Full opinion text
KATZMANN, Circuit Judge. The sexual abuse of children is heinous beyond words. It is intolerable as it is reprehensible. For that reason, justice demands that the perpetrators of such conduct be prosecuted to the fullest extent of the law, and that the penalties be appropriately severe for those whose actions are so destructive of young lives. The prosecution of child sexual abuse cases is challenging. With third-party witnesses often unavailable, these cases frequently hinge on judgments about credibility in which jurors must choose between contradictory stories proffered by the defendant and the complainants. Just as the complainants are entitled to effective advocacy, so too are those charged, especially given the consequences of conviction. Thus, we have underscored the importance of effective representation for defendants in child sexual abuse prosecutions. See generally Pavel v. Hollins, 261 F.3d 210 (2d Cir.2001); Lindstadt v. Keane, 239 F.3d 191 (2d Cir.2001). The teaching of the law in this Circuit is that defense counsel is obliged, wherever possible, to elucidate any inconsistencies in the complainant’s testimony, protect the defendant’s credibility, and attack vigorously the reliability of any physical evidence of sexual contact between the defendant and the complainant. Once again, before us now is someone convicted of child sexual abuse related crimes whose quality of trial representation causes us serious concern. Petitioner-appellant Louis Eze (“Eze”) seeks a writ of habeas corpus pursuant to 28 U.S.C. § 2254 vacating his conviction in New York Supreme Court, Erie County, of multiple counts pertaining to the alleged sexual abuse of his distantly-related nieces, twin girls by the names of Chendo and Nnedi Okongwu. Eze was convicted almost entirely based on the girls’ allegations at trial, supplemented with expert testimony substantiating their credibility and medical evidence suggesting that the girls may have been victims of sexual assault. Although defense counsel performed competently in certain respects, various apparent omissions of Eze’s counsel leave us troubled. Several obvious pieces of evidence that would have cast doubt upon Eze’s guilt were inexplicably neglected at trial, even though defense counsel was aware of them and their admissions apparently would not have interfered with the defense’s overall strategy. Further, virtually all the testimony that linked Eze to the crime related to the omissions to which we refer. Eze, of course, faces the heavy burden of showing ineffective assistance set forth in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), enhanced by the added hurdle posed by the highly deferential review accorded state court adjudications under the Anti-terrorism and Effective Death Penalty Act of 1996 (“AEDPA”), Pub.L. No. 104-132, 110 Stat. 1214 (1996) (codified at 28 U.S.C. § 2254(d)(1) (2000)). Notwithstanding these obstacles, if certain omissions cannot be explained convincingly as resulting from a sound trial strategy, but instead arose from oversight, carelessness, ineptitude, or laziness, we would find the quality of representation sufficiently deficient to grant the writ. With the record before us, however, we lack the benefit of an explanation of Eze’s trial counsel’s reasoning and cannot conclude now that no plausible trial strategy justified her actions. See Sparman v. Edwards, 154 F.3d 51, 52 (2d Cir. 1998) (stating that “a district court facing the question of constitutional ineffectiveness of counsel should, except in highly unusual circumstances, offer the assertedly ineffective attorney an opportunity to be heard and to present evidence, in the form of live testimony, affidavits, or briefs”); see also Bloomer v. United States, 162 F.3d 187, 194 (2d Cir.1998) (same). We therefore vacate the District Court’s denial of a writ of habeas corpus and remand for an evidentiary hearing at which Eze’s trial counsel be allowed to explain her trial strategy. The District Court then should determine whether such strategy can justify those omissions. BACKGROUND A. Eze’s State Trial Dominic Okongwu (“Okongwu”), Eze, and Joy Wosu (“Wosu”) were indicted by an Erie County Grand Jury on 68 counts related to the alleged sexual abuse of Chendo and Nnedi Okongwu. The three defendants were tried together in a seven day trial that commenced on November 12, 1993. Except as otherwise noted, the evidence adduced at trial established the following. Okongwu and his wife are Nigerian citizens with twin daughters, Chendo and Nnedi, who were born on September 12, 1984. At some point after Okongwu’s wife became ill and returned to Nigeria, Chen-do and Nnedi were placed under the foster care of Ms. Ollie McNair (“McNair”). In December 1990, a Family Court judge appointed Eze to supervise Okongwu’s weekend visitation of his daughters. The court order required Eze to be in the girls’ presence during the entirety of their visits with Okongwu and never to leave the girls unsupervised. The visits continued regularly every weekend until December 1991, at which time all contact between Okong-wu and his daughters was terminated. On December 12, 1991, McNair entered the girls’ bedroom and found Nnedi lying on top of Chendo in a manner imitative of sexual contact. When McNair asked the girls what they were doing, Nnedi responded that “she was doing it because that’s the same thing their father does to her when she goes to his house.” McNair notified Child Protective Services, which commenced an investigation, and McNair received explicit instructions to keep Ok-ongwu away from the children. At this point, the girls made no mention to McNair of Eze’s involvement in any sexual abuse. At trial, the girls testified to three instances of sexual abuse, with each incident bearing striking similarities. The first occasion occurred in June 1991, following a community celebration known as the June-teenth Festival. The girls testified that Eze picked them up at their foster home, and brought them to the festival with Ok-ongwu and Wosu. On direct examination, both girls recalled that after the festival they returned to Okongwu’s house; Eze and Wosu then left, at which point Okong-wu, now alone with the girls, sexually abused them in his basement. On cross examination, however, Chendo testified that Eze and Wosu remained in Okongwu’s house during the sexual abuse and that they came downstairs at one point. According to Chendo, Okongwu instructed them to shower and go downstairs to the basement and wait for him. The girls testified that, upon joining them, Okongwu directed Chendo to lie down on two mattresses that he had placed together, while he bound Nnedi by tying her arms and waist to a chair and taping her mouth shut. Both girls stated that, with Nnedi restrained, Okongwu inserted his penis into Chendo’s vagina and forced Chendo to perform fellatio on him. The girls testified that Okongwu then had the girls switch roles. They said that Okongwu untied Nnedi and untaped her mouth, and bound Chendo to the chair in the same manner he had restrained her sister. The girls stated that Okongwu at this point engaged in oral sex and then intercourse with Nnedi. Afterwards, Okongwu ihstructed them, “you better not tell or I’ll kill you,” while brandishing a knife, belt, and scissors. According to the trial testimony, Okongwu then told both girls, who were now bleeding, to take a