FOR PUBLICATION

UNITED STATES COURT OF APPEALS

FOR THE NINTH CIRCUIT

C. L., an individual, No. 19-56074 Plaintiff-Appellant, D.C. No. v. 8:18-cv-00475-

DOC-DFM

DEL AMO HOSPITAL, INC., a California corporation; DOES, 1–10, inclusive, OPINION Defendants-Appellees.

Appeal from the United States District Court for the Central District of California David O. Carter, District Judge, Presiding

Argued and Submitted December 11, 2020 Pasadena, California

Filed March 30, 2021

Before: Ronald M. Gould and Ryan D. Nelson, Circuit Judges, and Brian M. Cogan, * District Judge.

Opinion by Judge Gould

* The Honorable Brian M. Cogan, United States District Judge for the Eastern District of New York, sitting by designation. SUMMARY **

Americans with Disabilities Act

The panel vacated the district court’s judgment, after a bench trial, in favor of the defendant in an action seeking injunctive relief under Title III of the Americans with Disabilities Act (“ADA”), which prohibits discrimination in “places of public accommodations,” including hospitals.

Plaintiff C.L., who has been diagnosed with post- traumatic stress disorder and other conditions, obtained a dog named Aspen, intending it to be her service dog. Because enrolling in a full training course to provide Aspen with formal certification was not a viable option for C.L., she began self-training the dog. C.L. sought inpatient treatment at Del Amo Hospital’s National Treatment Center. When she asked the Center if she could bring Aspen with her as a service dog, Del Amo denied the dog admission. The district court entered judgment in favor of Del Amo on the ground that Aspen did not qualify as a service animal under the ADA.

The panel held that the district court erred as a matter of law by effectively imposing a certification requirement for C.L.’s dog to be qualified as a service animal. The panel held that the ADA prohibits certification requirements for qualifying service dogs for three reasons: (1) the ADA defines a service dog functionally, without reference to specific training requirements; (2) Department of Justice regulations, rulemaking commentary, and guidance have

** This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader.

C. L. V. DEL AMO HOSPITAL 3

consistently rejected a formal certification requirement; and (3) allowing a person with a disability to self-train a service animal furthers the stated goals of the ADA, for other training could be prohibitively expensive.

The panel remanded for the district court to consider whether C.L.’s testimony regarding her self-training of Aspen, coupled with expert testimony, was sufficient to show that Aspen was more likely than not a qualified service dog at the time of trial.

COUNSEL

Christopher H. Knauf (argued) and Alexandra M. Robertson, Disability Rights Legal Center, Los Angeles, California; Jennifer E. Mathis, Bazelon Center for Mental Health Law, Washington, D.C.; Celia McGuinness and Anthony E. Goldsmith, Derby McGuinness & Goldsmith LLP, Oakland, California; for Plaintiff-Appellant.

Raul L. Martinez (argued), Lewis Brisbois Bisgaard & Smith LLP, Los Angeles, California; Scott D. Buchholz and Moira S. Brennan, San Diego, California; for Defendants- Appellees.

Gina F. Elliott (argued), Adam R. Lawton, and Brian J. Springer, Munger Tolles & Olson LLP, Los Angeles, California, for Amici Curiae National Disability Rights Network, American Association of People With Disabilities, Autistic Self Advocacy Network, Disability Rights California, Mental Health America, National Council on Independent Living, and Psychiatric Service Dog Partners. GOULD, Circuit Judge:

In 1990, Congress enacted the Americans with Disabilities Act (“ADA”) to eliminate the discrimination that persons with disabilities faced in essential facets of everyday life. 42 U.S.C. § 12101. Title III of the ADA prohibits discrimination in “places of public accommodations,” including hospitals. Id. §§ 12181– 12182. Plaintiff-Appellant C.L. (“C.L.”), who survived years of abuse at the hands of her family and a romantic partner, has been diagnosed with post-traumatic stress disorder (“PTSD”), dissociative identity disorder (“DID”), anxiety, and depression. As a result of these conditions, C.L. experiences hypervigilance, PTSD-related nightmares and flashbacks, severe anxiety in public spaces and while bathing, and has difficulty remaining focused and engaged in daily tasks. To mitigate the symptoms of her disability, C.L. obtained Aspen, a 16-pound bichon-poodle mix, intending Aspen to be her service dog. Because enrolling in a full training course to provide Aspen with formal certification was not a viable option for C.L., she began self- training Aspen to perform specific tasks she thought would ameliorate her disability and decrease her isolation.

