiness of the hearing on the admissibility of Dr. Bebensee’s testimony.
Counsel for the defense also requested “some kind of instruction to the jury” explaining what they had been doing outside the presence of the jury and why the *1265doctor would not be testifying despite defense counsel’s earlier statements indicating that she would be. Id. at 203. The court agreed and explained to the jury in general terms what had happened.
I have ordered and ruled that a Dr. Beben[s]ee, who was a witness that you heard described earlier on in the trial, will not be permitted to testify in this case and the reason I did deals with issues of law as compared to issues of fact.
Dr. Beben[s]ee, having been disqualified from testifying by the Court — I don’t know if that’s the correct word. I shouldn’t say disqualified — I just have ruled that she will not testify.
Id. at 208.
In a subsequent motion for a mistrial, defense counsel argued that the court’s statement that Dr. Bebensee was “disqualified” adversely “reflect[ed] on [defense counsel’s] credibility,” ROA, Vol. 11 at 12, because he had told the jury during opening statements that she would be testifying. He further stated that because Detective Betz had been allowed to opine about the consistency of the Victim’s statements, Dr. Bebensee should have been allowed to
present[ ] evidence quite to the contrary ... that these inconsistencies in his statement, of which Dr. Bebensee was aware of each and every one of the statements in words that [the Victim] has said on the record or in a police report, she would have testified that, in fact, these were inconsistencies.
Id. at 10.
D. Conviction and Sentence
Following these rulings, the case was submitted to the jury. During its deliberations, the jury asked whether the Victim would have faced legal consequences if he had recanted. After the court informed the jury that it could not answer the question, defense counsel renewed his motion for a mistrial, arguing that recantation was a fact issue about which Dr. Bebensee would have testified if allowed. Defense counsel asserted that she would have testified that
the literature says that in false claims or false accusations of sexual assault on a child made by a child victim — or alleged child victim, I should say — that recanting is extremely rare and that, in fact, it is much more common that victims recant in the cases where they actually were sexually assaulted.
Id. at 106. The motion for mistrial was denied. The jury found Petitioner guilty, and he received a life sentence as a habitual offender.
E. Direct Appeal
Petitioner appealed his conviction to the Colorado Court of Appeals, arguing, among other things, that the trial court abused its discretion and violated his constitutional rights by excluding Dr. Be-bensee’s testimony, and violated his constitutional right to confront witnesses by limiting the cross-examination of Detective Betz. People v. Morris, No. 94CA0214, slip op. at 1, 5 (Colo.Ct.App. July 25, 1996). The appellate court upheld the trial court’s rulings. It rejected the Petitioner’s confrontation claim. It also determined that the lower court had not abused its discretion in excluding Dr. Bebensee’s testimony, although it did not mention the Petitioner’s constitutional challenge to the exclusion. Id. at 3-6. The Court of Appeals wrote:
The trial court ... determined that the expert witness’ testimony that certain statements given by the victim and by other witnesses during the course of the investigation and at prior proceedings were inconsistent was unnecessary and would not be helpful to the jury in assessing the credibility of the victim. Further, the court determined that the *1266essence of her testimony was an opinion on the victim’s truthfulness as to specific statements made by him and, therefore, was not admissible under [Colorado Rule of Evidence] 608. We agree with the trial court’s reasoning....
Unlike [the Colorado decisions] upon which defendant relies, defendant’s expert did not observe or interview the victim or any of the witnesses and was not going to testify as to her observations of the victim’ [sic] demeanor and patterns of behavior. Nor would her testimony have included only brief references to actual statements made by the victim.
Instead, her testimony was solely based on prior statements of the victim and other persons involved in this incident. Indeed, the expert’s testimony was not such that its indirect effect would have been either to corroborate or to discredit the victim’s credibility, but rather, would have been a direct expression of the expert’s opinion as to the victim’s credibility on prior occasions. Thus, the trial court did not abuse its discretion in excluding the expert witness’ testimony.
Id. at 3-4 (internal citations omitted) (emphasis added). Petitioner’s petition for certiorari was denied by the Colorado Supreme Court.
F. Habeas Petition
Petitioner then sought habeas relief in federal district court under 28 U.S.C. § 2254. The district court granted his petition, finding that (1) “[t]he Colorado Court of Appeals decision was based on an unreasonable determination of the facts in light of the evidence received and excluded at the trial,” Mem. Op. and Order at 40, and (2) “the trial court’s cumulative errors [in excluding Dr. Bebensee’s testimony and limiting the cross-examination of Detective Betz] denied [Petitioner’s] constitutional right [under the Due Process Clause of the Fourteenth Amendment] to present a defense by generating sufficient skepticism about [the Victim’s] testimony to create a reasonable doubt as to whether the assault occurred,” id. The court asserted that the trial transcript “demonstrate^] how devastating the ruling excluding all testimony from Dr. Bebensee was and how it undermined the legitimacy of the effort to persuade the jury that the police had accepted the victim’s statements at face value without conducting an adequate investigation to validate the filing of the charges.” Id. at 41.
