When the crime is child sexual abuse, one of the more difficult to detect and prosecute, a conviction hinges often on the words of children. What makes this case troubling is that the children did not testify at trial. On the basis of hearsay statements, supported by minimal corroborating evidence at best, a state jury convicted William and Kathleen Swan of sexually abusing their three-year-old daughter and her friend. The Swans have exhausted their state court remedies and appeal the district court's denial of their habeas corpus petition.

Our main issue is whether admission of the children's hearsay statements violated the Swan's Sixth Amendment Confrontation Clause rights. The Swans also argue that the state withheld information that a key reporting witness had once been sexually abused, that newly discovered medical evidence indicated that their daughter was not molested and that they received ineffective assistance of counsel. We affirm the dismissal of the habeas petition.

The sexual abuse charges stemmed from statements made by the Swans' daughter, B.A., and her three-year-old friend, R.T., to day-care center workers. The two children attended a day-care facility managed by Cindy Bratvold. She had hired Lisa Conradi as the center's new part-time assistant. The disturbing allegations arose on Conradi's second day of work. She told B.A. to keep her dress covered over her tights, reminding her that no one should look at or touch her "private parts." Conradi said that B.A. responded "Uh-huh, Mommy and Daddy do." After further questioning, the child allegedly told Conradi about games with her parents involving sexual acts.

Conradi alerted Bratvold, who called Child Protective Services, a state agency. Bratvold then spoke with B.A., who allegedly told her that the games sometimes included her friend, R.T. Two CPS caseworkers arrived and talked with B.A., but the interview was inconclusive. They ended it when Kathleen Swan arrived to take her daughter home.

R.T. came to the center the next day, but B.A. did not. Bratvold asked R.T. about the Swans and what types of games they played together. R.T. allegedly described activities similar to what B.A. had disclosed, including genital touching and oral sex.

These initial statements to the day-care workers were the most damaging. The Washington Supreme Court reviewed in detail the trial testimony about these events. State v. Swan, 114 Wash.2d 613, 790 P.2d 610, 616-18 (1990), cert. denied, 498 U.S. 1046, 111 S. Ct. 752, 112 L. Ed. 2d 772 (1991). We need not repeat those findings, except to say that the implications are grave and alarming. The children demonstrated precocious sexual knowledge, describing multiple episodes of abuse by the Swans, which, if believed by a jury, would warrant conviction.

After interviews with a CPS caseworker, the police were called and the children placed in protective custody. The state charged both Swans with two counts of statutory rape. Superior Court Judge Ellington conducted pretrial hearings to determine whether the young girls were competent to testify. After observing them, the judge concluded that, because of their youth and inability to answer questions in court, they could not satisfy the competency requirements. The Swans do not contest this ruling.

The state introduced the children's statements to the day-care workers under Washington's statutory child sexual abuse hearsay exception, RCW Sec. 9A.44.120.1 Under the same exception, the state introduced other disputed hearsay evidence, including the children's disclosures to the caseworker, B.A.'s statements to her foster mother and R.T.'s statements to her father and a police detective.

Before the admission of each statement, Judge Ellington, assisted by counsel, conducted extensive preliminary examinations of the reporting witnesses outside the presence of the jury. Once satisfied that a statement met the reliability and corroboration requirements of the statutory exception, the court allowed it into evidence. The jury returned guilty verdicts.

The Washington Court of Appeals reversed, holding that the statements lacked sufficient corroboration as required by the hearsay statute. The State Supreme Court disagreed and reinstated the convictions.

The Swans petitioned the district court for a writ of habeas corpus, arguing that: (1) the admission of the hearsay statements violated the Confrontation Clause; (2) the state withheld favorable Brady evidence; namely, that day-care worker Conradi had been sexually abused; (3) newly discovered evidence showed that their daughter's hymen is intact; and (4) they received ineffective assistance by retained counsel. The court adopted the Report and Recommendation of the magistrate judge and denied the petition on summary judgment.