Before and after obtaining Aspen, C.L. sought inpatient treatment at Defendant-Appellee Del Amo Hospital’s (“Del Amo”) National Treatment Center. When C.L. asked the Center if she could bring Aspen with her as her service dog, Del Amo denied Aspen admission, concluding that the dog’s presence would interfere with C.L.’s therapy. In the underlying suit, C.L. challenged Del Amo’s practice of denying admission to Aspen as a violation of Title III of the ADA and California’s Unruh Civil Rights Act. C.L. is undisputedly a person with a disability, and Del Amo is a

C. L. V. DEL AMO HOSPITAL 5

place of public accommodation. After a bench trial, the district court determined that Aspen does not qualify as a service dog under the ADA. In this appeal, C.L. challenges the district court’s judgment in favor of Del Amo. We have jurisdiction under 28 U.S.C. § 1291, and we vacate and remand.

We hold that the district court erred by effectively imposing a certification requirement for C.L.’s dog to be qualified as a service animal under the ADA. We vacate and remand for the district court to reconsider whether Aspen was a qualified service dog at the time of trial, and if Aspen is a service dog, whether Del Amo has proved its affirmative defense of fundamental alteration.

I

C.L. is a speech-language pathologist with a master’s degree in Speech and Language Pathology and a Ph.D. in Education. As a child, C.L. endured years of physical, psychological, and sexual abuse. She escaped her family’s abuse at age 17, but then experienced a 10-year abusive relationship before escaping and starting therapy in 1995. By 1996, she had been diagnosed with PTSD and DID, and started taking medication. Since then, C.L. has been diagnosed with anxiety and depression.

C.L.’s psychiatric condition limits major life activities, such as interacting with others, self-care, and sleeping. As a result of her PTSD, C.L. has a heightened startle response and hypervigilance—she continually checks whether things in her environment are dangerous and finds it difficult to remain focused and engaged in daily tasks. Additionally, having people unexpectedly in her presence induces anxiety. Taking a shower is particularly challenging due to her history of abuse while bathing. C.L. has PTSD-related nightmares and flashbacks that are sometimes so severe that she tries not to sleep at all to avoid them.

In 2011, after an event triggered a traumatic memory, C.L.’s mental health deteriorated and her flashbacks, anxiety, and depression increased. C.L. felt suicidal. Dr. Michael Foust, C.L.’s therapist, encouraged C.L. to obtain a service dog to manage her symptoms of PTSD and decrease her isolation.

C.L. began researching the possibility of obtaining a service dog. First, she purchased a book called Training Your Own Psychiatric Service Dog, by a service dog trainer named Katie Gonzalez, to help her understand the tasks performed by psychiatric service dogs and whether one could meet her needs. Katie Gonzalez is the director of Little Angels Service Dogs (“Little Angels”), a nonprofit service dog training organization. Gonzalez has trained service dogs, including psychiatric service dogs, for twenty years, and has published several books on training service dogs. Through her research, C.L. learned that because she was living on Supplemental Security Income, she could not afford to pay for a trained dog. A trained dog would cost at least $15,000. 1 C.L. conducted further research and conferred with a service dog training agency about what dog breed might best meet her needs.

In August 2013, C.L. obtained Aspen—a 16-pound bichon-poodle mix that was then 8 weeks old—to be her

1 Little Angels fundraises $38,000 to cover the cost to train each service dog and then provides the trained service animal to an approved applicant with a disability on its waiting list for a comparatively minimal deposit of at least $500. At the time of trial, the waiting list was approximately 180–200 approved applicants, and the estimated time for an approved applicant to receive a trained service dog was ten years.

C. L. V. DEL AMO HOSPITAL 7

service dog. At that time, she did not yet know what tasks she would want Aspen to perform. She took several dog- training classes at a general dog-training facility, Wags & Wiggles, where she learned how to train Aspen for general socialization and good behavior in public. 2 C.L. used the methods she learned at the dog-training facility and Katie Gonzalez’s service dog-training book to begin training Aspen to perform specific tasks. C.L.’s method, as taught by the classes, included positive reinforcement and verbal acknowledgement of successfully performed tasks, while extinguishing inaccurate or inappropriate behavior. C.L. also used a clicker as a positive reinforcement tool.

In 2013, C.L. trained Aspen to perform specific tasks to mitigate symptoms of her disability:

1. Waking from Nightmares: C.L. trained Aspen to wake her from nightmares by standing on her or licking her face. This task interrupted the nightmares, thereby improving her sleep and reducing the amount of distress she experienced following a nightmare. C.L. testified that by the end of 2013, Aspen was consistently performing the task of waking C.L. from nightmares and not waking her for any other purpose.

2. Grounding: C.L. experiences flashbacks and anxiety. C.L. trained Aspen to place herself in a particular position on C.L.’s lap and apply deep pressure while facing forward. This “grounds” C.L. in the present. C.L. testified that Aspen was

2 C.L. continued to attend classes periodically at Wags & Wiggles through the spring of 2019. Wags & Wiggles is certified in dog training, but not service dog training. performing this task consistently by the end of 2013.

3. Alert for People Approaching: C.L. trained Aspen to alert her that someone is approaching outside her sightline, alleviating C.L.’s symptoms of hypervigilance and improving her ability to focus on tasks at hand.

