Molina v. Astrue

FOR PUBLICATION
UNITED STATES COURT OF APPEALS
FOR THE NINTH CIRCUIT

EDILIA MOLINA, 
Plaintiff-Appellant, No. 10-16578
v.
 D.C. No.
4:09-cv-00327-BPV
MICHAEL J. ASTRUE, Commissioner
of Social Security Administration, OPINION
Defendant-Appellee.

Appeal from the United States District Court
for the District of Arizona
Bernardo P. Velasco, Magistrate Judge, Presiding

Argued and Submitted
October 24, 2011—San Francisco, California

Filed April 2, 2012

Before: Susan P. Graber and Sandra S. Ikuta, Circuit Judges,
and Lewis A. Kaplan,* Senior District Judge.

Opinion by Judge Ikuta;
Partial Concurrence and Partial Dissent by Judge Graber

*The Honorable Lewis A. Kaplan, Senior United States District Judge
for the Southern District of New York, sitting by designation.

3501
 MOLINA v. ASTRUE 3505

COUNSEL

John A. Gravina, Esq. (argued), Tuscon, Arizona, for
plaintiff-appellant Edilia Molina.

Michael A. Thomas (argued), Assistant United States Attor-
ney, Denver, Colorado, for defendant-appellee Michael J.
Astrue, Commissioner of the Social Security Administration.
3506 MOLINA v. ASTRUE
OPINION

IKUTA, Circuit Judge:

Edilia Molina appeals the district court’s decision affirming
the Social Security Commissioner’s denial of her application
for disability insurance benefits and supplemental security
income under the Social Security Act. She argues that the
administrative law judge (ALJ) erred by giving inadequate
weight to the opinions of her primary care provider, by
improperly determining she was not credible, and by rejecting
the testimony of her family members without comment. We
have jurisdiction pursuant to 28 U.S.C. § 1291, and we affirm.

I

On September 22, 2006, Molina filed claims for disability
benefits and supplemental security income, alleging that she
suffered from panic and anxiety attacks1 and had been unable
to work since September 7, 2002, when she last held a job as
an assembler for Texas Instruments. According to Molina’s
primary care provider, physician’s assistant Molly Wheel-
wright, Molina had a “longstanding panic disorder,” and had
tried several anxiety medications and antidepressants, includ-
ing Lorazepam, over the years. Wheelwright also noted that
Molina had to leave work and visit the emergency room more
than once because of “severe panic attacks.” After September
7, 2002, Molina’s alleged disability onset date, Wheelwright’s
records showed that Molina continued to have symptoms of
panic disorder, although as of February 2005, Molina stated
she was stable on Lorazepam. Although Wheelwright fre-
quently encouraged Molina to seek counseling for her anxi-
ety, Molina declined to do so, citing various concerns
including unpleasant past experiences with psychiatrists and
1
On appeal, Molina does not challenge the ALJ’s findings that her phys-
ical impairments do not restrict her ability to work, so we do not discuss
them here.
 MOLINA v. ASTRUE 3507
a belief that the local mental health clinic was for drug
addicts.

A few weeks after Molina filed her application with the
Social Security Administration, she contacted a clinician at a
nearby mental health clinic for intake and evaluation. The cli-
nician wrote that Molina was “motivated” as well as “calm,
orderly, [and] normally responsive.” Molina was diagnosed
with post-traumatic stress disorder and panic disorder. She
failed to show up for her follow-up appointment, and told a
clinician in June 2007 that she no longer wanted counseling
services.

Dr. Hunter Yost, a psychiatrist, examined Molina in
November 2006 and reported that Molina “ha[d] some partial
symptoms” of anxiety. After conducting a mental status
examination, Dr. Yost concluded that Molina was alert and
oriented, maintained good eye contact, and “did not appear
excessively anxious.” She spoke coherently and fluently,
smiled appropriately, and was congenial. She did not have an
elated or depressed mood, and there was no evidence of a
thought disorder. Dr. Yost diagnosed her with panic disorder
without agoraphobia, and wrote that her condition was con-
trolled with Lorazepam and that she was able to do self-
calming measures to help with panic attacks when she was out
in public. He concluded that her disorder had not and would
not impose any limitations on her ability to work. A state Dis-
ability Determination Services staff physician reviewed Moli-
na’s file in December 2006 and reached similar conclusions.

In December 2006, Wheelwright wrote that Molina was
feeling generally well and described her anxiety episodes as
intermittent. In February 2008, however, in a Mental Impair-
ment Questionnaire for Molina’s disability application,
Wheelwright wrote that Molina had “panic disorder, agora-
phobia” with an “unsure” prognosis. In the grid for “mental
abilities and aptitudes needed to do unskilled work,” Wheel-
wright checked “unable to meet competitive standards” next
3508 MOLINA v. ASTRUE
to “maintain regular attendance,” “deal with normal work
stress,” and “complete a normal workday and workweek with-
out interruptions from psychologically based symptoms,”
among other categories. Although the form instructed her to
explain these assessments and provide medical or clinical
findings to support them, Wheelwright did not do so. As for
Molina’s functional limitations, Wheelwright checked “ex-
treme” next to the boxes for “restriction of activities of daily
living,” “difficulties in maintaining social functioning,” and
“difficulties in maintaining concentration, persistence, or
pace.” In separate medical notes, Wheelwright wrote that
Molina was “very disabled by her condition,” because she
“[b]ecomes easily panicked in many situations even through-
out the day” and is “[u]nable to work as a result.” She noted,
however, that a psychiatrist would “likely need to supple-
ment” her evaluation.

At her hearing before the ALJ, Molina testified that she had
been helping care for her two grandchildren since 2003,
including walking them to school in the morning and picking
them up in the afternoon. Molina had cared for the younger
granddaughter full-time from 2003 until she was old enough
to start school in 2005. A state agency paid her for her ser-
vices until 2007. Molina said she regularly attended church,
went shopping with her sister, attended family barbeques, and
went for short walks in the neighborhood. Molina also sub-
mitted two function reports in which she wrote that she drove
a car in the immediate area and went shopping once a month
but liked to get in and out quickly because she did not like to
be around a lot of people. She drove when she had to but felt
a lot of anxiety and would “only go very short distances.” She
went out to eat with friends or family occasionally, but once
her condition began, she would experience panic attacks that
made it difficult for her to swallow her food and would have
to go outside and wait for the others to finish their meals. She
did not like to go out because she did not like to be around
a lot of people and would experience panic attacks; these
panic attacks were characterized by anxiety, shortness of
 MOLINA v. ASTRUE 3509
breath, dizziness, sweating, and an elevated heart rate. She
liked to watch TV and eat dinner in her room.

Molina testified that her last job as an assembler for Texas
Instruments involved performing wire bonding and inspec-
tion. She worked in a large room with four or five other peo-
ple, and her job did not require her to speak to anyone unless
there was a problem with her machine, in which case she
would ask the setup operator for help. At her lunch break she
would take her food to her car and eat by herself. She said she
was not able to return to this job because she could not toler-
ate the presence of other people. When she worked, she would
get anxiety and panic attacks and would have to go to the
infirmary to lie down or to the hospital. These attacks would
come on without warning; her heart would start to beat very
hard, and she would have difficulty breathing and feel fearful.