A. Standard of Review and the Presumption of Correctness

We begin our analysis mindful that this is a habeas corpus proceeding, not direct review of a criminal conviction. The Swans have already had the opportunity to litigate their claims in the state courts. Washington's highest court upheld their convictions. Different principles apply on collateral review, constraining the role of a federal appellate court. As the Supreme Court recently reminded us, we may not second-guess the state courts:

Direct review is the principal avenue for challenging a conviction. "When the process of direct review ... comes to an end, a presumption of finality and legality attaches to the conviction and sentence. The role of federal habeas proceedings, while important in assuring that constitutional rights are observed, is secondary and limited. Federal courts are not forums in which to relitigate state trials."

Brecht v. Abrahamson, --- U.S. ----, ----, 113 S. Ct. 1710, 1719, 123 L. Ed. 2d 353 (1993) (quoting Barefoot v. Estelle, 463 U.S. 880, 887, 103 S. Ct. 3383, 3391-92, 77 L. Ed. 2d 1090 (1983)).

We review de novo the decision to deny the habeas petition and, to the extent we need to review the district court's factual findings, the clearly erroneous standard applies. Hendricks v. Zenon, 993 F.2d 664, 668 (9th Cir.1993).

Under 28 U.S.C. Sec. 2254(d), we accord a presumption of correctness to the state courts' factual findings. But this presumption does not apply to the state courts' resolution of mixed questions of law and fact. Acosta-Huerta v. Estelle, 954 F.2d 581, 585 (9th Cir.1992). Whether the hearsay statements were sufficiently reliable to be admitted without violating the Confrontation Clause is a mixed question. SeeUnited States v. Owens, 789 F.2d 750, 757-58 (9th Cir.1986) (admission of hearsay statement by assault victim who later suffered memory loss), rev'd on other grounds, 484 U.S. 554, 108 S. Ct. 838, 98 L. Ed. 2d 951 (1988); see alsoMyatt v. Hannigan, 910 F.2d 680, 685 (10th Cir.1990) (hearsay declarations of child sexual abuse victim).

Consequently, we accord deference to the state courts' factual findings regarding the timing, manner and circumstances of the hearsay statements. We review de novo the ultimate determination that the Swans' Confrontation Clause rights were not violated.

The Confrontation Clause and the hearsay rule are not coextensive. Although both protect similar values, each sets independent prohibitions on admissibility. SeeOhio v. Roberts, 448 U.S. 56, 62-65, 100 S. Ct. 2531, 2537-39, 65 L. Ed. 2d 597 (1980). The Clause does not necessarily bar the admission of hearsay statements. Most evidence that falls under a recognized hearsay exception may be admitted without confrontation because of its presumed trustworthiness. But the Clause may prohibit introducing some evidence that otherwise would be admissible under a hearsay exception. Idaho v. Wright, 497 U.S. 805, 813-14, 110 S. Ct. 3139, 3145-46, 111 L. Ed. 2d 638 (1990).

A statement falling under an exception will also be admissible under the Clause if the prosecution demonstrates the unavailability of the declarant and that the statement bears adequate "indicia of reliability." Roberts, 448 U.S. at 66, 100 S.Ct. at 2539.

The Swans do not contest the trial court's determination that the children were incompetent to testify and "unavailable" for hearsay purposes.2 The crux of this appeal is whether the incriminating statements bore sufficient "indicia of reliability" to withstand scrutiny under the Clause.

The reliability requirement is satisfied if a statement falls within a "firmly rooted hearsay exception" or if it is supported by "a showing of particularized guarantees of trustworthiness." Wright, 497 U.S. at 818, 110 S.Ct. at 3148. The trial court admitted the statements under Washington's child sexual abuse hearsay exception. Enacted in 1982, this exception is relatively new and not firmly rooted.

As the statements were admitted under a nontraditional exception, the state, as proponent of the evidence, had the burden to demonstrate reliability by showing "particularized guarantees of trustworthiness." The proof is based on consideration of the totality of the circumstances but "the relevant circumstances include only those that surround the making of the statement and that render the declarant particularly worthy of belief." Id. at 819, 110 S.Ct. at 3148.