In 2014, C.L. attended a two-day seminar at Little Angels Service Dogs’ facility in San Diego. The seminar was the first course in a three-seminar series, where C.L. learned about how to select an appropriate service dog, the laws and regulations related to having a service dog, and basic training concepts. Specifically, C.L. learned additional techniques for using positive reinforcement and correcting unwanted behaviors such as the “leash tug” and an approach that involved tapping the dog in the hind area. Later that year, Little Angels offered two additional seminars for training one’s own service dog. C.L. says she did not attend them because they offered training in tasks that she did not need her dog to perform, such as turning on a light or opening a door. Moreover, she could not afford the tuition or the cost of traveling from her home in Santa Ana to San Diego. Nevertheless, she continued to communicate with Little Angels trainers via email and telephone to discuss Aspen’s progress, receive feedback, and get her training questions answered as she continued training.

C.L. testified to training Aspen to accomplish the following additional tasks in 2014 and 2015:

1. Interrupt Self-Injurious Behavior: C.L. trained Aspen to interrupt behaviors such as cutting and banging her head against a wall. For example, when C.L. is banging her head against a wall,

C. L. V. DEL AMO HOSPITAL 9

Aspen places herself between C.L. and the wall. Although C.L.’s therapist also proposed strategies for interrupting self-injurious behavior, including use of ice or a rubber band to safely provide a sensory distraction, C.L. testified that Aspen is much more effective than these strategies.

2. Cornering: C.L. trained Aspen to go around a corner ahead of C.L. and alert her if someone is approaching. Being alerted to the presence of people before she sees them alleviates C.L.’s anxiety and hypervigilance.

3. Boundary Control: C.L. trained Aspen to create a boundary with her body between C.L. and other people, enabling her to spend more time in public.

4. Alert for Medication: C.L. trained Aspen to alert her when her anxiety is increasing, even before C.L. becomes conscious of it herself.

5. Standing Guard by the Shower: C.L. has difficulty showering due to past sexual abuse, so she trained Aspen to sit in a specific location outside the bathroom door and to come get her if someone approaches.

On thirteen separate occasions, C.L. sought inpatient treatment at Del Amo’s National Treatment Center. Only seven of those admissions, which took place between September 2015 and August 2017, were the subject of C.L.’s claims in her initial complaint. The National Treatment Center program specializes in treatment of patients who have experienced trauma. During those seven admissions, Del Amo denied C.L.’s request to bring Aspen with her. The hospital denied C.L.’s request because Del Amo clinicians determined that Aspen’s presence in the Center would interfere with C.L.’s therapy by allowing her to rely on Aspen rather than learn coping skills.

On March 23, 2018, C.L. filed a complaint in the Central District of California challenging Del Amo’s practice of denying admission to Aspen on the seven previous occasions as well as the ongoing denials of admission to Aspen. C.L. filed suit under Title III of the ADA, 42 U.S.C. § 12182(a), and California’s Unruh Civil Rights Act, Cal. Civ. Code § 51(b). The parties filed cross-motions for summary judgment.

In its order on summary judgment, the district court found that Del Amo did not dispute that C.L. is a “person with a disability” within the meaning of the ADA. The court denied Del Amo’s motion in its entirety and granted C.L.’s motion in part, holding that Del Amo is a “place of public accommodation” under Title III of the ADA. The court denied C.L.’s pre-trial motion in limine requesting adjudication of whether Aspen was, in fact, a service dog. The district court initially rejected C.L.’s request for a bench trial, but it then acquiesced when C.L. waived any claim for damages and stated that she would be pursuing only injunctive relief at trial.

A bench trial took place from July 23 to 26, 2019. Two issues were presented at trial: (1) whether Aspen was a service dog, and (2) whether allowing Aspen to participate in C.L.’s hospitalization would “fundamentally alter” the psychiatric services that were being offered to C.L. during those hospitalizations under 42 U.S.C. § 12182(b)(2)(A).

C. L. V. DEL AMO HOSPITAL 11

At trial, C.L. testified to training Aspen to perform the specific tasks outlined above. Because C.L. trained Aspen in these tasks herself, there was no other witness to testify to the details of her training and how Aspen helped her manage her disability from 2015 to 2017.

Katie Gonzalez testified as an expert in the training of service dogs, the use of service dogs for psychiatric disabilities, and the laws and regulations for service dogs as applied to the work service dogs perform. Gonzalez testified that interrupting nightmares, alerting for people approaching, grounding, interrupting self-harm, cornering, boundary control, alerting for medication, and standing guard in particular places are typical tasks for psychiatric service dogs.