Five family members submitted statements on Molina’s
behalf. Molina’s sister Mary stated that Molina was scared to
get out on her own, could not stand to be around a lot of peo-
ple, and did not feel safe by herself. She would go to her room
when people came to the house. Her sister Frances De La
Cerda wrote that Molina’s anxiety was “so severe that she is
not able to go anywhere without having a panic attack” and
that she “had to quit working because the anxiety affected her
mental abilities.” According to her brother-in-law, David
Chenoweth, Molina did “not like to travel anywhere except
for short trips in the daytime,” and she seemed to be most
comfortable in her room. Her sister Barbara Molina wrote that
Molina could not be around people or go out to eat because
she tended to get very nervous and would take her food home
with her. Molina’s sister-in-law, Theresa Molina, said that
when Molina had a panic attack, she would get very shaky
and sweaty and feel faint, and that she preferred not to go
anywhere for fear that she would have one of her attacks.

Following the hearing, the ALJ issued a written opinion
holding that Molina was not disabled because she could per-
3510 MOLINA v. ASTRUE
form her past relevant work as an assembler. In reaching this
conclusion, the ALJ discussed Dr. Yost’s evaluation, the
Mental Impairment Questionnaire filled out by Wheelwright,
and Molina’s testimony at the hearing. The ALJ adopted Dr.
Yost’s conclusion regarding the severity of Molina’s mental
impairments. She found that Molina’s subjective allegations
regarding the severity of her pain and ability to work were not
credible to the extent they conflicted with Dr. Yost’s evalua-
tion. She considered but discounted Wheelwright’s opinions
because they were “quite conclusory” and provided “very lit-
tle explanation of the evidence relied on in forming those
opinions.” Additionally, they did not appear to be supported
by Molina’s objective medical condition and were inconsis-
tent with Dr. Yost’s opinion. The ALJ also noted that Molina
had cared for her granddaughters throughout much of the
alleged disability period, and that she had not made any
efforts to seek counseling from 2003 to 2005. The district
court affirmed.

On appeal, Molina argues that the ALJ erred in reaching
this conclusion because (1) the ALJ should have given more
weight to Wheelwright’s opinions, (2) the ALJ improperly
evaluated Molina’s credibility, and (3) the ALJ erred in reject-
ing the testimony of Molina’s family members without com-
ment. We consider each of these arguments in turn.

II

For purposes of the Social Security Act, a claimant is dis-
abled if the claimant is unable “to engage in any substantial
gainful activity by reason of any medically determinable
physical or mental impairment which can be expected to
result in death or which has lasted or can be expected to last
for a continuous period of not less than 12 months.” 42 U.S.C.
§ 423(d)(1)(A). In order to determine whether a claimant
meets this definition, the ALJ employs a five-step sequential
evaluation. Parra v. Astrue, 481 F.3d 742, 746 (9th Cir.
2007); 20 C.F.R. §§ 404.1520(a), 416.920(a). In brief, the
 MOLINA v. ASTRUE 3511
ALJ considers whether a claimant is disabled by determining:
(1) whether the claimant is “doing substantial gainful activi-
ty”; (2) whether the claimant has a “severe medically deter-
minable physical or mental impairment” or combination of
impairments that has lasted for more than 12 months; (3)
whether the impairment “meets or equals” one of the listings
in the regulations; (4) whether, given the claimant’s “residual
functional capacity,” the claimant can still do his or her “past
relevant work”; and (5) whether the clamant “can make an
adjustment to other work.” 20 C.F.R. §§ 404.1520(a),
416.920(a). The claimant bears the burden of proof at steps
one through four. Parra, 481 F.3d at 746.

We review the district court’s order affirming the ALJ’s
denial of social security benefits de novo, Tommasetti v.
Astrue, 533 F.3d 1035, 1038 (9th Cir. 2008), and reverse only
if the ALJ’s decision was not supported by substantial evi-
dence in the record as a whole or if the ALJ applied the wrong
legal standard, Stone v. Heckler, 761 F.2d 530, 531 (9th Cir.
1985). Substantial evidence “ ‘means such relevant evidence
as a reasonable mind might accept as adequate to support a
conclusion.’ ” Valentine v. Comm’r Soc. Sec. Admin., 574
F.3d 685, 690 (9th Cir. 2009) (quoting Desrosiers v. Sec’y of
Health & Human Servs., 846 F.2d 573, 576 (9th Cir. 1988)).
The evidence must be “more than a mere scintilla,” but may
be less than a preponderance. Id. (quoting Desrosiers, 846
F.2d at 576). Even when the evidence is susceptible to more
than one rational interpretation, we must uphold the ALJ’s
findings if they are supported by inferences reasonably drawn
from the record. Tommasetti, 533 F.3d at 1038. Finally, we
may not reverse an ALJ’s decision on account of an error that
is harmless. See Stout v. Comm’r, Soc. Sec. Admin., 454 F.3d
1050, 1055-56 (9th Cir. 2006). “[T]he burden of showing that
an error is harmful normally falls upon the party attacking the
agency’s determination.” Shinseki v. Sanders, 556 U.S. 396,
409 (2009).
3512 MOLINA v. ASTRUE
III

We first turn to Molina’s argument that the ALJ erred in
discounting Wheelwright’s opinions about Molina’s residual
functional capacity. The ALJ rejected Wheelwright’s views
on the grounds that Wheelwright’s opinions were “quite con-
clusory,” provided very little explanation of the evidence
relied on, were not supported by Molina’s objective medical
condition, and were inconsistent with the opinion of Dr. Yost,
the examining psychiatrist.

[1] In order to reject the testimony of a medically accept-
able treating source, the ALJ must provide specific, legitimate
reasons based on substantial evidence in the record. Valentine,
574 F.3d at 692. However, only licensed physicians and cer-
tain other qualified specialists2 are considered “[a]cceptable
medical sources.” 20 C.F.R. § 404.1513(a). Physician’s assis-
tants are defined as “other sources,” § 404.1513(d), and are
not entitled to the same deference, see § 404.1527; SSR 06-
03p. The ALJ may discount testimony from these “other
sources” if the ALJ “ ‘gives reasons germane to each witness
for doing so.’ ” See Turner v. Comm’r of Soc. Sec., 613 F.3d
1217, 1224 (9th Cir. 2010) (quoting Lewis v. Apfel, 236 F.3d
503, 511 (9th Cir. 2001)).