The district court’s opinion states that the Colorado courts “unfairly characterized the proposed testimony of Dr. Beben-see as an opinion on credibility beyond the limitations of Rule 608 of the Colorado Rules of Evidence,” and “incorrectly applied Rule 702.” Mem. Op. and Order at 42. The court based this determination on its conclusion that both state courts confused the testimony offered by Dr. Beben-see on the “validation criteria” with the “statement validity analysis” discussed by the prosecutor during Dr. Bebensee’s voir dire. Id. The district court further stated that “Dr. Bebensee drew a distinction between the analysis of the validity of the victim’s statements and the need for the use of generally accepted methods of investigation before deciding that the claim is sufficiently valid to warrant criminal prosecution.” Id. It concluded that the proffered testimony was “almost entirely parallel with the ‘validation criteria’ used by a psychotherapist testifying as an expert witness for the prosecution and approved by the Colorado Court of Appeals in People v. Aldrich, 849 P.2d 821, 827 (Colo.Ct.App.1992).” Id.
The district court wrote:
The proffer of Dr. Bebensee as an expert witness was less than clear. Counsel failed to articulate the critical dis-*1267tinetion between opinions as to the credibility of the testimony given by witnesses and an assessment of the validity of the police investigation leading to the prosecution of the criminal case.
Id. This failure, the court asserted, led to the state courts’ misapplication of Colorado Rule of Evidence 702 in this case. In the district court’s opinion,
the jurors should not be expected to have sufficient experience with circumstances comparable to those presented at this trial to enable them to rely solely on their developed knowledge of human nature to warrant the conclusion that their common sense gives them an adequate basis for their analysis of the evidence.
The prejudicial effect of the limitations on cross-examination and the exclusion of the defendant’s expert witness was aggravated by the trial court’s denial of defense objections to the direct and rebuttal questioning of Detective Betz. The jury was given this police officer’s opinions that the investigation was properly conducted and that [the Victim’s] accounts of what happened were substantially consistent in all material aspects while the defendant was prevented from attacking the reasonableness of both conclusions. Moreover, the jury was led to believe that the detective had special training and qualifications in the investigation of child sexual abuse claims.
Id. at 43.
II. Analysis
A. Standard of Review
Because Petitioner filed his habeas petition after the effective date of the Antiter-rorism and Effective Death Penalty Act of 1996 (AEDPA), our review of this case is governed by the revised standard of review set out in the AEDPA amendments to 28 U.S.C. § 2254. AEDPA dictates that if a claim was “adjudicated on the merits” in state court, a petitioner is entitled to federal habeas relief only if the state court decision was (1) “contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States” or (2) “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d)(1)-(2).
The district court and the parties have assumed, without discussing the matter, that this deferential standard applies. We agree in part. The Colorado Court of Appeals’ opinion expressly addressed the Petitioner’s confrontation claim, so we exercise deferential review of that decision. The opinion did not, however, mention the Petitioner’s claim that the exclusion of Dr. Bebensee’s testimony deprived him of his constitutional right to present defense evidence, a matter raised in an 11-page portion of his opening brief on appeal. When the state court addresses the great bulk of the issues raised by the petitioner’s brief in that court but omits to address a particular claim, we have inferred that the claim was not decided “on the merits” in the state court. Duckett v. Mullin, 306 F.3d 982, 991 n. 1 (10th Cir.2002). That is the situation here. Therefore, we do not apply the deferential review set forth in 28 U.S.C. § 2254(d) with respect to that claim.
On the other hand, another provision of AEDPA, 28 U.S.C. § 2254(e)(1), instructs us to presume that state court factual findings are correct unless rebutted by clear and convincing evidence. This presumption will apply here because the state courts made certain findings concerning Dr. Bebensee’s proposed testimony, even though the courts were addressing only state law.
*1268As for our standard of review of the federal district court decision, in general “we review the district court’s factual findings under a clearly erroneous standard and its legal conclusions de novo.” Valdez v. Ward, 219 F.3d 1222, 1230 (10th Cir.2000). But “[w]hen [, as here,] the district court’s findings [of fact] are based merely on a review of the state record, we do not give them the benefit of the clearly erroneous standard but instead conduct an independent review.” Smallwood v. Gibson, 191 F.3d 1257, 1264 n. 1 (10th Cir.1999).
Finally, we emphasize that our review under § 2254 is confined to alleged errors of federal law. Estelle v. McGuire, 502 U.S. 62, 67-68, 112 S.Ct. 475, 116 L.Ed.2d 385 (1991). In particular, we do not concern ourselves with whether the state courts’ rulings were contrary to Colorado law.
B. Cross-Examination of Detective Betz
Before addressing the heart of Petitioner’s claim — the exclusion of Dr. Bebensee’s testimony — we discuss the cross-examination of Detective Betz. Petitioner argues not only that his confrontation rights were violated by restrictions on his cross-examination of Betz, but also that those restrictions exacerbated the prejudice from exclusion of Dr. Bebensee’s testimony.