Gonzalez also testified to observing Aspen and C.L. as a dog-handler unit in June 2019. Gonzalez testified that Aspen was a fully trained service dog at the time she observed C.L. and Aspen in June 2019. Gonzalez observed C.L. with Aspen at multiple locations, a restaurant and a shopping center, and the meeting lasted one to one-and-a-half hours. She instructed C.L. to demonstrate some of Aspen’s trained tasks and examined how the dog reacted to people and its disposition. She also considered Aspen’s breed and size to evaluate appropriateness for the tasks Aspen was trained to perform. Gonzalez also considered the methods by which C.L. trained Aspen. She testified that the techniques C.L. described using to train Aspen are the same techniques that she and Little Angels’ trainers teach in the seminars, and as described in the Training Your Own Psychiatric Service Dog book and the Little Angels Seminar Booklet. Gonzalez interviewed C.L. regarding how C.L. trained Aspen and concluded that Aspen’s tasks were trained tasks, rather than mere behaviors. Relevant to this appeal, Gonzalez testified that Little Angels offers a “certification” for service dogs, based in part on protocols developed by Assistance Dogs International (“ADI”). ADI is a private, nonprofit organization of professional service dog trainers. ADI developed the protocols for service dog training as best practices by its members. Gonzalez emphasized, however, that the legal standard for determining whether a dog is a service dog under the ADA is different and more basic than the Little Angels certification or the ADI protocols. A company like Little Angels can of course set its standards for certifying a service dog at a level that is more rigorous than that required by the ADA.

Gonzalez testified that she could not certify Aspen under ADI’s protocols because further steps would be required for her to do so, e.g., submitting the dog’s veterinary records, remaining in contact with Gonzalez every month, and completing all of Little Angels’ seminars and assessments. Gonzalez observed C.L. and Aspen working together and concluded that the unit exhibited both classical and operant conditioning—forms of training that are common in the industry and used by Little Angels’ trainers. Based on the foregoing, Gonzalez testified that even though C.L. did not complete the steps required for Little Angels certification, C.L. had successfully trained Aspen as a service dog.

On Del Amo’s fundamental alteration defense, Del Amo’s psychiatrist expert, Dr. Anna Solt, testified that having the dog present during hospitalization would make it “too tempting for her to try to utilize the service dog instead of implementing tools that are taught within the program.” Dr. Solt also testified that there is “no scientific way to know that someone is having a nightmare.”

C. L. V. DEL AMO HOSPITAL 13

After the four-day bench trial, the district court entered judgment in favor of Del Amo on the grounds that C.L. had not shown Aspen was or is a service dog. The court did not reach the question of whether Del Amo had proved its affirmative defense of “fundamental alteration.” C.L. timely appealed.

While this appeal was pending, on February 7, 2020, several organizations moved pursuant to Federal Rule of Appellate Procedure 29(a) for leave to file an amicus curiae brief in support of C.L. The organizations are recognized authorities in the field of disability rights. We granted the motion, and the amicus brief was filed on November 6, 2020. 3

II

We first address the proper standard of review for C.L.’s claims. We review de novo questions of law or mixed questions of law and fact. See OneBeacon Ins. Co. v. Has Inds., Inc., 634 F.3d 1092, 1096 (9th Cir. 2011); Lim v. City of Long Beach, 217 F.3d 1050, 1054 (9th Cir. 2000). We review a district court’s findings of fact for clear error. Fed. R. Civ. Proc. 52(a)(6); Anderson v. Bessemer City, 470 U.S. 564, 566 (1985).

C.L. frames the issues on appeal as two legal errors by the district court: (1) imposing a service dog “certification” requirement that the ADA prohibits, and (2) concluding incorrectly that C.L.’s uncontroverted and unimpeached testimony requires corroboration to satisfy the ADA. C.L. contends that these erroneous legal rulings were the sole bases for the district court’s finding that Aspen does not

3 The court thanks the amicus parties for their views. qualify as a service dog. Del Amo disputes this characterization, contending that the issue before us is “whether Aspen was a trained service dog and whether Aspen could perform tasks directly related to C.L.’s disability,” a factual finding resting exclusively on the district court’s “Findings of Fact.” According to Del Amo, the district court based its factual conclusion that Aspen is not a service dog on credibility grounds.

Based on our reading of the district court’s Findings of Fact and Conclusions of Law, we frame the certification issue as presented by C.L. and review the claim de novo. The plain language of the district court’s decision suggests that it relied on Katie Gonzalez’s inability to “certify” Aspen under Little Angels’ standards to reject C.L.’s claim that Aspen is a service dog.