[2] Wheelwright did not qualify as a medically acceptable
treating source because she was a physician’s assistant, see 20
C.F.R. § 404.1513(d)(1), and the record does not show that
she worked under a physician’s close supervision, see Gomez
v. Chater, 74 F.3d 967, 971 (9th Cir. 1996) (holding that a
nurse practitioner could be considered a medically acceptable
source where she worked under a physician’s close supervi-
sion such that she acted as the physician’s agent).3 The ALJ
2
These are limited to licensed or certified psychologists, licensed
optometrists, licensed podiatrists, and qualified speech-language patholo-
gists. 20 C.F.R. § 404.1513(a).
3
In holding that a nurse practitioner could be an acceptable medical
source, Gomez relied in part on language in 20 C.F.R. § 416.913(a)(6),
 MOLINA v. ASTRUE 3513
gave several germane reasons for discounting Wheelwright’s
opinions in favor of the conflicting testimony of Dr. Yost, and
these reasons were substantiated by the record. First, Wheel-
wright’s opinion as to the impact of Molina’s mental impair-
ments on her ability to work consisted primarily of a
standardized, check-the-box form in which she failed to pro-
vide supporting reasoning or clinical findings, despite being
instructed to do so. We have held that the ALJ may “permissi-
bly reject[ ] . . . check-off reports that [do] not contain any
explanation of the bases of their conclusions.” Crane v. Sha-
lala, 76 F.3d 251, 253 (9th Cir. 1996); see also Holohan v.
Massanari, 246 F.3d 1195, 1202 (9th Cir. 2001) (“[T]he regu-
lations give more weight to opinions that are explained than
to those that are not.”). Second, Wheelwright’s opinions that
Molina “[b]ecomes easily panicked in many situations even
throughout the day” and is “unable to work as a result,” were
conclusory and conflicted with her earlier assessment that
Molina’s anxiety attacks were only intermittent. Finally, as
the ALJ held, Wheelwright’s opinion was inconsistent with
that of Dr. Yost, who specialized in the relevant field of psy-
chiatry, and whose opinion was therefore entitled to greater
weight. See Holohan, 246 F.3d at 1202 (“[T]he regulations
give more weight to . . . the opinions of specialists concerning
matters relating to their specialty over that of nonspecial-
ists.”); Smolen v. Chater, 80 F.3d 1273, 1285 (9th Cir. 1996)
(holding that the ALJ should have given greater weight to a
physician with the expertise that was most relevant to the
patient’s allegedly disabling condition). In fact, Wheelwright
herself indicated that her evaluation would need to be supple-
mented by a psychiatrist. Accordingly, the ALJ did not err in

which stated that “[a] report of an interdisciplinary team that contains the
evaluation and signature of an acceptable medical source is also consid-
ered acceptable medical evidence.” See Gomez, 74 F.3d at 971. This regu-
latory section has since been repealed. We need not, however, address
Gomez‘s continued vitality because Wainwright acted alone.
3514 MOLINA v. ASTRUE
discounting Wheelwright’s opinion where it conflicted with
Dr. Yost’s evaluation.4

IV

Molina next challenges the ALJ’s conclusion that Molina’s
allegations regarding the severity of her symptoms and their
effect on her ability to work were not credible. Molina argues
that the ALJ erred by basing her conclusion on activities of
daily living that were not transferable to a work setting, and
in considering Molina’s failure to make any efforts to seek
counseling in 2003, 2004, or 2005. Again, we disagree.

In assessing the credibility of a claimant’s testimony
regarding subjective pain or the intensity of symptoms, the
ALJ engages in a two-step analysis. Vasquez v. Astrue, 572
F.3d 586, 591 (9th Cir. 2009). First, the ALJ must determine
whether there is “ ‘objective medical evidence of an underly-
ing impairment which could reasonably be expected to pro-
duce the pain or other symptoms alleged.’ ” Id. (quoting
Lingenfelter v. Astrue, 504 F.3d 1028, 1036 (9th Cir. 2007)).
If the claimant has presented such evidence, and there is no
evidence of malingering, then the ALJ must give “ ‘specific,
clear and convincing reasons’ ” in order to reject the claim-
ant’s testimony about the severity of the symptoms. Id. (quot-
ing Lingenfelter, 504 F.3d at 1036). At the same time, the
ALJ is not “required to believe every allegation of disabling
pain, or else disability benefits would be available for the ask-
ing, a result plainly contrary to 42 U.S.C. § 423(d)(5)(A).”
Fair v. Bowen, 885 F.2d 597, 603 (9th Cir. 1989). In evaluat-
ing the claimant’s testimony, the ALJ may use “ ‘ordinary
4
Some language in the ALJ’s opinion suggests the ALJ may have con-
sidered Wheelwright to be a medically acceptable source; specifically, the
ALJ discussed the deference generally given to treating sources under 20
C.F.R. § 404.1527(d)(2). Even assuming the ALJ erred in this regard, the
error was harmless because the ALJ gave specific, legitimate reasons
based on substantial evidence in the record for discounting Wheelwright’s
testimony. Valentine, 574 F.3d at 692.
 MOLINA v. ASTRUE 3515
techniques of credibility evaluation.’ ” Turner, 613 F.3d at
1224 n.3 (quoting Smolen, 80 F.3d at 1284). For instance, the
ALJ may consider inconsistencies either in the claimant’s tes-
timony or between the testimony and the claimant’s conduct,
id.; “ ‘unexplained or inadequately explained failure to seek
treatment or to follow a prescribed course of treatment,’ ”
Tommasetti, 533 F.3d at 1039 (quoting Smolen, 80 F.3d at
1284); and “whether the claimant engages in daily activities
inconsistent with the alleged symptoms,” Lingenfelter, 504
F.3d at 1040. While a claimant need not “ ‘vegetate in a dark
room’ ” in order to be eligible for benefits, Cooper v. Bowen,
815 F.2d 557, 561 (9th Cir. 1987) (quoting Smith v. Califano,
637 F.2d 968, 971 (3d Cir. 1981)), the ALJ may discredit a
claimant’s testimony when the claimant reports participation
in everyday activities indicating capacities that are transfer-
able to a work setting, see Morgan v. Comm’r Soc. Sec.
Admin., 169 F.3d 595, 600 (9th Cir. 1999); Fair, 885 F.2d at
603. Even where those activities suggest some difficulty func-
tioning, they may be grounds for discrediting the claimant’s
testimony to the extent that they contradict claims of a totally
debilitating impairment. See Turner, 613 F.3d at 1225; Valen-
tine, 574 F.3d at 693.

[3] Here, the ALJ found that Molina’s claimed inability to
tolerate even minimal human interaction was inconsistent
with her daily activities throughout the disability period. This
finding is supported by substantial evidence in the record. The
ALJ could reasonably conclude that Molina’s activities,
including walking her two grandchildren to and from school,
attending church, shopping, and taking walks, undermined her
claims that she was incapable of being around people without
suffering from debilitating panic attacks. It was also reason-
able for the ALJ to conclude that Molina’s daily activities
involved a degree of human interaction that was analogous to
that required by her past relevant work, which involved work-
ing at a machine by herself in a large room that was occupied
by only four or five other people, with whom she did not gen-
erally need to interact.
3516 MOLINA v. ASTRUE
[4] Moreover, the ALJ supported her conclusion that
Molina was not credible on the additional grounds that Moli-
na’s allegations were undermined by her demeanor and pre-
sentation as described by Dr. Yost, and inconsistent with
other medical evidence in the record. The first finding is sup-
ported by Dr. Yost’s report that Molina was alert and ori-
ented, maintained good eye contact, did not appear
excessively anxious, spoke coherently and fluently, smiled
appropriately, and was congenial. The second is supported by
the conclusions of both Dr. Yost and the state examining phy-
sician that Molina’s anxiety disorder was not severe and that
she was able to control it with Lorazepam and other self-
calming measures. Because the ALJ’s adverse credibility
determination was supported by specific, clear, and convinc-
ing reasons, we uphold it.

