In Doe, Sex Offender Registry Bd. No. 151564 v. Sex Offender Registry Bd., 456 Mass. 612 (2010) (Doe No. 151564), the Supreme Judicial Court remanded this case to the Sex Offender Registry Board (SORB) to consider numerous scientific and statistical studies regarding the effect of the age of the plaintiff (Doe) on his dangerousness and likelihood to re-offend and whether Doe was entitled to expert witness funds. On remand, a SORB hearing examiner denied Doe’s request for funds for an expert and engaged in her own unguided review of the scientific and statistical studies. The hearing examiner reaffirmed SORB’s earlier decision to classify Doe as a level three sex offender, and on Doe’s appeal to the Superior Court pursuant to G. L. c. 30A, see G. L. c. 6, § 178M, a judge affirmed that decision. We conclude that Doe was entitled to funds for an expert to inform both his own presentation and the hearing examiner’s classification review. We therefore vacate the judgment of the Superior Court and remand the case to SORB for further proceedings consistent with this opinion.
Background. Doe was born on March 26, 1946. In June, 2001, in Maine, he sexually abused a six year old child who was not a family member. Doe pleaded guilty in Maine to one count of unlawful sexual conduct in violation of Me. Rev. Stat. Ann. tit. 17A, § 255(1)(C). In Doe No. 151564, 456 Mass, at 613, the Supreme Judicial Court concluded that this conviction was a “like violation” when compared to the Massachusetts crime of indecent assault and battery on a child under fourteen years of age, G. L. c. 265, § 13H, and therefore required Doe to register as a sex offender in Massachusetts, where he was then living.
In the original classification decision in 2007 (original decision), the hearing examiner stressed that Doe “sexually abused an extravulnerable, extrafamilial, six year old child on multiple *3occasions [the victim estimated twenty times].”1 As reported by the victim, Doe sexually assaulted her while he bathed her, rubbing her vagina with his fingers and fondling her chest and buttocks. He also took her from her bed during the night, again touching her vagina and putting her hand on his penis to try to make her masturbate him. On one occasion, he ejaculated. He kissed her on the lips during the assaults and threatened to hurt her baby sister if she told anyone.
In the original decision, the hearing examiner found that Doe had a prior criminal history, having been convicted of three counts of credit card fraud and one count of assault and battery by means of a dangerous weapon (a stick) in 1997. He also had prior probation violations, first in 1998 and then in 2006, when he was found in violation of his sex offense probation for failing to register as a sex offender.2
The hearing examiner considered Doe’s then-current cir*4cumstances, as a resident of a homeless shelter where he was described as “very encouraging to other guests in their substance abuse and homelessness struggles,” and where he was reported by one registered psychiatric nurse as having been “substance free for approximately nine years.”* *3 A case manager at the shelter reported that Doe had been in a relationship with a woman for four and one-half years. His health problems, including diabetes, cardiovascular disease, arthritis, and erectile dysfunction, were also described. The hearing examiner noted, however, that “[a]lthough [Doe] submitted documentation concerning his health, it does not attest that his illness affects his ability to offend sexually. In addition, his offense did not involve penile penetration of his [v]ictim, thus erectile dysfunction (which may also be transitory affliction and/or medically treated), is not conclusive of his ability to reoffend.”
After considering the facts and factors discussed above, the hearing examiner in her original decision concluded that Doe “pose[d] a high risk of reoffense and a high degree of dangerousness” and required him to register as a level three sex offender. She also concluded that “[i]n three more years, with established stability in the community, [Doe] may be a candidate for an administrative review by [SORB] to determine the appropriateness of reclassifying his level.”
In Doe No. 151564, 456 Mass, at 623, the Supreme Judicial Court vacated the judgment and remanded the decision to SORB for a new classification hearing. Most relevant to this appeal, the court concluded that the hearing examiner erred in making a classification determination without considering the effect of Doe’s age, which was sixty-one at the time of the original hearing, on his dangerousness and likelihood to reoffend. Id. at 621. The court further concluded that the hearing examiner erred in making her decision without considering “evidence of *5numerous scientific and statistical studies, published during the last decade, that conclude that age is an important factor in determining the risk of recidivism and that such risk diminishes significantly as an offender ages,” particularly for offenders over age sixty, like Doe. Ibid.4 These studies were authored by “many of the same authorities on whom [SORB] relies in its regulations.” Id. at 622. Additionally, the court noted that at least six States have recognized the importance of age as a classification factor for sex offenders. Ibid.