Del Amo contends that even though the district court relied on the fact that Aspen was not and could not have been certified under Little Angels standards, the court also relied on proper grounds, i.e., credibility. This contention is belied by the district court order. First, the court repeatedly emphasized certification standards in its judgment. Second, in laying out its conclusions of law, the court stated: “Gonzalez testified that she believes Aspen is a service animal as of June 2019,” but that testimony was “contradicted” because Gonzalez “is still not willing to certify C.L. and Aspen as a service dog and handler team.” The court then concluded: “Accordingly, the Court finds that Plaintiff has not met her burden to show . . . that Aspen is currently a trained service dog.” In other words, Gonzalez’s expert testimony that Aspen was a trained service dog as of June 2019—the only expert testimony implicating service dog training—was not accepted by the court because Gonzalez’s inability to certify Aspen necessarily

C. L. V. DEL AMO HOSPITAL 15

contradicted that claim. If the inability to certify Aspen “contradicts” an opinion that Aspen is a trained service dog, as the district court opined, then the court improperly considered certification to be a legally necessary standard for assessing whether Aspen was a trained service dog. Because the court declined to give weight to C.L.’s testimony on its own and considered her testimony uncorroborated, the court’s conclusion that Aspen was not a trained service dog hinged on the supposed contradiction in Gonzalez’s testimony.

Contrary to Del Amo’s contentions, the court did not challenge or raise questions about C.L.’s credibility, and it made no findings of fact whatsoever on the substance of C.L.’s testimony, her demeanor at trial, or any impeaching evidence. Nor did the court make credibility findings as to Gonzalez, other than noting—apparently based on the court’s conflation of the standard for a service dog under the ADA and Little Angels’ certification standards—that her testimony contradicted itself. Del Amo points to portions of C.L.’s testimony it believes are inconsistent or not credible, but because these contentions are post-hoc rationalizations of the district court decision—rather than an accurate representation of the district court’s express findings and conclusions—we decline to affirm on credibility grounds and review C.L.’s claim as legal error. See Alpha Distrib. Co. v. Jack Daniel Distillery, 454 F.2d 442, 452–53 (9th Cir. 1972) (declining to decide contested factual claim where the district court’s findings of fact were unclear); Fed. R. Civ. Proc. 52(a)(6). 4

4 Rule 52(a) requires: “In an action tried on the facts without a jury or with an advisory jury, the court must find the facts specially and state its conclusions of law separately.” The Rule “requires the district court’s We next address the scope of C.L.’s requested relief. Because she requests only injunctive relief, 5 the relevant question is whether Aspen was a trained service dog at the time of trial: 2019.

III

We next address what appears to be a question of first impression for our Circuit 6—whether it is a misinterpretation of the ADA to effectively require a service dog to meet formal certification requirements. We hold that the ADA prohibits certification requirements for qualifying service dogs for three reasons: (1) the ADA defines a service dog functionally, without reference to specific training requirements, (2) Department of Justice (“DOJ”) regulations, rulemaking commentary, and guidance have consistently rejected a formal certification requirement, and (3) allowing a person with a disability to self-train a service animal furthers the stated goals of the ADA, for other training could be prohibitively expensive. Thus, the district

findings to ‘be explicit enough to give the appellate court a clear understanding of the basis of the trial court’s decision, and to enable it to determine the ground on which the trial court reached its decision.’” Zivkovic v. S. Cal. Edison Co., 302 F.3d 1080, 1090 (9th Cir. 2002) (quoting Alpha Distrib., 454 F.2d at 453). 5 C.L.’s initial complaint requested damages, but at the time of trial, C.L. had waived any claim for damages and was pursuing only injunctive relief. Though Del Amo devotes much of its answering brief to Aspen’s status as a service dog from 2014 to 2018, that evidence is not dispositive as to whether Aspen is currently a trained service dog that should be allowed to accompany C.L. to Del Amo during future admissions. 6 Even outside of our Circuit, the case law in this area is sparse and primarily exists at the district court level.

C. L. V. DEL AMO HOSPITAL 17

court erred by imposing a heightened requirement on Aspen that is inconsistent with the ADA.

A

The ADA’s implementing regulations define a service animal as “any dog that is individually trained to do work or perform tasks for the benefit of an individual with a disability,” including a psychiatric disability, where the work or tasks are “directly related to the individual’s disability.” 28 CFR § 36.104. The regulations do not specify by whom the dog must be trained. Rather, the statute defines a service dog by the outcome of training—what the dog is capable of doing to ameliorate an individual’s disability. The language also makes clear that the dog’s capabilities must be trained for that purpose; a well-trained companion animal that happens to alleviate a person’s anxiety would not suffice, see id., but a dog trained by the individual to perform certain tasks to alleviate that anxiety would. Our view is consistent with district courts that have considered the question and found that the ADA does not require a service dog to perform a particular number of trained tasks or amount of work. See, e.g., Green v. Hous. Auth. of Clackamas Cnty., 994 F. Supp. 1253, 1256 (D. Or. 1998) (“[T]here is no requirement as to the amount or type of work a service animal must provide for the benefit of the [person with a disability].”). There must be some evidence of individual training to distinguish the service animal from the ordinary pet. For example, a dog may naturally jump up in its owner’s lap, which would constitute a behavior— however, if the owner trains the dog to sit in her lap in a particular position and only in response to certain triggers related to the owner’s disability, then the dog has been trained and has ceased to merely behave in a way that dogs naturally do. Here, Gonzalez, an expert in service dog training, explained that the process of training a dog to mitigate symptoms of disability may involve reinforcing some natural behaviors and extinguishing other behaviors until the dog is consistently performing the desired task. And Aspen had been trained to perform several tasks well beyond the normal behavior of a pet, such as licking C.L.’s face to wake her from a nightmare, interrupting self- injurious behavior, cornering, boundary control and other specific trained tasks.