shower. Eze later brought the girls back to McNair’s home. Chendo said she did not tell McNair about the abuse because she was scared that her father would kill her if she did. The second incident about which the girls testified occurred on or about September 12, 1991, the date of their seventh birthday. According to the girls, Eze picked them up from McNair’s house and, after stopping off at Okongwu’s house, they attended a birthday party with Ok-ongwu- and Wosu at Okongwu’s friend’s house. Similar to the June 1991 incident, Okongwu, Eze, Wosu, and the girls went to Okongwu’s house following the party. Upon arrival, Okongwu told the girls to get undressed and go downstairs to the basement. Chendo testified that Eze remained upstairs talking on the telephone and saw her walking downstairs naked. Nnedi, however, testified that Eze and Wosu left the house, leaving them alone with their father. The girls stated that Okongwu soon came downstairs, naked, and what followed mirrored the June 1991 incident. Okongwu again arranged the mattresses next to each other and engaged in coerced sexual acts with his daughters. Chendo testified that he first told her to lie down, while he tied Nnedi to a chair and taped her mouth. Okongwu proceeded to have sexual intercourse and oral sex with Chendo, ignoring her pleas for him to stop. Chendo testified that she then switched roles with her sister, as Nnedi was forced to have sexual intercourse and oral sex with Okongwu, while Chendo was bound. Chendo recalled that her father again threatened them, stating, “you better not tell or I’ll kill you.” Nnedi testified to similar events, but according to Nnedi, her father first had sexual intercourse and oral sex with her, while her sister was bound, and then the girls switched roles. Nnedi also testified that her father again told her that if she were to tell anyone what happened, he would kill her. The testimony indicated that the girls again showered, at Okongwu’s instruction, and then had a birthday cake with Eze and Wosu. Eze returned the girls to McNair’s home that evening. Nnedi explained that she did not tell McNair about what happened because she was scared of her father. The final incident of abuse adduced at trial occurred in November 1991 during Thanksgiving. This time Okongwu joined Eze to pick up the girls, and they brought the girls for Thanksgiving dinner to Ok-ongwu’s house, where Wosu joined them. This occasion, however, bears a critical distinction from the previous incidents in that both girls testified that Eze and Wosu participated in the actual sexual abuse. The girls stated that Okongwu instructed them to undress and wait for him in the basement. Chendo said that Eze, Wosu, and Okongwu, who were naked, joined the girls in the basement. Chendo testified that, as Nnedi was tied up with her mouth taped, Eze had sexual intercourse with her as the others watched. Wosu then “tried to stick her private into our private very deep” and put her hands into Chendo’s vagina. Finally, Okongwu had sexual intercourse with Chendo, while Wosu cheered him on, encouraging him to “go deeper.” Once again, the girls switched roles. Chendo stated that she was bound with her mouth taped, and the three now abused Nnedi. Chendo recalled that Ok-ongwu first had sexual intercourse with Nnedi and forced her to perform oral sex. Wosu then put “[h]er vagina in [Nnedi’s] private” and then placed her hand in her vagina. Chendo testified that Eze followed by engaging in sexual intercourse and oral sex with Nnedi. Nnedi’s recollection of the events varied slightly. Nnedi recalled that her father first forced her to have sexual intercourse and oral sex with him. Nnedi testified that Eze also had sexual intercourse with her and Wosu used her hand to touch Nnedi’s vagina. Nnedi said that she was screaming as this transpired, and that Eze told her that if she revealed what happened, they would kill her. Nnedi testified that her sister was then abused by the three defendants. First, she said that Okongwu had sexual intercourse and oral sex with Chendo. Nnedi testified that Eze next had sexual intercourse and oral sex with Chendo, and then Wosu put her hand in Chendo’s vagina. Both girls indicated that following this incident they were bleeding and went upstairs to wash up. The prosecution also offered evidence that Okongwu threatened the girls after they came forward with allegations of abuse. Chendo testified that, in October 1992, well after they had informed McNair of the abuse and the investigation had commenced, Okongwu would come to their bus stop and threaten them, saying “you better not tell or I’ll kill you.” According to Chendo, these threats occurred at least three or four times. After the girls informed McNair of these threats, McNair contacted the Board of Education and requested their bus stop be switched. The prosecution’s only physical evidence of sexual abuse came from Dr. Stephen Lazoritz. Dr. Lazoritz examined both girls on January 6, 1992 and found evidence of sexual abuse. Dr. Lazoritz’s conclusions relied heavily on the girls’ abnormal and attenuated hymens. Dr. Lazoritz explained that, in children their age, the hymenal opening should be no more than seven millimeters. Both Chendo and Nnedi, however, had hymenal openings that measure nine millimeters by five millimeters. With regard to Chendo, Dr. La-zoritz found that she had less hymen tissue than would be expected, demonstrating that some hymen membrane had eroded away. Dr. Lazoritz also found scar tissue on Chendo’s hymen at the six o’clock position, which suggests injury from penetration or attempted penetration, as well as a band of scar tissue at the seven o’clock position. Dr. Lazoritz concluded “beyond a reasonable degree of medical certainty that [Chendo] was sexually abused. In fact, based on these findings, I can diagnose that sexual abuse had occurred [independent of an allegation of such abuse].” Dr. Lazoritz was less certain that Nnedi had been sexually abused and concluded that “[i]f Nnedi made a statement that she was sexually abused, I would say, with a reasonable degree of medical certainty, that these findings were consistent with that abuse.” On cross examination, Dr. Lazoritz agreed that his findings could have been caused by trauma other than sexual abuse and that the girls’ throat cultures revealed nothing abnormal. In addition, Dr. Lazo-ritz acknowledged that, when he examined the children in January 1992, Child Protective Services had informed him that there was an allegation of sexual abuse. The most critical point elicited during Dr. La-zoritz’s cross examination was that he had examined Chendo in '1988, at which point he made findings regarding her attenuated hymen and scar tissue similar to those he made in January 1992. This line of questioning raised the serious possibility that Chendo’s abnormally large hymenal opening in 1992 existed prior to the alleged abuse in 1991. Because of the similarities between his 1992 findings and his 1988 findings, Dr. Lazoritz could not state with any degree of certainty that Chendo had been sexually abused at any time after 1988. Although Eze’s counsel also attempted to question Dr. Lazoritz on Nnedi’s medical records from 1988, the trial judge refused to allow counsel to elicit this testimony, sustaining the prosecution’s objection that Eze’s counsel was “doing this through the wrong witness.” The defense