The Supreme Court, in Wright, identified several factors bearing on the reliability of a child's hearsay statements: (1) spontaneity and consistent repetition; (2) mental state of the declarant; (3) use of terminology unexpected of a child of similar age; and (4) lack of motive to fabricate. Id. at 821-22, 110 S.Ct. at 3149-50 (citing cases). But no mechanical test prevails and "courts have considerable leeway in their consideration of appropriate factors." Id. 497 U.S. at 822, 110 S. Ct. at 3150.3

Despite this latitude, other corroborating evidence may not be considered in assessing the hearsay's reliability. This would permit "admission of a presumptively unreliable statement by bootstrapping on the trustworthiness of other evidence at trial." Id. at 823, 110 S.Ct. at 3150. It also presents dangers because of the problem of selective reliability.4

1. Relevant factors and impermissible bootstrapping

We consider first the girls' initial statements to the day-care workers. In evaluating the admissibility of these remarks, the trial court made three reliability findings of the type explicitly approved by the Court in Wright: (1) neither girl had a motive to lie; (2) both girls described sexual acts in specific terms, using age-appropriate language; and (3) most of the statements were spontaneous responses to open-ended questions or to no questions at all. The court also found that the girls had a reputation for truthfulness and that the disclosures were made to adults who had relationships of trust with the girls. These latter factors do not seem inappropriate, given Wright's emphasis that trial courts have considerable leeway in making reliability determinations.

But the court relied improperly upon one additional factor. It observed that the girls "made the statements on two consecutive days without the children having discussed the matter and giving basically the same content to the statements, I think is a strong indicator of reliability." This is impermissible bootstrapping as the court referenced other evidence in finding cross-corroboration of each child's statements.5 "To be admissible under the Confrontation Clause, hearsay evidence used to convict a defendant must possess indicia of reliability by virtue of its inherent trustworthiness, not by reference to other evidence at trial." Id. at 822, 110 S. Ct. at 3150 (emphasis added).

We do not analyze separately the findings of the Washington Supreme Court as it agreed generally with the trial court's determinations. SeeSwan, 790 P.2d at 628-30. The district court conducted an independent review of the transcripts. It agreed with the state courts that (1) the statements were spontaneous; (2) neither girl had a motive to lie; (3) they did not have a tendency to lie; and (4) most significant to the district court, the statements included "a description of oral intercourse which was so specific, and showed such inappropriate and precocious knowledge, that it would be difficult to see this knowledge as anything other than the result of personal experience."

But the court also made the same bootstrapping mistake as did the state courts. It found relevant that "[t]he allegations made by the girls were consistent, both with one another's reports and with each girl's later reports of the same activities." (emphasis added).

Nevertheless, we do not conclude that the bootstrapping errors of the state courts and the district court tainted their reliability determinations. Neither the state courts nor the district court indicated that cross-corroboration was a primary reason for finding the statements reliable; indeed, the district court found most significant the knowledge of sexual acts unexpected of children of that age.

We turn next to the other reliability findings. In particular, the Swans dispute that the statements were spontaneous. They say that the day-care workers prompted the girls' initial remarks. According to her preliminary testimony, Conradi told B.A. that no one should touch her private parts and the girl responded "Mommy and Daddy do." Although this statement did not come out of the blue, it was not made in response to any question posed by Conradi. Other statements made by B.A. at that time seemed impulsive, such as when, while playing a game of peekaboo, the girl said that "My daddy puts his penis in my mouth and icky milk comes out."

Similarly, R.T.'s statements were made in response to Bratvold's open-ended questions, such as "What kinds of games do you play with Bill [Swan]?" This contrasts with the leading questions the Supreme Court criticized in Wright. See 497 U.S. at 810, 110 S.Ct. at 3144 ("Do you play with daddy? Does daddy play with you? Does daddy touch you with his pee-pee? Do you touch his pee-pee?").6 We find no clear error in the state and district courts' spontaneity determinations.

We also find no clear error in the finding that the girls had no motive to lie. The Swans do not suggest any motive. They contend instead that the day-care workers, and particularly Conradi, were biased. That, however, does not suggest that the girls themselves had a reason to lie.

Next, the Swans argue that the courts erred in finding that the girls were generally not prone to lie. They note that R.T.'s father testified that she occasionally had make-believe companions and pretended to have gone somewhere the day before when she had not. And she told the police that "Jerry," possibly a reference to her father Gerald, put "marbles" in her genital area. She said that she had related the same information to Bratvold, who denied that R.T. had made such a statement to her.