[5] We also reject Molina’s claim that the ALJ erred in
relying on Molina’s failure to seek or follow prescribed treat-
ment. First, the ALJ did not expressly place any weight on
this factor in discounting Molina’s credibility. But to the
extent the ALJ implicitly considered Molina’s failure to fol-
low Wheelwright’s advice that she seek counseling, the ALJ
did not err. We have long held that, in assessing a claimant’s
credibility, the ALJ may properly rely on “ ‘unexplained or
inadequately explained failure to seek treatment or to follow
a prescribed course of treatment.’ ” Tommasetti, 533 F.3d at
1039 (quoting Smolen, 80 F.3d at 1284); Fair, 885 F.2d at
603. According to agency rules, “the individual’s statements
may be less credible if the level or frequency of treatment is
inconsistent with the level of complaints, or if the medical
reports or records show that the individual is not following the
treatment as prescribed and there are no good reasons for this
failure.” SSR 96-7p.5 Moreover, a claimant’s failure to assert
5
Social Security Rulings (SSRs) “do not carry the ‘force of law,’ but
they are binding on ALJs nonetheless.” Bray v. Comm’r Soc. Sec. Admin.,
554 F.3d 1219, 1224 (9th Cir. 2009). They “ ‘reflect the official interpreta-
tion of the [SSA] and are entitled to some deference as long as they are
consistent with the Social Security Act and regulations.’ ” Id. (alteration
in original) (quoting Avenetti v. Barnhart, 456 F.3d 1122, 1124 (9th Cir.
2006)).
 MOLINA v. ASTRUE 3517
a good reason for not seeking treatment, “or a finding by the
ALJ that the proffered reason is not believable, can cast doubt
on the sincerity of the claimant’s pain testimony.” Fair, 885
F.2d at 603. Here, the record was filled with evidence that
despite Wheelwright’s repeated efforts to persuade Molina to
seek psychiatric treatment for her anxiety disorder, she failed
to do so until after she applied for disability benefits.
Although Molina provided reasons for resisting treatment,
there was no medical evidence that Molina’s resistance was
attributable to her mental impairment rather than her own per-
sonal preference, and it was reasonable for the ALJ to con-
clude that the “level or frequency of treatment [was]
inconsistent with the level of complaints.” SSR 96-7p. Moli-
na’s reliance on Pate-Fires v. Astrue, 564 F.3d 935, 945 (8th
Cir. 2009), is misplaced, because in that case, the “evidence
overwhelmingly demonstrate[d] Pate-Fires’s noncompliance
was attributable to her mental illness,” id. at 946, whereas
there was no similar evidence here.6

V

Finally, Molina argues that the ALJ erred by failing to
properly discuss the testimony of Molina’s family members.

A

[6] Lay testimony as to a claimant’s symptoms or how an
impairment affects the claimant’s ability to work is competent
evidence that the ALJ must take into account. Nguyen v.
6
We also reject Molina’s argument that the ALJ erred by failing to com-
ply with SSR 82-59, which provides that an ALJ may deny benefits to a
claimant who has a disability if the claimant unjustifiably fails to follow
prescribed treatment that is “clearly expected to restore capacity to engage
in any [substantial gainful activity].” This rule is not applicable here,
because the ALJ determined that Molina was not disabled, and Molina’s
failure to seek treatment (to the extent the ALJ considered it at all) was
merely a factor in the ALJ’s credibility determination. See Roberts v. Sha-
lala, 66 F.3d 179, 183 (9th Cir. 1995).
3518 MOLINA v. ASTRUE
Chater, 100 F.3d 1462, 1467 (9th Cir. 1996); Dodrill v. Sha-
lala, 12 F.3d 915, 919 (9th Cir. 1993). We have held that
competent lay witness testimony “cannot be disregarded with-
out comment,” Nguyen, 100 F.3d at 1467, and that in order to
discount competent lay witness testimony, the ALJ “must
give reasons that are germane to each witness,” Dodrill, 12
F.3d at 919. We have not, however, required the ALJ to dis-
cuss every witness’s testimony on a individualized, witness-
by-witness basis. Rather, if the ALJ gives germane reasons
for rejecting testimony by one witness, the ALJ need only
point to those reasons when rejecting similar testimony by a
different witness. See Valentine, 574 F.3d at 694 (holding that
because “the ALJ provided clear and convincing reasons for
rejecting [the claimant’s] own subjective complaints, and
because [the lay witness’s] testimony was similar to such
complaints, it follows that the ALJ also gave germane reasons
for rejecting [the lay witness’s] testimony”). The applicable
regulations are in accord; they require the ALJ to consider tes-
timony from family and friends submitted on behalf of the
claimant, see 20 C.F.R. §§ 404.1529(c)(3), 404.1545(a)(3),
but do not require the ALJ to provide express reasons for
rejecting testimony from each lay witness, see id.; see also
SSR 06-03p (recognizing that “there is a distinction between
what an adjudicator must consider and what the adjudicator
must explain in the disability determination or decision”).

[7] Here, the ALJ stated that the rationale for her credibil-
ity determination “include[d] reference to . . . the third party
statements submitted in support of the claimant.” This state-
ment establishes that the ALJ reviewed the lay witness testi-
mony in the record, but it does not provide a reason for
discounting the testimony. The ALJ gave reasons for rejecting
Molina’s testimony regarding her symptoms that were equally
relevant to the similar testimony of the lay witnesses, and that
would support a finding that the lay testimony was similarly
not credible. But under our rule that lay witness testimony
“cannot be disregarded without comment,” Nguyen, 100 F.3d
at 1467, the ALJ erred in failing to explain her reasons for
 MOLINA v. ASTRUE 3519
disregarding the lay witness testimony, either individually or
in the aggregate. Nevertheless, as discussed below, that error
was harmless.

B

[8] We have long recognized that harmless error principles
apply in the Social Security Act context. Stout, 454 F.3d at
1054 (collecting cases). We have, for example, deemed errors
harmless where the ALJ misstated the facts about the claim-
ant, but we were able to conclude from the record that the
ALJ would have reached the same result absent the error. See
Curry v. Sullivan, 925 F.2d 1127, 1131 (9th Cir. 1990) (hold-
ing that the ALJ’s error in stating that the claimant was 50
years old and had a G.E.D. was harmless given that the claim-
ant was in the “closely approaching advanced age” category
regardless of the error and was able to communicate in the
English language). We have also deemed legal errors harm-
less where it was clear they did not alter the ALJ’s decision.
For instance, in Brawner v. Secretary of Health & Human
Services, we held that an ALJ’s error in classifying the claim-
ant’s past relevant work as “light” was inconsequential where
the record supported the ALJ’s determination that the claim-
ant was able to perform other light work and therefore was
not disabled. 839 F.2d 432, 434 (9th Cir. 1988) (per curiam).
And in Matthews v. Shalala, we held that even if the ALJ
erred in failing to mention the claimant’s inability to stay in
one position to the vocational expert, such error was harmless
because the claimant had not shown that he was unable to
return to his previous job as a receiving clerk and inspector,
which, the record demonstrated, involved a combination of
sitting and standing. 10 F.3d 678, 681 (9th Cir. 1993).