B

The DOJ has consistently stated—in regulations, rulemaking commentary, and official department guidance—that a service animal within the meaning of the ADA must be individually trained to perform tasks related to an individual’s disability, but the animal need not be formally certified. The test is a functional one: can the dog consistently help the person with a disability meet the challenges of life by assisting in the person’s activities of daily living?

In enacting the ADA, Congress explained that one purpose is “to ensure that the Federal Government plays a central role in enforcing the standards established in this chapter.” 42 U.S.C. § 12101(b)(3). To that end, Congress gave the Attorney General the responsibility to promulgate regulations implementing the provisions of Title III of the ADA. 42 U.S.C. § 12186(b). “To flesh out the details of [Title III’s] general rule, Congress charged the Attorney General with the task of promulgating regulations clarifying how public accommodations must meet these statutory obligations.” United States v. AMC Ent., Inc., 549 F.3d 760, 763 (9th Cir. 2008).

C. L. V. DEL AMO HOSPITAL 19

DOJ regulations and commentary make clear that individuals may self-train service animals without obtaining formal certification. The DOJ’s administrative guidance regarding the “public accommodations” provision is “entitled to deference.” Bragdon v. Abbott, 524 U.S. 624, 646 (1998) (citing Chevron, U.S.A., Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 844 (1984)). The current regulations were made final after the DOJ published a notice of proposed rulemaking in June 2008. See Nondiscrimination on the Basis of Disability by Public Accommodations and in Commercial Facilities, 73 Fed. Reg. 34508, 34515 (proposed June 17, 2008). With regard to qualifying service animals, the DOJ wished to amend the definition of “service animal” to exclude some species (such as rabbits) and to exclude emotional support animals, but it also sought to formalize the agency’s longstanding position that people with psychiatric and mental disabilities can use service animals. Id. at 34515–16, 34521. After receiving comments, the DOJ issued its final regulations in September 2010, and the regulations took effect in March 2011. See Nondiscrimination on the Basis of Disability by Public Accommodations and in Commercial Facilities (“Final Rule”), 75 Fed. Reg. 56236, 56237 (Sept. 15, 2010); 76 Fed. Reg. 13286, 13288 (Mar. 11, 2011) (making technical corrections).

The district court’s ruling that Aspen was not a service animal in part because she could not be certified under ADI standards is contrary to multiple aspects of the 2010 regulations. First, the regulations state that a person with a “psychiatric . . . or other mental disability” may benefit from the use of service animals. 28 C.F.R. § 36.104. Trained tasks can include “preventing or interrupting impulsive or destructive behaviors.” Id. As other courts have noted, a dog can be trained to aid a person with a disability without formal schooling. See Bronk v. Ineichen, 54 F.3d 425, 430– 31 (7th Cir. 1995) (holding that the district court erred as a matter of law by providing a jury instruction from which the “jury could logically infer . . . that without school training, a dog cannot be a reasonable accommodation,” where a reasonable accommodation is defined by statute as “facilitat[ing] a [person with a disability’s] ability to function”); Green, 994 F. Supp. at 1256.

The DOJ’s commentary accompanying its rulemaking confirms that persons with disabilities need not secure formal training and may self-train their animals. In fact, the DOJ considered but specifically rejected a recommendation submitted by multiple commenters to adopt “formal training requirements for service animals.” Final Rule, 75 Fed. Reg. at 56272 (rejecting this approach and concluding that DOJ “will not impose any type of formal training requirements or certification process”). The DOJ justified its decision by noting that a certification requirement would increase the cost of acquiring a service animal, thus limiting access to such animals for individuals with limited financial resources, and that suggested training standards were too “lengthy and detailed.” Id. The DOJ also expressed an intention not to “unnecessarily impede individual choice” in light of “the diverse needs and preferences of individuals with disabilities.” Id. at 56266. Declining to impose any kind of rigid training requirement, the DOJ emphasized that “individuals with disabilities may be capable of training, and some have trained, their service animal to perform tasks or do work to accommodate their disability.” Id. Thus, the district court’s conclusion that Aspen’s failure to receive certification weighed against her claim for relief cannot be reconciled with the permissive approach adopted by the DOJ. It is enough if a service dog had been trained to perform specific tasks that will consistently aid a person with

C. L. V. DEL AMO HOSPITAL 21

a disability by making them more able to perform necessary tasks and enjoy activities of daily living.