For us to grant relief on the confrontation claim, we must find that the state court’s affirmance of the limitations on cross-examination of Betz either “involved an unreasonable application of clearly established Federal law” or “was based on an unreasonable determination of the facts....” 28 U.S.C. § 2254(d)(l)-(2). We do not so find.
In reviewing Petitioner’s confrontation claim, we must recognize that the constitutional right to cross-examine prosecution witnesses is not unlimited. “[T]rial judges retain wide latitude insofar as the Confrontation Clause is concerned to impose reasonable limits on ... cross-examination based on concerns about, among other things, harassment, prejudice, confusion of the issues, the witness’ safety, or interrogation that is repetitive or only marginally relevant.” Delaware v. Van Arsdall, 475 U.S. 673, 679, 106 S.Ct. 1431, 89 L.Ed.2d 674 (1986). In this case the trial judge’s limitations on cross-examination appear reasonable.
First, consider Betz’s testimony concerning inconsistencies in the Victim’s statements. Although Betz testified on direct that the statements were consistent, the judge halted the prosecutor’s questioning on the matter as soon as an objection was raised by defense counsel. When defense counsel began to enter the same territory on cross-examination, the judge treated the matter as he had before, sustaining the prosecutor’s objection. More importantly, all the inconsistencies were in the trial record, so defense counsel could point them out to the jury in final argument. What was relevant was whether the Victim made inconsistent statements; we see no relevance to Betz’s opinion regarding whether the statements were inconsistent.
Turning next to the cross-examination about Betz’s training, we disagree with Petitioner’s characterization of the trial judge’s rulings as “precluding] inquiry into an entire area of relevant cross-examination.” Aple. Br. at 59 (quoting United States v. Atwell, 766 F.2d 416, 419 (10th Cir.1985)). On direct examination Betz had said merely that he had 22 years’ experience as a police officer “with training and experience in the investigation of crime” and “training in interviewing child witnesses.” Defense counsel was permitted to ask whether Betz had attended seminars or workshops, how long ago the *1269training had occurred, and whether he had been trained by Social Services. The only question regarding training to which objection was sustained was one asking who had trained him not to tape record the interview with the Victim. Moreover, defense counsel was able to ask a number of questions that suggested the sloppiness and incompleteness of Betz’s investigation, thereby undermining any claim that his training gave him special expertise.
The questions cited by Petitioner to which objections were sustained did .not relate to the adequacy of Betz’s training but rather appear to have been intended to elicit his opinion regarding potential weaknesses in child sexual abuse cases in general: (1) “Can you tell us from your training and experience what are some of the dangers and pitfalls that confront an investigator in a sexual assault on a child case?” ROA, Vol. 10, at 240.(2) “In your training, are you taught about what possible motives for false reporting should be — you should be on the lookout for?” Id. at 241.(3) “And are you also aware that recanting is not typical in these types of cases where false reports are made?” Id. at 240. The trial court explained:
[W]hat [Betz] did is relevant. What he didn’t do or might have done is not relevant for you. You can bring an expert in and the expert can be critical of what they did do. And the expert, assuming that one comes in, can say what they did fails to meet a standard. And I don’t have any problem with that concept, but to sit here and probe the area that you’re doing through this witness, I sustain the objection.
Id. at 242. Defense counsel made no effort to convince the court that Betz possessed the requisite expertise to answer the questions (assuming they were otherwise proper).
As for the questions regarding Betz’s omissions in the investigation — e.g., “Did you investigate whether there was any pornography[?]” id. at 249 — they were of marginal relevance, because Betz had already fully described his (very limited) investigation. See Van Arsdall, 475 U.S. at 679, 106 S.Ct. 1431 (Confrontation Clause does not prohibit exclusion of interrogation that is “only marginally relevant.”) Also, defense counsel could make the same point in jury argument.
Finally, Petitioner complains that the trial judge prohibited two questions regarding investigative guides. Betz testified on cross-examination that he had discarded reference materials he had used in “these investigations” and that he did not know whether there were investigative guides “in the field.” Defense counsel then asked whether during the investigation of this case Betz was “in possession of any books or literature or any kind of materials that provided [him] with a checklist of how to approach the investigation” or whether he had “reified] on any guide when [he was] investigating this case.” The prosecutor objected that the questions were irrelevant. The judge sustained the objections. Perhaps we would have permitted this questioning. Nevertheless, the questions had only marginal relevance. Defense counsel’s cross-examination established that Betz had done virtually nothing to check the Victim’s story. The apparent purpose of this line of questioning was merely to emphasize that point. If Petitioner’s counsel expected to obtain something more useful than that from the questions, he should have so informed the trial judge, through a proffer or otherwise.
Based on this review of the restrictions on Petitioner’s cross-examination of Betz, we hold that the Colorado courts’ ruling on Petitioner’s confrontation claim did not “involve[ ] an unreasonable application of clearly established federal law,” nor was it *1270“based on an unreasonable determination of the facts.” 28 U.S.C. § 2254(d)(l)-(2).