In addition, several of our cases have held that an ALJ’s
error was harmless where the ALJ provided one or more
invalid reasons for disbelieving a claimant’s testimony, but
also provided valid reasons that were supported by the record.
See Bray, 554 F.3d at 1227; Carmickle v. Comm’r, Soc. Sec.
3520 MOLINA v. ASTRUE
Admin., 533 F.3d 1155, 1162-63 (9th Cir. 2008); Batson v.
Comm’r of Soc. Sec. Admin., 359 F.3d 1190, 1195-97 (9th
Cir. 2004). In this context, we have said that an error is harm-
less so long as there remains substantial evidence supporting
the ALJ’s decision and the error “does not negate the validity
of the ALJ’s ultimate conclusion.” Batson, 359 F.3d at 1197;
see also Carmickle, 533 F.3d at 1162.

[9] Although we have expressed different formulations of
the harmless error rule depending on the facts of the case and
the error at issue, we have adhered to the general principle
that an ALJ’s error is harmless where it is “inconsequential to
the ultimate nondisability determination.” Carmickle, 533
F.3d at 1162; Tommasetti, 533 F.3d at 1038; Robbins v. Soc.
Sec. Admin., 466 F.3d 880, 885 (9th Cir. 2006); Stout, 454
F.3d at 1055. In other words, in each case we look at the
record as a whole to determine whether the error alters the
outcome of the case.

We applied these harmless error principles in Stout, where
the ALJ had erred by neglecting to comment on competent lay
witness testimony. Stout, 454 F.3d at 1054. Specifically, the
claimant’s sister and brother-in-law (who worked with the
claimant for 15 years) testified that the claimant was unable
to accomplish the simplest of tasks without near-constant
supervision. Id. at 1053. We emphasized that this testimony
about how the claimant’s mental impairments affected his
ability to work was “uncontradicted” and “consistent with
medical evidence.” Id. In holding that the ALJ’s silent disre-
gard of the testimony was not harmless, Stout explained: “The
[vocational expert] specifically opined that constant supervi-
sion is unacceptable in competitive employment; yet, the ALJ
articulated no reasons for dismissing the uncontradicted lay
testimony indicating Stout needed such supervision to per-
form even simple tasks.” Id. at 1056. Because there was noth-
ing in the ALJ’s opinion addressing the limitations described
by the lay witnesses, we were left “with nothing to review to
determine whether the error materially impacted the ALJ’s
 MOLINA v. ASTRUE 3521
ultimate decision.” Id. We explained that “where the ALJ’s
error lies in a failure to properly discuss competent lay testi-
mony favorable to the claimant, a reviewing court cannot con-
sider the error harmless unless it can confidently conclude
that no reasonable ALJ, when fully crediting the testimony,
could have reached a different disability determination.” Id.
Applying this rule, we determined that the ALJ’s failure to
consider the testimony at issue could not be deemed “inconse-
quential to the ultimate nondisability determination,” id. at
1055, because the testimony identified limitations not consid-
ered by the ALJ, was uncontradicted by anything in the
record, and was highly probative of Stout’s inability to work
in a competitive environment.

Our subsequent decision in Robbins applied the Stout rule
that the ALJ may not disregard lay witness testimony about
the claimant’s limitations if the ALJ has not validly rejected
those limitations and their existence would alter the ultimate
nondisability determination. See Robbins, 466 F.3d at 885. In
Robbins, we held that all of the ALJ’s reasons for rejecting
the claimant’s allegations of subjective pain were legally
invalid. Id. at 883-85. We then addressed the ALJ’s failure to
discuss the corroborative lay witness testimony of the claim-
ant’s son. Id. at 885. As in Stout, the lay witness testimony
was unrebutted by anything in the record; in fact, it was sub-
stantially similar to the testimony of the claimant’s daughter,
which the ALJ found “generally credible.” Robbins, 466 F.3d
at 890 (O’Scannlain, J., dissenting). Robbins reasoned that “if
credited, the testimony of Robbins’s son adds substantial
weight not only to Robbins’s claim, but also to the testimony
of Robbins’s wife and daughter, which support Robbins’s
claim.” Id. at 885 (maj. op.). In this situation, the court held,
“[b]ecause the ALJ did not make a legally sufficient adverse
credibility finding with regard to Robbins’s own testimony,
we cannot say with respect to [the son’s] testimony that ‘no
reasonable ALJ, when fully crediting the testimony, could
have reached a different disability determination.’ ” Id. (quot-
3522 MOLINA v. ASTRUE
ing Stout, 454 F.3d at 1056).7 In other words, given that the
ALJ had not validly rejected the claimant’s excess pain testi-
mony, the lay witness’s corroborative testimony (if credited
by the ALJ) would alter the ALJ’s disability decision. There-
fore, the ALJ’s failure to mention the testimony was not
harmless. By the same token, Robbins suggested that the
ALJ’s failure to discuss the son’s testimony would have been
harmless if the ALJ had provided legally sufficient reasons
for rejecting the claimant’s own testimony. See id.; see also
Valentine, 574 F.3d at 694 (indicating that it is not harmful
error for the ALJ to fail to discuss lay witness testimony
where the ALJ has provided sufficient reasons for rejecting
similar testimony).

In claiming that the ALJ’s error here is not harmless,
Molina relies heavily on Stout’s statement that “where the
ALJ’s error lies in a failure to properly discuss competent lay
testimony favorable to the claimant, a reviewing court cannot
consider the error harmless unless it can confidently conclude
that no reasonable ALJ, when fully crediting the testimony,
could have reached a different disability determination.”
Stout, 454 F.3d at 1056. Plucking this language out of context,
she claims that an ALJ’s failure to give individualized reasons
for rejecting a lay witness’s testimony that would be material
standing alone is per se prejudicial, even if the ALJ gave well-
supported reasons for rejecting similar testimony. Under this
interpretation of Stout, if ten lay witnesses submitted state-
ments that the claimant is unable to lift more than ten pounds
and thus cannot return to her previous work, and the ALJ
properly rejected the testimony of nine of them because it was
belied by the medical evidence in the record, the ALJ’s failure
7
Although Robbins held that the ALJ’s failure to discuss the son’s testi-
mony was reversible error, the ALJ’s decision would have required rever-
sal in any event due to the ALJ’s failure to give legally sufficient reasons
for rejecting the claimant’s own testimony.
 MOLINA v. ASTRUE 3523
to comment on the tenth witness’s testimony would require
reversal.8

[10] We disagree. Stout made clear that its holding was
“consistent with our prior harmless error cases” in the Social
Security context. Id. Interpreting Stout as creating a rule that
the ALJ’s failure to expressly reject any facially material lay
witness testimony is per se prejudicial would run afoul of our
settled rule that we will not reverse for errors that are “incon-
sequential to the ultimate nondisability determination.” See
Carmickle, 533 F.3d at 1162 (“Likewise, in Stout, after sur-
veying our precedent applying harmless error in social secur-
ity cases, we concluded that ‘in each case, the ALJ’s error . . .
was inconsequential to the ultimate nondisability determina-
tion.’ ” (quoting Stout, 454 F.3d at 1055)) (emphasis omitted);
Tommasetti, 533 F.3d at 1038; Robbins, 466 F.3d at 885.
Where lay witness testimony does not describe any limitations
not already described by the claimant, and the ALJ’s well-
supported reasons for rejecting the claimant’s testimony apply
equally well to the lay witness testimony, it would be incon-
sistent with our prior harmless error precedent to deem the
ALJ’s failure to discuss the lay witness testimony to be preju-
dicial per se. See also Valentine, 574 F.3d at 694 (holding that
if an ALJ gave germane reasons for rejecting the claimant’s
testimony, those reasons are equally germane to similar testi-
mony by a lay witness.); Lewis, 236 F.3d at 512 (stating that
the ALJ “noted arguably germane reasons for dismissing the
8
According to the dissent, Molina’s interpretation of Stout does not
create a per se rule because “we still must decide whether the testimony
affected the disability determination.” Dis. op. at 3532-33. We disagree.
Claimants do not generally submit lay witness statements unless they sup-
port their disability claim, and thus such statements will naturally affect
the disability determination if considered on their own. Because Molina’s
proposed interpretation of Stout would preclude the reviewing court from
considering undiscussed lay witness statements in the context of the record
as a whole, it would effectively create a per se rule of prejudice whenever
the ALJ fails to give an individualized reason for rejecting such state-
ments.
3524 MOLINA v. ASTRUE
[lay] testimony, even if he did not clearly link his determina-
tion to those reasons”).