The district court’s decision also creates tension with a related regulation. Section 36.302 specifies the arrangements that public accommodations must make for service animals. See 28 C.F.R. § 36.302. Subsection (c)(6) permits public accommodations to ask only two questions to determine whether an animal is a service animal: (1) whether “the animal is required because of a disability,” and (2) “what work or task the animal has been trained to perform.” 28 C.F.R. § 36.302(c)(6). The public accommodation is expressly prohibited from “requir[ing] documentation, such as proof that the animal has been certified, trained, or licensed as a service animal.” Id. 7 The DOJ observed that requiring individuals with disabilities to carry around documentation “would be unnecessary, burdensome, and contrary to the spirit, intent, and mandates of the ADA.” See Final Rule, 75 Fed. Reg. at 56272. We conclude that the district court’s implicit rule creates a mismatch between the information a public accommodation may request and the conditions that a service animal must satisfy to be qualified.

Consistent with the clear language of the regulations and commentary, many district court decisions both within and outside our Circuit have declined to apply a certification obligation. See Green, 994 F. Supp. at 1255–56; Riley v. Bd. of Comm’rs of Tippecanoe Cnty., No. 14-CV-00063, 2017

7 Del Amo contends that this provision of the regulations is a “different question” from the one presented here, because the issue here is not that C.L. did not have proper documentation for Aspen. True enough, but as C.L. points out, it would be nonsensical for the ADA to effectively require certification when public accommodations are prohibited from asking for proof of it. WL 4181143, at *5 (N.D. Ind. Sept. 21, 2017); Cordoves v. Miami-Dade County, 92 F. Supp. 3d 1221, 1230 (S.D. Fla. 2015); Rose v. Springfield–Greene Cnty. Health Dep’t, 668 F. Supp. 2d 1206, 1214–15 (W.D. Mo. 2009) (“There are no requirements as to the amount or type of training that a service animal must undergo, nor the type of work or assistance that a service animal must provide, but the animal be trained to perform tasks or do work for the benefit of a [person with a disability].”); see also Prindable v. Ass’n of Apartment Owners of 2987 Kalakaua, 304 F.Supp.2d 1245, 1256 (D. Haw. 2003) (noting that “there are no federally- mandated animal training standards”).

The DOJ has conveyed the same views in its technical assistance manual and other guidance documents. Such materials may properly serve as authoritative sources of interpretive guidance. See, e.g., Bay Area Addiction Rsch. & Treatment, Inc. v. City of Antioch, 179 F.3d 725, 732 n.11 (9th Cir. 1999) (stating that DOJ’s ADA Technical Assistance Manual “must also be given substantial deference and will be disregarded only if ‘plainly erroneous or inconsistent with the regulation’” (citation omitted)). The DOJ Technical Assistance Manual on the ADA defines service animals by the tasks they are able to perform, and not by reference to a particular training protocol. See U.S. Dep’t of Justice, Technical Assistance Manual on the ADA (1994), available at https://www.ada.gov/taman3.html. Though the manual identifies that “[a] number of States have programs to certify service animals,” it instructs that private entities “may not insist on proof of State certification before permitting the entry of a service animal to a place of public accommodation.” Id. Similarly, the DOJ’s document titled “Frequently Asked Questions about Service Animals and the ADA,” which provides guidance on the service animal provisions, makes clear that the ADA considers self-training

C. L. V. DEL AMO HOSPITAL 23

to be a viable option. See U.S. Dep’t of Justice, “Frequently Asked Questions about Service Animals and the ADA” (2015), available at https://www.ada.gov/regs2010/service_ animal_qa.pdf.

C

Finally, we conclude that the district court erred because effectively requiring certification would hinder the goals of the ADA.

The ADA was signed into law on July 26, 1990. Congress sought to eliminate the discrimination faced by people with disabilities in essential facets of everyday life, including at public accommodations. 42 U.S.C. § 12101(a)(2)–(3). Accordingly, one of the ADA’s goals was “to provide a clear and comprehensive national mandate for the elimination of discrimination against individuals with disabilities.” Id. at § 12101(b)(1). The flexibility to self- train a service animal for an individual’s specific needs furthers the ADA’s goal of helping people with disabilities live with “equality of opportunity, full participation, independent living, and economic self-sufficiency.” Id. at § 12101(a)(7).