The court further concluded that the hearing examiner erred in denying Doe’s request for expert funds solely on the ground that SORB did not intend to rely on expert testimony, as this decision was inconsistent with Doe, Sex Offender Registry Bd. No. 89230 v. Sex Offender Registry Bd., 452 Mass. 764, 774 (2008), a decision which the court applied retroactively. Doe No. 151564, 456 Mass, at 624 n.8. The court directed that on remand, the hearing examiner should “consider whether funds for an expert are necessary given the individual circumstances of Doe’s case.” Id. at 624. See Doe, Sex Offender Registry Bd. No. 89230 v. Sex Offender Registry Bd., supra at 775 (“[T]he burden will be on the sex offender to identify and articulate the reason or reasons, connected to a condition or circumstance special to him, that he needs to retain a particular type of expert”).
In 2010, on remand from the Supreme Judicial Court, the hearing examiner issued a “supplemental decision” on Doe’s appeal from his classification (supplemental decision). Without holding a second evidentiary hearing, and based on her own review of the scientific and statistical studies, the hearing examiner concluded that “[a]t best, all of the research and articles submitted suggest that the correlation between age and risk of recidivism varies depending on the type of offender (child molester versus rapist/adult), the victim pool (incest versus extrafamilial or stranger), and is fairly inconclusive as applied to the facts in [Doe’s] particular case.” The hearing examiner stressed the fact that Doe “was in his mid-fifties when he first offended[;] thus the assertion of decreased arousal
*6and [increased] self-control is less persuasive.[5] Nor are there any known restrictions regarding his potential access to children.”
In the supplemental decision, the hearing examiner denied Doe’s request for funds in the amount of $5,000 for “the consultation and testimony of an expert witness, namely a psychologist or psychiatrist to interview the petitioner, review his records, and provide a determination as to his risk of recidivism and degree of dangerous[ness], if any.” The hearing examiner denied this request on the ground that Doe did not “identify and articulate the reasons, connected to a condition or circumstance special to him,” as required by Doe, Sex Offender Registry Bd. No. 89230 v. Sex Offender Registry Bd., 452 Mass, at 775. She rejected Doe’s argument that the expert was needed to address the issue of age as it relates to recidivism. According to the hearing examiner, “[e]yen if an expert so opined, a hearing examiner would not necessarily be bound to accept such conclusions, especially in light of the submitted research which is less than decisive regarding the effect of age on an extrafamilial child molester. Age is not a ‘disorder’ ’particular to him’ nor is it beyond common knowledge or understanding requiring expert explanation.”
Standard of review. As we have noted, an appeal from SORB’s final classification and registration requirements proceeds pursuant to G. L. c. 30A. See G. L. c. 6, § 178M. “We reverse [SORB’s] decision if it was ‘(a) [i]n violation of constitutional provisions’; ‘(b) [i]n excess of the statutory authority or jurisdiction’ of the board; ‘(c) [b]ased upon an error of law’; ‘(d) [m]ode upon unlawful procedure’; ‘(e) [unsupported by substantial evidence’; ‘(f) [unwarranted by facts found by the court . . . where the court is constitutionally required to make independent findings of fact’; or ‘(g) [arbitrary or capricious, an abuse of discretion, or otherwise not in accordance with law.’ See G. L. c. 30A, § 14(7). In conducting our review, we ‘give due weight to the experience, technical competence, and *7specialized knowledge’ of the board. Id.” Doe No. 151564, 456 Mass, at 614-615.
Discussion. To determine whether the hearing examiner erred in her consideration of the scientific and statistical studies on the effect of age on Doe’s likelihood to reoffend and in her denial of expert funds, we must consider not only the guidance set out in Doe No. 151564, but also that provided in a subsequent decision of the court, Doe, Sex Offender Registry Board No. 205614 v. Sex Offender Registry Board, 466 Mass. 594 (2013) (Doe No. 205614), which was issued after the hearing examiner’s supplemental decision.
Doe No. 205614 involved the classification of a female sex offender. The court emphasized that “in sex offender classification proceedings, the offender bears no burden of proof . . . . Rather, the burden is on SORB to undertake a ‘sound application of [the risk-classification] factors to derive a true and accurate assessment of an offender’s potential for reoffending.’ ” 466 Mass, at 605, quoting from Doe, Sex Offender Registry Bd. No. 136652 v. Sex Offender Registry Bd., 81 Mass. App. Ct. 639, 656 (2012). Indeed, “[t]he application of accurate and up-to-date information, including all known and empirically validated risk factors, thereby ensures that hearing examiners have the tools they need to arrive at individualized classification determinations. Such determinations must be grounded in a corpus of objective facts and data, necessarily dynamic and evolving to revise collective understanding of the risk that various individuals pose to the public.” Doe No. 205614, supra at 609.