never revisited the issue of Nnedi’s 1988 medical records. Jan Henry, the Director of Program Services for Child and Adolescent Treatment Services, testified for the prosecution as an expert witness on the psychology of child sexual abuse. Henry’s testimony provided a backdrop against which the jury could assess the children’s testimony and account for changes in their stories over time. Prior to trial, the defense expressed concern regarding Henry’s expert testimony because she had been involved in a 1988 Family Court proceeding with the Okongwu family and had interviewed the girls after the 1991 allegations of abuse. The prosecutor represented to the trial court that Henry’s testimony would be limited to explaining child sexual abuse syndrome “without in any way assessing the believability of Nnedi and Chendo Ok-ongwu in this trial.” The court agreed, limiting Henry’s testimony to the behavioral characteristics of children that have been sexually abused. The jury therefore did not learn about Henry’s interviews with the girls, including the girls’ statement to Henry that they wanted to see Eze so they could ask Eze whether he knew “what our father did to us.” Henry discussed child sexual abuse syndrome, a theory that describes a series of phases experienced by a sexually abused child, including a child’s behavior during abuse and after the abuse has ended. Henry explained that as the abuse continues to progress, a child begins to realize that something is wrong, and therefore the abuser often takes measures to insure that the victim will not tell anyone, often through threats, bribes, or brute force. A victimized child, Henry stated, may seem happy on the outside as a defense mechanism, and often seems to allow abuse to continue because of his or her powerless position. Consequently, according to Henry, a child will keep the abuse private until, for some reason, the secret is broken. Henry testified that a child may disclose accidentally because, for example, the child may “start doing things with other kids, repeating the behavior that they have experienced, and someone sees it and says how is it that you learned this, why are you doing this, and then they’ll disclose what’s happened to them.” After this initial disclosure, however, the child may again suppress the abuse upon realizing that undesirable consequences arise following disclosure, such as being forced to retell the story many times or causing family disharmony. Henry also explained the reasons why a child may provide inconsistent, yet reliable, stories of the abuse. If a child’s story is too consistent, it may have been memorized or programmed. In addition, a young child may initially only disclose part of the abuse as a test to gauge the reaction, and if the partial story receives a positive reaction, the child may then disclose more. Henry also discussed her assessment program for determining the veracity of a child’s allegation of abuse. She testified, without objection from the defense, that out of approximately fifty-to-one hundred cases, she has encountered only one instance of a false allegation. On cross examination, Henry acknowledged that a child’s mental process could change the facts surrounding the abuse and add people who were not actually there. She recounted a case in which the child stated that his or her parents were present during the abuse; it was later determined, however, that the child’s parents had not in fact been present, but instead the child wished that they had been. Henry agreed that it is possible that a child’s story could be influenced by adults and that, although therapists have methods to control an adult’s input, “you can only surmise, you don’t know for sure.” Henry also acknowledged that her evaluation relies on medical examinations and that the accuracy of her evaluation would be compromised if the medical information proved to be flawed. She admitted as well that the child sexual abuse syndrome study on which she relied during direct examination has been the subject of criticism because it was based on a small sample size. Finally, Henry was cross-examined on her statement that only one out of the fifty-to-one hundred cases she has worked on involved a false allegation. In the other cases, there were no judicial rulings establishing that the children told the truth. The defense case consisted of the testimony of Wosu, Eze, Okongwu, Andrew Greenlee (Wosu’s fiancé), and character witnesses. Wosu, an acquaintance of Eze’s, testified that she met Okongwu, Chendo, and Nnedi in 1991 at Nigerian Association meetings. Wosu denied being with the children during the alleged incidents that occurred in June 1991 and November 1991. She explained that she attended the June-teenth Festival in 1991, but did not go with Okongwu, Eze, Chendo, and Nnedi. Wosu also stated that she did not go to Okong-wu’s house for Thanksgiving, but instead had dinner with her fiancé that evening. Wosu testified that she was with the children on their seventh birthday in September 1991, but they only went to a restaurant to celebrate. Wosu also stated that she had only been to Okongwu’s house on one occasion, October 26,1991. Eze testified to his supervision of the children during Okongwu’s visitation. Eze said that, pursuant to the Family Court order, he never left Okongwu alone with the children during his supervision. Eze recalled that he and Okongwu took the girls to the Juneteenth Festival in June 1991. Although Wosu did not come with them to the festival, Eze testified that she also may have been there. Regarding the September 1991 incident, Eze testified that there was not a party for the girls’ birthday; instead he, Okongwu, and Wosu took the girls to a restaurant. After eating, he returned the girls to McNair’s. For Thanksgiving in 1991, Eze brought the girls to Okongwu’s house for dinner, but Wosu was not present. Eze testified that the girls remained under his surveillance throughout the day, and did not think that Okongwu ever went down to the basement. On cross examination, the prosecution brought out inconsistencies in Eze’s statements following his arrest regarding Wosu’s presence with the children. After being arrested, Eze supposedly stated that Wosu was with the girls several times, but on direct Eze testified that Wosu was with them only once. The prosecution also stressed the long and late hours that Eze worked in 1991, implying that he needed to sleep during the day on the weekends when he was charged with supervising Ok-ongwu’s visitation. Eze, however, maintained that he could adequately supervise the children on three hours of sleep. Finally, the prosecution impeached Eze’s credibility by suggesting that Eze had lied on his application to become a naturalized citizen in 1990 when he stated that he had lived in the same residence with his wife for three years. The prosecution referred to an affidavit from his wife, filed with the Immigration and Naturalization Service (“INS”), swearing that Eze moved out in December 1987 and that they had not lived together since. Eze responded that his wife’s statement was not true, and that although he had moved out of the house after they had a fight, he moved back for a while. Okongwu substantiated Wosu and Eze’s testimony. Okongwu explained that in 1991, he would see his daughters by having Eze pick them up from McNair’s home. Okongwu testified that he and Eze brought the