The testimony of R.T.'s father is not dispositive. Children pretend. This does not reveal a tendency to prevaricate. R.T.'s later statement about "Jerry" suggests a possible problem with lying. Whether the Swans may rely on a later statement to undermine the reliability of R.T.'s earlier statements, however, is unclear. Wright forbids using other corroborating evidence at trial to show that an initial hearsay statement is reliable. The issue here is subtly different: can arguably noncorroborating evidence be used to demonstrate the unreliability of a hearsay statement? Wright's caution against reference to other evidence at trial suggests not.7

Regardless, the district court considered this later statement but discounted it. The court reasoned that the accusation was ambiguous and that, by that time, several adults had questioned R.T. and the spontaneity surrounding her answers had faded. This determination was not clearly erroneous.

The Swans say that R.T.'s faulty performance at the competency hearing precludes a finding that her hearsay statements had sufficient guarantees of trustworthiness.8

The Supreme Court rejected a similar argument in Wright. Id. 497 U.S. at 825, 110 S. Ct. at 3151-52 ("the Confrontation Clause does not erect a per se rule barring admission of prior statements of a declarant who is unable to communicate to the jury at the time of trial"). The Swans would distinguish Wright, arguing that there the trial court found the child incapable only of communicating to the jury but here the incompetency problems were broader and more damaging.

The trial court recognized the distinction between R.T.'s ability to testify in a courtroom setting and to tell the truth at the time of the declarations. It clearly considered her incompetent only as to the former. We defer to its finding.

The Swans say that the statements were unreliable because the day-care workers, especially Conradi, were pre-disposed to find child abuse. Amicus supports this position. Amicus essentially makes a policy argument that in cases involving nontraditional hearsay exceptions, the trial court should be vigilant in assessing reliability by making a preliminary evaluation of the reporting witness' credibility. Whatever the merits of this view, a federal habeas court may not prescribe evidentiary rules for the states. We may grant relief only if there is serious constitutional error.

We do not read the Confrontation Clause as mandating a threshold assessment of the reporting witness's credibility before the admission of hearsay evidence. The Supreme Court has limited the reliability inquiry required by the Clause to whether the circumstances surrounding the making of the statement "render the declarant particularly worthy of belief." Id. 497 U.S. at 820, 110 S. Ct. at 3149 (emphasis added). Unlike the declarant, the reporting witness is subjected to cross-examination and the requirements of an oath. SeeUnited States v. Hinkson, 632 F.2d 382, 385 (4th Cir.1980) (discussing trustworthiness requirement included in the analogous residual hearsay exception, Fed.R.Evid. 803(24)). See also John E.B. Myers, Evidence in Child Abuse and Neglect, Sec. 7.45, at 249, 262, 264 (2d ed. 1992) (reporting witness' credibility not a reliability factor that surrounds the making of the statement). The jury, not the trial judge, must weigh the reporting witness's credibility.

Obviously the trial court must make some inquiries, such as deciding whether a witness is competent to testify. But the Confrontation Clause does not require the court to take basic credibility determinations from the jury.

In admitting the girls' statements to the CPS caseworker, the trial court found indicia of reliability because (1) more than one person was present when the statements were made; (2) some responses were spontaneous; (3) neither girl had a motive to lie and they had relationships of trust with the adults. It noted that leading questions had been asked of B.A. but concluded that direct questions are sometimes appropriate for difficult child witnesses and that the girl's answers were consistent with her other spontaneous statements. The district court agreed with these findings.

Although we are troubled by the use of leading questions,9 this does not necessarily render B.A.'s responses untrustworthy. SeeUnited States v. George, 960 F.2d 97, 100 n. 2 (9th Cir.1992); Myers, supra, Sec. 4.5 at 229-39. The consistency of the girl's statements with her earlier statements is the critical reliability factor here. We conclude that it provided a particularized guarantee of trustworthiness.

2. Statements to R.T.'s father and to police detective

These statements had a low level of reliability because, when allegedly made, several adults had already questioned the girls. But we agree with the district court that any error in admitting the statements was harmless because the evidence was "insignificant in impact when compared to the initial disclosures of the girls, and did not add anything new."

In admitting this hearsay evidence, the trial court found indicia of reliability because (1) B.A. had no motive to lie; (2) she had the character for truthfulness; and (3) the statements were "basically spontaneous." The district court made similar findings.