Applying these standards, the court in Doe No. 205614, 466 Mass, at 606, noted that “authoritative evidence was introduced suggesting that the guidelines, developed as they were from studies of male offenders, could not predict accurately the recidivism risk of a female offender, and that such risk could not be evaluated without examining the effect of gender.” The court pointed to a 2005 study,6 coauthored by Dr. R. Karl Hanson, one of the experts cited in SORB’s guidelines, that found *8that female sex offenders “exhibited a one per cent sexual recidivism rate, as compared to a 13.4 per cent sexual recidivism rate in males,” and a 2007 study that cautioned against the extrapolation of assessment tools that have “been developed for and normed on male sex offenders.” Ibid, (citations omitted). Speaking more broadly the court remarked: “This potential for the frustration of individualized risk assessment is particularly conspicuous where the growing scientific consensus suggests specific factors that have quantifiable effects on recidivism rates, such as age and gender” (emphasis added). Id. at 608. The court concluded that it was arbitrary and capricious for SORB to fail to “evaluate the evidence of the effect of gender” in its classification determination, particularly when it did not cite any studies to the contrary. Id. at 607-608.
In Doe No. 205614, the court also decided that the plaintiff was entitled to expert witness funds. By presenting scientific and statistical evidence suggesting that the guidelines could not accurately predict the recidivism rate of female offenders, the plaintiff had identified “a particular characteristic or condition peculiar to her that the guidelines do not appear to contemplate and that an available expert is qualified to address. It was therefore an abuse of discretion to deny [the plaintiff’s] motion for expert-witness funds, where [she] established that the hearing examiner would not otherwise have competent evidence before him to assess fully [her] risk of reoffense.” 466 Mass, at 610.
In the instant case, Doe suggests that SORB is repeating the errors that it made in Doe No. 205614. Doe claims that here, as in Doe No. 205614, statistical and scientific studies were introduced demonstrating that age, like gender, has a significant quantifiable effect on recidivism rates. As in that case, Doe here relies on the work of researchers respected and referenced by SORB, particularly Dr. R. Karl Hanson, who observed in his study previously cited by the court that “five year recidivism-rates of offenders over 60 was only 2%, compared with 14.8% for offenders less than 40.” Hanson, Does Static-99 Predict Recidivism Among Older Sexual Offenders, 18 Sexual Abuse: A Journal of Research and Treatment, at 351 (2006). Doe also cites numerous studies corroborating the significance of the ef*9feet of age on recidivism rates.7 He claims that the hearing examiner’s dismissal of this evidence and denial of funds for an expert witness was therefore an abuse of discretion and arbitrary and capricious.
SORB contends on appeal that unlike in Doe No. 205614, the hearing examiner on remand did consider the scientific and statistical studies but found them unpersuasive because Doe committed his first offense in his fifties, when he abused an ex-trafamilial child. SORB emphasizes that the hearing examiner’s discounting or dismissal of the studies was not arbitrary or capricious given the methodological limitations of the studies establishing the significance of age as a factor for recidivism, the different conclusions drawn in the studies for different types of sex offenders, and the focus in the studies on age of reoffense rather than age of first offense.8 SORB further asserts that there was no abuse of discretion in the denial of funds for an expert witness because no expert explanation was required.
Our own review of the scientific and statistical studies in the record reveals that they are technical and complex.9 They *10emphasize a significant correlation between age and risk of offense,10 particularly for much older offenders, but stress the importance of distinguishing the different types of sex offenders and the different statistics associated with each as they age.* 11 They also reflect careful analysis and acknowledgment of both what is known and unknown about the effect of age on different types of sex offenders, where the results of the different studies align and differ,12 and where further research is required for more definitive determinations, including for older sexual offenders.13 The importance of expert assistance and guidance *11for the proper interpretation and understanding of these studies, and their application to individual plaintiffs, particularly an ex-trafamilial offender who is fifty-four at the time of his first offense and sixty-one at the time of classification, is readily apparent. The expert witness will inform both the plaintiff’s presentation and the hearing examiner’s analysis.
As the court in Doe No. 205614, 466 Mass, at 609, makes clear, “[a]t the very least, an expert could have offered assistance as to the possible limitations of applying SORB’s risk factors to [the petitioner].” An expert could have addressed the disputes in this case concerning the current state of scientific knowledge and consensus regarding the recidivism rates for extrafamilial offenders in their early sixties, whether that knowledge accounts or discounts for different ages of first reported offenses, including advanced ages, and how Doe’s varied medical conditions factor into the age calculus.14 The hearing examiner should not simply conclude, as she did here, that she was not required to accept any particular expert testimony. Of course, Doe is entitled to consideration, not acceptance, of the expert testimony, but the hearing examiner cannot even consider testimony she refuses to hear.