Molina’s interpretation of Stout would also run afoul of the
Supreme Court’s recent decision in Shinseki v. Sanders. See
Sanders, 556 U.S. at 409. In Sanders, the Supreme Court
rejected a test developed by the Federal Circuit for determin-
ing whether the Department of Veterans Affairs’s failure to
give the disability claimant certain statutorily required notice
was harmless. Id. at 399. The Federal Circuit had held that the
Veteran’s Court should presume any such error is prejudicial
and reverse unless the Department could make a specified show-
ing.9 Id. at 403-04, 407. According to the Supreme Court, this
test conflicted with the Veteran Court’s statutory duty to “take
due account of the rule of prejudicial error,” id. at 406 (quot-
ing 38 U.S.C. § 7261(b)(2)), which the Supreme Court held
“requires the Veterans Court to apply the same kind of
‘harmless-error’ rule that courts ordinarily apply in civil
cases,” id. In other words, courts must review cases “ ‘without
regard to errors’ that do not affect the parties’ ‘substantial
rights.’ ” Id. at 407 (quoting 28 U.S.C. § 2111, which codifies
the federal harmless error rule for civil cases).

In expounding on this principle, Sanders focused on the
error of relying on “mandatory presumptions and rigid rules,”
reasoning that such reliance frustrates Congress’s express
preference for determining harmlessness by “case-specific
application of judgment, based upon examination of the
record.” Id. at 407 (citing Kotteakos v. United States, 328 U.S.
750, 759-60 (1946)). The use of presumptions, the Court
explained, “exhibit[s] the very characteristics that Congress
sought to discourage,” because it prevents the court “from
resting its conclusion on the facts and circumstances of the
9
Specifically, the Veterans Court was required to find the error harmful
unless the Department could demonstrate “(1) that the claimant’s ‘actual
knowledge’ cured the defect or (2) that the claimant could not have
received a benefit as a matter of law.” Id. at 407.
 MOLINA v. ASTRUE 3525
particular case,” and therefore requires the reviewing court to
find the error prejudicial “even if that court, having read the
entire record, conscientiously concludes the contrary.” Id. at
408. The Court held that the Federal Circuit’s use of a pre-
sumption “increase[d] the likelihood of reversal in cases
where, in fact, the error is harmless.” Id. at 409. Such a pre-
sumption only “encourages abuse of the judicial process and
diminishes the public’s confidence in the fair and effective
operation of the judicial system.” Id. While the Court
acknowledged that “courts may sometimes make empirically
based generalizations about what kinds of errors are likely, as
a factual matter, to prove harmful,” id. at 411, it emphasized
that courts should “hesita[te] to generalize too broadly about
particular kinds of errors when the specific factual circum-
stances in which the error arises may well make all the differ-
ence,” id. at 412. Such generalizations are better left to the
lower court that “sees sufficient case-specific raw material” to
enable it to draw empirical conclusions. Id. at 412.

Molina’s interpretation of Stout as creating a rule that an
ALJ’s failure to give individual reasons for rejecting a lay
witness’s material testimony is per se prejudicial cannot be
reconciled with the reasoning in Sanders. First, like the Fed-
eral Circuit’s presumption, it would prevent the reviewing
court from making a “case-specific application of judgment,
based upon examination of the record,” id. at 398, and would
require the reviewing court to reverse in many cases where it
is “obvious from the record . . . that the error made no differ-
ence,” id. at 407. Second, Molina’s interpretation of Stout is
not an empirically-based generalization “about what kinds of
errors are likely, as a factual matter, to prove harmful.” Id. at
411. Molina has not suggested any basis for concluding that
an ALJ’s failure to discuss a lay witness’s testimony
expressly is likely to affect the outcome in situations where
the testimony is similar to other testimony that the ALJ val-
idly discounted, or where the testimony is contradicted by
more reliable medical evidence that the ALJ credited. More-
over, as the Supreme Court pointed out, we are not well situ-
3526 MOLINA v. ASTRUE
ated to make such generalizations, since we are removed from
the “case-specific raw material.” Id.

Finally, if we interpreted Stout the way Molina asks us to,
we would ignore the Supreme Court’s instruction that we are
“to apply the same kind of ‘harmless-error’ rule that courts
ordinarily apply in civil cases.” Id. at 406.10 That harmless
error rule, as codified in 28 U.S.C. § 2111, requires us to
“give judgment after an examination of the record without
regard to errors or defects which do not affect the substantial
rights of the parties.” 28 U.S.C. § 2111; see also Sanders, 556
U.S. at 407. Interpreting § 2111, we have held that a district
court’s erroneous exclusion of evidence does not warrant
reversal unless the error “more probably than not tainted the
verdict.” Engquist v. Or. Dep’t of Agric., 478 F.3d 985, 1009
(9th Cir. 2007), aff’d 553 U.S. 591 (2008); see also Haddad
v. Lockheed Cal. Corp., 720 F.2d 1454, 1459 (9th Cir. 1983)
(“Just as the verdict in a civil case need only be more proba-
bly than not true, so an error in a civil trial need only be more
probably than not harmless.”).11 We have deemed several fac-
tors important in determining harmlessness, including
whether the evidence at issue was “cumulative of other com-
petent testimony,” Haddad, 720 F.2d at 1460; see also id. at
10
Although Sanders interpreted a different harmless error statute spe-
cific to the Veterans Administration, see 38 U.S.C. § 7261(b)(2), we have
already held that the same harmless error standard applies to Social Secur-
ity cases because § 7261 merely “import[ed] the Administrative Procedure
Act rule and the general rule for federal cases.” McLeod v. Astrue, 640
F.3d 881, 887 (9th Cir. 2011).
11
In Obrey v. Johnson, we held that “[t]he party benefitting from the
error has the burden of persuasion, and ‘in cases of equipose, we
reverse.’ ” 400 F.3d 691, 701 (9th Cir. 2005) (quoting United States v.
Seschillie, 310 F.3d 1208, 1214-15 (9th Cir. 2002)). We do not for present
purposes need to decide whether this holding survives Sanders, which
held that in ordinary civil cases “the party that ‘seeks to have a judgment
set aside because of an erroneous ruling carries the burden of showing that
prejudice resulted.’ ” Sanders, 556 U.S. at 409 (quoting Palmer v. Hoff-
man, 318 U.S. 109, 116 (1943)). Regardless of which party bears the bur-
den of persuasion, we conclude that the error was harmless in this case.
 MOLINA v. ASTRUE 3527
1456 (“Since this evidence is cumulative of other evidence in
the record and the record contains no evidence to the contrary,
we presume that the improper admission of this testimony had
no effect on the court’s decision.”); Obrey v. Johnson, 400
F.3d 691, 701 (9th Cir. 2005) (concluding that the district
court’s erroneous exclusion of relevant witness testimony was
not harmless because the evidence “was not merely tangential
or cumulative”); Bank of the West v. Commercial Credit Fin.
Servs., 852 F.2d 1162, 1166 n.2 (9th Cir. 1988) (concluding
that the district court’s erroneous exclusion of a witness’s
declaration was harmless because the declaration was “cumu-
lative of other evidence”), and the “overall strength of [the]
case” against the appealing party, see Boyd v. City & Cnty. of
S.F., 576 F.3d 938, 949 (9th Cir. 2009); United States v.
Morales, 108 F.3d 1031, 1040 (9th Cir. 1997) (en banc).
Molina’s interpretation of Stout would conflict with the harm-
less error standard we apply in civil cases because it would
prevent us from considering these crucial factors. If the ALJ’s
failure to address lay witness testimony that is material on its
face is always prejudicial, it would be irrelevant whether such
testimony is cumulative of other testimony that the ALJ prop-
erly rejected. Cf. Robbins, 466 F.3d at 885 (finding reversible
error “[b]ecause the ALJ did not make a legally sufficient
adverse credibility finding with regard to” similar testimony
from the claimant himself); Valentine, 574 F.3d at 694. It
would be equally irrelevant whether substantial evidence
credited by the ALJ contradicts the lay witness testimony on
material points. Cf. Stout, 454 F.3d at 1053-57 (finding
reversible error where the lay witness testimony was “uncon-
tradicted” and “consistent with medical evidence”).