C. Exclusion of Dr. Bebensee’s Expert Opinion Testimony
The next question presented for review is whether the district court erred in granting habeas relief on the basis of Petitioner’s claim that the trial court unconstitutionally excluded the testimony of Dr. Bebensee. In both his state appellate briefs and his habeas petition, Petitioner asserted that the exclusion of Dr. Beben-see’s testimony violated his right to present a defense under the Due Process and Compulsory Process Clauses of the United States Constitution.
The district court ruled that the state courts, by excluding Dr. Bebensee’s testimony, had violated Petitioner’s constitutional right “to present a cogent defense,” in violation of “the fundamental fairness guaranteed by the Due Process clause of section 1 of the Fourteenth Amendment to the United States Constitution.” Mem. Op. and Order at 41. The court held that this exclusion resulted from the state courts’ “unreasonable determination of the facts in light of the evidence received and excluded at the trial.” Id. at 40 (quoting § 2254(d)(2)). In so ruling, the district court stated:
The [appellate] court failed to recognize that the trial court’s cumulative errors denied [Petitioner’s] constitutional right to present a defense by generating sufficient skepticism about [the Victim’s] testimony to create a reasonable doubt as to whether the assault occurred. The Court of Appeals repeated the errors of the trial court in separating the questions of the proper scope of cross-examination of the detective who verified the charge from the relevance of opinion testimony about the adequacy of the investigation. Accordingly, this court is not required to defer to the decision of the Colorado Court of Appeals and must make an independent evaluation of the claims that [Petitioner] was convicted after a trial that was fundamentally unfair because he was denied the opportunity to present a cogent defense.
Id. at 40-41.
1. The Constitutional Right to Present a Defense
We take our principal guidance from the Supreme Court’s recent decision in United States v. Scheffer, 523 U.S. 303, 118 S.Ct. 1261, 140 L.Ed.2d 413 (1998), which affirmed a military court’s exclusion of polygraph evidence in a court martial. Before turning to Scheffer, however, we review three earlier Supreme Court decisions distinguished by Scheffer. Comparing these decisions to Scheffer is instructive.
The first of the three cases is Washington v. Texas, 388 U.S. 14, 87 S.Ct. 1920, 18 L.Ed.2d 1019 (1967). In Washington a defendant on trial for murder wished to call a witness who would have testified that he, not the defendant, had killed the victim. Id. at 16, 87 S.Ct. 1920. But because the witness had already been convicted in connection with the murder, the trial judge ruled his testimony inadmissible under a Texas statute that prohibited people charged as coparticipants in the same crime from testifying on each other’s behalf. Id. at 16-17, 87 S.Ct. 1920.
The Supreme Court held that the disqualification of the witness violated the defendant’s constitutional right “to have compulsory process for obtaining witnesses.” Id. at 18, 87 S.Ct. 1920. The Court wrote:
The right to offer the testimony of witnesses, and to compel their attendance, if necessary, is in plain terms the right to present a defense, the right to present the defendant’s version of the facts as well as the prosecution’s to the jury so it may decide where the truth lies.
*1271Id. at 19, 87 S.Ct. 1920 (emphasis added). Pointing out that “the Sixth Amendment was designed in part to make the testimony of a defendant’s witnesses admissible on his behalf in court,” id. at 22, 87 S.Ct. 1920, the Court explained that the Sixth Amendment bars “arbitrary rules that prevent whole categories of defense witnesses from testifying on the basis of a priori categories that presume them unworthy of belief.” Id.
In Chambers v. Mississippi, 410 U.S. 284, 93 S.Ct. 1038, 35 L.Ed.2d 297 (1973), a defendant charged with murder attempted to show at trial that another man, McDonald, was the guilty party. Two eviden-tiary rulings prevented the defendant from fully developing this theory. First, the trial court held that Mississippi’s “voucher” rule barred the defendant from cross-examining and impeaching McDonald, whom he had called as a witness. Id. at 295, 93 S.Ct. 1038. Second, the trial court excluded the testimony of three of McDonald’s friends, to whom McDonald had confessed to the killing, on the ground that the testimony was hearsay. Id. at 298, 93 S.Ct. 1038.
The Supreme Court held that, taken together, the trial court’s rulings “denied [the defendant] a trial in accord with traditional and fundamental standards of due process.” Id. at 302, 93 S.Ct. 1038. With regard to the application of the common-law voucher rule, the Court asserted that “[t]he availability of the right to confront and to cross-examine those who give damaging testimony against the accused” is so important that it should not be interpreted as hinging on the “technicality” of which party called the witness. Id. at 297-98, 93 S.Ct. 1038. As for the application of the hearsay rule, the Court wrote:
The testimony rejected by the trial court here bore persuasive assurances of trustworthiness and thus was well within the basic rationale of the exception for declarations against interest. That testimony also was critical to Chambers’ defense. In these circumstances, where constitutional rights directly affecting the ascertainment of guilt are implicated, the hearsay rule may not be applied mechanistically to defeat the ends of justice.
Id. at 302, 93 S.Ct. 1038.