We conclude that in these circumstances, the hearing examiner’s decision to deny Doe expert assistance was an abuse of discretion and that her own unguided interpretation, application, and ultimate dismissal of the complex scientific and statistical studies on the effect of age on recidivism rates was arbitrary *12and capricious. The judgment of the Superior Court affirming SORB’s classification of Doe as a level three sex offender is vacated, and the matter is remanded to SORB for further proceedings consistent with this opinion.15
So ordered.
As required by G. L. c. 6, § 178K(1), SORB “has promulgated guidelines ‘for determining the level of risk of reoffense and the degree of dangerousness posed to the public’ ” in connection with a sex offender’s classification hearing. Doe, Sex Offender Registry Bd. No. 89230 v. Sex Offender Registry Bd., 452 Mass. 764, 772 (2008). See 803 Code Mass. Regs. § 1.40(l)-(24) (2002) (guidelines). In the original decision, the hearing examiner noted the relevant statutory and regulatory provisions that she applied to the facts. See, e.g., G. L. c. 6, § 178K(Z>)(i), and 803 Code Mass. Regs. § 1.40(7) (2002) (describing greater pool of victims when offender has assaulted victim outside family), and 803 Code Mass. Regs. § 1.40(9)(c)(4) (2002) (defining extravulnerable victims as including victim under age of ten); and G. L. c. 6, § 178K(a)(ii), and 803 Code Mass. Regs. § 1.40(2) (2002) (describing greater risk posed by offender who manifests his compulsive behavior by engaging in continuing course of sexual misconduct). The hearing examiner cross-referenced her findings of fact with statutory and regulatory citations throughout her decision.
In her 2007 decision, the hearing examiner also factored in Doe’s failure to participate in sex offender treatment, even though no sex offender treatment was available to him, and his denial of his sex offenses, without considering his guilty plea. In Doe No. 151564, 456 Mass, at 625, the Supreme ludicial Court concluded that “the board may not draw a negative inference from the absence of sex offender treatment where such treatment was neither available nor required during Doe’s incarceration, parole, or probation.” In addition, the court concluded that the hearing examiner erred when she failed to consider Doe’s guilty plea in analyzing “whether, or to what degree, the offender accepts responsibility that he committed his sex offenses,” a relevant factor in the board’s analysis. Ibid., quoting from 803 Code Mass. Regs. § 1.40(9)(c)(13) (2002). On remand, the hearing officer stated that she “[did] not consider lack of sex offender treatment as an aggravating factor.” As for the guilty plea, on remand she found “insufficient evidence to draw a conclusion that his plea *4was a ‘genuine expression of accountability,’ ” but at the same time she found insufficient evidence “to conclude that he is in denial . . . [given that] he did admit to (although arguably minimized) many of the surrounding details.” Doe has not raised any issue in this appeal regarding the hearing officer’s reconsideration of these two factors.
The hearing examiner stated that “[w]hile it is commendable that [Doe] has reportedly been sober for nine years, ... he committed his sex offense [within that time period], in June of 2001.”
The court cited statistics from Public Safety and Emergency Preparedness Canada, Validity of Static-99 with Older Sexual Offenders, at 10 (2005). Doe No. 151564, 456 Mass, at 622. .
In support of her conclusion, the hearing examiner cited a study by Fazel, Sjostedt, Langstrom, and Grann, Risk Factors for Criminal Recidivism in Older Sexual Offenders (2006) (speculating that the reasons for a reduction in older men’s recidivism risks may have to do with “lower sexual arousal in older men, increased self-control with age, and changes in victim access”).
Cortoni & Hanson, Correctional Service of Canada, A Review of Recidivism Rates of Adult Female Sexual Offenders (2005).
See, e.g., Barbaree, Langton, Blanchard, & Boer, Predicting Recidivism in Sex Offenders Using the SVR-20: The Contribution of Age-at-Release, 7 International Journal of Forensic Mental Health 47, 62 (2008) (Barbaree) (“These studies show remarkable uniformity in the slope of the regression line representing reductions in rates of recidivism over age”).
See, e.g., Hanson, Age and Sexual Recidivism: A Comparison of Rapists and Child Molesters, at 16 (Dept, of the Solicitor General Canada) (2001) (“Another consideration is that offenders’ ages were measured at time of release and not the onset of offending”).