The spontaneity of the statements is questionable. They occurred some five months after B.A.'s initial disclosure and after many interviews. We are not persuaded by the district court's reasoning that attention surrounding the earlier allegations had faded. Among other things, B.A. had been seeing a therapist and doubtless was asked about the earlier events, even if indirectly.

But again the critical factor is the consistency of the statements with B.A.'s earlier allegations. Under the totality of the circumstances, we find that it provided an adequate sign of reliability.

Amicus argues that "a criminal defendant cannot be convicted on the basis of hearsay alone." Even if hearsay can pass the admissibility threshold, says Professor Nesson, it is inherently deficient, without other corroborating evidence, for reaching a conclusion of guilt beyond a reasonable doubt.

We decline to address this argument because it is raised for the first time on appeal and not by the Swans. Although they challenged properly the sufficiency of the evidence in the trial court by a motion to dismiss at the close of the evidence, they did not raise this issue in their habeas petition.

Generally, we do not consider on appeal an issue raised only by an amicus. Sanchez-Trujillo v. INS, 801 F.2d 1571, 1581 n. 9 (9th Cir.1986) (amicus may not frame the issues for appeal); Preservation Coalition, Inc. v. Pierce, 667 F.2d 851, 862 (9th Cir.1982) (same). Discretionary exceptions exist where a party attempts to raise the issue by reference to the amicus brief. SeeUnited States v. Van Winrow, 951 F.2d 1069, 1072 (9th Cir.1991); Toussaint v. McCarthy, 801 F.2d 1080, 1106 n. 27 (9th Cir.1986), cert. denied, 481 U.S. 1069, 107 S. Ct. 2462, 95 L. Ed. 2d 871 (1987). And we have reached the issue where it involves a jurisdictional question or touches upon an issue of federalism or comity that could be considered sua sponte. SeeStone v. City and County of San Francisco, 968 F.2d 850, 855-56 (9th Cir.1992) (federalism and comity), cert. denied, --- U.S. ----, 113 S. Ct. 1050, 122 L. Ed. 2d 358 (1993); Miller-Wohl Co. v. Commissioner of Labor and Industry, 694 F.2d 203, 204 (9th Cir.1982) (jurisdiction); Chadha v. INS, 634 F.2d 408, 411-12 (9th Cir.1980) (same), aff'd, 462 U.S. 919, 103 S. Ct. 2764, 77 L. Ed. 2d 317 (1983).10

The Swans did not adopt amicus' argument by reference in their brief and none of the other exceptions apply. The issue has been waived.11

The Swans contend that the state violated their due process rights when it suppressed evidence that Conradi, the day-care worker who was the key reporting witness, had been sexually abused. The district court rejected this contention without a hearing. It found no evidence that the state knew Conradi had been sexually abused, and concluded that this was not a material fact of exculpatory value.

During a pretrial deposition, defense counsel asked Conradi whether she had ever been abused. The prosecutor advised her not to answer unless relevancy could be shown. No attempt was made either to show relevancy or to require an answer to the question. The Swans argue that the prosecutor's advice demonstrates that the state knew of Conradi's past victimization or, at least, that the court should have granted an evidentiary hearing to determine whether the prosecutor knew, but did not disclose, that she had been abused.

The Swans were entitled to an evidentiary hearing only if they alleged facts that, if proved, would entitle them to relief and if they did not receive a full and fair evidentiary hearing in the state court. SeeGreyson v. Kellam, 937 F.2d 1409, 1412 (9th Cir.1991). We review for an abuse of discretion the decision to deny an evidentiary hearing. Id.

The government must disclose evidence favorable to a defendant and material to either guilt or punishment. United States v. Streit, 962 F.2d 894, 900 (9th Cir.) (citing United States v. Bagley, 473 U.S. 667, 674, 105 S. Ct. 3375, 3379, 87 L. Ed. 2d 481 (1985) and Brady v. Maryland, 373 U.S. 83, 87, 83 S. Ct. 1194, 1196-97, 10 L. Ed. 2d 215 (1963)), cert. denied, --- U.S. ----, 113 S. Ct. 431, 121 L. Ed. 2d 352 (1992). Evidence is "material" only if a reasonable probability exists that, had it been disclosed, the result would have been different. United States v. Kennedy, 890 F.2d 1056, 1058 (9th Cir.1989), cert. denied, 494 U.S. 1008, 110 S. Ct. 1308, 108 L. Ed. 2d 484 (1990). Evidence undermining the credibility of a government witness must be disclosed when the reliability of the witness may be determinative of the defendant's guilt or innocence. United States v. Brumel-Alvarez, 991 F.2d 1452, 1458 (9th Cir.1993).