The third decision, Rock v. Arkansas, 483 U.S. 44, 49, 107 S.Ct. 2704, 97 L.Ed.2d 37 (1987), concerned the application of a rule barring “hypnotically refreshed testimony.” The defendant, charged with manslaughter following her husband’s death from gunshot wounds, recalled certain exculpatory details about the shooting only after being hypnotized. The trial court limited the defendant’s testimony to those facts that she remembered prior to hypnosis, id. at 47-48, 107 S.Ct. 2704, and the Arkansas Supreme Court affirmed this decision, announcing a per se rule excluding hypnotically refreshed testimony. Id. at 48-49, 107 S.Ct. 2704.
Reviewing the Arkansas rule, the Supreme Court emphasized that while “the right to present relevant testimony is not without limitation,” a state’s “restrictions of a defendant’s right to testify may not be arbitrary or disproportionate to the purposes they are designed to serve.” Id. at 55-56, 107 S.Ct. 2704. The Court concluded that Arkansas had “not shown that hypnotically enhanced testimony is always so untrustworthy and so immune to the traditional means of evaluating credibility that it should disable a defendant from presenting her version of the events for which she is on trial.” Id. at 61, 107 S.Ct. 2704.
The Court’s recent decision in Scheffer differed from the preceding three cases in that it affirmed the exclusion of the disputed evidence. Of particular interest to our analysis, Scheffer was the first of this line *1272of cases to address opinion evidence, rather than factual testimony. Scheffer held that the exclusion of polygraph evidence did not violate the accused’s constitutional right to present a defense. The accused, an Air Force airman facing drug charges, maintained at his court-martial that he had not knowingly consumed any drugs. 523 U.S. at 306, 118 S.Ct. 1261. He sought to introduce evidence of a polygraph test that supported his contention, but the trial judge excluded the evidence under Military Rule of Evidence 707, which prohibits the introduction of polygraph results. Id. at 306-07, 118 S.Ct. 1261.
The Supreme Court began its discussion by noting that “[a] defendant’s right to present relevant evidence is not unlimited, but rather is subject to reasonable restrictions.” Id. at 308, 118 S.Ct. 1261. A rule is suspect only if it is “ ‘arbitrary’ or ‘disproportionate to the purposes [it is] designed to serve’ ” and exclusion of evidence is “unconstitutionally arbitrary or disproportionate only where it ... in-fringefs] upon a weighty interest of the accused.” Id. (quoting Rock, 483 U.S. at 56, 107 S.Ct. 2704).
The Court held that excluding polygraph evidence “is a rational and proportional means of advancing the legitimate interest in barring unreliable evidence.” Id. at 312, 118 S.Ct. 1261. In addition, Rock, Washington, and Chambers were distinguishable in that the “[t]he exclusions of evidence ... declared unconstitutional in those cases significantly undermined fundamental elements of the defendant’s defense.” Id. at 315, 118 S.Ct. 1261. In Washington, the Court noted, “ ‘the State arbitrarily denied [the defendant] the right to put on the stand a witness who was physically and mentally capable of testifying to events that he had personally observed.’ ” Id. at 316, 118 S.Ct. 1261 (quoting Washington, 388 U.S. at 23, 87 S.Ct. 1920). In Rock, “the rule [barring hypnotically refreshed recollection] deprived the jury of the testimony of the only witness who was at the scene and had firsthand knowledge of the facts” and also infringed upon the “particularly significant” interest of the defendant “in testifying in her own defense.” Id. at 315, 118 S.Ct. 1261. As for Chambers, its holding was confined to the specific “ ‘facts and circumstances’ presented in that case.” Id. at 316, 93 S.Ct. 1038 (quoting Chambers, 410 U.S. at 303, 93 S.Ct. 1038).
In contrast, said Scheffer, “Rule 707 does not implicate any significant interest of the accused.” Id. at 316-17, 118 S.Ct. 1261. At the court-martial, “the court members heard all the relevant details of the charged offense from the perspective of the accused.” Id. at 317, 118 S.Ct. 1261. The rule excluding polygraph evidence did not keep the defendant “from introducing any factual evidence,” but prevented him only “from introducing expert opinion testimony to bolster his own credibility.” Id.
2. The Trial Court’s Ruling
Before applying these principles to the specifics of our case, we must examine further the nature of Dr. Bebensee’s testimony: what would her testimony have been, and how would it have been relevant? Because defense counsel repeatedly referred to Dr. Bebensee’s opinion that the investigation was conducted improperly, we begin by considering the relevance of such an opinion.
In the abstract, whether the government conducted a thorough, professional investigation is not relevant to what the jury must decide: Did the defendant commit the alleged offense? Juries are not instructed to acquit the defendant if the government’s investigation was superficial. “Conducting a thorough, professional investigation” is not an element of the government’s case. For an investigatory *1273lapse to be relevant, there must be some specific reason why it raises doubt about the defendant’s guilt.
This point may readily be overlooked because of the frequency with which investigative techniques are at issue. For example, when, unlike in this case, the chief issue at trial is whether the accused is the person who committed the crime, the defense may point to lapses in the investigation to explain why law enforcement officers failed to uncover evidence of the “true” culprit. Cf. United States v. Foster, 982 F.2d 551, 553-54 (D.C.Cir.1993) (Cross-examination to show that officer had not followed standard police surveillance techniques — such as use of camera— could suggest officer’s motive to compensate for that failure by expressing excessive confidence in his identification of defendant).