A full understanding of the studies requires specialized training. See, for example, this excerpt from Barbaree, supra at 52: “Table 2 presents mean age-at-release (and standard deviations and Ns) for offender subgroups who received different scores on each SVR-20 item. We conducted ANOVA on each item, partitioning the 2 df mean square into its single df components (the linear component compared mean age-at-release of the group assigned a ‘0’ with the group assigned a ‘2’ ... . The quadratic trend can be thought of as a test of the deviation from linearity in relationship between age-at-release and risk score,” or this passage from Thornton, Age and Sexual Recidivism: A Variable Connection, 18 Sexual Abuse: A Journal of Research and Treatment 123, 130 (2006) (Thornton): “The picture changes again, however, when interaction effects are fitted. Adding the linear effect of age by criminal history product terms did not lead to a significant improvement in fit [Log Likelihood Ratio x2(2, n = 752) = 0.111, ns]. Adding the quadratic effect of age by criminal history product terms also failed to show significant improvement in fit ... . However, adding the cubic age by criminal history interac*10tian terms did lead to a significant improvement in the fit of the logistic regression equation [Log Likelihood Ratio x2(2, n = 752) = 10.51, p = 0.006]. The Wald coefficient for the cubic age by sexual sentencing occasions interaction was significant.”
See, e.g., Barbaree, supra at 61 (discussing “the now well-established relationship between age-at-release and recidivism; offenders who are older when they are released are less likely to re-offend”); Wollert, Low Base Rates Limit Expert Certainty When Current Actuaríais Are Used to Identify Sexually Violent Predators, 12 Psychol. Pub. Pol’y & Law, 56, 69 (2006) (“No doubt exists as to the presence of a very strong effect of age on sexual recidivism”).
See, e.g., Thornton, supra at 210 (“This overall trend disguised considerable variation in the nature of the relationship [between age at release and sexual recidivism] within different groups of offenders” including offenders with differing numbers of “prior sexual sentencing occasions”); Hanson, Age and Sexual Recidivism, supra at 9 (“Even though the effect of age was linear in the total sample, the relationship with age was different for the rapists [and] extrafamilial child molesters .... The recidivism rate of the extrafamilial child molesters showed relatively little decline until after age 50”); Prentky & Lee, Effect of Age-at-Release on Long Term Sexual Re-Offense Rates in Civilly Committed Sexual Offenders (2007) (discussing “marked differences” in reoffense patterns for rapists and child molesters as they age).
See, e.g., Thornton, supra at 125 (“Taken together this literature seems to indicate a consensus that the 18-24 age band represents a higher risk .... Additionally, there would seem to be a consensus that the 60+ age band represents a lower risk. The picture between these extremes seems more uncertain and controversial. In what might be called the middle years of life [25-59], some actuarial scales indicate little age-related change in recidivism while others suggest meaningful change takes place”); Barbaree, supra at 62 (“And, except for the Prentky and Lee study, these studies show fair consistency in the rate of recidivism reported at similar age groups”); Barbaree & Blanchard, Sexual Deviance over the Lifespan, Sexual Deviance, at 53 (2008) (discussing disagreements concerning significance of age at release versus age of first offense).
See, e.g., Thornton, supra (discussing limitations in the studies); Hanson, Does Static-99 Predict Recidivism Among Older Sexual Offenders, supra at 352 (“Sexual offenders risk assessments are likely to improve as researchers *11identify the reasons for the lower recidivism rates among older sexual offenders”); Hanson, Age and Sexual Recidivism: A Comparison of Rapists and Child Molesters, at 16 (Dept, of the Solicitor General Canada 2001) (“[R]esearch has yet to examine the extent to which reductions in sexual recidivism risk should be expected for older offenders who remain in good health”); Saari & Saari, Actuarial Risk Assessment with Elderly Sex Offenders: Should it be Abandoned? Sex Offender L. Rep. 68, 75 (Aug./Sept. 2002) (“There are reasons to think that the low rate of recidivism among those over age 60 in [2001 study by Hanson, Age and Sexual Recidivism] may be partly a function of a biased sample. Moreover, these offenders were only followed for an average of about seven years, and the rate of offending among sex offenders, especially among child molesters, has been observed to continue for up to 25 years”); Prentky & Lee, supra (“Clearly, much more research is required on subtype-differentiated samples of child molesters”).
See Hanson, Static-99, supra at 352 (“One question for further research is the extent to which the observed declines in recidivism potential are mediated by poor health”).
We decline to require SORB to appoint a different hearing examiner on remand.