We also note Sanders’s warning that the harmless error test
in civil cases should not be as stringent as that in criminal
cases. As the Supreme Court explained, “the fact that the
Government must prove its case beyond a reasonable doubt
[in criminal cases] justifies a rule that makes it more difficult
for the reviewing court to find that an error did not affect the
outcome of a case. But in the ordinary civil case that is not
3528 MOLINA v. ASTRUE
so.” Sanders, 556 U.S. at 410-11 (internal citation omitted);
see also Bank of Nova Scotia v. United States, 487 U.S. 250,
255-56 (1988) (“It would be inappropriate to devise a rule
permitting federal courts to deal more sternly with nonconsti-
tutional harmless errors than with constitutional errors that are
likewise harmless.”). But Molina interprets Stout as setting a
higher standard than is required to protect a criminal defen-
dant’s constitutional rights. Whereas in criminal cases, “an
otherwise valid conviction should not be set aside if the
reviewing court may confidently say, on the whole record,
that the constitutional error was harmless beyond a reasonable
doubt,” Delaware v. Van Arsdall, 475 U.S. 673, 681 (1986)
(emphasis added), Molina’s interpretation would prevent the
reviewing court from considering whether the ALJ had given
reasons for rejecting the undiscussed lay witness testimony
that were supported by substantial evidence in the context of
the record as a whole. And whereas in criminal cases, the
reviewing court is directed to consider factors such as
“whether the testimony was cumulative, the presence or
absence of evidence corroborating or contradicting the testi-
mony of the witness on material points . . . , and . . . the over-
all strength of the prosecution’s case,” id. at 684, Molina’s
interpretation would, as discussed, prevent us from consider-
ing these important factors at all.

The dissent suggests that Molina’s proposed rule is appro-
priate because a reviewing court cannot make independent
credibility determinations. Dis. op. at 3532. Therefore, the
dissent argues, if the ALJ fails to make an individual determi-
nation of the credibility of each lay witness, we must presume
that the lay witness is credible and give full effect to the wit-
ness’s testimony.12 We do not disagree with the general prop-
12
The dissent analogizes to cases in the immigration context holding
that we presume that an asylum petitioner is credible if the immigration
judge did not make an explicit adverse credibility finding. Dis. op. at 3532
(citing Abovian v. INS, 219 F.3d 972, 978 (9th Cir.), as amended by 228
F.3d 1127, 1127 (9th Cir. 2000)); see also Kalubi v. Ashcroft, 364 F.3d
 MOLINA v. ASTRUE 3529
osition that it is the ALJ’s prerogative to determine the
credibility of witnesses. Our role as a reviewing court is lim-
ited in this regard, both because the administrative judge is
generally in a unique position to observe the witnesses’
demeanor and conduct firsthand, see Universal Camera Corp.
v. NLRB, 340 U.S. 474, 496 (1951),13 and because we may not
uphold an agency’s decision on a ground not actually relied
on by the agency, see SEC v. Chenery Corp., 332 U.S. 194,
196 (1947). But for the same reasons, if an ALJ has provided
well-supported grounds for rejecting testimony regarding
specified limitations, we cannot ignore the ALJ’s reasoning
and reverse the agency merely because the ALJ did not
expressly discredit each witness who described the same limi-
tations. Further, where the ALJ rejects a witness’s testimony
without providing germane reasons, but has already provided
germane reasons for rejecting similar testimony, we cannot
reverse the agency merely because the ALJ did not “clearly
link his determination to those reasons.” Lewis, 236 F.3d at
512. Even when an agency “explains its decision with ‘less
than ideal clarity,’ ” we must uphold it “if the agency’s path
may reasonably be discerned.” Alaska Dep’t of Envtl. Conser-
vation v. EPA, 540 U.S. 461, 497 (2004) (quoting Bowman
Transp., Inc. v. Arkansas-Best Freight Sys., 419 U.S. 281, 286
(1974)). Giving full effect to the ALJ’s reasoning does not
require us to consider whether the undiscussed lay witness
was credible or not credible. Rather, it is Molina’s proposed

1134, 1137-38 (9th Cir. 2004). But see 8 U.S.C. § 1158(b)(1)(B)(iii)
(2008) (providing that “if no adverse credibility determination is explicitly
made, the applicant or witness shall have a rebuttable presumption of
credibility on appeal” (emphasis added)); Su Hwa She v. Holder, 629 F.3d
958, 964 n.5 (9th Cir. 2010). In our view, these immigration cases are
unhelpful because they do not address the issue before us, namely, when
an administrative judge’s failure to expressly comment on a witness’s tes-
timony is prejudicial.
13
Here, of course, the lay witness testimony was in the form of written
statements, not live testimony given under oath, so firsthand observations
are irrelevant to the credibility determination.
3530 MOLINA v. ASTRUE
interpretation of Stout that would require the reviewing court
to make a credibility determination where the ALJ had not.
Specifically, under Molina’s proposed rule, the reviewing
court must deem the undiscussed lay witness testimony to be
both credible and disproportionately important, so as to com-
pel the conclusion that the ALJ’s decision was not supported
by substantial evidence. By contrast, our decision today
accords the proper deference to the agency factfinder by
applying the “highly deferential standard” that requires us to
affirm the ALJ’s decision if supported by “ ‘such relevant evi-
dence as a reasonable mind might accept as adequate to sup-
port a conclusion.’ ” Valentine, 574 F.3d at 690 (quoting
Desrosiers, 846 F.2d at 576). A reviewing court’s refusal to
consider whether the ALJ’s reasoning applies to undiscussed
lay witness testimony is contrary not only to our case law
holding that errors are harmless if they are “inconsequential
to the ultimate nondisability determination,” see, e.g., Car-
mickle, 533 F.3d at 1162, but also to the long-settled rule that
we will not set aside the denial of a disability claim unless
“the Secretary’s findings are not supported by substantial evi-
dence in the record as a whole,” Stone, 761 F.2d at 531
(emphasis added).