girls to the Juneteenth Festival in 1991. Wosu did not come with them, although Okongwu recalled seeing her at the festival. Okongwu explained that they took the girls back to the foster home immediately after the festival. As to the September 1991 incident, Okongwu testified that Eze picked up his daughters from the foster home and brought the girls to his house. Afterwards, Okongwu and Eze picked up Wosu, and they all went to a restaurant. Like Eze, Okongwu denied attending the party the girls mentioned and ever being in his basement alone with the girls. Okongwu also explained that he did not know the location of the girls’ bus stop and that he worked the day shift, suggesting that it would have been impossible for him to threaten the girls while they were waiting for their bus. On cross examination, Okongwu said that he had no idea why his daughters would hold a grudge against him and fabricate their allegations. In addition, the prosecution brought out that his apartment was not searched until two months after he learned of the allegations, suggesting that he had the opportunity to remove the box containing the knife, belt, rope, and tape. The remaining witnesses sought to enhance the credibility of the defendants. Greenlee, Wosu’s fiancé, testified the he was visiting Wosu on Thanksgiving 2001 and produced an airplane ticket for his trip. Greenlee offered an alibi for Wosu, claiming they were eating dinner that evening when the alleged abuse occurred. On cross examination, the prosecution stressed Greenlee’s bias in testifying on behalf of his fiancé and questioned his lack of a dinner receipt. Moreover, although Greenlee found out about the charges against Wosu in December 1992, he did not contact the police regarding her alibi and first came forward at trial. In addition, two character witnesses testified to Okongwu and Eze’s good reputation in the Nigerian community. On November 26, 1993, the jury found the defendants guilty on all remaining charges. Eze was convicted of eight counts of rape in the first degree, N.Y. Penal L. §§ 20.00, 130.35[1], [3], eight counts of sodomy in the first degree, id. §§ 20.00, 130.50[1], [3], four counts of incest, id. §§ 20.00, 255.25, six counts of sexual abuse in the first degree, id. '§§ 20.00, 130.65[1], [3], and four counts of endangering the welfare of a child, id. §§ 20.00, 260.10[1]. Eze was sentenced to an aggregate prison term with a minimum of 35% years and a maximum of 107 years. B. State Appeal Eze appealed his conviction on numerous grounds. On July 14, 1995, the Appellate Division, Fourth Department, affirmed Eze’s conviction and concluded, as relevant for our purposes, that “[t]he record does not support the contentions of the defendant that ... he was denied effective assistance of counsel.” People v. Eze, 217 A.D.2d 987, 631 N.Y.S.2d 268, 268 (4th Dep’t 1995). The Appellate Division did not discuss the merits of Eze’s ineffective assistance claim and cited only the New York Court of Appeals decision, People v. Baldi, 54 N.Y.2d 137, 444 N.Y.S.2d 893, 429 N.E.2d 400 (1981). Eze, 631 N.Y.S.2d at 268. The New York Court of Appeals denied Eze leave to appeal on September 26, 1995. People v. Eze, 86 N.Y.2d 841, 634 N.Y.S.2d 451, 658 N.E.2d 229 (1995) (Table). In June 1996, Eze moved in Erie County Court to vacate his conviction, or in the alternative to set aside his sentence, again raising a claim of ineffective assistance. Because the claim already had been decided-on the merits by the Appellate Division on direct appeal, the court was required to deny the motion under New York criminal procedure law. See N.Y.Crim. Proc. Law § 440.10[2] (McKinney 2002) (“[T]he court must deny a motion to vacate a judgment when: (a) The ground or issue raised upon the motion was previously determined on the merits upon an appeal from judgment, unless since the time of such appellate determination there has been a retroactively effective change in the law controlling such issue .... ”). The Appellate Division, Fourth Department, denied Eze leave to appeal on January 6, 1997. Eze also applied to the Appellate Division, Fourth Department, for a writ of error coram nobis, arguing ineffective assistance of appellate counsel. The Appellate Division rejected this application on September 30,1997. C. Federal Habeas Proceedings In April 1997, Eze filed a petition for a writ of habeas corpus in the Western District of New York (William M. Skretny, Judge) claiming ineffective assistance of counsel. In his memorandum of law supporting his petition, Eze detailed approximately twenty allegations of supposedly deficient trial counsel performance. The District Court denied Eze’s petition on April 24, 1999. The court determined that many of Eze’s allegations of ineffective assistance were “little more than ‘second guessing’ of counsel’s strategic decisions, and/or criticism of aspects of counsel’s performance that clearly had no impact on the outcome.” As to the remainder of the allegations, the court found that only four merited discussion: 1) the decision not to object to Henry’s expert testimony and the lack of an effective cross examination of Henry; 2) the failure to question Dr. Lazo-ritz on Nnedi’s 1988 medical examination and failure to call the physicians who performed that examination; 3) the decision not to move to dismiss a juror who attended the same church as the children; and 4) the failure to object to questions regarding Eze’s INS application and to the jury charge on prior bad acts. The District Court concluded that these allegations either lacked merit or were unlikely to have had any impact on the verdict. The District Court also rejected Eze’s claim that he was denied effective assistance of appellate counsel. On July 9, 2001, we granted a certificate of appealability on (1) whether there was a state court adjudication on the merits of Eze’s federal ineffective assistance claim so as to trigger 28 U.S.C. § 2254(d)(l)’s standard of review, and (2) whether, under the applicable standard of review, Eze is entitled to habeas relief on his claim of ineffective assistance of trial counsel, with specific regard to trial counsel’s failure to challenge the medical evidence consistent with Lindstadt v. Keane, 239 F.3d 191, 201 (2d Cir.2001). On July 9, 2001, we issued a supplemental certificate of appealability on whether the respondent has waived § 2254(d)(1)’s standard of review. Discussion We review de novo the District Court’s denial of Eze’s petition for a writ of habeas corpus. See Ryan v. Miller, 303 F.3d 231, 245 (2d Cir.2002). I. The Standard of Review of the State Court Decision AEDPA changed the landscape of federal habeas corpus review by “significantly curtailing] the power of federal courts to grant the habeas petitions of state prisoners.” Lainfiesta v. Artuz, 253 F.3d 151, 155 (2d Cir.2001), cert. denied, 535 U.S. 1019, 122 S.Ct. 1611, 152 L.Ed.2d 625 (2002); see also Bell v. Cone, 535 U.S. 685, 122 S.Ct. 1843, 1849, 152 L.Ed.2d 914 (2002) (noting that AEDPA “modified a federal habeas court’s role in reviewing state prisoner applications in order to prevent federal habeas ‘retrials’ and to ensure that state-court convictions are given effect to the extent possible under law”). Under AEDPA, (d) An application for a writ of habeas corpus on behalf of a person in custody pursuant to the judgment of a State court shall not be granted with respect to any