A certification requirement would have negative consequences for persons with psychiatric disabilities who rely on service animals. Research shows the significant impact service animals can have on the quality of life of persons with such disabilities. See Anne Ruff & Adriana Fortune, Emerging Duties Under Unsettled Disability Law: Web Access and Service Animals in Health Care, 11 J. Health & Life Sci. L. 80, 100 (2017) (recognizing that service animals are increasingly being used to assist persons with psychiatric disabilities rather than physical disabilities). Specifically, service animals have been shown to help “individuals with autism, post-traumatic stress disorder, and anxiety.” Id. Service dogs in particular have been “associated with clinically significant reductions in [PTSD] symptoms” compared to usual care alone. See Marguerite E. O’Haire & Kerri E. Rodriguez, Preliminary Efficacy of Service Dogs as a Complementary Treatment for Posttraumatic Stress Disorder in Military Members and Veterans, 86 J. Consulting & Clinical Psychol. 179, 184 (2018); see also Terry K. Crowe et al., Veterans Transitioning from Isolation to Integration: A Look at Veteran/Service Dog Partnerships, 40 Disability & Rehabilitation 2953, 2956 (2018) (finding that service dogs decrease isolation because they “were alerted to the veterans’ anxiety,” and the animals would “nudge[] or alert[] (cue[]) the veterans to leave the store”).

C.L.’s case illustrates the benefits a trained service dog can provide to an individual with PTSD. C.L. testified that she is able to go into public “much, much more” because of Aspen’s training in alerting her to the presence of other people. She also feels comfortable going grocery shopping with Aspen’s assistance, which she was not able to do before. Enrolling in a training course to obtain a certification is not always a viable option for persons like C.L., who had not been able to work before obtaining Aspen. The district court placed great weight on the fact that Gonzalez would not certify Aspen as a service dog under standards set by Assistance Dogs International, a private trade association, but Gonzalez explained that she could not certify Aspen under this standard unless C.L. attended three seminars and provided proof of disability from a medical provider. Those burdens were in addition to the $900 tuition, plus any travel or other expenses that may be required to attend multiple- day sessions. Moreover, C.L. testified that she did not attend all three sessions of Little Angels’ seminar in part because

C. L. V. DEL AMO HOSPITAL 25

they were focused on tasks she did not require Aspen to perform. Importantly, the DOJ declined to adopt formal training requirements precisely because the needs of each individual with a disability vary greatly. See 73 Fed. Reg. at 34524.

Thus, a ruling that service animals cannot be qualified under the ADA if an expert is not able to certify the animal based on the standards of a private organization would have the effect of denying legally protected access to public accommodations for persons who—like C.L.—need service animals to mitigate the effects of their disabilities in these spaces. And from a pragmatic standpoint, there is no industry-wide consensus on the proper certification standards. For example, Gonzalez testified that her organization starts with the general certification framework of another organization, but then adds additional standards. It is unclear how a person like C.L. could reliably choose between these various standards, none of which the DOJ endorses, to ensure the “certification” will be judicially recognized. Importing a certification requirement would not create certainty for whether a dog is truly a service animal. Instead, it would multiply litigation over which certifications are judicially valid. Under the ADA, the proper focus is on whether a service animal will consistently and reliably help a person with a disability in performing activities of daily living.

IV

Because we hold that the district court erred as a matter of law by imposing a certification requirement, we need not delve deeply into C.L.’s second claim of legal error— whether corroboration is required if a person self-training their service dog gives unrebutted testimony. According to C.L., the district court discounted C.L.’s testimony regarding Aspen’s training without giving explicit reasons for doing so. The court stated: “Other than C.L.’s assertions, there is no evidence in the record as accepted by the Court that Aspen was trained to perform, and could perform, the outlined [service dog] tasks”; and “The sole evidence Plaintiff put forth that Aspen was a service animal during this time period is her own testimony that she trained Aspen to perform tasks.” (emphasis added). Read plainly, these statements imply that the court had accepted C.L.’s testimony, but without additional accepted testimony, C.L.’s own assertions were insufficient to show that Aspen was a trained service dog.

But because the district court erred by improperly discounting Gonzalez’s testimony on certification grounds, and Gonzalez’s expert testimony that Aspen was a “service dog” corroborated C.L.’s testimony and thus provided “additional accepted testimony,” we need not decide whether unimpeached testimony is necessarily insufficient to prove that a dog is a service animal. We remand for the district court to consider whether C.L.’s testimony regarding her self-training of Aspen, coupled with Katie Gonzalez’s expert testimony, was sufficient to show that Aspen was “more likely than not” a qualified service dog at the time of trial. See Sanchez v. Monumental Life Ins. Co., 102 F.3d 398, 404 (9th Cir. 1996) (describing the preponderance standard for establishing a fact as “more likely than not”).

V

For these reasons, we hold that the district court erred by requiring Aspen to meet formal certification standards to qualify as a service dog, because such a requirement is inconsistent with the ADA. We remand for the district court to reconsider C.L.’s claims consistent with what we have said in this opinion.

C. L. V. DEL AMO HOSPITAL 27

VACATED AND REMANDED. 8

8 Pursuant to General Order 4.5(e), Del Amo Hospital shall bear the costs of this appeal.