Nevertheless, the common admissibility of inadequate-investigation evidence does not relieve the courts of making case-specific determinations of relevance. Sometimes such evidence is relevant; sometimes not. Thus, in the case at bar it would have been fair game for the Petitioner to argue at his trial that the failure of the police to autopsy the cat raises doubt about the Victim’s account of how the cat died. But it is not apparent why it would be relevant whether the police filed charges without first conducting a thorough investigation.
Accordingly, when Petitioner complains about not being able to put on evidence that the investigation was improperly conducted, he must explain the relevance of the proffered-but-excluded evidence. It was not enough for Dr. Bebensee to state that she would testify that the investigation was not conducted in accordance with widely accepted standards. The Petitioner needed to explain what relevant inference could be drawn from that fact. All that defense counsel said at trial was that the faulty investigation indicated that the charge was invalid. The trial judge grappled with this issue in deciding whether to admit Dr. Bebensee’s testimony. Ultimately, his understanding was that the purpose of Dr. Bebensee’s testimony regarding the inadequacy of the investigation was to support her inference (based on an analysis of the Victim’s statements, among other things) that the accusation was not valid; in other words, the purpose was to show that the Victim was not telling the truth. To repeat the trial judge’s observations:
[T]he sum of this is that the proffered testimony really is not a matter for expert opinion. I think in very large part all it does is go to [the Victim’s] credibility. I don’t think that expert testimony is necessary to evaluate [the Victim’s] credibility. It is certainly suspect. It’s certainly an area of fair argument in this case.
There are lots of inconsistencies. There are lots of reasons why the jury might not think that his testimony was credible, but I don’t think that it is at all appropriate or necessary for an expert witness to come in here and tell this jury what result they should reach. In other words, to tell the jury her conclusions that the information gathered is not credible.
ROA, Vol. 16 at 194.
A different trial judge may have viewed the proffer differently, or at least may have found a component of Dr. Bebensee’s proposed testimony that was being offered as something other than the basis for an expression of opinion on the Victim’s credibility. But the trial judge’s view of the proffer, which was seconded by the Colorado Court of Appeals, is entitled to great deference. Under 28 U.S.C. § 2254(e)(1):
In a proceeding instituted by an application for a writ of habeas corpus by a *1274person in custody pursuant to the judgment of a State court, a determfyution of a factual issue made by a State court shall be presumed to be correct. The applicant shall have the burden of rebutting the presumption of correctness by clear and convincing evidence.
There was ample support in the record for the trial judge’s decision that Dr. Be-bensee’s proposed testimony amounted to an expression of opinion on the Victim’s truthfulness in making his accusation against Petitioner. The clear import of the proffer of Dr. Bebensee’s testimony is that she had examined the statements of the various witnesses, reviewed the conduct of the investigation, and concluded that the charge was not a valid one. Her assessment was not founded on any physical evidence, but on an analysis of what was said. In the context of this case, her assertion that the charge was not valid was tantamount to asserting that the Victim was not truthful. Dr. Bebensee would not be testifying to a conclusion from which the jury could infer that the Victim was lying; no inference would be necessary here, only the mental task of translating the statement that “the charge is not valid” into the plain English statement that “the Victim is not telling the truth.” The task of translation was made easier for the trial judge by various comments by defense counsel: “Ultimately, she will be asked to give an opinion on how this case was investigated and whether or not this case points towards a valid accusation or a false allegation or accusation” (opening statement); “Dr. Bebensee will give an opinion on whether or not she believes that this is a valid accusation, and it’s my — I anticipate she will be telling you that this is not a valid claim, that the story, the statements don’t track” (opening statement); Dr. Bebensee will “give her opinion on the stand ... whether the information that is provided through the investigation indicates a valid claim of sexual assault or not” (pre-proffer). Dr. Bebensee herself echoed this characterization of her testimony: “[T]he purpose of the criteria [is] to determine better if the child is credible or is not credible” (proffer).
Petitioner’s brief to this court actually supports the trial judge’s view because it clarifies (unintentionally) that the thrust of Dr. Bebensee’s testimony was to challenge the Victim’s veracity. For example, the brief asserts, “Pointing out discrepancies among statements that are significant in light of accepted criteria may cause a fact-finder to question one or more of the statements, but it does not tell the factfin-der which account to believe.” Aple. Br. at 38. The “accepted criteria,” however, are no more than criteria for determining whether the person who made the statements is being truthful. To assert that an inconsistency is “significant” is functionally equivalent to asserting that a blood pressure reading on a polygraph is “significant.” Testimony regarding the criteria can be contrasted with, say, an engineer’s testimony that a witness’s account of an accident is inconsistent with the laws of physics. When Dr. Bebensee states that an inconsistency between two statements is “significant,” she is simply using shorthand to say that someone who utters such an inconsistency is probably lying; she is not saying that the statement is inconsistent with other evidence concerning the events at issue or violates some scientific principle. (To the extent that she is saying that her validation techniques are scientifically based, we address that matter later. That is a different issue from the issue of whether her opinion is, in essence, an opinion regarding the Victim’s veracity.)