claim that was adjudicated on the merits in State court proceedings unless the adjudication of the claim- (1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States; or (2) resulted in a decision that 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 (2000). In an apparent attempt to elude the heavy burden AED-PA presents, Eze makes two arguments for why § 2254(d)(l)’s deferential standards should not apply. First, Eze contends that the respondent waived the standards by failing to raise them in the District Court. Second, Eze argues that the Appellate Division merely adjudicated his ineffective assistance claim under state law, and therefore did not adjudicate his federal claim on the merits so as to trigger § 2254(d)(1). A. Whether § 2254(d)(l)’s Standard of Review Was Waived Eze maintains that the respondent waived the statute’s standard of review by failing to reference the standard in its papers before the District Court. Eze therefore urges us to apply the pre-AED-PA de novo review of a state court adjudication. See Pavel v. Hollins, 261 F.3d 210, 215 (2d Cir.2001) (“Under [pre-AEDPA law], pure questions of law are reviewed de novo, as are mixed questions of law and fact; state court factual findings are presumed correct absent certain circumstances.”) (internal citations, quotation marks, and alterations omitted). The gravamen of Eze’s waiver argument is that procedural defenses can be waived if not raised by the defendant. AEDPA’s standard of review, however, is not a procedural defense, but a standard of general applicability for all petitions filed by state prisoners after the statute’s effective date presenting claims that have been adjudicated on the merits by a state court. See Brown v. Artuz, 283 F.3d 492, 498 n. 2 (2d Cir.2002) (“AED-PA’s standards for reviewing state court findings and conclusions apply to any petition filed ... after April 24, 1996, the AEDPA’s effective date.”); Mancuso v. Herbert, 166 F.3d 97, 101 (2d Cir.1999) (stating that “AEDPA is applicable to all petitions filed after its effective date”), cert. denied, 527 U.S. 1026, 119 S.Ct. 2376, 144 L.Ed.2d 779 (1999). The statute contains unequivocally mandatory language. See 28 U.S.C. § 2254(d) (instructing that a state prisoner’s petition for a writ of habeas corpus “shall not be granted with respect to any claim that was adjudicated on the merits in State court proceedings unless the adjudication of the claim — (1) resulted in a decision that was contrary to, or involved an unreasonable application of clearly established Federal law, as determined by the Supreme Court of the United States; or (2) resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding”) (emphasis added). Therefore, if the Appellate Division adjudicated Eze’s federal ineffective assistance claim on the merits, we must apply AEDPA deference. B. Adjudication of Eze’s Federal Claim on the Merits Eze argues that the Appellate Division failed to adjudicate his claim on the merits because its decision cited the New York test for ineffective assistance from People v. Baldi, 54 N.Y.2d 137, 444 N.Y.S.2d 893, 429 N.E.2d 400 (1981), but failed to mention the federal test from Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). See People v. Eze, 217 A.D.2d 987, 631 N.Y.S.2d 268, 268 (4th Dep’t 1995). According to Eze, because Baldi and Strickland set forth different tests for ineffective assistance, see infra U.A., the Appellate Division did not adjudicate his federal ineffective assistance claim. AEDPA’s deferential standard is limited to claims that were “adjudicated on the merits in State court proceedings.” 28 U.S.C. § 2254(d). If a state court has failed to adjudicate a claim on the merits, “we apply the pre-AEDPA standards, and review de novo the state court disposition of the petitioner’s federal constitutional claims.” Aparicio v. Artuz, 269 F.3d 78, 93 (2d Cir.2001) (citing Washington v. Schriver, 255 F.3d 45, 55 (2d Cir.2001)). To adjudicate a claim on the merits, the state court “need not mention the argument raised or cite relevant case law,” Brown, 283 F.3d at 498, or even “explain[ ] its reasoning process,” Sellan v. Kuhlman, 261 F.3d 303, 311 (2d Cir.2001). Rather, a state court adjudicates a claim on its merits by “(1) disposing] of the claim ‘on the merits,’ and (2) reducing] its disposition to judgment.” Id. at 312. Whether a claim has been disposed on its merits turns on: “ ‘(1) what the state courts have done in similar cases; (2) whether the history of the case suggests that the state court was aware of any ground for not adjudicating the case on the merits; and (3) whether the state court’s opinion suggests reliance upon procedural grounds rather than a determination on the merits.’ ” Id. at 314 (quoting Mercadel v. Cain, 179 F.3d 271, 274 (5th Cir.1999)). Since Sellan, “we have given a broad reading to state court dispositions.” Norde v. Keane, 294 F.3d 401, 410 (2d Cir.2002). We have explained that the “state court need only dispose of the petitioner’s federal claim on substantive grounds, and reduce that disposition to judgment. No further articulation of its rationale or elucidation of its reasoning process is required.” Aparicio, 269 F.3d at 93-94. In fact, an issue may be considered to be adjudicated on its merits “even when the state court does not specifically mention the claim but uses general language referable to the merits.” Norde, 294 F.3d at 410. Eze alerted the Appellate Division to the federal nature of his claim by citing Strickland in his pro se supplemental brief. Therefore, the Appellate Division was apprised of Eze’s federal claim and the governing federal law when it rejected Eze’s ineffective assistance claim. The Appellate Division held that “[t]he record does not support the contentions of the defendant that ... he was denied effective assistance of counsel.” People v. Eze, 631 N.Y.S.2d at 268. This language demonstrates that the Appellate Division disposed of the claim on substantive grounds. See Aparicio, 269 F.3d at 94 (concluding that the Appellate Division adjudicated the petitioner’s claims on the merits because “there is nothing in its decision to indicate that the claims were decided on anything but substantive grounds”). Therefore, under Sellan and its progeny, the Appellate Division adjudicated Eze’s federal ineffective assistance claim on the merits so as to trigger § 2254(d)(1)’s review. II. The Merits of Eze’s Habeas Petition A. Whether the Appellate Division’s Decision Was Contrary to Clearly Established Federal Law Under AEDPA, a federal court shall issue a writ of habeas corpus if the state court adjudication “resulted in a decision that was contrary to ... clearly established Federal law, as determined by the Supreme Court of the United States.” 28 U.S.C. § 2254(d)(1). Eze argues that New York’s standard for ineffective assistance of counsel, which the Appellate Division applied on direct appeal, is “contrary to” the federal framework set forth in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). Eze correctly identifies Strickland as establishing clear federal law on ineffective assistance of counsel, as required under AED-PA. See Williams v. Taylor, 529 U.S. 362, 390-91, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (“It is past question that the rule set forth in Strickland qualifies as ‘clearly established Federal law, as determined by the Supreme Court of the United States.’”). We disagree, however, that New York’s standard is “contrary to” Strickland. The New York standard for ineffective assistance was announced in People v. Baldi, 54 N.Y.2d 137, 444 N.Y.S.2d 893, 429 N.E.2d 400 (1981). Under Baldi, “[s]o long as the evidence, the law, and the circumstances of a particular case, viewed in totality and as of the time of the representation, reveal that the attorney provided meaningful representation, the constitutional requirement will have been met.” Id. at 147, 444 N.Y.S.2d 893, 429 N.E.2d 400. The core of the Baldi standard is “whether [the] defendant received ‘meaningful representation.’ ” People v. Benevento, 91 N.Y.2d 708, 674 N.Y.S.2d 629, 697 N.E.2d 584, 587 (1998). The New York Court of Appeals has explained that Baldi’s “meaningful representation” component includes a prejudice inquiry, “which focuses on the ‘fairness of the process as a whole rather than [any] particular impact on the outcome of the case.’ ” People v. Henry, 95 N.Y.2d 563, 721 N.Y.S.2d 577, 744 N.E.2d 112, 114 (2000) (quoting Benevento, 674 N.Y.S.2d 629, 697 N.E.2d at 588). To prevail under Strickland, a defendant must show that counsel’s representation “fell below an objective standard of reasonableness” based on “prevailing professional norms” and that “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” 466 U.S. at 688, 694, 104 S.Ct. 2052. The Strickland Court defined a “reasonable probability” as “a probability sufficient to undermine confidence in the outcome.” Id. at 694, 104 S.Ct. 2052. To prevail, a defendant must establish both of Strickland’s prongs because, otherwise, “it could not be said that the sentence or conviction ‘resulted from a breakdown in the adversary process that rendered the result of the process unreliable,’ and the sentence or conviction should stand.” Cone, 122 S.Ct. at 1850 (quoting Strickland, 466 U.S. at 687, 104 S.Ct. 2052). New York courts have acknowledged that the Baldi standard for ineffective assistance is “somewhat different” from the Strickland test. People v. Claudio, 83 N.Y.2d 76, 607 N.Y.S.2d 912, 629 N.E.2d 384, 385-86 (1993); see Henry, 721 N.Y.S.2d 577, 744 N.E.2d at 114 (“This Court has previously recognized the differences between the Federal and State tests for ineffectiveness .... ”). Whereas both tests contain a prejudice component, the touchstone of the New York test is “the fairness of the process as a whole,” Benevento, 674 N.Y.S.2d 629, 697 N.E.2d at 588, while the federal test considers the outcome of the proceeding for the defendant, Strickland, 466 U.S. at 694, 104 S.Ct. 2052. With these two standards and their modest differences in mind, the determinative question becomes whether the Appellate Division rendered “a decision that was contrary to” the federal standard established in Strickland. 28 U.S.C. § 2254(d)(1). The Supreme Court has interpreted § 2254(d)(l)’s “contrary to” clause as permitting a federal court to “grant the writ if the state court arrives at a conclusion opposite to that reached by [the Supreme] Court on a question of law or if the state court decides a case differently than [the Supreme] Court has on a set of materially indistinguishable facts.” Williams, 529 U.S. at 412-13, 120 S.Ct. 1495; accord Cone, 122 S.Ct. at 1850; see Lainfiesta, 253 F.3d at 155. In fact, we need not look further than two of our recent decisions that addressed this precise issue. In Lindstadt, the Appellate Division, in denying the petitioner’s appeal, similarly “d[id] not reference Strickland; instead, it relie[d] on a standard articulated in People v. Baldi.” 239 F.3d at 198. We held in Lindstadt that “[t]he standard applied by the state court [i.e., the Baldi standard] is not ‘diametrically different, opposite in character or nature, or mutually opposed’ to the standard articulated in Strickland.” Id. (quoting Williams, 529 U.S. at 405, 120 S.Ct. 1495). We reached the identical conclusion in Loliscio v. Goord, 263 F.3d 178 (2d Cir.2001). In Loliscio, after the state court applied the same New York standard for ineffective assistance, we determined that the standard was not “contrary to” Strickland under § 2254(d)(1). Id. at 192-93 (citing Lindstadt, 239 F.3d at 198). Eze does not, and cannot, distinguish Lindstadt and Loliscio, but instead seems to speculate that those panels never seriously considered the question because there is no indication that the parties in those cases engaged in any adversarial testing of the issue. The fact remains, however, that both Lindstadt and Loliscio, which we are bound to follow, held that the Baldi test is not contrary to the Strickland test for purposes of § 2254(d)(1). See United States v. Santiago, 268 F.3d 151, 154 (2d Cir.2001) (instructing that we are compelled to follow the decisions of earlier panels unless they have “been called into question by an intervening Supreme Court decision or by one of this Court sitting in banc ”), cert. denied, 535 U.S. 1070, 122 S.Ct. 1946, 152 L.Ed.2d 849 (2002); accord In re Sokolowski, 205 F.3d 532, 534-35 (2d Cir.2000) (per curiam). B. Whether the Appellate Division Unreasonably Applied Clearly Established Federal Law Having concluded that Eze cannot prevail under AEDPA’s “contrary to” clause, we now turn to whether the Appellate Division decision “involved an unreasonable application ... of clearly established Federal law, as determined by the Supreme Court” in Strickland. 28 U.S.C. § 2254(d)(1); see Williams, 529 U.S. at 390, 120 S.Ct. 1495 (stating that federal habeas petitions alleging ineffective assistance are governed by Strickland). The Sixth Amendment guarantees persons charged with crimes the “Assistance of Counsel.” U.S. Const, amend. VI. “It has long been recognized that the right to counsel is the right to the effective assistance of counsel.” McMann v. Richardson, 397 U.S. 759, 771 n. 14, 90 S.Ct. 1441, 25 L.Ed.2d 763 (1970) (citations omitted). As discussed above, establishing ineffective assistance under Strickland entails showing that counsel’s performance “fell below an objective standard of reasonableness,” Strickland, 466 U.S. at 688, 104 S.Ct. 2052, and showing a “reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different,” id. at 694, 104 S.Ct. 2052. “Unless a defendant makes both showings, it cannot be said that the conviction ... resulted from a breakdown in the adversary process that renders the result unreliable.” Id. at 687, 104 S.Ct. 2052. Once both prongs are met, there is “sufficient indication that counsel’s assistance was defective enough to undermine confidence in a proceeding’s result.” Cone, 122 S.Ct. at 1850. AEDPA adds another hurdle in addition to the already heavy burden Eze faces under Strickland. See Sellan, 261 F.3d at 315. To prevail, Eze “must do more than show that he would have satisfied Strickland’s test if his claim were being analyzed in the first instance, because under § 2254(d)(1), it is not enough to convince a federal habeas court that, in its independent judgment, the state-court decision applied Strickland incorrectly.” Cone, 122 S.Ct. at 1852. Rather, Eze must show that the Appellate Division “applied Strickland to the facts of his case in an objectively unreasonable manner.” Id.; see Sellan, 261 F.3d at 315. We thus again turn to Williams, where the Supreme Court explained what it means for a state court to “unreasonably] appl[y] ... clearly established Federal law, as determined by the