[11] As we held in another recent social security case,
Sanders makes it “quite clear that no presumptions operate,”
and we must analyze harmlessness “in light of the circum-
stances of the case.” McLeod v. Astrue, 640 F.3d 881, 888
(9th Cir. 2011). Accordingly, we reject Molina’s interpreta-
tion of Stout as creating a per se rule of prejudice when the
ALJ fails to discuss lay witness testimony that is material if
considered by itself, and instead apply Stout according to our
generally applicable harmless error principles. In reaching this
conclusion, we join the Eighth Circuit’s well reasoned deter-
mination that an ALJ’s failure to comment upon lay witness
testimony is harmless where “the same evidence that the ALJ
referred to in discrediting [the claimant’s] claims also discred-
its [the lay witness’s] claims.” Buckner v. Astrue, 646 F.3d
549, 560 (8th Cir. 2011).
 MOLINA v. ASTRUE 3531
C

[12] Turning to the case at hand, we must consider whether
the ALJ’s failure to discuss the testimony from Molina’s fam-
ily members was “inconsequential to the ultimate nondisa-
bility determination” in the context of the record as a whole.
See Carmickle, 533 F.3d at 1162; Tommasetti, 533 F.3d at
1038; Robbins, 466 F.3d at 885; Stout, 454 F.3d at 1055.
Here, the ALJ failed to explain her reasons for rejecting the
lay witnesses’ testimony. That testimony, however, did not
describe any limitations beyond those Molina herself
described,14 which the ALJ discussed at length and rejected
based on well-supported, clear and convincing reasons. Spe-
cifically, the ALJ determined that Molina’s claim that her
anxiety disorder made her unable to work was contradicted by
her demeanor and presentation, by Dr. Yost’s evaluation, and
by her own testimony about her daily activities throughout the
alleged disability period. Because the ALJ had validly
rejected all the limitations described by the lay witnesses in
discussing Molina’s testimony, we are confident that the
ALJ’s failure to give specific witness-by-witness reasons for
rejecting the lay testimony did not alter the ultimate nondisa-
bility determination. Accordingly, the ALJ’s error was harm-
less.

VI

The ALJ did not err in weighing the evidence as she did or
in finding that Molina’s testimony regarding the severity of
her impairment was not credible. Although the ALJ erred in
failing to give germane reasons for rejecting the lay witness
testimony, such error was harmless given that the lay testi-
14
Contrary to the dissent, dis. op. at 3533, the letters from Molina’s fam-
ily members described the same limitations as Molina described in her
function reports and testimony before the ALJ, including her reluctance to
travel, her inability to finish her food at restaurants, and her physical
symptoms during a panic attack. See supra pp. 3508-09.
3532 MOLINA v. ASTRUE
mony described the same limitations as Molina’s own testi-
mony, and the ALJ’s reasons for rejecting Molina’s testimony
apply with equal force to the lay testimony. Applying the
principles set forth in our social security cases, as well as in
Sanders, we uphold the ALJ’s decision as supported by sub-
stantial evidence.

AFFIRMED.

GRABER, Circuit Judge, concurring in part and dissenting in
part:

I concur in Parts I through IV of the majority opinion and
also agree with the majority’s conclusion, in Part V, that the
ALJ erred in failing to comment in any way on the lay witness
testimony. But I dissent because that error was not harmless.

In my view, Stout v. Commissioner, Social Security Admin-
istration, 454 F.3d 1050, 1055-56 (9th Cir. 2006), means what
it says: When Stout requires our harmless error analysis to
proceed by “fully crediting” testimony ignored by the ALJ,
we must do exactly that. Id. at 1056 (emphasis added). That
rule simply reiterates the standard practice that we, as a
reviewing court, do not make independent credibility determi-
nations. When an administrative fact-finder has failed to com-
ment adversely on the credibility of testimony taken under
oath, we generally deem the testimony to be truthful for the
purpose of review. Cf. Abovian v. INS, 219 F.3d 972, 978 (9th
Cir.), as amended by 228 F.3d 1127, 1127 (9th Cir. 2000)
(“Where the [immigration judge] makes no credibility find-
ing, the petitioner’s credibility is presumed.”).

Treating ignored testimony as true does not, however, nec-
essarily resolve the issue in favor of remand or payment of
benefits and thus does not create a “per se rule of prejudice.”
Maj. op. at 3522-26 & n.8. Rather, we still must decide
 MOLINA v. ASTRUE 3533
whether the testimony affected the disability determination.
Stout, 454 F.3d at 1056. If the ignored testimony had no bear-
ing on that determination, the error is harmless.

Here, some of the ignored lay testimony related to Molina’s
subjective feelings of anxiety, nervousness, and fear. That is,
the testimony in part amounted to repetition of Molina’s sub-
jective assessment of her condition to her family members.
Crediting those statements means only accepting as true that
Molina expressed those feelings to her family members, not
necessarily that Molina actually experienced limitations as
severely as she claimed. Because the ALJ identified substan-
tial evidence—including medical opinions, Molina’s demea-
nor, and her daily activities throughout the disability period—
that was inconsistent with the level of impairment claimed by
Molina, no reasonable ALJ would have reached a different
disability determination when fully crediting lay testimony
that stood for nothing more than parroting of Molina’s subjec-
tive complaints.

But some of the lay testimony contained independent
observations of Molina’s behavior, which may well be consis-
tent with her claimed disability. Examples include her reluc-
tance to travel and her inability to sit in a restaurant for the
duration of a meal. Additional testimony discussed Molina’s
physical symptoms during a panic attack, such as shakiness
and sweating. That additional testimony bears on Molina’s
claimed impairments and does more than merely echo her
own complaints.

It is precisely because these external observations contain
details beyond those in Molina’s own testimony that we can-
not assume that any reasonable ALJ would discredit them on
the same basis that the ALJ in this case discredited Molina’s
own testimony. Indeed, this very case presents an answer to
the majority’s concerns that my reading of Stout will lead to
a remand whenever an ALJ fails to comment on lay witness
testimony. When the lay witness testimony is similar or iden-
3534 MOLINA v. ASTRUE
tical to properly discredited testimony (in this case, the testi-
mony parroting Molina’s own complaints), the error
ordinarily would be harmless. But when the lay witness testi-
mony contains different or additional material (in this case,
the testimony observing Molina’s behavior), the error is less
likely to be harmless.

Because I am not convinced that no reasonable ALJ, when
crediting that additional testimony, would have reached a dif-
ferent disability determination, Stout, 454 F.3d at 1056, I
respectfully dissent from the harmless error analysis. I would
remand the case to allow the ALJ, as fact-finder, to consider
in the first instance the credibility and import of the lay testi-
mony.