The trial judge’s characterization of the nature and purpose of Dr. Bebensee’s proposed testimony was a factual determination. See Davidson v. Bowersox, 288 F.3d 1076, 1078 (8th Cir.2002) (viewing state *1275court’s determination of content of proffer as a factual finding). Therefore, applying § 2254(e)(1), we adopt the trial judge’s characterization of the proffer, because Petitioner has failed to point to clear and convincing evidence to overcome the presumed correctness of the trial judge’s fact finding.
Characterizing Dr. Bebensee’s testimony as an expert opinion on the Victim’s credibility does not end our inquiry. To be sure, it ended the inquiry under Colorado law. Both the trial judge and appellate court ruled that Dr. Bebensee’s testimony must be excluded under Colorado Rule of Evidence 608 if it amounts to an opinion regarding the victim’s truthfulness. (This ruling is consistent with the weight of authority. See United States v. Charley, 189 F.3d 1251, 1267 (10th Cir.1999) (“Most courts that have considered the issue have concluded that expert testimony, based on the statements of the alleged victim, that sexual abuse in fact occurred is inadmissible under Fed.R.Evid. 702 (or similar military or state evidentiary rules) because, in such cases, the expert offering the opinion is merely vouching for the credibility of the alleged victim.”).) A state evidentiary rule, however, must sometimes yield to the constitutional right to present a defense. We now turn to that issue.
3. The Constitutionality of Excluding Dr. Bebensee’s Testimony
As previously stated, United States v. Scheffer is closely in point. In that case, eight members of the Court affirmed the exclusion of a polygrapher’s testimony regarding the truthfulness of the subject of a polygraph test. See 523 U.S. at 317 n. 13, 118 S.Ct. 1261 (emphasizing that polygraph testimony is opinion testimony, not factual evidence). But the Court did not hold that rules of evidence could always bar expert opinion on veracity. That is not to say that the Court did not come close to so ruling. Justice Thomas’s plurality opinion, joined on this matter by three other justices, said:
It is equally clear that Rule 707 [barring polygraph evidence] serves a second legitimate governmental interest: Preserving the court members’ core function of making credibility determinations in criminal trials. A fundamental premise of our criminal trial system is that “the jury is the lie detector.” United States v. Barnard, 490 F.2d 907, 912 (C.A.9 1973) (emphasis added), cert. denied, 416 U.S. 959, 94 S.Ct. 1976, 40 L.Ed.2d 310 (1974).
Id. at 312-13, 118 S.Ct. 1261.
This view, however, did not muster majority support. The basis for the majority opinion — the portion of Justice Thomas’s opinion joined by seven other justices— was that exclusion of the polygraph testimony was constitutionally permissible for a combination of two reasons. One, exclusion of the evidence did “not implicate any significant interest of the accused.” Id. at 316-17, 118 S.Ct. 1261. Two, the reliability of the evidence was questionable. Id. at 309-12, 118 S.Ct. 1261.
We therefore examine those factors here. We begin with the first factor— failure to “implicate any significant interest of the accused.” The term “significant interest” is used in a special sense in Scheffer. In ordinary usage, a party’s inability to use favorable polygraph testimony would be considered an injury to a “significant interest” of the party. The discussion in Scheffer, however, indicates that the “significant interest” of concern is the right to put on factual evidence. Scheffer emphasizes that “the court members [at the court-martial] heard all the relevant details of the charged offense from the perspective of the accused” and the accused was not precluded “from introducing any factual evidence.” Id. at 317, 118 S.Ct. 1261. The Court particularly *1276stressed that Scheffer was not restricted in testifying on his own behalf. Id. All that Scheffer was precluded from presenting was “merely ... expert opinion testimony to bolster his own credibility.” Id. In contrast, the defendant in Washington was denied the opportunity “ ‘to put on the stand a witness who was physically and mentally capable of testifying to events that he had personally observed[,]’ ” id. at 316, 118 S.Ct. 1261 (quoting Washington, 388 U.S. at 23, 87 S.Ct. 1920); in Rock the defendant was prohibited from testifying to her own version of events, id. at 315-16, 118 S.Ct. 1261; and in Chambers the defendant was barred from supporting his contention that the true culprit was a man named McDonald by cross-examining McDonald or questioning the men to whom McDonald had allegedly confessed, id. at 316, 118 S.Ct. 1261.
Under the Court’s meaning of “significant interest,” we doubt that any “significant interest” of Petitioner was implicated by exclusion of Dr. Bebensee’s testimony. Dr. Bebensee had no fact-based knowledge; in fact, she apparently had not even talked to anyone with firsthand knowledge. All she could contribute was opinion testimony — opinion testimony that amounted to a conclusion that the Victim’s account was not credible.
Moreover, in determining whether there has been a violation of the right to present a defense, the “significance” factor must be weighed against the second factor — the reliability of the evidence. An astrologer or palm reader may have testimony that could sway the jury, but no court could say that a defendant has the constitutional right to present such evidence.