The defense learned of the abuse from a post-trial interview of Conradi conducted by a defense investigator posing as a journalist. In the interview, Conradi said that she saw abuse everywhere in her neighborhood. She also claimed to have called Child Protective Services and "turned in at least 20 kids," who were abusing other children.

Although information about Conradi's abuse was not before the jury, it heard similar evidence undermining her credibility. She testified that her own children had been sexually abused and that she took a seminar to learn more about the subject, in part, because of their experience. During cross-examination, she admitted that she saw evidence of abuse "everywhere." On redirect, she clarified the statement by saying that she meant that child sexual abuse can be found in all social strata. Both defense attorneys argued in closing that she was predisposed to find sexual abuse because of her personal experience with it.

The jury knew that Conradi's children were sexually abused and heard argument that she was biased as a result. But apparently it credited her testimony. There is no reasonable likelihood that additional evidence of Conradi's own abuse would have altered the verdict.

Almost five years after the trial, Dr. Richard Soderstrom conducted a routine gynecological examination of B.A. The Swans argue that his findings, that she had a normal, intact hymen and that her introital opening (the opening into the vagina) could not previously have been 1 to 1.2 centimeters, constitutes newly discovered evidence justifying a new trial. They add that the evidence presented at trial regarding this issue was false and they were denied due process.

Newly discovered evidence is a ground for federal habeas corpus relief only when it bears on the constitutionality of an applicant's conviction, Herrera v. Collins, --- U.S. ----, ----, 113 S. Ct. 853, 860, 122 L. Ed. 2d 203 (1993), and would " 'probably produce an acquittal,' " Harris v. Vasquez, 949 F.2d 1497, 1523 (9th Cir.1990) (inner quotation omitted), cert. denied, --- U.S. ----, 112 S. Ct. 1275, 117 L. Ed. 2d 501 (1992).

We agree with the state that this new evidence is insignificant. The Swans were convicted of the statutory rape of R.T. even though physical evidence of sexual abuse as to her was nonexistent. The convictions were based upon the evidence of oral sex,12 which was substantially the same for B.A. as for R.T.

True, the jury may have convicted the Swans of raping B.A. based upon the physical evidence, but convicted them of raping R.T. because her out-of-court statements paralleled those of B.A. But even under this unlikely scenario, evidence about the intact hymen probably would not have produced an acquittal.

Nurse Practitioner Theodore Ritter said that when he examined B.A. he did not notice the presence or absence of the hymen. He said, however, that ordinarily an examiner cannot see into the vagina due to the presence of the hymen, but that he was able to see into B.A.'s vagina. He estimated that her introital opening was 1 to 1.2 centimeters, though he did not measure it. Another state witness, Dr. Carol Jenny, testified that Ritter's observations meant only that Ritter could see through the hymenal opening, and that an accurate measurement of the vaginal introitus is obtainable only with a measuring device, which Ritter did not use.

Dr. Lawrence Parris also testified for the state. He said that he examined B.A. and found no evidence of physical injury to her genitalia. He also said that he did not notice the presence or absence of a hymen.

Contrary to the Swans' contentions, the newly discovered evidence does not contradict materially the evidence presented at trial. At most, it could have been used to impeach Ritter's testimony concerning the size of the introital opening. It does not demonstrate that the state's evidence was false. It was merely equivocal.

The Swans argue that their trial attorneys' failure to discover that the state's medical evidence regarding B.A. was false constitutes ineffective assistance of counsel.

They must show deficient performance and prejudice. Strickland v. Washington, 466 U.S. 668, 687-90, 104 S. Ct. 2052, 2064-66, 80 L. Ed. 2d 674 (1984). In light of our conclusion that the new medical evidence would not with reasonable likelihood have altered the result, the Swans cannot show prejudice. Their ineffective assistance claim fails.

AFFIRMED.