Supreme Court,” 28 U.S.C. § 2254(d)(1). Justice O’Connor, writing for the majority on this issue, stressed that “an unreasonable application of federal law is different from an incorrect application of federal law.” Williams, 529 U.S. at 410, 120 S.Ct. 1495. To determine whether federal law was unreasonably applied, “a federal habeas court ... should ask whether the state court’s application of clearly established federal law was objectively unreasonable.” Id. at 409, 120 S.Ct. 1495; see Lainfiesta, 253 F.3d at 155. Under this objective assessment of unreasonableness, a state court unreasonably applies established federal law “if the state court identifies the correct governing legal principle from [the Supreme] Court’s decisions but unreasonably applies that principle to the facts of the prisoner’s case.” Williams, 529 U.S. at 413, 120 S.Ct. 1495; see Lainfiesta, 253 F.3d at 155. For a state court to unreasonably apply clearly established federal law under AEDPA, it is not enough that a federal court “concludes in its independent judgment that the relevant state-court decision applied clearly established federal law erroneously or incorrectly.” Williams, 529 U.S. at 411, 120 S.Ct. 1495; accord Gilchrist v. O’Keefe, 260 F.3d 87, 94 (2d Cir. 2001). Rather, “some increment beyond error is required.” Francis S. v. Stone, 221 F.3d 100, 111 (2d Cir.2000). This increment, however, “need not be great; otherwise habeas relief would be limited to state court decisions so far off the mark as to suggest judicial incompetence.” Id. (internal quotation marks and citation omitted); accord Aparicio v. Artuz, 269 F.3d 78, 94 (2d Cir.2001). The Appellate Division unfortunately did not enlighten us as to its reasoning for denying Eze’s ineffective assistance claim, stating only that “[t]he record does not support the contentions of the defendant that ... he was denied effective assistance of counsel.” People v. Eze, 217 A.D.2d 987, 631 N.Y.S.2d 268, 268 (4th Dep’t 1995). We have held that “when a state court fails to articulate the rationale underlying its rejection of a petitioner’s claim, and when that rejection is on the merits, the federal court will focus its review on whether the state court’s ultimate decision was an ‘unreasonable application’ of clearly established Supreme Court precedent.” Sellan, 261 F.3d at 311-12; accord Aeid v. Bennett, 296 F.3d 58, 62 (2d Cir.), cert. denied, — U.S.-, 123 S.Ct. 694, 154 L.Ed.2d 641 (2002). We therefore engage in a review of Eze’s counsel’s performance at trial and ascertain whether the Appellate Division’s decision could be reasonable under Strickland. 1. Acts and Omissions of Eze’s Trial Counsel Under Strickland’s first prong, counsel’s conduct falls to the level of being deficient if “counsel made errors so serious that counsel was not functioning as the ‘counsel’ guaranteed the defendant by the Sixth Amendment.” Strickland, 466 U.S. at 687, 104 S.Ct. 2052. Our scrutiny of counsel’s performance is “highly deferential” because “[i]t is all too tempting for a defendant to second-guess counsel’s assistance after a conviction or adverse sentence, and it is all too easy for a court, examining counsel’s defense after it has proved unsuccessful, to conclude that a particular act or omission of counsel was unreasonable.” Id. at 689, 104 S.Ct. 2052. We therefore must make “every effort ... to eliminate the distorting effects of hindsight, to reconstruct the circumstances of counsel’s challenged conduct, and to evaluate the conduct from counsel’s perspective at the time.” Id. Accordingly, we “indulge a strong presumption that counsel’s conduct falls within the wide range of reasonable professional assistance,” which forces the defendant to “overcome the presumption that, under the circumstances, the challenged action might be considered sound trial strategy.” Id. We also note that three defense attorneys, one for each defendant, jointly'tried the case. We take into account the tactics of all three defense attorneys in assessing the reasonableness of Eze’s counsel’s performance. For example, Eze’s counsel may have wisely decided not to repeat a line of questioning that had been pursued, or would be pursued, by Okongwu or Wosu’s counsel. See United States v. Nersesian, 824 F.2d 1294, 1321 (2d Cir. 1987) (reviewing criminal conviction involving multiple defendants and appellant’s counsel was “close to last in the order of questioning by numerous defense counsel. Counsel might very well have felt that there was little need for additional probing by the time it was his turn to cross-examine, or even that cross-examination at that point might have been counterproductive.”), cert. denied, 484 U.S. 1061, 108 S.Ct. 1018, 98 L.Ed.2d 983 (1988). Eze alleges ineffective assistance at five points at trial. Eze’s first three allegations pertain predominantly to the claimants’ believability, while the final two relate to Eze’s credibility. We review the merits of each of these allegations in turn, and then assess the cumulative effect of these alleged deficiencies. a. Challenges to Dr. Lazoritz’s Medical Conclusions Eze makes various allegations of ineffective assistance concerning Dr. Lazoritz’s testimony. Dr. Lazoritz was a critical witness for the prosecution’s case, as his testimony regarding his examinations of the girls and his subsequent medical conclusions enabled the jury to conclude that sexual activity occurred. This physical evidence corroborating the girls’ allegations of abuse was particularly incriminating in this case, which largely came down to conflicting testimony from the defendants and the alleged victims as to whether abuse occurred. See Pavel v. Hollins, 261 F.3d 210, 224 (2d Cir.2001) (‘When a sex abuse case boils down to such a ‘credibility contest,’ physical evidence will often be important.”). First, Eze argues that his counsel failed to adduce evidence indicating that the findings made with respect to Nnedi’s hymen in 1988 were similar to those made in 1992 by Dr. Lazoritz. After showing that Dr. Lazoritz’s finding as to Chendo’s hymenal attenuation were similar to those from his 1988 examination of Chendo, Eze’s counsel questioned Dr. Lazoritz about Nnedi’s 1988 medical examination. Eze’s counsel asked Dr. Lazoritz whether he was aware that Nnedi had been examined by Dr. Hornberg at the Children’s Hospital and then inquired about his findings regarding Nnedi’s hymenal tissue. The prosecution objected “because she’s doing this through the wrong witness. You can’t ask this witness what someone else found with medical records that are not in evidence.” The court sustained this objection, presumably because Dr. Lazoritz did not perform Nnedi’s earlier examination. After Dr. Lazoritz testified that he did not have any recent discussions with Dr. Hornberg and was unaware whether Dr. Hornberg was still at the Children’s Hospital, Eze’s counsel moved on to a different topic and the defense did not revisit Nnedi’s 1988 medical examination. The evidence of Nnedi’s 1988 examination was crucial because it would have cast doubt as to whether Nnedi’s physical condition pre-existed the alleged abuse in 1991. The defense, however, did not introduce evidence of Nnedi’s 1988 examination, either through the testimony of the physician who examined Nnedi in 1988 or through a certified copy of the medical report. This failure to introduce this evidence, without any plausible