Here, there is no compelling reason to credit the reliability of Dr. Bebensee’s proffered opinion regarding the validity of the Victim’s allegations. As the trial judge said:
[Dr. Bebensee] would testify that inappropriate interview and investigative procedures were used, which brings into question the accuracy of the information gathered. -
I assume that in this area she would testify as to what procedures were used, what procedures should have been used, why the difference between the “should have been used” and the “was used” led to, in her opinion, conclusions lacking in validity. This of course draws into question whether or not any research that is scientific research has been accomplished or demonstrated which would indicate that one line of techniques would more frequently result in a valid determination than another.
In this area ... there has not been any demonstration that such conclusion would be scientific.
Her own publication and process, which is apparently involved in a process of peer review at this time, is admittedly, by her own statement, not based on controlled experiments.
ROA, Vol. 16 at 193-94. We find no scientific support in the record for the proposition that Dr. Bebensee’s techniques for evaluating the validity of an accusation of child sexual abuse provide superior results to those of an informed jury. Indeed, there are no supporting scientific studies whatsoever in the record, either from the trial court, the state appeal, or the habeas proceeding in district court (although we doubt that the trial judge’s ruling could be challenged on the basis of later-produced studies).
The district court believed that “the jurors should not be expected to have sufficient experience with circumstances comparable to those presented at this trial to enable them to rely solely on their developed knowledge of human nature to warrant the conclusion that their common *1277sense gives them an adequate basis for their analysis of the evidence.” Mem. Op. and Order at 43. Perhaps expert opinion could have improved the jury’s ability to evaluate the Victim’s testimony. But before that possibility can be used to set aside the state court’s exclusion of such evidence, there must be considerably more scientific support than the assertion of the expert herself. Indeed, in Scheffer the Court affirmed the exclusion of polygraph evidence despite a study proffered by the accused that “polygraph testing is reliable more than 90 percent of the time.” 523 U.S. at 310 n. 6, 118 S.Ct. 1261. It was enough that “[t]o this day, the scientific community remains extremely polarized about the reliability of polygraph techniques.” Id. at 309, 118 S.Ct. 1261.
In sum, as we understand the law set forth by the Supreme Court, the exclusion of Dr. Bebensee’s testimony did not deny Petitioner’s constitutional right to present a defense.
We add one further comment regarding the exclusion of her testimony. The district court expressed its view that exclusion of the testimony was contrary to Colorado precedent. It noted that similar evidence was admitted in People v. Aldrich, 849 P.2d 821 (Colo.Ct.App.1992). Of course, relief can be granted under 28 U.S.C. § 2254 only for violations of federal law, not state law. Estelle, 502 U.S. at 67-68, 112 S.Ct. 475. Nonetheless, we could grant relief if a state applied its rules of evidence in a discriminatory fashion, unfairly preventing “a defendant from presenting evidence that is critical to his defense.” Romano v. Gibson, 239 F.3d 1156, 1166 (10th Cir.2001). For this reason, we have reviewed the matter. We find no basis for relief here.
The sole challenge to the evidence in Aldrich was hearsay. The dictum of the court expressing approval of the expert testimony has not been cited in later Colorado reported decisions. On Petitioner’s state appeal, the court distinguished Aid-rich on the ground that Dr. Bebensee
did not observe or interview the victim or any of the witnesses and was not going to testify as to her observations of the victim’ [sic] demeanor and patterns of behavior. Nor would her testimony have included only brief references to actual statements made by the victim. Instead, her testimony was solely based on prior statements of the victim and other persons involved in this incident. Indeed, the expert’s testimony was not such that its indirect effect would have been either to corroborate or to discredit the victim’s credibility, but rather, would have been a direct expression of the expert’s opinion as to the victim’s credibility on prior occasions. Thus, the trial court did not abuse its discretion in excluding the expert witness’ testimony. Cf. People v. Aldrich, supra (expert testimony allowed when expert did not give opinion that the victim’s allegations were truthful).
Colo. Ct.App. Op. at 4. To the extent that the Colorado Court of Appeals opinion in Petitioner’s case failed to distinguish Aid-rich persuasively, the opinion can be read as disapproving the Aldrich dictum on which Petitioner relies. We see no deprivation of due process arising from a change in judicial interpretation of Colorado rules of evidence, particularly when the opinion in Petitioner’s case is rational and not foreclosed by entrenched precedent. Cf. Rogers v. Tennessee, 532 U.S. 451, 121 S.Ct. 1693, 149 L.Ed.2d 697 (2001) (retroactive judicial abrogation of day-and-a-year rule for homicide prosecutions did not violate due process).
Thus, even assuming that the argument has been properly preserved, we reject the contention that Colorado has applied its rules of evidence “unfairly to prevent a *1278defendant from presenting evidence that [was] critical to his defense.” Romano, 239 F.3d at 1166.
III. Conclusion
We reverse the district court’s grant of a writ of habeas corpus to the Petitioner.
Judge McWILLIAMS concurs in the judgment.