justification, appears to be a significant dereliction by the defense. Eze also contends that his attorney failed to bring out various inconsistencies in Dr. Lazoritz’s testimony with respect to Chendo. Eze notes that whereas Dr. La-zoritz testified at trial that Chendo’s 1992 findings were “diagnostic” of sexual abuse, his report merely stated that those findings were “consistent with” sexual abuse. Although it may have been preferable for Eze’s counsel to pursue this line of questioning, the failure to do so does not rise to the level of deficient representation. “[T]he conduct of examination and cross-examination is entrusted to the judgment of the lawyer, and an appellate court on a cold record should not second-guess such decisions unless there is no strategic or tactical justification for the course taken.” United States v. Luciano, 158 F.3d 655, 660 (2d Cir.1998); see Nersesian, 824 F.2d at 1321 (“Decisions whether to engage in cross-examination, and if so to what extent and in what manner, are similarly strategic in nature.”). Eze’s counsel conducted a cross examination of Dr. Lazoritz that elicited several important points helpful to the defendant, including that other trauma could have caused the girls’ condition, that Chendo’s 1988 findings resembled those from 1991, and that Dr. Lazoritz had been informed that there was an allegation of sexual abuse prior to conducting his examination. Eze next argues that his counsel failed to enlighten the jury about the extent to which the medical community had called into question the method used by Dr. La-zoritz to conclude that sexual abuse occurred. In particular, Eze notes that his counsel failed to bring out that medical child sex abuse experts were questioning the significance of enlarged hymenal openings. Closely related to this objection is Eze’s claim that his counsel should have called a medical expert to rebut Dr. Lazo-ritz’s methodology and conclusions. We are especially concerned, in light of our holding in Lindstadt, with defense counsel’s failure to impeach the underlying medical grounds on which Dr. Lazoritz based his conclusion that the girls had been abused. The defendant in Lindstadt was convicted of sexually abusing his young daughter based on the testimony of his daughter, his estranged wife, and two (one medical and one psychological) expert witnesses. 239 F.3d at 193. The medical expert offered testimony that carried probative strength comparable to that of Dr. Lazoritz’s testimony. The expert in Lind-stadt testified to his findings from a physical examination of the alleged victim; these findings were the only physical evidence the prosecution introduced. Id. at 201. The expert stated that his findings were indicative of sexual abuse, a conclusion he based on studies that were never introduced into evidence. Id. at 201-02. Because the defendant’s counsel did not even request copies of these studies, he was unable to cross-examine the medical expert effectively on them. Id. Nor was there evidence that the defense contacted an expert of its own “either to testify or (at least) to educate counsel on the vagaries of abuse indicia.” Id. at 201. Had counsel done so, he would have discovered a bevy of medical literature that called the expert’s conclusions into doubt. Id. at 202. We thus concluded in Lindstadt that “counsel’s failure to consult an expert, failure to conduct any relevant research, and failure even to request copies of the underlying studies relied on by [the expert] contributed significantly to his ineffectiveness.” Id. We rejected arguments that counsel’s cross examination was otherwise satisfactory, observing that the cross examination “was hamstrung by counsel’s lack of familiarity with the studies upon which [the expert] was presumably relying.” Id. “[T]he result was ruinous” because these studies supposedly ruled out any explanation for the physical findings other than abuse. Id. In Pavel, we similarly determined that defense counsel’s failure to call a medical expert to testify about the prosecution’s physical evidence contributed significantly to our conclusion that the representation was constitutionally deficient. 261 F.3d at 223. The defendant in Pavel was accused of sexually abusing his seven and five year old sons. Id. at 211, 214. The prosecution’s medical expert testified, based on her review of the records of the boys’ physical examinations, that the physical findings were consistent with the boys’ allegations that their father had anally sodomized them. Id. at 215. The expert, however, also testified that some of the findings could have been attributed to benign causes, such as diarrhea. Id. Defense counsel neither called a medical expert to rebut the prosecution’s expert nor consulted an expert prior to trial. Id. at 223. We explained that the case boiled down to a “credibility contest” and that “when a case hinges all-but-entirely on whom to believe, an expert’s interpretation of relevant physical evidence (or the lack of it) is the sort of ‘neutral, disinterested’ testimony that may well tip the scales and sway the fact-finder.” Id. at 224. We counseled in Pavel that, because of the particular importance of physical evidence in child sexual abuse cases that turn into credibility contests, “physical evidence should be a focal point of defense counsel’s pre-trial investigation and analysis.” Id. Moreover, “because of the ‘vagaries of abuse indicia,’ such pre-trial investigation and analysis will generally require some sort of consultation with an expert.” Id. Consultation with an expert was crucial in Pavel for two reasons: 1) counsel had neither the education nor the experience necessary to evaluate the evidence and “make for himself a reasonable, informed determination as to whether an expert should be consulted or called to the stand”; and 2) there was an “obvious, commonsense mismatch” between the physical evidence and the allegations such that a “reasonably professional attorney” would have consulted and been ready to call an expert to address the inconsistencies. Id. In light of these and other trial errors, we found counsel’s performance to be deficient. Id. at 225-26. A lesson to be learned from Lindstadt and Pavel is that when a defendant is accused of sexually abusing a child and the evidence is such that the case will turn on accepting one party’s word over the other’s, the need for defense counsel to, at a minimum, consult with an expert to become educated about the “vagaries of abuse indicia” is critical. See id. at 224; Lindstadt, 239 F.3d at 201. The importance of consultation and pre-trial investigation is heightened where, as here, the physical evidence is less than conclusive and open to interpretation. It is undisputed that the defense did not question the basis for Dr. Lazoritz’s conclusions and did not call a medical expert to testify. Indeed, there is no indication that counsel even consulted with an expert prior to trial. This is extremely troubling because, prior to trial, various questions had been raised in the medical community that could have been used to call Dr. Lazo-ritz’s conclusions into doubt. Dr. Lazoritz partly based his conclusions on an examination of the girls’ hymens, specifically focusing on measurements of their enlarged hymenal openings. Contemporaneous medical studies raised questions about the reliability of hymenal examinations as the basis for determining the occurr